The classical estrogen receptor (ER) mediates genomic as well as rapid nongenomic estradiol responses. In case of genomic responses, the ER acts as a ligand-dependent transcription factor that regulates gene expression in estrogen target tissues. In contrast, nongenomic effects are initiated at the plasma membrane and lead to rapid activation of cytoplasmic signal transduction pathways. Recently, an orphan G protein-coupled receptor, GPR30, has been claimed to bind to and to signal in response to estradiol. GPR30 therefore might mediate some of the nongenomic estradiol effects. The present study was performed to clarify the controversy about the subcellular localization of GPR30 and to gain insight into the in vivo function of this receptor. In transiently transfected cells as well as cells endogenously expressing GPR30, we confirmed that the receptor localized to the endoplasmic reticulum. However, using radioactive estradiol, we observed only saturable, specific binding to the classical ER but not to GPR30. Estradiol stimulation of cells expressing GPR30 had no impact on intracellular cAMP or calcium levels. To elucidate the physiological role of GPR30, we performed in vivo experiments with estradiol and G1, a compound that has been claimed to act as selective GPR30 agonist. In two classical estrogen target organs, the uterus and the mammary gland, G1 did not show any estrogenic effect. Taken together, we draw the conclusion that GPR30 is still an orphan receptor.
ESTRADIOL HAS BEEN implicated in the regulation of a variety of physiological processes such as reproduction, mammary gland development, bone turnover, cardiovascular function, and neuroprotection (1). The involvement of the estrogen receptors (ERs) ERα and ERβ in these processes has been demonstrated in mouse models in which either ERα or ERβ or both receptors have been inactivated by gene targeting (2). Different molecular mechanisms seem to contribute to the diverse biological effects of estradiol. In the classical mode of action, estradiol binds to the ER that acts as a ligand-dependent transcription factor and regulates gene expression in target tissues. These rather slow genomic effects are sensitive toward inhibitors of transcription and translation. In contrast, nongenomic estrogen effects are insensitive toward these inhibitors and lead to rapid activation of cytoplasmic signal transduction pathways, such as phosphatidylinositol 3-kinase or ERK activation, cAMP elevation, or increases in intracellular calcium levels (3). There is considerable evidence that the classical ER can associate with the plasma membrane and transmit the rapid nongenomic responses after estradiol stimulation (4). Nevertheless, because some estradiol effects cannot be inhibited by the classical ER antagonist ICI182780 (ICI), it has been repeatedly speculated that other ERs might exist (5).
Recently, an orphan G protein-coupled receptor (GPCR), GPR30, has been claimed to bind to and signal in response to estradiol and thus might be partly responsible for the mediation of nongenomic estradiol effects (6, 7). GPR30 was cloned by several groups a decade ago (8–12). Due to sequence homology with the IL-8 receptor and other chemoattractant receptors, it has been assumed that the natural ligand of GPR30 would be a peptide (8, 12). However, none of the tested peptides provoked a response in cells transfected with GPR30 (8). In ERα-negative breast cancer cells, GPR30 mediated ERK activation in response to estradiol by transactivation of the epidermal growth factor (EGF) receptor. Via a Gβγ-dependent, pertussis-sensitive pathway, GPR30 activated matrix metalloproteinases that released surface-bound proheparin-binding EGF-like growth factor. Subsequently, the EGF receptor was activated and stimulated ERK phosphorylation. Interestingly, ER antagonists such as ICI also stimulated ERK activation in GPR30-positive cells, activating the same signal transduction pathway as estradiol (13). In addition, GPR30 also activated the adenylate cyclase, leading to an attenuation of EGF-induced ERK activation over time (14). In both studies mentioned above, it remained unclear whether GPPR30 acted as a direct estradiol receptor or worked in concert with other unknown receptor proteins (13, 14). A few years later, GPR30 was identified as a novel estradiol receptor by two groups independently (6, 7). Prossnitz and co-workers (6) demonstrated that cells transiently transfected with GPR30 bind estradiol. Estradiol stimulation provoked an increase in intracellular calcium in COS-7 cells transiently transfected with GPR30. This GPR30-mediated calcium increase was blocked by EGF receptor inhibitors and was partly sensitive toward pertussis toxin (6). The ER antagonists tamoxifen and ICI acted as full agonists on GPR30 (6). Because GPR30 transactivates the EGFR it has been speculated that not only the classical ER but also GPR30 might be involved in the uterine response after estradiol treatment (15). Intrauterine injections of estradiol increased EGF concentrations and stimulated EGF receptor activation (16). Moreover, neutralizing EGF antibodies prevented uterine epithelial cell proliferation in response to estradiol (17).
Apart from the fact that the physiological significance of GPR30 is still unknown, there is some controversy regarding GPR30’s function as an ER. In contrast to previously published data (13, 14), Levin and co-workers (18) could not demonstrate cAMP or ERK activation in GPR30-positive, ER-negative breast cancer cells. Moreover, in ER-positive, GPR30-positive MCF-7 cells, nongenomic estradiol responses were blocked by ICI and were dependent on ERα. Silencing of GPR30 function in these cells had no effect on estradiol-induced cAMP elevation and ERK activation (18). There is also some dispute regarding the cellular localization of GPR30. One group reported localization of GPR30 in the endoplasmic reticulum (6); the other found GPR30 to be expressed in the plasma membrane (7). A third group reported that cellular localization of GPR30 depends on the cell type and type of GPR30-tag used for analysis (19).
The present study was performed to unravel the cellular localization of GPR30 and to clarify its role as a potential ER. Moreover, we wanted to obtain first insights into the biological role of GPR30 and therefore used the selective GPR30 agonist G1 (20) in vivo to analyze its effects in two classical estrogen target organs, the uterus and the mammary gland.
