Hetero-oligomeric Complex between the G Protein-coupled Estrogen Receptor 1 and the Plasma Membrane Ca2+-ATPase 4b*

Background: GPER/GPR30's actions are unclear. Results: GPER/GPR30 and PMCA4b constitutively interact via PDZ-binding motifs. This inhibits PMCA but enhances GPER/GPR30 activity. GPER/GPR30 activation further suppresses PMCA activity via tyrosine phosphorylation. Conclusion: GPER/GPR30 and PMCA4b form a physical and functional complex. Significance: GPER/GPR30-PMCA4b interactions mediate cross-talk between GPER/GPR30 and Ca2+ signaling. The new G protein-coupled estrogen receptor 1 (GPER/GPR30) plays important roles in many organ systems. The plasma membrane Ca2+-ATPase (PMCA) is essential for removal of cytoplasmic Ca2+ and for shaping the time courses of Ca2+-dependent activities. Here, we show that PMCA and GPER/GPR30 physically interact and functionally influence each other. In primary endothelial cells, GPER/GPR30 agonist G-1 decreases PMCA-mediated Ca2+ extrusion by promoting PMCA tyrosine phosphorylation. GPER/GPR30 overexpression decreases PMCA activity, and G-1 further potentiates this effect. GPER/GPR30 knockdown increases PMCA activity, whereas PMCA knockdown substantially reduces GPER/GPR30-mediated phosphorylation of the extracellular signal-related kinase (ERK1/2). GPER/GPR30 co-immunoprecipitates with PMCA with or without treatment with 17β-estradiol, thapsigargin, or G-1. Heterologously expressed GPER/GPR30 in HEK 293 cells co-localizes with PMCA4b, the main endothelial PMCA isoform. Endothelial cells robustly express the PDZ post-synaptic density protein (PSD)-95, whose knockdown reduces the association between GPER/GPR30 and PMCA. Additionally, the association between PMCA4b and GPER/GPR30 is substantially reduced by truncation of either or both of their C-terminal PDZ-binding motifs. Functionally, inhibition of PMCA activity is significantly reduced by truncation of GPER/GPR30's C-terminal PDZ-binding motif. These data strongly indicate that GPER/GPR30 and PMCA4b form a hetero-oligomeric complex in part via the anchoring action of PSD-95, in which they constitutively affect each other's function. Activation of GPER/GPR30 further inhibits PMCA activity through tyrosine phosphorylation of the pump. These interactions represent cross-talk between Ca2+ signaling and GPER/GPR30-mediated activities.

Removal of cytoplasmic Ca 2ϩ following agonist-induced Ca 2ϩ signals is important for cell functions by controlling the dynamics of Ca 2ϩ signals. For example, in vascular smooth muscle cells, the enhanced rate of cytoplasmic Ca 2ϩ removal causes faster relaxation, allowing for a more dynamic control of vascular tone (17). In addition, cytoplasmic Ca 2ϩ removal prevents Ca 2ϩ overload and its consequences. In the opposite direction, reductions in the rate of cytoplasmic Ca 2ϩ removal prolong the time courses of Ca 2ϩ -dependent processes (18,19). The plasma membrane Ca 2ϩ -ATPase (PMCA) is a major Ca 2ϩ extrusion mechanism in many cell types. In vascular endothelial cells, PMCA has been suggested to be responsible for up to 50% of the removal of cytoplasmic Ca 2ϩ (20). PMCA4b is the predominant isoform in vascular endothelial cells, smooth muscle cells, and human embryonic kidney (HEK) 293 cells (21). PMCA activity is regulated by its interactions with many other proteins. Calmodulin (CaM) tightly regulates the pump's activity through high affinity interaction with the PMCAs (22,23). A significant number of other proteins interact with PMCA via its PDZ-binding domain in the C terminus, including members of the MAGUKs, such as PSD-95/SAP90, SAP97/hDlg, SAP102, and PSD-93/chapsyn-110, and neuronal nitric-oxide synthase (24 -28). Interaction with PSD-95 facilitates plasma membrane targeting and function of PMCA4b (27). These interactions are important for anchoring and targeting of PMCA to the plasma membrane and regulate its roles in signaling.
It is completely unknown whether GPER/GPR30 affects cellular Ca 2ϩ homeostasis via modulation of PMCA activity. We have begun to test the idea that significant cross-talk exists between GPER/GPR30 signaling and Ca 2ϩ signaling via interactions between this receptor and components of the Ca 2ϩ signaling machinery. In this study, we describe physical interactions and mutual functional influences between GPER/GPR30 and PMCA4b in the vascular endothelium. Multiple experimental paradigms performed under basal conditions, receptor agonism, and gene silencing in primary porcine aortic endothelial cells (PAECs) or with overexpression of wild type and modified versions of GPER/GPR30 and PMCA4b in HEK 293 cells demonstrated the following: 1) GPER/GPR30 activation inhibits PMCA activity by promoting tyrosine phosphorylation of the pump; 2) GPER/GPR30 and PMCA4b constitutively and physically interact; 3) this interaction inhibits PMCA activity independently of phosphorylation but promotes GPER/GPR30 activity; and 4) interaction occurs via their C-terminal PDZbinding domains and their association with PSD-95. The implications of these results on Ca 2ϩ -dependent activities and GPER/GPR30 signaling are discussed.

Experimental Procedures
Cell Isolation and Culture-PAECs were obtained as described previously (29). Briefly, the intima of freshly isolated porcine thoracic aortas was gently scraped and resuspended in M-199 medium (Caisson Laboratories, Logan, UT) containing 10% newborn calf serum (Fisher) and 1% penicillin/streptomycin (MP Biomedicals, Solon, OH). Primary vascular smooth muscle cells (VSMCs) were subsequently obtained from the same vessels as described previously (30). Human embryonic kidney (HEK) 293 cells (ATCC) were cultured in DMEM containing 5-10% fetal bovine serum. Cells were cultured in a 37°C incubator containing 5% CO 2 humidified air. The medium was frequently renewed.
