Heterodimerization of the prostaglandin E2 receptor EP2 and the calcitonin receptor CTR

G protein-coupled receptors (GPCRs) have been found to form heterodimers and modulate or fine-tune the functions of GPCRs. However, the involvement of GPCR heterodimerization and its functional consequences in gonadal tissues, including granulosa cells, have been poorly investigated, mainly due to the lack of efficient method for identification of novel GPCR heterodimers. In this paper, we identified a novel GPCR heterodimer between prostaglandin E2 (PGE2) receptor 2 (EP2) and calcitonin (CT) receptor (CTR). High-resolution liquid chromatography (LC)-tandem mass spectrometry (MS/MS) of protease-digested EP2-coimmunoprecipitates detected protein fragments of CTR in an ovarian granulosa cell line, OV3121. Western blotting of EP2- and CTR-coimmunoprecipitates detected a specific band for EP2-CTR heterodimer. Specific heterodimerization between EP2 and CTR was also observed by fluorescence resonance energy transfer analysis in HEK293MSR cells expressing cyan- and yellow-fluorescent protein-fused EP2 and CTR, respectively. Collectively, these results provided evidence for heterodimerization between EP2 and CTR. Moreover, Ca2+ mobilization by CT was approximately 40% less potent in HEK293MSR cells expressing an EP2-CTR heterodimer, whereas cAMP production by EP2 or CT was not significantly altered compared with cells expressing EP2- or CTR alone. These functional analyses verified that CTR-mediated Ca2+ mobilization is specifically decreased via heterodimerization with EP2. Altogether, the present study suggests that a novel GPCR heterodimer, EP2-CTR, is involved in some functional regulation, and paves the way for investigation of novel biological roles of CTR and EP2 in various tissues.


Introduction
Most receptors of neurotransmitters, neuropeptides, and hormones are G protein-coupled receptors (GPCRs), and their pharmacological properties are targets of drug development [1]. GPCRs are not only present as monomers and homodimers but form heterodimers with other GPCRs [2,3]. GPCR heterodimerization has been found to alter ligand binding affinity, signal transduction, and desensitization of GPCRs [2, 3] and to participate in pathological processes [4,5], pharmacological profiles [6][7][8] and species-specific biological events [9] in vitro and in vivo. Moreover, several orphan GPCRs have been shown to serve as protomers (i.e. partner GPCRs) in GPCR heterodimers and to cause specific functional a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 cells were maintained at 37˚C under 5% CO 2 as previously described [11]. After being washed with PBS, cells were collected and centrifuged. The resultant pellets were stored at -80˚C until use.

RT-PCR analysis
Total RNA was extracted and purified using Sepasol-RNA I Super G (Nacalai Tesque, Kyoto, Japan), treated with TURBO DNase I (Ambion, Austin, USA), and reverse-transcribed to the cDNA using oligo(dT) 20 and Superscript III (Thermo Fisher Scientific, Waltham, USA). PCR was performed as previously described [25] using the primers listed in S1 Table. Membrane preparation and Western blotting Cell pellets were homogenized using a Polytron homogenizer in homogenizing buffer (HB, 10 mM Tris-HCl, 1 mM EDTA, 10 mM MgCl 2 , 11% sucrose, pH 8.0) containing 1 × complete protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). After sonication, the homogenate was incubated with 250 μg/ml RNase A (Qiagen, Valencia, USA) on ice for 30 min, releasing ribosomal proteins from the membrane fraction, as described [26]. The homogenate was centrifuged twice at 8,000 × g at 4˚C for 10 min each, and then, the supernatant was further centrifuged at 100,000 × g at 4˚C for 1 hr. The precipitate was washed, resuspended, sonicated in HB, and solubilized by adding an equal volume of 2 × radioimmunoprecipitation assay buffer (RIPA, 50 mM Tris-HCl, 300 mM NaCl, 2% NP-40, 2% sodium deoxycholate (SDC), 0.2% sodium dodecyl sulfate (SDS)) containing 2 × protease inhibitor cocktail and rotating at 4˚C for 1 hr. Aliquots of 15-μg membrane proteins were electrophoresed on polyacrylamide gels and transferred to nitrocellulose membranes. The blot was incubated with 1 μg of each rabbit monoclonal anti-EP2 (ab167171) or anti-CTR (ab11042) antibody (Abcam, Tokyo, Japan) as a primary antibody. The normal rabbit IgG (sc-2027, Santa Cruz Biotechnology, Santa Cruz, USA) and anti-Na + /K + ATPase α1 antibody (ab7671, Abcam) were used as a negative and a loading control of membrane proteins, respetively. After the incubation with horseradish peroxidase-linked secondary antibody (NA9340V, GE healthcare, Buckinghamshire, UK), signals were detected using an ECL system (GE healthcare) as previously described [11].