Materials and Methods
Chemicals and antibodies
17β-Estradiol and 5′-bromo-2′-deoxyuridine (BrdU) were purchased from Sigma Chemical Co. (St. Louis, MO), G1 was synthesized in the laboratories of Bayer Schering Pharma AG (Berlin, Germany). [2,4,6,7-3H]estradiol (70 Ci/mmol) was from PerkinElmer (Norwalk, CT). Fluorescent phallotoxins from Invitrogen (Carlsbad, CA) were used to stain actin fibers. Rabbit polyclonal antibodies directed against the C terminus of GPR30 were a generous gift from Eric Prossnitz. In addition, the following primary and secondary antibodies were used: polyclonal rabbit GPR30 antibody (Lifespan, Seattle, WA; LS4272 and LS4290), monoclonal pan-cadherin antibody (CH-19; Abcam, Cambridge, MA), monoclonal KDEL antibody (Abcam), monoclonal anti-Golgi 58K protein/formiminotransferase cyclodeaminase antibody (Sigma), Alexa 488-conjugated secondary goat antimouse and antirabbit antibody (Invitrogen), and Alexa 568-conjugated secondary goat antirabbit antibody (Invitrogen).
Cell lines
COS-7 [Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ), Braunschweig, Germany] HEK293 (American Type Culture Collection, Rockville, MD), and HEC50 cells (generous gift from Eric Blasko, Berlex, Seattle, WA) were cultured in DMEM supplemented with 2 mm glutamine and 10% fetal calf serum (FCS). CHO cells (DSMZ) were grown in the same medium with the addition of 1% sodium pyruvate, whereas MDA-MB231 cells (American Type Culture Collection) were cultured in DMEM/Ham’s F-12 medium supplemented with 10% FCS and 2 mm glutamine. Transient transfections were performed 24 h after seeding cells using FuGENE (Roche, Indianapolis, IN) according to the manufacturer’s instructions. Cells were maintained in phenol red-free media with 5% charcoal-stripped serum 24–48 h before experimentation.
Generation of expression vectors
For the cellular expression of untagged GPR30, full-length human (NM_001505) and murine GPR30 cDNA (NM_029771) were cloned in the pSG5 vector using PCR. GPR30-green fluorescent protein (GFP) fusion proteins carrying GFP either at the N or the C terminus of GPR30 were cloned using the pENTR Directional TOPO cloning kit (Invitrogen) and the Gateway system (Invitrogen). The vector pcDNA-Dest47 was used to express human or murine GPR30 C-terminally tagged with GFP, whereas the vector pcDNA3.1/NT-GFP allowed for the cellular expression of GPR30 N-terminally tagged with GFP. All constructs were confirmed by sequencing.
Double immunofluorescence staining
Cells seeded on round coverslips (12 mm) were stained 24–48 h after transfection and maintenance in phenol red-free serum containing 5% charcoal-stripped serum. Cells endogenously expressing GPR30 were maintained in their respective growth media containing 10% FCS. After fixation with 3% paraformaldehyde, cells were rinsed with PBS and incubated for 45 min at room temperature in 0.5% Triton X-100 in PBS with 3% BSA. Cells were then incubated in a mixture of polyclonal GPR30 antiserum (antiserum obtained from Eric Prossnitz diluted 1:5000 to 1:10,000 or 1 μg/ml LS4272) together with one of the following monoclonal antibodies diluted in 3% BSA for 2 h at room temperature: pan-cadherin (1:500), KDEL (5 μg/ml), or Golgi (diluted 1:100). After washing with PBS, cells were incubated with the respective secondary antibodies diluted in 3% goat serum for 1 h. GPR30 immunostaining was visualized with an Alexa 568-conjugated goat antirabbit antibody (1:1000), and immunoreactivity of the respective monoclonal antibody was achieved by incubation with Alexa 488-conjugated goat antimouse antiserum (1:750). In case of GPR30/actin double staining, the actin staining was performed first with fluorescent phallotoxins according to the manufacturer’s instructions. GPR30 immunostaining was done in a second step using an Alexa 488-conjugated goat antirabbit antibody as secondary antibody. Coverslips were mounted using Vectashield containing 4′,6-diamidino-2-phenylindole. Cells were examined with a Zeiss Axioplan 2 microscope equipped with specific filters for immunofluorescence analysis and an AxioCam camera. Stainings were analyzed using the Axiovision software.
Binding assays
Saturation binding assays were performed in 96-well plates using COS-7 cells transfected with GPR30 or ERα and HEC50 cells that had been kept in phenol red-free medium with 5% charcoal-stripped serum for 24 or 48 h. A total of 150,000 cells per well were incubated with increasing doses of tritiated estradiol (0.5–100 nm) in the absence (total binding) or presence (unspecific binding) of a 1000-fold excess of unlabeled estradiol for 1 h at room temperature in phenol red-free DMEM containing 0.5% BSA. The incubation was terminated by filtration using a Filtermate Harvester (Packard Downers Grove, IL). Filters were rapidly washed with 20 mm Tris (pH 7.5) and dried. Labeled 17β-estradiol bound to the cells was quantified by β-scintillation counting of the filters. For determination of specific estradiol binding, unspecific estradiol binding was subtracted from total binding. Specific estradiol binding was plotted against the tracer concentration, curves were fitted using Sigmaplot. Each point was done in triplicate. Binding assays were performed using tagged and untagged human as well as murine GPR30. Expression of GPR30 under the experimental conditions was analyzed by quantitative RT-PCR and immunostaining.
