The mechanism underlying the crosstalk between multiple G protein–coupled receptors remains poorly understood. We previously reported that prostaglandin E receptor EP1 facilitates dopamine D1 receptor signaling in striatal slices and promotes behavioral responses induced by D1 receptor agonists. Here, using human embryonic kidney (HEK)-293T cells expressing D1 and EP1, we have analyzed the mechanism underlying EP1-mediated facilitation of D1 receptor signaling. Fluorescent immunostaining showed that EP1 and D1 receptors are partly colocalized in the cells, and coprecipitation experiments revealed a molecular complex of EP1 and D1 receptors. Treatment of the cells with 17S,17,20-dimethyl-2,5-ethano-6-oxo-PGE1 (ONO-DI-004), an EP1-selective agonist, enhanced cAMP production induced by D1 agonists (±)-6-chloro-2,3,4,5-tetrahydro-1-phenyl-1H-3-benzazepine hydrobromide (SKF-81297) and 6-chloro-2,3,4,5-tetrahydro-1-(3-methylphenyl)-3-(2-propenyl)-1H-3-benzazepine-7,8-diol hydrobromide (SKF-83822). Although this facilitative effect of EP1 stimulation was not affected by pharmacologic blockade of EP1-induced Ca2+ increase, it was blocked by overexpression of Gtα as a Gβγ scavenger. Consistently, depletion of adenylyl cyclase (AC) 7, a Gβγ-sensitive AC isoform, abolished the facilitative action of EP1 on D1-induced cAMP production. Notably, neither Gtα overexpression nor AC7 depletion affected cAMP production induced by D1 stimulation alone. In contrast, depletion of AC6, another AC isoform, reduced cAMP production induced by D1 stimulation alone, but spared its facilitation by EP1 stimulation. Collectively, these data suggest that, through complex formation with D1, EP1 signaling directs the D1 receptor through Gβγ to be coupled to AC7, an AC isoform distinct from those used by the D1 receptor alone, in HEK-293T cells.
G protein–coupled receptors (GPCRs) are among the primary targets for therapeutics in clinical use (Prinster et al., 2005; Milligan, 2009; Siehler and Milligan, 2011). They often form a heteromeric complex, but the mechanism underlying the crosstalk between multiple GPCRs remains poorly understood. Classically, it was assumed that each type of GPCR should be invariably coupled to a specific G protein and its effectors (Milligan, 2009). Given this assumption, the crosstalk of multiple GPCRs has been viewed as the sum of their signaling pathways (Panetta and Greenwood, 2008). The discovery of GPCR heteromers and the understanding of their actions and functions have greatly challenged this classic view of GPCR signaling (Prinster et al., 2005; Milligan, 2009; Fuxe et al., 2010). The concept of GPCR heteromers was originally substantiated by the discovery of dimer formation of class C GPCRs, such as GABAB receptors (Jones et al., 1998; Kaupmann et al., 1998; White et al., 1998; Kuner et al., 1999) and taste receptors (Nelson et al., 2001; Zhao et al., 2003; Temussi, 2009), which is required for proper expression and function of these receptors. Later experiments have suggested that class A GPCRs, many of which are neurotransmitter receptors, can also form heteromers in the heterologous system and in native tissue (Pin et al., 2007). In many cases, one protomer of these GPCR heteromers can modulate the signaling efficacy of the other protomer (Siehler and Milligan, 2011). For example, the adenosine A2A receptor forms a heteromeric complex with the dopamine D2 receptor, and the activation of the former receptor reduces the ligand binding of the latter receptor (Ferré et al., 1991). In another example, the serotonin 2A receptor forms a complex with the metabotropic glutamate receptor 2 through a specific transmembrane domain (González-Maeso et al., 2008). The activation of each protomer of this heteromer negatively regulates G protein–signaling of the other protomer (Fribourg et al., 2011).
Recent studies of newly identified heteromers of dopamine receptors have expanded a role for the GPCR heteromer in activating distinct signaling pathways from each protomer (So et al., 2005; Rashid et al., 2007; Kern et al., 2012). Dopamine receptors are divided into two families: D1-like receptors and D2-like receptors, which are primarily coupled to Gs-mediated cAMP increase and Gi-mediated cAMP decrease, respectively (Missale et al., 1998). However, activation of the D1 receptor in the heteromer with the D2 receptor can evoke a Gq-mediated Ca2+ increase only if the D2 receptor is simultaneously activated (So et al., 2005; Rashid et al., 2007). In another example, the D2 receptor forms a heteromer with the ghrelin receptor, growth hormone secretagogue receptor 1a, in the heterologous system and in central nervous system neurons (Kern et al., 2012). It was reported that stimulation of the D2 receptor activates a Gβγ subunit–dependent rise in intracellular Ca2+ in neuroblastoma cells only when growth hormone secretagogue receptor 1a is coexpressed (Kern et al., 2012). These examples illustrate a novel mode of signaling crosstalk in which the coupling of a certain GPCR to a downstream effector is altered by the presence of another GPCR. However, the molecular mechanism underlying such crosstalk remains elusive.