Molecular Biology-Total mRNA was obtained from primary PAECs, VSMCs, HEK 293 cells, and primary human umbilical vein endothelial cells (ATCC) using Promega's ImProm-II reverse transcription system (Promega). cDNAs of human PMCA4b and GPER/GPR30 were then reverse-transcribed from HEK 293 cells' mRNA. A BamHI and an XbaI restriction site was introduced into the N-and C-terminal ends, respectively, of human PMCA4b and GPER/GPR30 by PCR amplification. These sequences were then incorporated in a pcDNA3.1 mammalian expression vector. C-terminal fusions PMCA4b-DsRed2 and GPER/GPR30-DsRed2 were then constructed by inserting DsRed2 (Clontech) via a linker (IDYDVLDYAGS) to the C-terminal end of PMCA4b and GPER/GPR30. The GPER/ GPR30-ECFP fusion was constructed by replacing the EYFP moiety and the CaM-binding linker sequence of a previously published GPER/GPR30 biosensor (BSGPER(330 -351)) (30) with the entire GPER/GPR30 sequence. The N-terminal fusions DsRed2-GPER/GPR30 and ECFP-PMCA4b were generated by inserting DsRed2 or ECFP between the HindIII and BamHI restriction sites upstream of PMCA4b or GPER/GPR30 in the pcDNA3.1 plasmids. FLAG (MDYKDDDDK) and HA (MYPYDDVPDYA) tag sequences were PCR-generated using forward and reversed primers designed to span the entire tag length, with a KpnI and a BamHI restriction site added to the N-and C-terminal ends of the tag sequences, respectively. These FLAG and HA tags were then inserted upstream of the N terminus of GPER/GPR30 or PMCA4b, respectively, in the pcDNA3.1 vector. The FLAGtagged GPER/GPR30 and HA-tagged PMCA4b mutants with deletions of the four C-terminal residues (SSAV for GPER/ GPR30 and ETSV for PMCA4b) were subsequently generated using standard molecular biology techniques. All constructs were verified by DNA sequencing (University of Missouri-Columbia DNA Core Facility).
Transfection-Oligonucleotides and plasmids encoding various wild-type or mutant constructs were transfected into primary PAECs, HEK 293 cells using the Transit-2020 or Transit-Oligo transfection kits (Mirus Bio LLC) as per the manufacturer's instructions.
Gene Silencing of GPER/GPR30, PMCA, and PSD-95-GPER/GPR30 was knocked down in primary PAECs using an antisense oligonucleotide directed against the porcine GPER/ GPR30 sequence. For identification of transfected cells during PMCA activity measurement, GPER/GPR30 antisense and scrambled oligonucleotides were both conjugated with TAMS, a red fluorescent marker. PSD-95 antisense and scrambled sequences were based partly on previously published data, with some adjustment to match porcine sequences (31). All oligomer nucleotides were synthesized by Integrated DNA Technologies (Coralville, IA). The antisense and scrambled oligo sequences for these targets are shown in Table 1.
For experiments measuring PMCA activity in GPER/GPR30 knockdown cells, transfection was performed on PAECs plated on 60-mm dishes containing number 1.5 coverslips. Two days after transfection, the coverslips were processed for imaging to measure PMCA activity, and the remaining cells on the dishes were collected for Western blotting to verify effective knockdown of GPER/GPR30. Cells on coverslips were first loaded with fura-2/AM. Following loading, cells expressing GPER/ GPR30 antisense or scrambled oligo were detected and marked using TAMS fluorescence emission at 610 nm in response to excitation at 570 nm. Imaging cube was then switched for measurement of PMCA activity using fura-2 fluorescence on pre-marked TAMS (ϩ) cells. After the imaging experiments, the same coverslips were washed and fixed with 0.4% Triton X-100 for 10 min, followed by incubation overnight with a rabbit anti-GPR30 antibody (N-15, SC-48525, Santa Cruz Biotechnology). The coverslips were then incubated with anti-rabbit FITC-conjugated secondary antibody for 1 h. Following washing, GPER/GPR30's levels of detection were verified by FITC emission intensity at 525 nm, with excitation at 490 nm. In addition, remaining cells from the plates were lysed for immunoblotting for GPER/GPR30. Following electrophoresis, the membrane was cut between the levels of vinculin (130 kDa) and GPER/GPR30 (38 kDa), and the fragments were simultaneously probed, respectively, for vinculin and GPER/GPR30. Following development, the lower fragment (initially probed for GPER/ GPR30) was stripped and reprobed for ␤-actin. Densitometric values of vinculin and ␤-actin were averaged and used as loading control for those from GPER/GPR30 bands. This combined approach guaranteed that PMCA activity was performed on control or GPER/GPR30-knocked down cells.
Measurement of Intracellular Free Ca 2ϩ -Primary PAECs were plated on number 1.5 glass coverslips (Fisher) and grown in phenol red-free medium until subconfluence with or without treatments prior to experiment. Cells were incubated with 4 M fura-2/AM (Invitrogen) in culture medium for 30 min at 37°C. The dye was then removed, and cells were equilibrated in Ca 2ϩcontaining modified Tyrode's buffer (composition in mM: 150 NaCl, 2.7 KCl, 1.2 KH 2 PO 4 , 1.2 MgSO 4 , 10 HEPES, 1 CaCl 2 , pH 7.4) for 15 min at room temperature. Excitation of fura-2 alternated between 340 nm and 380 nm for 50 ms each per 500-ms cycle from an ultra-high speed wavelength switcher (Lambda DG-4, Sutters Instruments). The excitation switching lapse between the two wavelengths was 1 ms. Emission light was collected at 510 nm through an ultra-rapid filter wheel (Lambda-10B, Sutters Instruments) and processed by an electron-multiplying charge-coupled digital camera (DU-885, Andor Technology). Pairs of images were analyzed using the Imaging Workbench 6.0 software (INDEC Biosystems). The ratio of emission fluorescence intensities collected following the excitation at 340 nm and 380 nm was used as a measure of intracellular Ca 2ϩ concentration. For some experimental paradigms, free intracellular Ca 2ϩ concentrations were calculated using Equation 1, where the K d value is 224 nM for fura-2; R is the observed ratio fluorescent signal during the experiment. S f and S b represent the emission intensities collected at 510 nm corresponding to the Ca 2ϩ -free and Ca 2ϩ -bound states of fura-2. We avoided potential errors associated with using fixed values for R min and R max by determining these values in individual cells in each experiment. We first determined the average relationship between R min , R max , and the basal R basal value, namely the R value obtained in Ca 2ϩ -free medium from unstimulated cells. Following dye loading, R basal values were obtained. The same cells were then incubated on stage with 20 M BAPTA/AM (Tocris Bioscience, Ellisville, MO) for 30 min to obtain R min values. R max values from the same cells were next obtained by adding 5 M ionomycin and 10 mM CaCl 2 . Comparing the average (n ϭ 100) absolute R min values with the observed fura-2 ratios in the same cells in Ca 2ϩ -free medium prior to addition of BAPTA/AM (R basal ) and the average R max value, Equation 2 was obtained.