Analysis of calcium signaling and cAMP formation in response to estradiol
For calcium measurements, CHO cells transiently expressing GPR30 and MDA-MB231 cells were seeded into black clear-bottom 384-well plates (BD-Falcon, Heidelberg, Germany) with cell densities of 12,000 (CHO) and 10,000 (MDA-MB231) per 50 μl per well. The next day, the culture medium was removed and the cells were washed twice with 80 μl assay buffer (1× Hanks’ balanced salt solution, 20 mm HEPES, pH 7.4). Cells were resuspended in 20 μl assay buffer and loaded at 20 C for 90 min (CHO) or 60 min (MDA-MB231) with 20 μl Calcium 3 Express Assay Kit solution (Molecular Devices, Sunnyvale, CA). Baseline fluorescence (excitation 488 nm, emission 510–570 nm bandpass) was recorded for 10 sec using a fluorescence imaging plate reader (FLIPR3; Molecular Devices). Test compounds (10 μl) diluted in assay buffer containing 0.1% BSA and 1% dimethylsulfoxide (DMSO) (final DMSO concentration was 0.2%) were added, and fluorescence intensity was recorded for 230 sec. Cellular stimulation resulted in transient increases in fluorescence intensity that reflected increases in calcium currents. Relative fluorescence units were calculated by subtracting the respective baseline fluorescence from the values obtained after cellular stimulation and plotted against the concentrations of the test compounds.
For quantification of cAMP accumulation, MDA-MB231 cells and transfected COS-7 cells were seeded into 384-well plates (Greiner Bio-One, Frickenhausen, Germany; black) at a density of 5000 cells per well. Twenty-four hours later, cells were rinsed with Krebs Ringer buffer and incubated for 120 min with 20 μl test and reference compounds in buffer containing 0.75 mm isobutylmethylxanthine. Cells were lysed, and cAMP formation was determined using a femtomolar cAMP homogenous time-resolved fluorescence (HTRF) assay kit according to the manufacturer’s instructions (CisBio International, Camarillo, CA; 62 AM1PEC). Fifty microseconds after excitation at 337 nm, Europium cryptate fluorescence and time-resolved fluorescence resonance energy transfer signals were measured at 620 and 665 nm, respectively, using a RubyStar fluorometer (BMG Labtechnologies, Offenburg, Germany). Results are expressed as HTRF ratio [(fluorescence665 nm/fluorescence620 nm) × 104]. Please note that a decrease in the HTRF ratio indicates an increase in intracellular cAMP formation (see Fig. 3).
GPR30 is not activated by estradiol. A, Normalized expression levels of GPR30 under the experimental conditions in the diverse cellular systems used for estradiol binding or signaling in response to estradiol are depicted. Note the huge induction of GPR30-mRNA upon transfection of COS-7 and CHO cells and the low expression levels in MDA-MB231 and HEC50 cells. B, Specific binding of estradiol to COS-7 cells transfected with ERα (○), human GPR30 (▴), or empty vector (•) and to HEC50 cells (▵). C, Formation of cAMP was analyzed in COS-7 cells transfected with empty vector (white bars) or with GPR30 (black bars) after stimulation with forskolin (10 μm), α- or β-estradiol (10 nm), or G1 (100 nm). Only forskolin, but not G1 or β-estradiol, enhanced cAMP formation in COS-7 cells. Please note that as described in Materials and Methods, an increase in cAMP is reflected by a decreased HTRF ratio. D, MDA-MB231 cells were stimulated with increasing doses of PGE2 (white bars), forskolin (gray bars), or estradiol (black bars). Only PGE2 and forskolin, but not estradiol, provoked cAMP formation. E, CHO cells were transfected with GPR30 and stimulated with increasing doses of ionomycin (•), ATP (○), estradiol (▴), or G1 (▵), and relative fluorescence intensity reflecting intracellular calcium increase was measured. F, Calcium currents in response to the same stimuli were analyzed in MDA-MB231 cells. Only ionomycin and ATP, but not G1 or estradiol, stimulated intracellular calcium increases in both cellular systems (E and F). Please note that the symbols for G1 (▵) hide the symbols for estradiol (▴). For each readout, four to 10 independent experiments were performed. One representative experiment is depicted. •, P < 0.05 (two-sided Student’s t test).
Animals
C57BL/6 mice (Charles River Laboratories, Wilmington, MA) were maintained on a 14-h light, 10-h dark cycle and provided with food and water ad libitum. All animal procedures were run according to German animal welfare law with the permission of the District Government of Berlin.
Uterine growth assays
For uterine growth and mammary gland assays, mice weighing 20–22 g were ovariectomized at the beginning of the sixth week of age. Two weeks after ovariectomy, animals were injected once sc with vehicle (ethanol/arachis oil 1:10, vol/vol), 100 ng estradiol, or G1 (100 ng to 100 μg). Animals were killed 6 and 18 h after hormonal stimulation (n = 8 per group and time point). Two hours before being killed, animals were injected ip with BrdU (70 mg/kg body weight) dissolved in PBS. Mice were killed by cervical dislocation. One uterine horn was fixed in 4% buffered formalin and embedded in paraffin. BrdU immunostaining using the mouse monoclonal anti-BrdU antibody from Dako (Carpinteria, CA; M0744) was performed as described previously (21). The other uterine horn was rapidly frozen in liquid nitrogen and processed for RNA extraction. Plasma samples were precipitated with acetonitrile, and G1 was determined by liquid chromatography with tandem mass spectrometry.
Mammary gland whole-mount assay
Two weeks after ovariectomy, the animals were treated for 3 wk daily sc with vehicle, 100 ng 17β-estradiol, or 2 μg G1 dissolved in ethanol/arachis oil (1:10, vol/vol) (n = 8 mice per group). The left inguinal mammary gland was removed, spread on a glass slide, and fixed for 48 h at room temperature in Carnoy’s fixative (ethanol, chloroform, glacial acetic acid, 6 parts of ethanol, 3 parts of chlorofom, and 1 part of glacial acetic acid, vol/vol). After staining in carmine alum (0.2% carmine alum, 0.5% aluminum potassium sulfate, 1 crystal of thymol), the mammary glands were dehydrated, cleared in xylene, and stored in ProTaqstura (Quartett GmbH, Berlin, Germany). The dorsal two thirds of the right inguinal mammary gland were fixed in 4% formalin at 4 C overnight and processed for BrdU immunostaining (21). The number of BrdU-positive ductal epithelial cells was evaluated in four complete transverse mammary gland sections per animal. The ventral third of the right inguinal mammary gland (without lymph node) was rapidly frozen in liquid nitrogen for gene expression analysis.