It is known that dopamine receptor signaling is modulated by several neurotransmitters and neuromodulators (Svenningsson et al., 2004; Nishi et al., 2011). Prostaglandin E2 (PGE2), a bioactive lipid derived from arachidonic acid, is one of these substances (Furuyashiki and Narumiya, 2011) and exerts its functions through one of four GPCRs named EP1, EP2, EP3, and EP4 (Hirata and Narumiya, 2011). We previously reported that EP1 and D1 receptors are coexpressed in striatal projection neurons, and that PGE2-EP1 signaling augments D1-induced phosphorylation of dopamine- and cAMP-regulated neuronal phosphoprotein-32 at its Thr34 residue in striatal slices and promotes hyperlocomotion induced by D1 agonists (Kitaoka et al., 2007). However, the mechanism underlying this EP1 action remains unknown.
In this study, we analyzed the mechanism underlying the facilitative effect of EP1 on D1 receptor signaling in human embryonic kidney (HEK)-293T cells expressing EP1 and D1. We have found that EP1 and D1 receptors form a complex in these cells, and that EP1 activation facilitates D1-induced cAMP production through Gβγ subunits and adenylyl cyclase (AC) 7, a Gβγ-sensitive AC isoform (Sadana and Dessauer, 2009). Our findings also suggest that distinct AC isoforms are employed for cAMP production by D1 receptors alone, and for cAMP production induced by D1 receptors in the D1-EP1 complex.
Materials and Methods
Plasmids for mammalian expression of the αl subunit of transducin (Gtα), the β1 subunit of G protein (Gβ), the γ2 subunit of G protein (Gγ, transcript variant 1) under the cytomegalovirus promoter were obtained from OriGene Technologies (Rockville, MD). A plasmid for mammalian expression of the mouse EP2 receptor under the cytomegalovirus promoter was obtained by inserting the open reading frame of EP2 into the EcoRI site of pCMS–enhanced green fluorescent protein (EGFP) (Clontech, Mountain View, CA) (Matsuoka et al., 2003).
A plasmid for mammalian expression of mouse dopamine D1 receptor fused at its N terminus to the signal sequence (SS) derived from influenza hemagglutinin (Guan et al., 1992) and the FLAG peptide was generated as follows. First, to remove the open reading frame of EGFP in pEGFP-C1 (Clontech), pEGFP-C1 was digested with AgeI and BsrGI. The remaining portion of pEGFP-C1 was fused with the oligo DNA containing the AgeI site, the Kozak sequence, SS, the FLAG peptide, and the BsrGI site (5′–ACC GGT CCA CCA TGA AGA CGA TCA TCG CCC TGA GCT ACA TCT TCT GCC TGG TAT TCG CCG ACT ACA AGG ACG ATG ATG ACG CCT GTA CA–3′). The resultant plasmid is named pSS-FLAG-C1. Then, the open reading frame of the D1 receptor from the second amino acid attached with BsrGI and PstI at its N and C termini, respectively, was generated from pBluescript KS(+) containing the open reading frame of mouse D1 receptor (Riken, Saitama, Japan) by conventional polymerase chain reaction (PCR) using the two following primers. The forward primer contains the BsrGI site and the N-terminal portion of D1 receptor (5′–TAA TGT ACA GCT CCT AAC ACT TCT ACC ATG G–3′). The reverse primer contains the C-terminal portion of D1 receptor and the PstI site (5′–CGT TTC TGC AGA ACC CAA TAT TCA GGT TGA ATG CTG–3′). Finally, this PCR fragment was digested with BsrGI and PstI and ligated to the fragment of pSS-FLAG-C1 after BsrGI and PstI digestion.