Similarly, Equation 3 was obtained between S f and S basal , which is the fluorescence emission intensity at 510 nm at the beginning of the experiment with cells in Ca 2ϩ -free medium in response to excitation at 380 nm.
No statistical differences (n ϭ 100 cells from five separate pilot experiments) were observed between the measured R min and S f values versus the calculated R min and S f values using Equations 2 and 3. Because R basal and S basal values were readily measured at the beginning of every experiment, and corresponding R max and S b values were also easily obtained by the end of each imaging time course by adding high concentrations of ionomycin and Ca 2ϩ , free Ca 2ϩ concentrations in individual cells could be calculated from Equations 1-3 with relatively high reliability. Measurement of PMCA Activity in Living Cells-Cells were prepared as described above in the Ca 2ϩ imaging section. Thapsigargin (1 M) was added to nominally Ca 2ϩ -free buffer to deplete the endoplasmic reticulum of Ca 2ϩ . Ca 2ϩ influx was initiated by the addition of 1.5 mM CaCl 2 with or without specified concentrations of G-1. When peak influx was reached, the extracellular medium was replaced by one containing 5 mM BAPTA and 150 mM N-methylglucamine in place of NaCl, to block the Na ϩ /Ca 2ϩ exchanger and abolish Ca 2ϩ entry. The rate of decay in intracellular Ca 2ϩ concentration now only reflected the rate of Ca 2ϩ extrusion via the PMCA, and it was determined by fitting the time course of the apparent free Ca 2ϩ concentration to the mono-exponential Equation 4.
In previous pilot studies, a number of fitting approaches were tested, including linear, mono-, bi-, and tri-exponential equations; the mono-exponential equation provided the most reliable data with the best residuals (18,19). The extrusion rates of Ca 2ϩ are presented as relaxation times (). Because the activity of PMCA is intrinsically Ca 2ϩ -dependent, comparisons were made only among cells in which the free Ca 2ϩ concentrations at the beginning of the Ca 2ϩ extrusion time course were in the same ranges (18,19). Absolute Ca 2ϩ values in individual cells were calculated as described above. Relative PMCA activity was expressed as the inverse of the relaxation time.
Confocal Microscopy-Confocal microscopy was performed using a Leica TCS SP8 confocal microscopy system (University of Iowa). HEK 293 cells plated on number 1.5 coverglass were transiently transfected with wild-type or mutant GPER and PMCA4b fused at their N terminus with DsRed2 or ECFP, respectively, followed by mounting with Prolong Gold Antifade Reagent (Fisher) 24 h prior to imaging. RGB images were analyzed using the ImageJ software (rsb.info.nih.gov) for visualization of separately labeled proteins.
Western Blotting-Cells were lysed in a buffer containing (in mM) 25 Tris-HCl, 148 NaCl, 97.6 NaF, 27.8 Na 4 P 2 O 7 , 271.8 Na 3 VO 4 , Triton X-100 (1% v/v), trypsin inhibitor (45 M), 1:200 protease inhibitor mixture (Sigma), and 2 g/ml PMSF. Protein concentrations were determined using the Pierce BCA protein assay kit (Thermo Fisher Scientific, Rockford, IL). Following SDS-PAGE, proteins were transferred to a PVDF membrane (Thermo Scientific, Rockford, IL). The membranes were then blocked with 5% BSA overnight at 4°C. Primary antibodies were applied in Tris-buffered saline containing Tween 20 (TBST) for 1 h at room temperature. Membranes were subsequently incubated with an appropriate secondary antibody in TBST. Enhanced chemiluminescence (GE Healthcare, Buckinghamshire, UK) was used for film development and visualization of protein bands. Films were then scanned using Chemidoc XRS Imager (Bio-Rad), and densitometric values of bands were quantified using Image Lab 5.0 software (Bio-Rad). Densitometric values of the proteins of interest were corrected for corresponding values of the simultaneously probed loading control protein.
Co-immunoprecipitation-Co-immunoprecipitation was performed using protein A/G (Thermo Scientific) as per the manufacturer's instructions. Cell lysis was performed on monolayer cells without trypsinization to avoid complications having to do with shear during the process. Cells were lysed on ice for 15 min, and the lysate was centrifuged at 21,000 ϫ g for 5 min at 4°C. Following pre-clearing, the protein content of the eluant was determined using the BCA assay (Pierce). Three milligrams of total cellular proteins were then rocked with the resin-antibody conjugates for 3 h in 5-ml columns at 4°C prior to eluting. Following electrophoresis and transfer, membranes were cut between the levels of the bait and the prey proteins prior to incubation with separate primary antibodies. Densitometric values of the baits were corrected for those of the preys using the ImageLab 5.0 software (rectangular volume tool) for analysis.
Statistical Analysis-Data are expressed as means Ϯ S.E. Statistical analysis was performed using Student's t test, assuming unequal variances between control and treated groups. Statistical significance was determined as p Ͻ 0.05.

Results
GPER/GPR30 Activation Inhibits PMCA Activity-We first verified the expression of GPER/GPR30 in a number of vascular cells and cell lines, including primary PAECs and VSMCs, primary human umbilical vein endothelial cells and HEK 293 cells. Total mRNAs were isolated from these cells, followed by RT-PCR to amplify a segment of the submembrane domain 4 of GPER/GPR30 (amino acids 330 -375). Total lysate from these cells were probed for GPER/GPR30 using the N-15 antibody. Fig. 1A shows both mRNA (upper panel) and protein expression levels (lower panel) of GPER/GPR30 in these cell types.