Gene expression analysis by quantitative RT-PCR
Uterine and mammary gland RNA was isolated after homogenization of tissues in guanidinium thiocyanate (22). Five micrograms of RNA were digested with deoxyribonuclease I and reverse transcribed with random hexamers using the SuperScript III First-Strand Synthesis System (Invitrogen). Real-time TaqMan PCR analysis was performed using the ABI Prism 7700 Sequence Detector System according to the manufacturer’s instructions (PE Applied Biosystems, Foster City, CA). Prevalidated probes and primers for murine INDO (indoleamine-pyrrole 2,3-dioxygenase) (catalog no. Mm00492586_m1), cytokeratin 18 (catalog no. Mm01601702_g1), Wnt-4 (catalog no. Mm00437341_m1), frizzled-2 (catalog no. Mm02524776_s1), IGF-I (catalog no. Mm00439561_m1), cyclin E1 (catalog no. Mm00432367_m1), calpactin (catalog no. Mm00501457_m1), BIP (catalog no. Mm00517691_m1), LTF (lactotransferrin) (catalog no. Mm00434787_m1), PR (progesterone receptor) (catalog no. Mm00435625_m1), and TBP (TATA-box-binding-protein) (catalog no. Mm00446973_m1) were purchased from PE Applied Biosystems. Relative mRNA levels were calculated by the comparative cycle threshold method. In the mammary gland, INDO expression levels were normalized to Cytokeratin 18, whereas uterine gene expression was normalized to TBP. Expression of GPR30 in transiently transfected cells and in human cell lines endogenously expressing GPR30 was determined using custom-made primers and probes. The sequence of the forward primer was ATG GGA CAG GTG AGC TCT CG, the reverse primer was ATC AGG ATG TTG CCC ACA AAG, and the probe was AGC ACC AGC AGT ACG TGA. Cellular expression of human GPR30 was normalized to human TBP (catalog no. Hs00427620_m1).
Statistical analysis
Data are depicted as mean ± sd and were analyzed by two-sided Student’s t test.
Results
Subcellular localization of GPR30
First, we checked the specificity of the anti-GPR30 antibodies employed, i.e. the Lifespan antibody LS4272 and the polyclonal anti-GPR30 antiserum obtained from Eric Prossnitz. In COS-7 cells that were transiently transfected with C-terminally GFP-tagged human GPR30, the GFP-fusion protein was expressed in the cytoplasm (Fig. 1A). Immunostaining with the antibody LS4272 (Lifespan) demonstrated that only the cells that expressed the fusion protein (green signals in Fig. 1A) stained positively with the antibody (red signals in Fig. 1B). No specific staining was obtained when unspecific IgG was used instead of the primary antibody (Fig. 1C), thus underlining the specificity of the anti-GPR30 antibody LS4272. Transient transfection of COS-7 cells with N-terminally GFP-tagged human GPR30 again revealed cytoplasmic localization of the fusion protein (green signals in Fig. 1D). All cells that expressed the fusion protein stained positively with the polyclonal GPR30 antiserum obtained from Eric Prossnitz (see red signals in Fig. 1E). No specific signal was observed when preimmune serum was used (Fig. 1F). As demonstrated in Fig. 1, A, B, D, and E, none of the employed antibodies delivered positive immunostaining for GPR30 in untransfected cells. Interestingly, permeabilization of fixed cells was required to obtain positive signals with any GPR30 antiserum used, including antibodies directed against the N terminus of GPR30 (LS4290, data not shown). To clarify the precise subcellular localization of GPR30, we performed double immunostaining with GPR30 antibodies and markers for the plasma membrane, the Golgi apparatus, and the endoplasmic reticulum. In these studies, we used COS-7 cells transiently transfected with untagged human GPR30 to exclude any missorting or misexpression of GFP-GPR30 fusion proteins. With untagged human GPR30, we observed the same cytoplasmic immunoreactivity (Fig. 1G) as with GFP-GPR30-fusion proteins (Fig. 1, A and D). GPPR30 (red signals in Fig. 1, G and H) was not expressed in the Golgi apparatus (green signals in Fig. 1H). There was no colocalization of GPR30 (red signal in Fig. 1, I and J) with the plasma membrane marker cadherin (green signal in Fig. 1J). However, double immunostaining with the endoplasmic reticulum marker KDEL (green signals in Fig. 1, K and M) and the GPR30 antibody demonstrated that GPR30 (red signals in Fig. 1L) localized to the endoplasmic reticulum (yellow signal in Fig. 1M). The subcellular localization of GPR30 protein was independent of the transfected DNA amounts (data not shown). We also analyzed the localization of untagged GPR30 in transfected HEK293 cells. As demonstrated by double immunostaining with the anti-KDEL marker (green signal in Fig. 1, N and P), GPR30 protein (red signal in Fig. 1O) localized to the endoplasmic reticulum in transfected HEK293 cells (yellow signal in Fig. 1P). In MDA-MB231 and HEC50 cells (resembling an aggressive human endometrial cancer cell line), which express GPR30 endogenously (6), only the antibody obtained from the Prossnitz lab, but not the commercially available antibodies, delivered positive immunostaining. In MDA-MB231 cells, GPR30 (green signals in Fig. 2, B and C) did not colocalize with submembrane-expressed actin (red signals in Fig. 2, A and C), demonstrating that GPR30 did not localize to the plasma membrane (Fig. 2C). However, there was colocalization of the endoplasmic reticulum marker anti-KDEL (green signal in Fig. 2D) with GPR30 (red signal in Fig. 2E) as evidenced by the yellow signals in Fig. 2F. Please note that the signal intensity for GPR30 varied from cell to cell in the MDA-MB231 cell line (Fig. 2E). In HEC50 cells, GPR30 (red signal in Fig. 2, H and N) did not colocalize with cadherin (green signal in Fig. 2, G and I) or the Golgi apparatus (green signal in Fig. 2, M and O) but was expressed in the endoplasmic reticulum (yellow signal in Fig. 2L) as evidenced by costaining with antibodies against the endoplasmic reticulum marker KDEL (green signal in Fig. 2, J and L) and GPR30 (red signal in Fig. 2K). Taken together, our results demonstrated that GPR30 localized to the endoplasmic reticulum of transiently transfected cells and cells endogenously expressing GPR30.