A plasmid for mammalian expression of the human β2-adrenergic receptor (ADRB2) fused at its N terminus to SS and the FLAG peptide was generated as follows. First, we obtained cDNA encoding ADRB2 attached with BsrGI and HindIII sites at its N and C termini, respectively, from total RNA of HEK-293 cells by reverse transcription (RT)-PCR using the two following primers. The forward primer contains the BsrGI site and the N-terminal portion of ADRB2 sequence (5′–ATC TAG TGT ACA GGG CAA CCC GGG AAC GGCA–3′). The reverse primer contains the C-terminal portion of ADRB2 sequence and the HindIII site (5′–AGT ATT AAG CTT TTA CAG CAG TGA GTC ATT TGT ACT ACA–3′). This PCR fragment was digested with BsrGI and HindIII and ligated to the fragment of pSS-FLAG-C1 after BsrGI and HindIII digestion.
A plasmid for mammalian expression of mouse EP1 receptor fused at its N terminus to SS and the hemagglutinin (HA) peptide was generated as follows. First, pEGFP-C1 was digested with NheI and AgeI, and the resultant fragment was fused to the oligo DNA containing the NheI site, the Kozak sequence, the HA peptide sequence, and the AgeI site (5′–GCT AGC CCA CCA TGA AGA CGA TCA TCG CCC TGA GCT ACA TCT TCT GCC TGG TAT TCG TAT CCT TAC GAC GTT CCG GAC TAC GCA ACC GGT–3′). This plasmid is named pSS-HA-EGFP-C1. Then, the open reading frame of EP1 from the second amino acid attached with AgeI and KpnI at its N and C termini, respectively, was generated from the plasmid for EP1 expression (Watabe et al., 1993) by conventional PCR using the two following primers. The forward primer contains the AgeI site and the N-terminal portion of the EP1 sequence (5′–ATC GAA CCG GTA GCC CCT GCG GGC TTA A–3′). The reverse primer contains the C-terminal portion of EP1 sequence and the KpnI site (5′–GCT CAC CAT GGA GGC ACA GTC GAG GCT G–3′). Finally, this PCR fragment was digested with AgeI and Acc65I and ligated to the fragment of pSS-HA-EGFP-C1 after AgeI and Acc65I digestion.
The sequences of the open reading frames of the resultant plasmids were confirmed by conventional DNA sequencing using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Carlsbad, CA).
Cell Culture and Transfection.
HEK-293T cells (American Type Culture Collection, Manassas, VA) were maintained on Type I collagen-coated dishes (Iwaki Glass, Tokyo, Japan) in Dulbecco’s modified Eagle’s medium (Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (PAA, Etobicoke, ON, Canada) in a humidified atmosphere with 5% CO2 at 37°C. For cAMP assay, cells were plated on collagen I–coated 24-well dishes. For immunofluorescence, cells were plated on round coverslips of 12-mm diameter (Thermo Fisher Scientific, Waltham, MA). Plasmids were transfected to HEK-293T cells with Effectene (Qiagen, Alameda, CA), except for immunoprecipitation (Fig. 3B), calcium measurement (Fig. 4A), radioligand binding assay (Supplemental Fig. 3), and subcellular fractionation (Supplemental Fig. 4), in which FuGENE HD was used instead (Promega, Madison, WI). After DNA transfection, cells were maintained for 24–48 hours before each experiment. Small interfering RNAs (siRNAs) were transfected to HEK-293T cells for 48 hours with Lipofectamine RNAiMAX (Life Technologies). A prevalidated mixture of multiple siRNAs targeting human AC5, AC6, or AC7 (MISSION esiRNA) was obtained from Sigma-Aldrich (St. Louis, MO). Stealth RNAi siRNA negative control Med GC was used as a negative control for siRNA experiments (Life Technologies). The specificity and efficacy of the knockdown using these siRNAs in our experimental conditions were validated using quantitative RT-PCR (see Results).