Co-IP, SDS-PAGE, and in-gel protein digestion
Aliquots of 800-μg membrane proteins were precleared with protein G sepharose (GE healthcare) at 4˚C for 1 hr with rotation and centrifuged at 10,000 × g for 1 min. The supernatant was incubated overnight at 4˚C with 6-μg anti-EP2 antibody and protein G sepharose. The mixture was centrifuged at 4˚C at 10,000 × g for 1 min, and the beads were washed seven times with 1 × RIPA buffer and once with 0.1 × RIPA buffer. To denature proteins, the beads were incubated twice with 1.5 × sample buffer (2.5% SDS, 12.5% glycerol, 0.075% bromophenol blue, 2.5% β-mercaptoethanol) at 37˚C for 5 min each. Collected samples were subjected to SDS-PAGE and subsequent fragmentation. Another Co-IP followed by Western blotting was performed to verify the interaction between EP2 and CTR. Aliquots of 30-μg membrane proteins were co-immunoprecipitated with 1 μg of each anti-EP2 or anti-CTR antibody, and onetenth volume of collected samples (and 3 μg of input sample without Co-IP) were used for Western blotting using anti-CTR or anti-EP2 antibody, respectively. The sample prepared without Co-IP was used as a negative control.
Following SDS-PAGE and staining with Coomassie brilliant blue, the gel fragment corresponding to proteins of molecular mass 25-150 kDa was excised, cut into 1-mm cubes, and dehydrated with 100% acetonitrile (ACN) for 10 min. After drying in vacuo, the gel pieces were swollen in a minimum volume of digestion buffer (25 mM NH 4 HCO 3 , 0.1% SDC, 20 ng/ μl trypsin/Lys-C mix) for 15 min, covered with digestion buffer, and incubated overnight at 37˚C. Gel pieces were washed with MilliQ water and combined with the original supernatant. Proteolytic fragments were extracted three times from gel pieces by mixing in 50% ACN for 1 hr each. All extracts were treated with 0.5% trifluoroacetic acid, followed by centrifugation at 15,700 × g for 2 min. After drying the supernatants in vacuo, the residues were dissolved in MilliQ water and purified with YM-3 Microcon filters (Millipore, Bedford, USA).

LC-MS/MS analysis
The purified protein fragments were subjected to reverse-phase nano-liquid chromatography using an EASY-nLC 1000 system, followed by tandem mass spectrometry (LC-MS/MS) analysis using Orbitrap Elite (Thermo Fisher Scientific). The protein fragment solutions were first loaded onto a 75 μm × 2 cm C18 trap column (Acclaim PepMap100, 3 μm, 100Å) at 300 nl/ min and separated on a 50 μm × 15 cm C18 analytical column (Acclaim PepMap RSLC, 2 μm, 100Å) using a linear gradient of 15-30% ACN with 0.1% formic acid over 48 min. The LC eluent was introduced into the mass spectrometer by positive-mode nanospray ionization. Fullscan mass spectra over a range of 350-2,000 m/z were measured with Orbitrap at a resolution of 30,000, and the 15 most intense multivalent precursor ions were selected for data-dependent scanning. The MS/MS spectra were acquired with Velos Pro under conditions of 1 m/z isolation width, 35% normalized collision energy, and 10 ms maximum ion accumulation time. Data were analyzed with Xcalibur software 2.2.