Subcellular localization of GPR30 in transiently transfected cells. COS-7 cells were transiently transfected with C-terminally (A–C) or N-terminally GFP-tagged human GPR30 (D–F). Specificity of the GPR30 antibody LS4272 (A–C) and the GPR30 antiserum obtained from Eric Prossnitz was checked (D–F). All cells that expressed the GPR30-GFP fusion protein (green signals in A and D) stained positively with the GPR30 antisera (red signals in B and E) but not with unspecific IgG (C) or preimmune serum (F). For colocalization studies, untagged human GPR30 was expressed in COS-7 cells (G–M) or in HEK293 cells (N–P). GPR30 signals appeared in red, whereas signals derived from the monoclonal marker antibodies appeared in green. Any colocalization of GPR30 with the markers would lead to a yellow signal. GPR30 in transfected COS-7 cells was expressed in the cytoplasm (G, I, and L) and did not colocalize with the Golgi apparatus (green signals in H) or the plasma membrane (green cadherin signals in J). GPR30 localized to the endoplasmic reticulum in COS-7 cells (yellow signals in M), as evidenced by the colocalization of the endoplasmic reticulum marker KDEL (green signal in K and P) and GPR30 (red signal in L). In transfected HEK293 cells, GPR30 was also expressed in the endoplasmic reticulum (yellow signal in P) as shown by the colocalization of the KDEL signal (green signal in N and P) with GPR30 immunoreactivity (red signal in O).
GPR30 localizes to the endoplasmic reticulum in cells endogenously expressing the receptor. Double-immunofluorescence staining was performed for GPR30 and several subcellular markers in MDA-MB231 (A–F) and HEC50 cells (G–O). GPR30 immunoreactivity is depicted in green (B) in the case of actin costaining (A) and in red in all other cases (E, H, K, and N). GPR30 did not localize to the plasma membrane as evidenced by actin costaining in MDA-MB231 cells (A–C) but was expressed in the endoplasmic reticulum (F). In HEC50 cells (G–O), GPR30 did not localize to the plasma membrane (I); there was no colocalization of the green cadherin signals (G and I) with the red GPR30 signals (H and I). GPR30 was not expressed in the Golgi apparatus (O) because there was no colocalization of the green Golgi signals (M and O) with the red GPR30 signals (N and O). GPR30 was expressed in the endoplasmic reticulum of HEC50 cells (L). KDEL signals are shown in green in J and L, and GPR30 signals are depicted in red in K and L.
Estradiol binding to GPR30
To ensure that GPR30 was sufficiently expressed under our experimental conditions, we analyzed aliquots for receptor expression. There was a huge increase in GPR30 expression upon transfection (Fig. 3A) of CHO or COS-7 cells. Compared with transfected CHO or COS-7 cells, endogenous expression of GPR30 in HEC50 or MDA-MB231 cells was rather low (Fig. 3A). Saturation binding assays were performed using HEC50 cells or COS-7 cells transiently transfected with human GPR30 or empty vector. As a positive control, we used COS-7 cells transiently transfected with the classical ERα. As expected, there was specific, saturable estradiol binding to ERα (open circles in Fig. 3B). However, transfection of COS-7 cells with human GPR30 did not lead to specific estradiol binding. There was no difference in estradiol binding between COS-7 cells transiently transfected with empty vector (closed circles in Fig. 3B) or GPR30 (closed triangles in Fig. 3B). Similar results were obtained with transiently transfected U2OS cells or when GFP-tagged GPR30 or murine GPR30 was used (data not shown). There was also no specific binding of estradiol to HEC50 cells (open triangles in Fig. 3B). In summary, although GPR30 and ERα were strongly expressed in transfected cells, we observed specific, saturable estradiol binding only to the classical ER but not to GPR30.
Analysis of signal transduction in response to estradiol in cells expressing GPR30
To analyze whether GPR30 coupled to Gαs proteins, we examined cAMP formation in COS-7 cells transiently transfected with human GPR30 (black boxes in Fig. 3C) or empty vector (white boxes in Fig. 3C). Whereas the positive control (forskolin) stimulated cAMP formation in cells transfected with empty vector or GPR30, stimulation with 10 nm 17β-estradiol, 100 nm G1, or 10 nm 17α-estradiol (negative control) had no impact on cAMP formation in these cells when compared with vehicle effects (Fig. 3C). In MDA-MB231 cells that endogenously expressed GPR30, forskolin (gray bars in Fig. 3D) and prostaglandin E2 (PGE2) (white bars in Fig. 3D) stimulated cAMP formation with increasing doses, whereas estradiol (black bars in Fig. 3D) had no effect. To identify cells expressing higher endogenous levels of GPR30 than MDA-MB231 or HEC50 cells, we performed ribonuclease protection analysis with several human cell lines (data not shown). Among the cell lines analyzed, MCF-7 cells expressed by far the highest levels of GPR30. However, only forskolin stimulated cAMP formation in MCF-7 cells in a dose-dependent manner, whereas 17α-estradiol (negative control) and the presumptive GPR30 agonists 17β-estradiol, ICI, and G1 had no significant effect (supplemental Fig. 7a, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org).