Drugs used in this study were obtained from the following sources: Forskolin (Tocris Bioscience, Bristol, UK); (−)-isoproterenol and (±)-6-chloro-2,3,4,5-tetrahydro-1-phenyl-1H-3-benzazepine hydrobromide (SKF-81297) (Sigma-Aldrich); 6-chloro-2,3,4,5-tetrahydro-1-(3-methylphenyl)-3-(2-propenyl)-1H-3-benzazepine-7,8-diol hydrobromide (SKF-83822) (a kind gift from the National Institute of Mental Health, Bethesda, MD); 17S,17,20-dimethyl-2,5-ethano-6-oxo-PGE1 (ONO-DI-004) and 16s-9-deoxy-9β-chloro-15-deoxy-16-hydroxy-17,17-propano-19,20-didehydroprostaglandin E2 (ONO-AE1-259) (kind gifts from ONO Pharmaceuticals, Osaka, Japan). After cells were washed in Leibovitz's L-15 medium (Life Technologies) without serum, prewarmed L-15 medium containing appropriate agonists or their vehicles were applied to the cells at 37°C in the air for 5 minutes. To block intracellular Ca2+ increases (Fig. 4), cells were incubated in L-15 medium containing 10 μM 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid/acetoxymethyl ester (BAPTA-AM) (Life Technologies) or its vehicle (dimethylsulfoxide) for 30 minutes at 37°C before stimulation. As an alternative method for Ca2+ depletion, the cells were incubated in L-15 medium containing 1 mM EGTA and 100 nM thapsigargin (Life Technologies) for 30 minutes at 37°C. To block Gi signaling (Fig. 6), HEK-293T cells expressing EP1 and D1 receptors were incubated in serum-free Dulbecco’s modified Eagle’s medium with pertussis toxin (0.2 µg/ml; List Biologic Laboratories, Campbell, CA) overnight in a humidified incubator at 37°C with 5% CO2.
After L-15 medium containing agonists or vehicle was removed, cells were immediately lysed with 0.1 M HCl for 20 minutes at room temperature. cAMP levels in supernatants were measured using the cAMP EIA Kit (Cayman Chemical, Ann Arbor, MI) according to the manufacturer’s protocol. Supernatants were stored on ice if assayed on the same day, or kept at −80°C for longer storage. cAMP values were normalized to the maximal level of cAMP induced by forskolin (10 μM), except in the experiments with AC knockdown (Figs. 7 and 8). In these experiments, because knockdown of respective AC isoforms affected forskolin-induced cAMP responses to different extents, either the absolute concentration of cAMP in supernatants (Fig. 7) or those normalized to the cAMP level induced by 500 nM SKF-83822 (Fig. 8) were used for data analyses.
Fluorescent immunostaining was performed as previously described (Matsuoka et al., 2003). Briefly, cells were fixed with Dulbecco’s modified phosphate-buffered saline (D-PBS) (Nissui Pharmaceuticals, Tokyo, Japan) containing 4% paraformaldehyde (Polysciences, Warrington, PA) for 30 minutes at room temperature. After permeabilization in blocking buffer (D-PBS containing 2% goat serum, 1% bovine serum albumin, 0.01% Triton X-100, and 0.05% Tween-20) for 60 minutes at room temperature, the cells were incubated with both rat anti-HA peptide antibody (1:1,000 dilution; 3F10; Roche Diagnostics, Indianapolis, IN) and mouse anti-FLAG M2 antibody (1:1,000 dilution; Sigma-Aldrich) in blocking buffer. After washing in D-PBS containing 1% Tween-20, the cells were incubated with appropriate secondary antibodies conjugated with Alexa488 or Alexa555 (Life Technologies) at 1:200 dilution in the blocking buffer for 2 hours at room temperature. Finally, cover slips were washed in D-PBS and mounted with Prolong Gold Antifade Reagent (Life Technologies). Fluorescent images were acquired by TCS-SP5 confocal microscopy (Leica Microsystems, Nussloch, Germany) and processed using Image J software (NIH, Betheseda, MD) or Photoshop (Adobe, San Jose, CA) for illustrative purposes only.
At 24–48 hours after transfection, cells were resuspended and solubilized in the immunoprecipitation buffer (50 mM Tris-HCl, pH 6.8, 150 mM NaCl, 1 mM EDTA, and 1% fos-choline-14, N-tetradecylphosphocholine) with protease inhibitors (Complete Mini, EDTA-free; Roche Diagnostics) for 20 minutes at 4°C, followed by brief sonication on ice. Following centrifugation at 12,000g for 10 minutes at 4°C, FLAG-tagged D1 receptors (Fig. 3C) were precipitated with anti-FLAG M2 antibody-conjugated beads (Sigma-Aldrich) according to the manufacturer’s protocol. Then, precipitated proteins were eluted in 2 × SDS sample buffer (125 mM Tris HCl, pH 6.8, 20% glycerol, 4% SDS, 0.02% bromophenol blue, 50 mM dithiothreitol). For immunoprecipitation of EP1 receptors (Fig. 3B), cell lysates were precleared at 4°C for 10 minutes using Protein A/G beads (Cytosignal, Irvine, CA) in a 50% slurry. Beads were pelleted, and the protein concentration of the precleared lysate was determined by the BCA protein assay method (Thermo Fisher Scientific, Rockford, IL). The precleared cell lysate of 1 ml containing 500 µg protein was incubated overnight with rabbit anti-EP1 antibody (0.5 mg/ml; Cayman Chemical). The immune complex with EP1 antibodies was precipitated by the Protein A/G Magnetic IP Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Precipitated proteins were eluted in the sample buffer provided by the kit with the addition of 50 mM dithiothreitol and incubated at 4°C for 40 minutes, then at room temperature for 10 minutes prior to Western blotting.