To examine the potential role of GPER/GPR30 on Ca 2ϩ efflux, we first tested the effects of the GPER/GPR30 agonist G-1 (16) on PMCA activity in primary vascular endothelial cells. PMCA is known to be the key component of cytoplasmic Ca 2ϩ removal in these cells (20), which play an indisputable role in nitric oxide production and other vascular functions. We and others have successfully developed a protocol to measure PMCA activity in these cells (18 -20). This protocol involves initial depletion of intracellular Ca 2ϩ stores with the irreversible sarco/endoplasmic reticulum Ca 2ϩ -ATPase pump inhibitor thapsigargin in Ca 2ϩ -free medium, followed by activation of Ca 2ϩ entry by addition of extracellular Ca 2ϩ . PMCA activity is subsequently assessed as extracellular Ca 2ϩ is removed and Na ϩ -Ca 2ϩ exchanger prevented by substituting Na ϩ with Nmethyl-D-glucamine. A typical experiment time course is shown in Fig. 1B. Because PMCA activity is intrinsically Ca 2ϩdependent, namely the higher starting Ca 2ϩ values are associated with higher PMCA activity, Ca 2ϩ values at the beginning of the extrusion time courses were binned, and comparisons were made among cells with values in the same range. R min and R max values for individual cells were determined in each experiment so that the calculated Ca 2ϩ values reflect more precisely the free intracellular Ca 2ϩ concentrations (see "Experimental Procedures"). This binning approach helps minimize erroneous interpretation of effects on PMCA due to differences in total cytoplasmic Ca 2ϩ signals that might be due to other factors. To initially test the effect of GPER/GPR30 activation on PMCA activity, different doses of G-1 were added together with thapsigargin in this protocol. When PAECs with similar free Ca 2ϩ levels at the beginning of Ca 2ϩ extrusion, following removal of extracellular Ca 2ϩ , were binned together and rates of extrusion compared, G-1 clearly demonstrated a dose-dependent inhibitory effect on PMCA activity, as evidenced by increases in the Ca 2ϩ extrusion times. Fig. 1C summarizes the dose-dependent effect of G-1 on PMCA activity. Maximal effects were about 45% reduction in PMCA activity. Fig. 1D shows fits of average time courses from cells with free cytoplasmic Ca 2ϩ at the start of the Ca 2ϩ extrusion binned around 1000 nM from cells treated with a number of G-1 doses. Fig. 1, E and F, showed representative residuals (goodness of fits) of the mono-exponential fits for PMCA activity from control cells and cells treated with 1 M G-1. All other fits yielded equally good residuals (data not shown).

G-1 Inhibits Ca 2ϩ Efflux via Tyrosine Phosphorylation of
PMCA-A number of factors have been shown to inhibit PMCA activity. These include calpain-mediated cleavage (32), interaction with POST (partner of stromal interaction molecule 1) (33), or Src-dependent phosphorylation at Tyr-1176 in the C terminus of PMCA4b (34,35). Estrogen has been shown to activate Src-related tyrosine kinase activity and tyrosine phosphorylation of Shc adapter protein (1). As PMCA4b is the major PMCA isoform in endothelial cells (36), we considered the possibility that G-1 also inhibits PMCA activity by promoting tyro-sine phosphorylation of the pump. To test this possibility, we first verified that G-1 can promote tyrosine phosphorylation of the PMCA. Primary PAECs were treated with vehicle or G-1, with or without pretreatment with the Src kinase inhibitor PP2, which has been shown to effectively inhibit tyrosine phosphorylation of PMCA4b (34). Following lysis, PMCA was immunoprecipitated using the PMCA antibody 5F10, followed by probing of the pulldown fractions with anti-phosphotyrosine antibody PY20 (Thermo Scientific). Fig. 2A clearly shows that G-1 stimulated robust tyrosine phosphorylation of the PMCA and that inhibition of Src activity using PP2 prevented this effect. We next examined the effects of these treatments on PMCA activity. G-1 (0.5 M) produced the observed inhibitory effect as in Fig. 1, whereas pretreatment with Src kinase inhibitor PP2 prevented this effect (Fig. 2B). Fig. 2C shows average time courses of the Ca 2ϩ extrusion phases in these experiments. These data indicate that G-1 inhibits PMCA activity by promoting tyrosine phosphorylation of PMCA.
Heterologous Expression of GPER/GPR30 Decreases PMCA Activity-To further confirm that GPER/GPR30 inhibits PMCA activity, we heterologously expressed human GPER/GPR30 in HEK 293 cells. As shown in Fig. 1A, both PAECs and HEK 293 cells express GPER/GPR30 at the mRNA and protein levels. However, these overexpression experiments were performed in HEK 293 cells for relative ease of transfection of large plasmids as compared with primary endothelial cells. To confirm imaging was performed in cells expressing GPER/GPR30, the receptor was fused with the fluorescent protein DsRed 2 . As a first control, only the DsRed2 moiety was expressed, followed by loading with fura-2/AM. Fig. 3, A and B, shows fluorescence images of DsRed2 and fura-2 in the same cell population containing both DsRed2-transfected and nontransfected cells. Merged image (Fig. 3C) allows easy distinction of these two populations. PMCA activity was then performed as described in Fig. 1B and was compared between DsRed2-expressing and nonexpressing cells in the same microscopic field. There was no difference in Ca 2ϩ extrusion rates between these two groups, indicating that expression of DsRed2 alone did not interfere with PMCA activity (data not shown). The same approach was then performed for cells expressing GPER/GPR30-DsRed2 fusion (Fig. 3, D-F). Fig. 3G shows fits of Ca 2ϩ extrusion time courses between and mock-transfected cells and cells transfected with GPER/GPR30-DsRed 2 with or without treatment of 0.5 M G-1 prior to Ca 2ϩ extrusion. Interestingly, overexpression of GPER/GPR30 inhibited PMCA activity by 40% in the absence of agonist treatment, and this effect is significantly further pronounced by treatment with G-1 (Fig. 3H).