To determine whether GPR30 coupled to Gαq, we analyzed calcium currents in response to estradiol and G1. Please note that the final DMSO concentration was adjusted to 0.2% in all signaling experiments to control for vehicle effects. Neither estradiol (closed triangles, hidden by the open triangles) nor G1 (open triangles in Fig. 3E) increased intracellular calcium levels in CHO cells transiently transfected with GPR30. In contrast, ATP (open circles in Fig. 3E) and ionomycin (closed circles in Fig. 3E) stimulated calcium currents in these cells. Similar results were obtained with transiently transfected COS-7 cells (data not shown). Intracellular calcium levels were elevated in MDA-MB231 cells when stimulated with ATP (open circles in Fig. 3F) or ionomycin (closed circles in Fig. 3F) but remained unchanged after stimulation with increasing doses of estradiol (closed triangles hidden by the open triangles in Fig. 3F) or the GPR30 agonist G1 (open triangles in Fig. 3F). Similar results were obtained with HEC50 cells and MCF-7 cells (supplemental Fig. 7b). Time-resolved calcium currents obtained from MCF-7 cells after stimulation with fixed concentrations of ATP, ionomycin, estradiol, or G1 are depicted in supplemental Fig. 8. Ionomycin provoked a long-lasting increase in calcium currents, whereas ATP stimulated transient calcium currents. Estradiol and G1 did not increase calcium currents above vehicle levels (supplemental Fig. 8).
Taken together, these results show that estradiol did not stimulate cAMP or calcium elevation in cells transiently transfected with GPR30 or endogenously expressing this receptor.
Impact of GPR30 on uterine growth and mammary gland development in vivo
To elucidate the potential in vivo role of GPR30, we exploited the GPR30-selective agonist G1 in in vivo studies, focusing on the impact of GPR30-mediated signaling in uterine and mammary gland biology. Previously, it had been demonstrated that the early uterine response after estradiol stimulation was mediated in an ER-independent manner, probably involving a GPCR, and led to activation of the Wnt pathway (23). To test whether GPR30 could be this GPCR, we treated ovariectomized mice for 6 h with either vehicle, 100 ng estradiol or 100 ng to 100 μg G1. Estradiol stimulated Wnt-4 and IGF-I expression 6 h after application, and G1 had no effect on the expression of these genes, even at doses 1000-fold higher than those used for estradiol (Fig. 4, A and C). Compared with vehicle treatment, estradiol inhibited frizzled-2 expression, but again G1 had no effect (Fig. 4B). Please note that the observed down-regulation of frizzled-2 by estradiol was in contrast to published data describing an up-regulation of this gene in response to estradiol treatment (23). We observed up-regulation of frizzled-2 by estradiol only when using longer treatment periods than 6 h (data not shown). Six hours after drug administration, plasma levels were 38 pm in the case of estradiol (100-ng dose) and 33 nm in the case of G1 (100-μg dose). In addition, there was a linear relationship between G1 plasma levels and applied G1 dosages. To analyze whether G1 administration had any impact on uterine epithelial cell proliferation, ovariectomized mice were treated for 18 h as described above. Compared with vehicle treatment, estradiol strongly induced the S-phase cyclin, cyclin E1. G1, even at 1000-fold higher doses, had no significant effect (Fig. 4D). There were almost no proliferating uterine epithelial cells after vehicle or G1 treatment, but the vast majority of uterine epithelial cells entered S-phase after estradiol stimulation (Fig. 5). We also examined whether GPR30 could antagonize estradiol’s effects on uterine gene expression and epithelial cell proliferation. ICI inhibited the induction of calpactin, lactotransferrin (LTF), BIP, the progesterone receptor (PR), and cyclin E1 by estradiol, but G1 could not counteract the induction of these genes (supplemental Fig. 9). In addition, we did not observe any antagonistic effect of G1 on estradiol-stimulated uterine epithelial cell proliferation (supplemental Fig. 10). In a second set of experiments, we analyzed the impact of G1 on mammary gland development. Estradiol significantly stimulated epithelial cell proliferation (Fig. 6A) and induction of the target gene INDO (24) (Fig. 6B) and provoked ductal growth and endbud formation in the mammary glands (Fig. 6C). G1 did not show any significant effect in these readout paradigms compared with the control group (Fig. 6, A–C).
G1 does not influence the expression of classical estrogen target genes in the uterus. Two weeks after ovariectomy, C57BL/6 mice were injected sc with vehicle, 100 ng estradiol, or different doses of G1 (n = 8 per treatment group). Uterine expression of Wnt-4, frizzled-2, and IGF-I was analyzed 6 h after compound application, whereas cyclin E1 gene expression was analyzed 18 h after compound application using quantitative RT-PCR. Gene expression levels were normalized to TBP expression. ***, P < 0.005 (two-sided Student’s t test).
Estradiol, but not G1, stimulates uterine epithelial cell proliferation. Two weeks after ovariectomy, C57BL/6 mice were injected sc with vehicle, 100 ng estradiol, or different doses of G1 (n = 8 per treatment group). Uteri were processed for paraffin embedding and stained with anti-BrdU antibodies. Application of 100 ng estradiol for 18 h to ovariectomized mice led to proliferation of the majority of uterine epithelial cells, whereas almost no proliferating cells were observed after treatment with vehicle or G1 (100 ng to 100 μg).
The GPR30 agonist G1 does not stimulate estrogenic responses in the mammary gland. Two weeks after ovariectomy, C57BL/6 mice were treated sc for 3 wk with vehicle, 100 ng estradiol, or 2 μg G1 (n = 8 per treatment group). Estradiol, but not G1, stimulated mammary epithelial cell proliferation (A) and induced the target gene INDO (B). In the mammary gland, only rudimentary ducts were visible after vehicle or G1 treatment, whereas estradiol treatment stimulated ductal growth and endbud formation (C). *, P < 0.05; ***, P < 0.005 (two-sided Student’s t test).
In summary, using G1 as tool compound in vivo, we were not able to demonstrate any contribution of GPR30 to classical estrogenic responses in the uterus and the mammary gland.
Discussion
Conflicting data regarding the subcellular localization of GPR30 (6, 7, 19) as well as its signaling properties in response to estradiol (6, 7, 18, 25) prompted us to investigate whether GPR30 fulfills the criteria of a plasma membrane-bound estradiol receptor. Moreover, we were interested to obtain insight into the potential biological role of GPR30 by use of the GPR30-selective agonist G1 (20) in vivo. Our four main findings were as follows : 1) GPR30 localized to the endoplasmic reticulum, 2) GPR30 did not bind estradiol, 3) GPR30 did not signal in response to estradiol, and 4) use of the selective GPR30 agonist G1 in vivo did not show any involvement of GPR30-mediated signaling in classical estrogenic responses in the uterus or the mammary gland.