Western blotting was performed as previously described (Kitaoka et al., 2007). Briefly, protein samples dissolved in the SDS sample buffer were subjected to SDS-PAGE with a 10% polyacrylamide gel (Atto, Tokyo, Japan), followed by semi-dry transfer (Atto) onto a 0.45-μm polyvinylidene difluoride membrane (Millipore, Milford, MA). After blocking nonspecific binding with Tris-buffered saline (TBS; 50 mM Tris-Cl, pH 7.5, 150 mM NaCl) containing 3% skim milk (Difco, Becton Dickinson, Franklin Lakes, NJ), the membrane was incubated overnight at 4°C with either rabbit anti-EP1 antibody (1:100 dilution; Cayman Chemical), rat anti-D1 antibody (1:1,000 dilution; D2944; Sigma-Aldrich), or mouse anti–glyceraldehyde 3-phosphate dehydrogenase antibody (1:3,000 dilution; clone 6C5; Ambion, Carlsbad, CA). After several washes in TBS, the membrane was incubated with horseradish peroxidase–conjugated secondary antibody for IgG of appropriate species (1:5,000 dilution; GE Healthcare Biosciences, Pittsburgh, PA) in TBS containing 3% skim milk for 1 hour at room temperature. After several washes in TBS, the membrane was subjected to detection with ECL Plus or ECL Prime (GE Healthcare Biosciences).
Intracellular calcium concentration was measured with the Fluo-4-AM Direct Calcium Assay Kit (Life Technologies) according to the manufacturer’s protocol. Briefly, cells were incubated in the calcium assay buffer (Hanks’ balanced salt solution containing 0.8 mM MgCl2, 1.8 mM CaCl2, 0.1% bovine serum albumin, 2.5 mM probenecid, and Fluo-4-AM) for 30 minutes at 37°C. The fluorescent signals were measured with a fluorescent microplate reader (FlexStation 3; Molecular Devices, Sunnyvale, CA) at excitation of 494 nm and emission of 516 nm.
After 48-hour transfection, total RNA was extracted from HEK-293T cells using TRIzol (Life Technologies) according to the manufacturer’s protocol. cDNA was generated from the resultant total RNA using PrimeScript Reverse Transcriptase (TaKaRa Bio, Otsu, Shiga, Japan). The resultant cDNA was mixed with SYBR Premix ExTaq kit (TaKaRa Bio) and appropriate primers, and quantitative PCR was performed with Thermocycler C1000 (Bio-Rad, Hercules, CA). The sequences of the primers used in this study are commercially available (OriGene Technologies).
Statistical analyses were performed with GraphPad Prism 5 (GraphPad Software, La Jolla, CA). For pairwise comparisons, unpaired t tests were performed. For comparisons across three groups or larger and for two-factor analyses, one-way analysis of variance and two-way analysis of variance, respectively, followed by post hoc multiple comparison tests (Newman-Keuls or Bonferroni) were performed. P values less than 0.05 were considered to be statistically significant. Data are shown as mean ± S.E.M.
Stimulation of EP1 Enhances cAMP Production Induced by D1 Receptors in HEK-293T Cells.
To address the mechanism of how EP1 activation enhances dopamine D1 receptor signaling in striatal neurons (Kitaoka et al., 2007), we overexpressed both D1 receptors and EP1 receptors in HEK-293T cells, in which neither of these receptors is endogenously expressed, and examined the effect of ONO-DI-004, an EP1-specific agonist, on cAMP production induced by D1 stimulation. Treatment of the cells with SKF-81297, a D1 receptor agonist, at 30, 100, and 500 nM for 5 minutes increased cAMP production in a dose-dependent manner (Fig. 1A). Simultaneous application of ONO-DI-004, an EP1-specific agonist, at 1 and 10 μM significantly enhanced cAMP production induced by SKF-81297 in a dose-dependent manner (Fig. 1A). This result suggests that EP1 receptor activation enhances D1 receptor signaling in a cell-autonomous manner. Notably, treatment with ONO-DI-004 in the absence of SKF-81297 produced no significant increase in the cAMP levels (Fig. 1A), suggesting that EP1 activation alone cannot induce cAMP production.