Knockdown of GPER/GPR30 in Endothelial Cells Increases PMCA Activity-To further assess the effect of GPER/GPR30 on PMCA activity, we designed antisense oligonucleotides directed against porcine GPER/GPR30, and we compared PMCA activity in primary PAECs transfected with antisense or scrambled nucleotides. Both the antisense and scrambled oligonucleotides were conjugated with TAMS, a red fluorescent tag that would help identify cells transfected with the respective oligo prior to measuring PMCA activity using fura-2 fluorescence in the same field of cells. Transfection with GPER/GPR30 antisense resulted in a 70% reduction in GPER/GPR30 expression (Fig. 4A). Fig. 4B shows pictures of PAECs transfected with antisense (Fig. 4B, AS) or scrambled oligo (Fig. 4B, Scr) labeled with red fluorescence and a corresponding picture of fura-2 fluorescence in the same group of cells, ensuring that imaging was performed in cells expressing the respective oligos. To further verify GPER/GPR30 knockdown efficiency in cell populations in which PMCA activity was assessed, immunofluorescence of GPER/GPR30 was also performed on the same coverslips following the PMCA experiment (see "Experimental Procedures"). There was a clear reduction in FITC intensity in cells transfected with the GPER/GPR30 antisense (Fig. 4B, GPR30/AS) compared with those transfected with the scrambled oligo (Fig. 4B, GPR30/Scr). This is consistent with Fig. 4A, and it further confirms that PMCA activity was compared between cells with different levels of GPER/GPR30 expression. Fig. 4C shows the average time courses of Ca 2ϩ extrusion in cells transfected with GPER/GPR30 antisense or scrambled oligo. Knockdown of GPER/GPR30 apparently was associated with an ϳ2-fold increase in the rate of Ca 2ϩ extrusion via the PMCA. This result is quite consistent with the pharmacological data in Fig. 1 and overexpression data in Fig. 3.
PMCA Affects Function of GPER/GPR30 -The effect of GPER/GPR30 to inhibit PMCA activity raised the question whether PMCA reciprocally affects GPER/GPR30's function. Our initial task was to verify a parameter for GPER/GPR30 activity. Despite evidence that GPER/GPR30 was a G␣ s -associated GPCR (1), it has been shown that 17␤-estradiol failed to stimulate cAMP production in cells endogenously expressing GPER/GPR30 or heterologously overexpressing the receptor (37,38). More recently, it was shown that neither 17␤-estradiol nor GPER/GPR30 agonist G-1 stimulated cAMP production in cells overexpressing GPER/GPR30 and that GPER/GPR30 constitutively inhibits adenylyl cyclase-mediated cAMP produc-tion by interacting with the MAGUKs and AKAP5 (15). However, G-1 has been shown previously to increase ERK1/2 phosphorylation in many publications (1, 39 -41). We therefore first verified in our cell systems whether G-1-induced ERK1/2 phosphorylation could be utilized as a measure of GPER/GPR30 activity. In primary PAECs, G-1 shows a clear dose-dependent effect to stimulate ERK1/2 phosphorylation (Fig. 5A). It was noted that this effect is reduced in later passages of endothelial cells (data not shown). To further confirm the specificity of this effect, G-1-induced ERK1/2 phosphorylation was compared in primary PAECs transfected with GPER/ GPR30 antisense or scrambled oligo as described in Fig. 4A. Fig.  5B clearly shows that G-1-induced ERK1/2 phosphorylation was substantially reduced in GPER/GPR30 knockdown cells. To further confirm this possibility, we turned to overexpression experiments in HEK 293 cells. As noted in Fig. 1A, we observed that HEK 293 cells express GPER/GPR30 both at the mRNA and protein levels. Treatment of mock-transfected HEK 293 cells with G-1 triggered ERK1/2 phosphorylation; however, overexpression of GPER/GPR30 in these cells (Fig. 5C, upper panels) substantially increased G-1-induced ERK1/2 phosphorylation (Fig. 5C, lower panels). These data confirm that G-1induced ERK1/2 phosphorylation can be used as a parameter for GPER/GPR30 activity. Additionally, the middle panel in Fig. 5C, showing increased GPER/GPR30 in cells transfected with GPER/GPR30, together with data in Fig. 4A, showing that GPER/GPR30 expression is reduced in cells transfected with GPER/GPR30 antisense, confirmed the specificity of the GPER/ GPR30 antibody used.
Having confirmed a functional assay for GPER/GPR30 activity, we next tested whether PMCA knockdown would affect GPER/GPR30 function. An antisense sequence was designed to target all isoforms of PMCA in porcine endothelial cells (Table  1). Fig. 6A shows an ϳ60% reduction in total PMCA expression in primary PAECs transfected with the PMCA antisense. The effect of PMCA knockdown was now examined on G-1-induced ERK1/2 phosphorylation as an indicator of GPER/ GPR30 activity. G-1 robustly stimulated ERK1/2 phosphoryla-tion in primary PAECs (Fig. 6B, left panel), consistent with the data in Fig. 5. PMCA knockdown markedly reduced this effect (Fig. 6B, right panel). To verify that PMCA knockdown is associated with reduction in PMCA functions, we compared Ca 2ϩ extrusion time courses in cells transfected with the PMCA antisense or scrambled oligo (Fig. 6C). There was a clear reduction in PMCA activity in cells transfected with the PMCA antisense (Fig. 6D), indicating the functional efficiency of PMCA knockdown.