Protein identification
Raw files were processed using Proteome Discoverer 1.1 software incorporating the SEQUEST search algorithm. All protein sequences of Mus musculus were downloaded from RefSeq (December 2013) [27] and used for protein identification. Search criteria included precursor mass tolerance, 10 ppm; fragment mass tolerance, 1.2 Da; maximum missed cleavage site, 2; and dynamic modification, oxidation of methionine.

Plasmid construction and transfection
Open reading frames (ORFs) of EP2, CTR, and oxytocin receptor (OXTR) were amplified from the OV3121 by RT-PCR. The ORF of EP2 was subcloned in frame into 5'-end of the pcDNA4/V5 (Thermo Fisher Scientific) and pAmCyan1-N1 cyan fluorescent protein (CFP) vector (Clontech, Kyoto, Japan) at the EcoRI/XhoI and NheI/XhoI, sites, respectively. The CTR was subcloned into the pcDNA4/V5 and pZsYellow1-N1 yellow fluorescent protein (YFP) vector (Clontech) at the NheI/XhoI sites. The OXTR was subcloned into the pAmCyan1-N1 CFP vector at the EcoRI/NheI sites. Each construct was transiently transfected into HEK293MSR cells using Lipofectamine 2000 (Thermo Fisher Scientific), according to the manufacturer's instructions.

FRET
FRET was performed using a confocal laser microscopy as previously described with slight modification [11,12]. In brief, 1 μg of each expression vector (EP2-fused CFP (EP2-CFP), OXTR-fused CFP (OXTR-CFP), or CTR-fused YFP (CTR-YFP)) was transfected to HEK29 3MSR cells on the 35-mm glass-bottom dish. CFP, YFP, and FRET signals were visualized at the following day using confocal laser microscopy, Fluoview FV1000 (Olympus, Tokyo, Japan). CFP and YFP were excited with the 458-and 515-nm lines of an argon laser, respectively. The emitted fluorescence was collected at 475-500 nm for CFP and 530-630 nm for YFP and FRET signals. Dose response effects of EP2-expression levels on FRET were investigated as follows. 1 μg of CTR-YFP expression vector was transfected with 0, 0.01, 0.1, 0.5, and 1 μg of EP2-CFP expression vector to 1×10 5 HEK293MSR cells on the 35-mm glass-bottom dish. After 24 hr incubation, FRET signal was observed. The signal intensity was analyzed using Fiji software and data was represented as % maximal intensity.