It was reported that two classes of D1 agonists have different actions: SKF-81297 increases not only cAMP production but also phosphoinositide hydrolysis through the heterodimer of D1 and D2 receptors, whereas SKF-83822 only stimulates cAMP production through the homodimer of D1 receptors (Verma et al., 2010). Therefore, we examined whether EP1 activation could also enhance cAMP production induced by SKF-83822. Treatment with SKF-83822 at 100 and 500 nM for 5 minutes increased cAMP production (Fig. 1B). Since the cAMP responses at these two concentrations were similar, these cAMP responses are considered to be saturated. Simultaneous treatment with ONO-DI-004 further enhanced these saturated cAMP responses (Fig. 1B). These results suggest that EP1 stimulation enhances the maximal level of D1 receptor signaling in HEK-293T cells.
To confirm the specificity of ONO-DI-004 and SKF-83822 on EP1 and D1 receptors, respectively, we examined the effects of these drugs in HEK-293T cells that expressed either D1 or EP1 receptors alone. Thus, in HEK-293T cells that overexpressed D1 receptors alone, treatment with ONO-DI-004 at 10 μM failed to affect D1-induced cAMP response (Fig. 1C). In HEK-293T cells that overexpressed EP1 receptors alone, SKF-83822 did not increase cAMP levels (Fig. 1D). Thus, the effects of ONO-DI-004 and SKF-83822 are specific to EP1 and D1 receptors, respectively, in our experimental conditions.
EP1 Activation Is Specifically Coupled to D1 Receptor Signaling.
We next examined whether EP1 activation could enhance Gs-induced cAMP production regardless of the type of stimulated GPCR. For this purpose, we employed two GPCRs as examples. First, we examined the action of EP1 on cAMP production induced by EP2, another subtype of prostaglandin E receptor coupled to Gs. In HEK-293T cells that expressed both EP1 and EP2 receptors, ONO-AE1-259, an EP2-specific agonist, at 0.5 and 1.0 μM maximally increased cAMP production (Fig. 2A). These EP2-mediated cAMP responses were not augmented by simultaneous application of ONO-DI-004 at 10 μM (Fig. 2A). As another example, we examined the action of EP1 on cAMP response induced by the β2-adrenergic receptor. In HEK-293T cells that expressed both EP1 and ADRB2 receptors, treatment with isoproterenol at 100 and 500 nM increased cAMP production (Fig. 2B). These cAMP responses were also not augmented by ONO-DI-004 (Fig. 2B). Therefore, the EP1 receptor specifically facilitates D1 receptor signaling, rather than Gs-stimulated cAMP production in general.
EP1 and D1 Receptors Form a Heteromeric Complex in HEK-293T Cells.
Given that the stimulatory action of EP1 is specific to D1 receptors, we suspected that EP1 receptors form a heteromeric complex with D1 receptors. First, we examined whether these receptors are colocalized in HEK-293T cells by immunofluorescent staining. We used anti-HA antibody and anti-FLAG antibody to detect HA-tagged EP1 and FLAG-tagged D1, respectively. Using cells expressing either HA-tagged EP1 or FLAG-tagged D1, we confirmed that anti-HA antibody and anti-FLAG antibody selectively recognize HA-tagged EP1 and FLAG-tagged D1, respectively (Supplemental Fig. 1). In the cells overexpressing both EP1 and D1, EP1 signals were observed both at the periphery and around the nucleus of the cell, whereas D1 signals were localized mostly at the cell periphery (Fig. 3A, top). At a higher magnification, punctate signals for EP1 and D1 receptors were observed along the edge of the cell, and a considerable proportion of these signals were colocalized (Fig. 3A, bottom). However, not all D1 signals were colocalized with EP1 signals, and vice versa, suggesting that each of these receptors is present either alone or in complex with each other.