GPER/GPR30 Co-immunoprecipitates with PMCA in Vascular Endothelial Cells-The mutual functional effects between GPER/GPR30 and PMCA, and the fact that overexpression of GPER/GPR30 inhibits PMCA independently of agonist stimulation, raised the possibility that these proteins might physically interact in a hetero-oligomeric complex. We first tested this idea by performing reciprocal co-immunoprecipitation in primary endothelial cells under nonstimulated conditions or acute treatment with thapsigargin, endogenous GPER/GPR30 ligand   (Fig. 7A). Likewise, GPER/GPR30 was clearly detected in all PMCA pulldown fractions from all samples (Fig. 7B). The upper and lower immunoblots of PMCA or GPER/GPR30 in Fig. 7, A and B, were two fragments of the same SDS-PAGE membranes, guarantee-FIGURE 5. G-1-induced ERK1/2 phosphorylation as a parameter for GPER/ GPR30 function. A, effect of G-1 on ERK1/2 phosphorylation in primary PAECs. Primary PAECs were treated with the specified concentrations of G-1 for 15 min prior to cell lysis. Following immunoblotting for ERK1/2 phosphorylation, the same membrane was stripped and reprobed for total ERK1/2. Histogram represents average relative ratio of the densitometric values of ERK1/2 phosphorylation over total ERK1/2. B, effect of GPER/GPR30 gene silencing on G-1-induced ERK1/2 phosphorylation. Cells were transfected with the same GPER/GPR30 antisense-TAMS or scrambled oligo-TAMS as in Fig. 3. Two days after transfection, cells were treated with 100 nM G-1 for 15 min prior to lysis. Immunoblotting and data analysis were performed as in A. C, effect of G-1 on ERK1/2 phosphorylation in HEK 293 cells transfected with or without GPER/GPR30. HEK 293 cells were mock-transfected or transfected with GPER/GPR30. Cells were treated with 100 nM G-1 as indicated 24 h posttransfection. Lysates were probed for GPER/GPR30 and vinculin (fragments from the same membrane) or ERK1/2 phosphorylation, followed by stripping and reprobing for total ERK1/2. Histogram, relative ERK1/2 phosphorylation. Asterisks, p Ͻ 0.05 from mock-transfection, untreated sample. ?, p Ͻ 0.05 from mock-transfection, treated sample.

TABLE 1
Antisense and scrambled oligonucleotide sequences used for knockdown of GPER/GPR30, PMCA and PSD-95 Phosphorothioate bonds are denoted by asterisks, and the m indicates 2Ј-O-methyl RNA bases. TAMS, red fluorescent marker for identification of transfected cells during PMCA activity measurement.

Hetero-oligomeric Complex between GPER/GPR30 and PMCA4b
ing that the prey protein levels detected truly corresponded to the "input" levels of the bait proteins shown. These data clearly show that PMCA and GPER/GPR30 are in the same complex in cells under different scenarios, including basal condition. We did not see a significant difference in the association between GPER/GPR30 and PMCA among the different conditions (Fig.  7A). This appears to suggest a "constitutive" interaction between the two proteins. To further verify the specificity of the antibodies used, total cell lysate from the same treated cell samples were first probed using nonimmune rabbit antibody (upper panel, Fig. 7C), which showed no immunoreactivity. The same membrane was then stripped and reprobed using the rabbit anti-GPER/GPR30 antibody used in the co-immunoprecipitation on Fig. 7A, which now showed clear GPER/GPR30 bands (lower panel, Fig. 7C). Likewise, identical samples probed using nonimmune mouse antibody showed no immunoreactivity at the level of PMCA (upper panel, Fig. 7D); however, reprobing with mouse anti-PMCA antibody (clone 5F10) showed clear PMCA bands (lower panel, Fig. 7D). These results support the validity of the co-immunoprecipitation data in Fig. 7, A and B. In our studies, two anti-GPER/GPR30 antibodies were used, each with a different epitope. The N-15 antibody recognizes the N-terminal fragment of GPER/GPR30, although the H-300 antibody (SC-134576, Santa Cruz Biotechnology) recognizes amino acids 75-375 of the GPER/GPR30 sequence. To further confirm the specificity of these antibodies for their respective epitopes, we generated fusion proteins between EYFP and each submembrane domain of GPER/GPR30. These fusions were expressed in HEK 293 cells, followed by a series of Western blotting analyses. As seen in the left panel in Fig. 7E, probing with anti-GFP antibody showed bands compatible with the size of EYFP and the submembrane domains of GPER/GPR30 (ϳ27 kDa). Following stripping and reprobing with the N-15 antibody, the lower bands at the level of GPER/GPR30 in the left panel in Fig. 7E were no longer recognized, because this antibody should not recognize any of the submembrane domains of GPER/GPR30. In contrast, this panel in Fig. 7E shows clear bands at the level of ϳ38 kDa, the size of the endogenous nascent form of GPER/GPR30, consistent with data in Fig. 1A showing both mRNA and protein expression of GPER/GPR30 in HEK 293 cells. After another stripping and reprobing of the membrane with the H-300 antibody, both the endogenous  GPER/GPR30 and the heterologously expressed fragments that contain only the submembrane domains were now evident. Together with data in Figs. 4A and 5C, these data strongly validate the specificity of the antibodies used in our co-immunoprecipitation experiments and indicate that GPER/GPR30 and PMCA are constitutively in the same complex in primary PAECs.
Co-localization of PMCA4b and GPER/GPR30 -The data presented so far suggest a possibility of direct interaction between PMCA and GPER/GPR30. PMCA4b is the predominant isoform in endothelial cells, vascular smooth muscle, and HEK 293 cells (36). To further strengthen the possibility that PMCA4b and GPER/GPR30 physically interact, ECFP and DsRed2 were fused to the N termini of PMCA4b and GPER/ GPR30, respectively. The two fusions were then co-expressed in HEK 293 cells. Confocal microscopy scanning clearly showed co-localization of ECFP-PMCA4b and DsRed2-GPER/GPR30 (arrows, Fig. 8, A-C).
PMCA4b and GPER/GPR30 Interact via PSD-95-Our data so far clearly demonstrate mutual inhibitory effects between PMCA4b and GPER/GPR30. It has been demonstrated that both PMCA4b and GPER/GPR30 interact with the PSD-95 via their C-terminal PDZ-binding motifs (15,26,27,42). For both proteins, the interaction with PSD-95 was shown to facilitate plasma membrane targeting. Given this background, we hypothesized that the physical and functional interactions between PMCA4b and GPER/GPR30 are mediated in part by the anchoring effect of PSD-95. To test this idea, we first examined the effect of silencing PSD-95 in primary PAECs on the interaction between PMCA and GPER/GPR30. Fig. 9A shows expression of PSD-95 in primary vascular endothelial cells and the effect of PSD-95 gene silencing. A 50% reduction in PSD-95 expression was achieved. PMCA-GPER/GPR30 interaction was then examined in cells transfected with PSD-95 antisense or scrambled oligo using co-immunoprecipitation. An ϳ45% reduction was observed in the association between PMCA and GPER/GPR30 in cells in which PSD-95 was knocked down (Fig.  9B), indicating a role of PSD-95 in mediating the interaction between GPER/GPR30 and PMCA4b.