The finding that GPR30 localized to the endoplasmic reticulum in transiently transfected cells was in line with the results obtained by Revankar and co-workers (6). Other groups observed that GPR30 localized to the plasma membrane (7, 19). In these studies, cells expressing variously tagged GPR30 versions were used. Localization of these GPR30 fusion proteins seemed to depend on the cell type analyzed and the GPR30 tag used (19). It was reported that GPR30 localized to the endoplasmic reticulum in COS-7 cells transfected with GFP-GPR30 fusion proteins (19) but seemed to be expressed in the plasma membrane of HEK293 cells transfected with hemagglutinin-tagged GPR30 (7). Here, we analyzed the subcellular localization of tagged as well as untagged GPR30 versions. With three different antisera, two directed against the C terminus and one directed against the N terminus of GPR30, we observed GPR30 expression in the endoplasmic reticulum in a variety of transfected cells including HEK293 cells. Because positive immunostaining was achieved only after permeabilization of the fixed cells, even when an antibody against the N terminus of GPR30 was used, we concluded that GPR30 localized to the cytoplasm of transiently transfected cells. Our findings were further strengthened by identical results obtained from MDA-MB231 and HEC50 cells endogenously expressing GPR30. Interestingly, the presence or absence of serum (including estradiol) in the growth media had no impact on the cellular localization of GPR30. Of course, we cannot rule out the possibility that very little amounts of receptor protein localized to the plasma membrane but were below the limit of detection.
Our finding that GPR30 did not bind to estradiol and did not signal in response to estradiol is complementary to two other studies (18, 25). Pedram and co-workers (18) could not demonstrate cAMP elevation or ERK phosphorylation in GPR30-positive, ER-negative breast cancer cells (SKBR-3 cells), although other groups reported GPR30-mediated estradiol signaling in SKBR-3 cells (7, 20). In addition, all nongenomic estradiol effects in MCF-7 cells were blocked by ICI, and silencing of GPR30 in these cells had no impact on nongenomic estradiol responses. The only protein that bound estradiol in MCF-7 cells was the classical ERα, but not GPR30 (18). Others demonstrated that GPR30 expression in MCF-7 cells inhibited cellular growth independent from estrogen treatment (25). Based on these results, it was concluded that GPR30 is either a true orphan receptor or its ligand might be secreted into the medium by MCF-7 cells (25). Compared with transient transfection, very little expression of GPR30 was evident in MDA-MB231 and HEC50 cells in our studies. Therefore, it might not be too surprising that we did not detect any specific binding of radioactive estradiol to HEC50 or MDA-MB231 cells. However, there was very high expression of GPR30 in transfected COS-7 or CHO cells, comparable to the expression of the classical ERα that we used as control. Despite the high expression of GPR30 in transiently transfected cells, we observed specific saturable binding of estradiol only to ERα, but not to GPR30. Our results are in contrast to studies using a fluorescent Alexa-coupled estradiol derivative (6). Whereas radioactive estradiol is cell membrane permeable, use of Alexa-labeled estradiol required the permeabilization of cells with saponin to provide access to the intracellularly expressed receptor. Most likely, detergent treatment might have impact on the three-dimensional structure of GPR30 or other unknown proteins that start to bind the Alexa-estradiol conjugate under these conditions. Importantly, it has to be noted that saturable specific binding of Alexa-estradiol to GPR30 has never been analyzed in these studies (6).
We were not able to demonstrate cAMP formation, calcium elevation, or ERK activation (Otto, C., unpublished results) in cells endogenously expressing GPR30 (MDA-MB231, HEC50, and MCF-7 cells) or transiently transfected with GPR30 (COS-7 and CHO cells). These results are in clear contradiction to earlier studies (6, 7, 13) but support similar observations made by Pedram and co-workers (18) in cells endogenously expressing GPR30. This inconsistency of results might be explained by the fact that in many experiments analyzing ERK phosphorylation in response to GPR30 activation, the appropriate vehicle control, i.e. stimulation of cells with vehicle for the same time intervals as with compounds, is missing (13, 26, 27). Instead, untreated cells were often used as controls (13, 26, 27), and exceedingly high concentrations of estradiol up to 1 μm were applied, leading to the analysis of effects that are most likely not receptor mediated (26, 27). ERK phosphorylation is especially sensitive toward vehicle stimulation. When we stimulated serum-starved COS-7 cells (transiently transfected with either empty vector or GPR30) with estradiol or vehicle (i.e. 0.0001% DMSO), we observed the same time-dependent induction of ERK phosphorylation after vehicle and estradiol application, independent of the presence or absence or GPR30 (Otto, C., unpublished results).
A long-lasting stimulation of calcium currents (for more than 140 sec) after estradiol or G1 stimulation was described in COS-7 cells transiently transfected with GPR30 (20). This observation is puzzling. Such long-lasting stimulations of calcium currents are typical for agents such as ionomycin, which severely disturb the integrity of cellular membranes. In the case of estradiol or other receptor ligands, transient increases in calcium currents similar to those obtained after ATP stimulation would have been expected. The type and final concentration of vehicle applied in the experiments describing long-lasting calcium currents after estradiol application remain unclear (20).
The fact that several groups report conflicting results regarding estradiol binding to GPR30 and signaling in response to estradiol is reminiscent of the situation in a different field of nongenomic research: the so-called membrane progesterone receptors. These were thought to mediate nongenomic progesterone responses (28) until it was demonstrated that these receptors localized to the endoplasmic reticulum and did not act as progesterone receptors (29).