To directly analyze the complex formation of EP1 and D1 receptors, we examined whether these receptors could be coprecipitated. Although signals for EP1 or D1 were detected selectively in cell lysates expressing corresponding receptors, multiple bands were observed (Supplemental Fig. 2). To identify specific bands corresponding to intact HA-tagged EP1, we precipitated it with anti-EP1 antibody that recognizes the C terminus and detected it with anti-HA antibody that recognizes the N terminus. Similarly, to identify specific bands corresponding to intact FLAG-tagged D1, we precipitated it with anti-FLAG antibody that recognizes the N terminus and detected it with anti-D1 antibody that recognizes the C terminus. We decided to limit our analyses to the bands identified by this method. Immunoprecipitation from cell lysates of HEK-293T cells with EP1 antibodies precipitated D1 receptors only when both EP1 and D1 receptors were overexpressed (Fig. 3B). Conversely, immunoprecipitation from the same cell lysates with antibodies that recognized the FLAG tag fused to the D1 receptor precipitated EP1 receptors only in the presence of both EP1 and D1 receptors (Fig. 3C). These results suggest a complex formation of EP1 and D1 receptors in HEK-293T cells.
EP1-Induced Ca2+ Increase Is Not Required for the Facilitative Action of EP1 on D1 Receptor Signaling.
Since it is known that stimulation of one protomer may alter a ligand-binding profile of the other protomer (Ferré et al., 1991), we performed a saturation-binding analyses using [3H]-(R)-(+)-7-chloro-8-hydroxy-3-methyl-1-phenyl-2,3,4,5-tetrahydro-1H-3-benzazepine hydrochloride ([3H]SCH-23390), a radiolabeled ligand for the D1 receptor. However, EP1 stimulation did not affect either the maximum binding capacity or the dissociation constant of D1 receptors (Supplemental Fig. 3). Then we examined an involvement of intracellular signaling for the facilitative action of EP1 on D1 receptor signaling. EP1 receptor is primarily coupled to intracellular Ca2+ increase in various cell types (Hirata and Narumiya, 2011). Since intracellular Ca2+ increase can potentiate the activity of several isoforms of adenylyl cyclase, such as AC1, AC3, and AC8 (Sunahara et al., 1996; Siehler and Milligan, 2011), we examined whether EP1 activation could enhance D1 receptor signaling through intracellular Ca2+. To this end, we depleted intracellular Ca2+ using the two following methods. First, pretreatment with BAPTA-AM, a cell-permeable Ca2+ chelator, at 10 μM for 30 minutes completely blocked intracellular Ca2+ increase induced by ONO-DI-004 at 10 μM, as measured by Fluo-4-AM, a cell-permeable, fluorescent Ca2+ indicator (Fig. 4A). However, this Ca2+ depletion did not block the facilitative effect of EP1 on D1-induced cAMP increase (Fig. 4B). We also examined the effect of Ca2+ depletion by bath application of EGTA at 1 mM and thapsigargin at 100 nM for 30 minutes. Whereas Ca2+ depletion with this method also blocked the intracellular Ca2+ increase induced by ONO-DI-004 (Fig. 4C), this manipulation failed to disrupt the facilitative effect of EP1 on D1-induced cAMP production (Fig. 4D). Taken together, these findings suggest that EP1-mediated Ca2+ increase is not required for EP1 facilitation of D1 receptor signaling.
Gβγ Subunits Are Involved in the Action of EP1 in Facilitating D1-Induced cAMP Response.
In addition to AC isoforms that can be stimulated by intracellular Ca2+, other AC isoforms, such as AC2, AC4, and AC7, can be stimulated by Gβγ subunits in concert with Gαs subunits (Sunahara et al., 1996; Siehler and Milligan 2011). To evaluate the involvement of Gβγ subunits in EP1-mediated facilitation of D1 signaling, we depleted Gβγ subunits by overexpression of the Gα subunit of transducin (Gtα1) as a Gβγ scavenger. In this experiment, the cells are pretreated with 3-isobutyl-1-methylxanthine (IBMX) to exclude a confounding effect of stimulating cGMP phosphodiesterase, a primary downstream effector of Gtα (Tang and Gilman, 1991). First, we examined whether pretreatment with IBMX might affect the EP1 action on D1 receptor signaling. After pretreatment with IBMX, D1 stimulation by the addition of SKF-83822 at 1, 10, and 500 nM induced cAMP response in a dose-dependent manner (Fig. 5A). Simultaneous treatment with ONO-DI-004, an EP1 agonist, at 10 μM significantly increased this cAMP response by SKF-83822 at both concentrations of 10 and 500 nM (Fig. 5A). Thus, EP1 activation facilitates D1 receptor signaling in the absence or presence of IBMX. Then we examined the effect of Gβγ depletion on the EP1 action on D1 signaling. In HEK-293T cells that overexpressed Gtα in the presence of IBMX, SKF-83822 at 1, 10, and 500 nM induced cAMP increase in a dose-dependent manner. However, concurrent activation of EP1 by ONO-DI-004 at 10 μM failed to augment these D1-induced cAMP responses at any tested concentrations of SKF-83822 (Fig. 5, B and C). This finding suggests that Gβγ subunits are involved in the facilitative effect of EP1 on D1 receptor signaling.