PMCA4b and GPER/GPR30 Interact through Their C-terminal PDZ-binding Motifs-Both PMCA4b and GPER/GPR30 contain in their C terminus a type I PDZ-binding motif, which mediates their respective interaction with PSD-95. To test the idea that PMCA4b and GPER/GPR30 interact in part through these PDZ-binding motifs, we examined the effect of truncation of PMCA4b's PDZ-binding motif (ETSV) or GPER/ GPR30's (SSAV) on their association. HA-tagged wild-type PMCA4b or HA-tagged PMCA4b⌬ETSV was co-transfected with FLAG-tagged wild-type GPER/GPR30 or truncated GPER/ GPR30⌬SSAV into HEK 293 cells. The transfected PMCA4b was immunoprecipitated using anti-HA antibody, followed by probing of the respective membrane fragments for the HA tag (heterologous PMCA4b) or GPER/GPR30 using anti-FLAG antibody. Immunoblotting of equal amounts of total cell lysate shows that all samples expressed both HA-PMCA variants and FLAG-GPER/GPR30 variants as specified (Fig. 9C). Co-immunoprecipitation from the same samples clearly shows that truncation of the PDZ-binding motifs in GPER/GPR30 and/or PMCA4b significantly reduced their association in cells (Fig. 9,  D and E). Interestingly, removal of GPER/GPR30's PDZ-binding motif apparently has a much more profound effect, virtually abolishing the interaction between PMCA4b and GPER/ GPR30 (Fig. 9, D and E).
To examine the functional impact of the interaction between GPER/GPR30 and PMCA4b via their C-terminal PDZ-binding domains, we compared PMCA activity in cells transfected with either the FLAG-GPER/GPR30 or FLAG-GPER/GPR30⌬SSAV. Fig. 10A shows average Ca 2ϩ extrusion time courses in mock-transfected cells, cells expressing FLAG-GPER/GPR30, and cells expressing FLAG-GPER⌬SSAV. Again, only cells with similar free Ca 2ϩ values at the beginning of the extrusion time courses were binned for comparison ("Experimental Procedures"). PMCA activity was significantly reduced by ϳ40% in cells overexpressing wild-type GPER/ GPR30. Overexpression of the truncated version FLAG-GPER/ GPR30⌬SSAV significantly reduced the inhibitory effect on PMCA activity compared with the wild-type FLAG-GPER/ GPR30 (Fig. 10B), despite a small residual inhibitory effect. These functional data are consistent with the finding that truncation of the C-terminal PDZ-binding motif in GPER/GPR30 substantially reduced the interaction between GPER/GPR30 and PMCA4b (Fig. 9, C-E).

Discussion
In this study, we demonstrate physical and functional interactions between GPER/GPR30 and PMCA4b, a nonreceptor transmembrane protein with well established roles in the control of Ca 2ϩ homeostasis. Our data indicate that GPER/GPR30 can inhibit Ca 2ϩ efflux in the vascular endothelium via the PMCA by two independent mechanisms as follows: 1) PMCA tyrosine phosphorylation that occurs upon GPER/GPR30 activation, and 2) constitutive physical interaction between GPER/ GPR30 and PMCA4b. The link between agonist G-1 and GPER/ GPR30 was supported by two lines of evidence. First, G-1 promotes ERK1/2 phosphorylation in primary PAECs, and knockdown of GPER/GPR30 drastically reduced this effect (Fig.  4, A and B). Second, G-1 promotes ERK1/2 phosphorylation in HEK 293 cells, which also express GPER/GPR30 mRNA and protein, and overexpression of GPER/GPR30 in these cells substantially increases G-1-induced ERK1/2 phosphorylation (Fig.  4C). These results are consistent with previous studies showing that estrogen induced ERK1/2 phosphorylation via GPER/ GPR30 (1,43).
The observed effect of G-1 to inhibit PMCA activity via tyrosine phosphorylation of the pump is quite consistent with previous studies demonstrating a clear inhibitory role of phosphorylation of Tyr-1176 on PMCA4b activity (34,44,45). Activation PAECs were transfected as in A prior to co-immunoprecipitation (IP). PMCA pulldown fractions were probed for GPER/GPR30. Upper and lower immunoblots in all panels were fragments from the same respective SDS-PAGE membranes. AS, antisense; Scr, scrambled. C-E, effects of C-terminal PDZ-binding motifs in PMCA4b and GPER/GPR30 on their association. HEK 293 cells were co-transfected with HA-tagged PMCA4b or PMCA4b-⌬ETSV and FLAGtagged GPER/GPR30 or GPER/GPR30⌬SSAV as indicated. C, immunoblots of equal amounts of cell lysate from transfected samples showing expression of both heterologous proteins. D, transfected PMCA mutants from the same samples as in C were immunoprecipitated using anti-HA antibody; following SDS-PAGE, segments of the same membranes were probed for HA-PMCA4b (upper immunoblot) or FLAG-GPER/GPR30 (lower immunoblot). E, relative interaction between the expressed GPER/GPR30 and PMCA4b mutants. Histogram represents average ratios between the densitometric values of FLAG-GPER/GPR30 and HA-PMCA4b mutants. of GPER/GPR30 by G-1 treatment appears to solely inhibit PMCA activity via this mechanism, because inhibition of Src activity with PP2 completely prevented PMCA tyrosine phosphorylation and restored PMCA activity in cells treated with G-1 (Fig. 2).
Our data also provide several lines of evidence to show that GPER/GPR30 and PMCA4b physically interact via their PDZbinding motifs at the C terminus and a significant role of PSD-95 in facilitating this interaction. These include co-immunoprecipitation in primary PAECs, confocal microscopic evidence for the co-localization of GPER/GPR30 and PMCA4b, and co-immunoprecipitation in HEK 293 cells overexpressing epitope-tagged wild-type or PDZ-binding domain-truncated versions of the two proteins. We also provide functional data to demonstrate that these interactions affect both partners' activities. Interestingly, the inhibitory effect on PMCA activity due to this physical interaction appears to be constitutive and independent of GPER/GPR30 activation. Thus GPER/ GPR30 can clearly affect PMCA activity by two distinct, additive mechanisms.