Second messenger detection
Intracellular Ca 2+ mobilization and cAMP production were measured as previously described [11,12]. In brief, 1×10 6 HEK293MSR cells were spread on 60-mm dish at one day before transfection. 10 μg of pcDNA-based vector (5 μg of one GPCR-V5 and 5 μg of another) or mock vector were transfected into the cells. As for the evaluation of dose response effects of EP2-expressioon levels on Ca 2+ mobilization, 2.5 μg of CTR-V5 expression vector was transfected with 0, 0.5, 1, 2, and 2.5 μg of EP2-V5 expression vector to 5×10 5 cells on the 35-mm dish. After the 24 hr incubation, 5×10 4 cells per well were re-spread to the 96-well plate for the following assay. Prostaglandin E2 (PGE2, sc-201225, Santa Cruz Biotechnology) and salmon CT (sc-201167, Santa Cruz Biotechnology) were used for the assays. Real-time fluorescent measurement for Ca 2+ mobilization using a Ca5 kit (Molecular Devices, Sunnyvale, USA) and end point observation of cAMP production using CatchPoint Cyclic AMP Assay Kit (Molecular Devices) were performed on FlexStation II. Results are shown as means ± standard error of the mean (SEM) of three independent experiments. Emax values between monomer-and heterodimer-expressing cells were analyzed by Student's t test. P < 0.05 was defined as statistically significant.
Detection of known GPCR heterodimer heterologously expressed in HEK293MSR cells by the Co-IP-based LC-MS/MS method prompted us to identify a novel endogenous GPCR heterodimers in native tissues and cells using this experimental method. First, we investigated the GPCR expression in murine ovarian granulosa cell line, OV3121. RT-PCR analysis showed that PGE2 receptor 2, EP2 mRNA (Ptger2) was expressed in OV3121 cells (Fig 1A). Although EP2 was reported to be expressed in native granulosa cells of preovulatory follicles [15][16][17][18], whether EP2 forms GPCR heterodimer with other GPCRs and its functional consequences are not known. Subsequently, we utilized the Co-IP-based LC-MS/MS to identify GPCR protomers that heterodimerize with EP2 in OV3121 cells.
To identify endogenous GPCR heterodimers, we optimized three steps of the Co-IP-based LC-MS/MS analysis. First, in purifying membrane proteins, cell lysates were treated with RNase A to remove impurities such as ribosomal proteins and to enhance the solubility of membrane proteins [26]. Second, because rabbit monoclonal antibodies have shown extremely high affinity and specificity [28][29][30], in the Co-IP step, a rabbit monoclonal anti-EP2 antibody was used. Western blotting with this antibody detected specific band corresponding to EP2, whereas no band was observed using the normal rabbit IgG (Fig 1B, top). These results confirmed the expression of EP2 in OV3121 cells and the specificity of this antibody. Additionally, the expression of Na+/K+ ATPase α1 was indicated as a loading control (Fig 1B, bottom). Third, EP2-coimmunoprecipitates were serially digested with SDC to improve the accessibility of proteases to hydrophobic GPCR [31][32][33]. The resultant peptide fragments were subjected to nano-LC-Orbitrap MS/MS analysis, followed by database-referencing. These EP2-immunoprecipitates were found to contain various proteins, including some transporters, enzymes, non-GPCR receptors, and G proteins (S2 Table) as observed in proteomic analysis of other membrane proteins [31]. Notably, EP2-coimmunoprecipitates were found to contain fragments of one GPCR, CTR (S3 Fig), suggesting that EP2 and CTR form a heterodimer. The mRNA expression of CTR (Calcr) in OV3121 cells was confirmed by RT-PCR (Fig 1A). Western blotting also detected the bands of monomeric/multimeric CTR and verified the specificity of the anti-CTR antibody (Fig 1B). Thus, we examined the interaction between EP2 and CTR by another Co-IP-Western blotting using anti-EP2 and anti-CTR antibodies. The specific band of EP2-CTR heterodimer was detected in EP2-coimmunoprecipitates with anti-CTR antibody at 90 kDa (Fig 1C, left). Likewise, the identical band was also detected in CTR-coimmunoprecipitates with anti-EP2 antibody (Fig 1C, right), which was not observed in the corresponding amount of sample prepared without Co-IP (Fig 1C, IP-). This is consistent with the previous findings that non-obligatory GPCR heterodimers are not dissociated by SDS [34,35]. These results indicate that EP2 interacts with CTR in OV3121 cells.