From our in vitro experiments, we clearly conclude that GPR30 does not act as ER; however, we cannot exclude the possibility that GPR30 and ER act in concert in some signal transduction pathways, as was suggested previously (13, 14). Recently, it was demonstrated that G1 and estradiol induced in an estrogen response element-independent manner the c-fos gene in ovarian cancer cells. Fos gene induction by G1 or estradiol was blocked by ICI or by silencing of either ER or GPR30 (30). Because ICI was proposed to act as GPR30 agonist (6), the only interpretation of these results is that GPR30 and ER acted in the same signal transduction pathway. From the experimental design of this study, it remained unclear which receptor was acting upstream of the other receptor. Most importantly, direct experimental evidence that GPR30 acts as an ER in these cells is lacking (30).
The aim of the second part of our study was to elucidate the potential role of GPR30 in uterine or mammary gland biology, by exploiting the GPR30-selective agonist G1 (20) as a tool compound in vivo. It was speculated that GPR30 might be involved in the mediation of the uterine response after estradiol stimulation (15). In general, uteri of ovariectomized mice respond toward estradiol stimulation with cellular proliferation and the induction of target genes, such as lactotransferrin and the progesterone receptor. These uterine estradiol responses (17) as well as estradiol-induced ductal outgrowth and epithelial cell proliferation in the mammary gland can be blocked by the administration of neutralizing EGF antibodies (31). Notably, GPR30 (13), but also the classical ER (32), have been implicated in the transactivation of the EGF receptor pathway. Moreover, it was speculated that the early uterine response phase toward estradiol led to an activation of the canonical Wnt pathway that was independent of the classical ER and possibly involved a GPCR (23). Here, we demonstrated that G1 had no impact on classical estrogenic responses in the uterus and the mammary gland, such as epithelial cell proliferation, target gene induction, and endbud formation. In particular, and in contrast to estradiol, G1 was not able to activate the canonical Wnt pathway, ruling out the possibility that GPR30 might be the GPCR responsible for the early uterine responses after estradiol stimulation (23). It has to be taken into account that the relative binding affinity of estradiol to GPR30 is 20-fold lower than for ERα and that G1 binds GPR30 roughly with the same potency as estradiol (20). We therefore employed different doses of G1 covering a range of 1–1000 times the estradiol dose and leading up to 1000-fold higher exposure, i.e. to 1000-fold higher plasma levels, when compared with estradiol. Even at the highest dose, G1 did not exhibit any estrogenic effect although, according to our analysis, sufficiently high plasma levels for receptor activation were obtained. Nevertheless, the conclusion that GPR30 did not play any role in uterine and mammary gland biology might be premature, because we were not able to show calcium currents in response to G1 in MDA-MB231, HEC50, or MCF-7 cells or in COS-7 cells transiently transfected with GPR30. As long as there is no evidence that G1 competes for specific binding of radioactive estradiol in GPR30-expressing cells, we would be very careful with the conclusion that G1 is a GPR30 agonist. In addition, G1 might have effects completely unrelated to GPR30. For example, it was concluded that GPR30 contributed to the induction of thymic atrophy in mice in response to estradiol treatment (33). Whereas estradiol reduced thymic weight in intact female mice, G1 treatment had only a very mild effect (33). Using the same experimental conditions, we reproduced the estradiol effect on relative thymic weight but could not demonstrate any effect of G1 (supplemental Fig. 11). The major reason leading to the misinterpretation that G1 (and hence GPR30) might have impact on thymic atrophy is based on the experimental design of this study, i.e. the use of intact female mice that were not cycle controlled (33). In intact mice, high-dose estradiol treatment leads via negative hypothalamic feedback to the inhibition of endogenous estradiol production and produces constant estradiol levels due to external substitution of the hormone. Vehicle or G1-treated intact females were at random cycle and thus exhibited varying endogenous estradiol levels as evidenced by the strong variability of relative uterine weights in these two treatment groups (C. Otto, unpublished data). Therefore, the slight effect of G1 on absolute thymic weight (33) might be achievable if more mice in the G1 group than in the vehicle group were in the proestrous phase characterized by high estradiol levels, leading to ERα-mediated reduction in thymic weight.
Depending on whether G1 really is a GPR30 agonist or not, our in vivo studies with G1 result in two possible conclusions. Assuming that G1 is not a GPR30 agonist, we cannot learn anything from our in vivo studies about the physiological role of GPR30. Or, assuming that G1 is a GPR30 agonist, we would conclude that GPR30 is not crucially involved in mediating the effects of estradiol in the uterus and the mammary gland. This conclusion is further strengthened by the finding that GPR30-deficient mice seemed to be fertile and nursed their offspring (33). The phenotype of ERα-deficient mice (2) and the fact that the classical ERα can transactivate the EGF receptor pathway (32), which plays an important role in mediating estradiol’s effects in the uterus and the mammary gland, imply that GPR30 is not required for the mediation of estrogenic responses in these organs in vivo.
Taken together, the results from this study and other laboratories support the conclusion that GPR30 is still an orphan receptor. The classification of GPR30, by homology belonging to a family of peptide receptors, as an ER seems to be premature and has to be reconsidered.
Acknowledgments
We thank Prof. Eric Prossnitz for the generous gift of the GPR30 antiserum and the respective preimmune serum. We also thank E. Krahl, M. Sommer, D. Schwerdt, and C. Molitor for expert technical assistance. We are grateful to Dr. Tommaso Simoncini for helpful comments on the manuscript.
Disclosure Summary: C.O., B.R.-S., G.S., I.F., M.K., D.B., G.L., B.B., K.P., R.N., and K.-H.F. are employees of Bayer Schering Pharma AG.
Abbreviations:
- BrdU
5′-Bromo-2′-deoxyuridine
- DMSO
dimethylsulfoxide
- EGF
epidermal growth factor
- ER
estrogen receptor
- FCS
fetal calf serum
- GFP
green fluorescent protein
- GPCR
G protein-coupled receptor
- HTRF
homogenous time-resolved fluorescence
- ICI
ICI182780
- PGE2
prostaglandin E2.