Knockdown of PMCA resulted in a substantial reduction in GPER/GPR30-mediated ERK1/2 phosphorylation (Fig. 6, A and  B), indicating that PMCA-GPER/GPR30 interaction promotes GPER/GPR30 activity. It is noted, however, that the effect of PMCA knockdown on PMCA activity per se was not as pronounced (Fig. 6D). We currently do not have experimental data to explain this observation. It is unlikely for the increased cytoplasmic Ca 2ϩ as a result of PMCA knockdown to decrease G-1induced ERK1/2 phosphorylation, as MAPK functions in most cases are promoted rather than inhibited by cytoplasmic Ca 2ϩ signals. Nevertheless, it is not uncommon for trans-membrane proteins and GPCRs to be distributed at different subcellular locales, and speculatively, a small difference in the pools of PMCA participating in Ca 2ϩ extrusion and in interaction with GPER/GPR30 might explain this discrepancy. Further studies will be necessary to fully understand this observation.
Our data on the role of PDZ-binding motifs to promote interactions between GPER/GPR30 and PMCA4b are supported by a recent study demonstrating GPER/GPR30 interaction through its C-terminal PDZ-binding motif (SSAV) with the tandem PDZ domains 1 and 2 of PSD-95 in hippocampal tissue (42); this interaction promotes membrane targeting of GPER/GPR30 and increases the possibility of interactions between GPER/GPR30 and other receptors present in the hippocampus. In a different direction, GPER/GPR30 recently was nicely demonstrated to constitutively inhibit adenylyl cyclasemediated production of cAMP via PDZ domain-mediated interactions with AKAP5 (15). Interactions between GPCRs and nonreceptor transmembrane proteins through hetero-oligomeric complexes are increasingly recognized as an important regulatory input in receptor function (46). The most striking example of this type of interaction is the case of receptor activity-modifying proteins, which are single-pass trans-membrane proteins that interact with and regulate the functions of a number of GPCRs. For GPER/GPR30, RAMP3 was the first nonreceptor membrane protein shown to interact with it and to regulate its subcellular localization and cardioprotective effects (47). The role of PDZ domains in promoting homodimeric or heterodimeric interactions between GPCRs has also been shown, with the example of the endothelin receptors ET A and ET B (48). Our data demonstrate mutual functional impact between GPER/GPR30 and PMCA4b and highlight the importance of C-terminal PDZ-binding motifs in forming such complexes. Through these examples, the roles of C-terminal PDZbinding motifs in GPER/GPR30 and other GPCRs are being further demonstrated, and it is important to assess the multidirectional functional impact of such associations. This is particularly true considering that PDZ proteins such as PSD-95 contain multiple PDZ domains that could facilitate linkage among multiple partners in the same macromolecular complex.
Implications on Ca 2ϩ -dependent Activities-Our data demonstrate a novel mechanism whereby Ca 2ϩ efflux via the PMCA can be controlled: direct, constitutive interaction with a GPCR. Recently, a 10-pass transmembrane protein (partner of stromal interaction molecule 1-POST) was identified in complex with Stim1 and shuttles between the endoplasmic reticulum and plasma membrane to interact with the sarcoplasmic reticulum Ca 2ϩ -ATPase and the PMCA; similarly to GPER/GPR30, this interaction also decreases PMCA activity (33). Differently from GPER/GPR30, the interaction between POST and PMCA appears to be dynamic, although GPER/GPR30 seems to constitutively interact with PMCA4b. Clearly, inhibition of PMCA activity by GPER/GPR30 prolongs the time course of a particular Ca 2ϩ signal. When considering the potential impact of this effect on the activity of Ca 2ϩ -dependent proteins, the Ca 2ϩ sensitivity of their activation plays an important role. For example, we have previously reported EC 50 (Ca 2ϩ ) values of 422 Ϯ 23 and 130 Ϯ 5 nM for the activation of wild-type and doubly phosphorylated eNOS, respectively, in the presence of saturating CaM availability (49). If we take the example of the experiment in Fig. 1D and assume sufficient CaM to saturate eNOS in these cells, when GPER/GPR30 is activated, wild-type and doubly phosphorylated eNOS would still remain 50 and 100% active at the end of the measured time course. However, all else being equal, based on our previous data (49), the same eNOS species would only be Ͻ5% and ϳ20% active, respectively, at the same time point without GPER/GPR30's inhibitory effect on Ca 2ϩ efflux. This inhibitory effect on PMCA activity would therefore definitely enhance nitric oxide accumulation and contribute to the control of vascular tone. This is just an example of the potential impact that the inhibition of Ca 2ϩ efflux by GPER/ GPR30 could have on Ca 2ϩ -dependent activities.
Implications on GPER/GPR30 Studies-This study is the first to show a component of the Ca 2ϩ signaling machinery can affect the function of GPER/GPR30. PMCA4b has been shown to affect vascular functions via PDZ domain-mediated interactions. Specifically, PMCA4b interacts with the PDZ domain of neuronal NOS, reducing Ca 2ϩ available for the enzyme and leading to a decrease in NO production in the vascular smooth muscle cells (25). This study shows another instance in which PMCA4b potentially affects vascular function via its interaction with GPER/GPR30. The effect of PMCA4b to promote GPER/ GPR30-mediated ERK1/2 phosphorylation likely plays a role in the initiation of more long term or genomic outcomes of GPER/ GPR30 activation. In addition, GPER/GPR30 agonist G-1 has been shown to acutely trigger Ca 2ϩ signals (50,51). In many cases, these Ca 2ϩ signals were used as a parameter of receptor activation (2,3,16). Nevertheless, the mechanisms of these signals remain unknown. Arguably, the inhibitory effect of GPER/ GPR30 on Ca 2ϩ efflux can contribute to the acute total Ca 2ϩ signals and should be considered when using the total Ca 2ϩ mobilization signal as a parameter for receptor activity. Likewise, experimental paradigms that involve a change in PMCA expression or signaling might interfere with GPER/GPR30 signaling. A schematic of our findings is provided in Fig. 11.