To further validate the formation of EP2-CTR heterodimer, FRET analysis was performed. EP2-CFP, CTR-YFP, or both constructs were transfected into HEK293MSR cells. Both CFPand YFP-derived signals were observed when EP2-CFP and CTR-YFP were expressed alone or together, confirming that both GPCRs were sufficiently expressed on the plasma membranes (Fig 2). Furthermore, prominent FRET signals were detected exclusively in cells coexpressing EP2-CFP and CTR-YFP, but not in cells expressing either alone (Fig 2A). In contrast, no FRET signals were observed in OXTR-CFP-and CTR-YFP-coexpressing cells (Fig 2B), showing that the FRET signals were specific for EP2-CTR (Fig 2A). In addition, FRET signals were dependent on the ratio of EP2-CFP expression to CTR-YFP expression (S4 Fig). Taken together, these findings indicate that EP2 forms specific, constitutive heterodimers with CTR.
Co-IP-based LC-MS/MS analysis has been widely utilized to assess comprehensive interactions of proteins [23,24,[36][37][38][39][40]. To our knowledge, however, this report is the first to show the identification of a novel GPCR heterodimer by Co-IP-based LC-MS/MS. The ability of this method to detect GPCR protomers is particularly important in improving the specificity and sensitivity of detecting target proteins. In this context, utilizing high-specificity-rabbit monoclonal antibody and stringent detergents (SDS) in the Co-IP step was effective in detecting the specific interaction between EP2 and CTR. Moreover, serial protease-digestion with SDC during fragmentation of coimmunoprecipitates likely enhanced the sensitivity of detection. Multiepitope affinity purification in a buffer containing detergent of mild to intermediate stringency led to the identification of multiple non-GPCR-transmembrane proteins that form macromolecular complexes with native GPCR [24], suggesting that mixture of several antibodies in a mild stringent detergent may enhance the detection of GPCR heterodimers. Additional enrichment of plasma membrane fractions by density-gradient centrifugation may also remove impurities and enhance the specificity. However, it is noteworthy that a novel GPCR heterodimer, namely, EP2-CTR, was detected by our relatively simple procedure despite the presence of various non-GPCRs and impurities. The results therefore indicate the potential of this Co-IP-based LC-MS/MS method to identify other novel GPCR heterodimers in other cells or even in native tissues.
Subsequently, we evaluated the functional propensities of EP2-CTR in the second messenger production. EP2 is coupled with only Gs protein and elevates the cAMP production upon the PGE2 stimulation [15], whereas CTR are coupled with Gs and Gq proteins and induces both cAMP production and intracellular Ca 2+ mobilization in response to CT [41]. We thus examined the effect of the heterodimerization on the cAMP production using HEK293MSR cells. Neither CT nor PGE2 stimulation exhibited any significant alteration of cAMP production in heterodimer-expressing cells compared to monomer-expressing cells (Fig 3A and 3B). On the other hand in the intracellular Ca 2+ mobilization by CT, cells expressing EP2-CTR heterodimer showed significant suppression of intracellular Ca 2+ mobilization (Emax = 62.4% ± 5.2%, P < 0.01, EC 50 = 1.5 x 10 −7 M) in a CT dose-dependent fashion, compared with those expressing CTR alone (Emax = 100.0% ± 2.5%, EC 50 = 1.0 x 10 −7 M), (Fig 3C). Furthermore, the suppression of Ca 2+ mobilization was found to be correlated with the amounts of EP2-expression vector (Fig 3D). Taken together, these results, combined with the Co-IP (Fig 1 and S3  Fig) and FRET data (Fig 2 and S4 Fig), strongly indicate that EP2-CTR modulates intracellular CTR-induced Ca 2+ mobilization in response to CT (Figs 3 and 4), providing evidence for EP2-CTR heterodimerization.
Of particular interest is that heterodimerization of EP2 with CTR exclusively resulted in suppression of intracellular Ca 2+ mobilization by CTR (Figs 3 and 4), whereas CTR activates not only intracellular calcium mobilization but also cAMP production in response to CT [41]. These results are in good agreement with the findings that GPCRs adopt multiple active conformations which are specific to the respective G proteins and that these active conformations are not interconvertible [3, [42][43][44][45][46][47][48][49]. Furthermore, several protomers of GPCR heterodimers were shown to affect the adoption of a particular active conformation of a partner GPCR as an endogenous allosteric modulator [3, 11,12,[42][43][44][45][46][47][48][49]. In combination, the present results suggest that EP2 specifically suppresses the adoption of the active conformation of CTR specific to Gqcoupling via EP2-CTR heterodimerization.
In conclusion, we have identified a novel heterodimer, EP2-CTR using a Co-IP-based LC-MS/MS method and demonstrated the regulation of second messenger production in cultured cells. The present study paves the way for investigation of molecular mechanisms underlying a wide range of endocrine, neuroendocrine, and nervous systems involving GPCR heterodimers.