To examine whether the presence of free Gβγ subunits is sufficient to enhance D1 receptor signaling, we overexpressed the Gβ1 subunit and the Gγ2 subunit, a combination which is known to activate a Gβγ-sensitive AC isoform in vitro (Diel et al., 2006). However, overexpression of Gβ1γ2 subunits reduced the SKF-83822–induced cAMP response (Fig. 5D). Thus, overexpression of Gβ1γ2 subunits is not sufficient to mimic the facilitative effect of EP1 on D1-induced cAMP production.
Because actions of Gβγ subunits have been frequently associated with Gαi signaling (Federman et al., 1992), we next examined a possible involvement of Gαi subunits in the effect of EP1 on D1 signaling. However, pretreatment with pertussis toxin, a blocker of Gαi signaling, failed to alter the effect of ONO-DI-004 at 10 μM on D1-induced cAMP increase (Fig. 6).
AC7, a Gβγ-Stimulated AC Isoform, Mediates the Facilitative Effect of EP1 on D1-Induced cAMP Response.
Given the role for Gβγ subunits in the action of EP1 in D1 signaling as described above, we examined whether a Gβγ-stimulated AC isoform could mediate the facilitative effect of EP1 on D1-induced cAMP response. Among several AC isoforms that can be stimulated by Gβγ subunits, AC7 has been reported to be expressed in HEK-293 cells (Atwood et al., 2011) from which HEK-293T cells have been derived. Therefore, we analyzed the involvement of AC7 as well as that of AC5 and AC6, two other isoforms highly expressed in HEK-293 cells.
We first confirmed the mRNA expression of these AC isoforms in HEK-293T cells. Knockdown of these AC isoforms by transfection with siRNA targeting respective AC isoforms reduced mRNA levels of corresponding AC isoforms to approximately 20% of their mRNA levels with control siRNA in two independent experiments (20.1 and 21.6% for AC5 knockdown, 19.2 and 24.1% for AC6 knockdown, 20.0 and 29.5% for AC7 knockdown). We next examined whether any of these AC isoforms are involved in D1-induced cAMP response without EP1 activation. Knockdown of AC6 significantly, but partially, reduced cAMP increase induced by SKF-83822 at 500 nM (Fig. 7). In contrast, knockdown of either AC5 or AC7 did not significantly affect SKF-83822–induced cAMP increase (Fig. 7). This result suggests that D1 activation without simultaneous EP1 activation is coupled to specific AC isoforms, including AC6.
Using knockdown of respective AC isoforms, we next examined the AC isoform involved in the facilitative effect of EP1 on D1-induced cAMP response. In the cells transfected with control siRNA, simultaneous treatment with ONO-DI-004 at 10 μM significantly enhanced the cAMP response induced by SKF-83822 (Fig. 8A). Knockdown of AC5 did not affect D1-induced cAMP response with or without EP1 activation (Fig. 8B). Knockdown of AC6 that had suppressed cAMP response induced by D1 stimulation alone (Fig. 7) failed to alter the facilitative effect of ONO-DI-004 on this cAMP response (Fig. 8C). Surprisingly, knockdown of AC7, which had not affected cAMP response induced by D1 stimulation alone (Fig. 7), abolished the facilitative effect of ONO-DI-004 on this cAMP response (Fig. 8, D and E).
Although partial knockdown in our conditions might have failed to detect an AC isoform that plays a minor role, these results suggest that EP1 activation enhances D1-induced cAMP production through AC7, a Gβγ-sensitive AC isoform, whereas distinct AC isoforms including AC6 mediate cAMP production induced by D1 stimulation alone.
We previously reported that PGE receptor subtype EP1 enhances signaling of dopamine D1 receptors in striatal slices and promotes D1-induced hyperlocomotion (Kitaoka et al., 2007). In the present study, we examined the mechanism underlying this facilitative effect of EP1 on D1 receptor signaling using HEK-293T cells. Stimulation of EP1 facilitates cAMP production induced by D1 receptor agonists in a manner independent of EP1-induced Ca2+ increase. Our findings suggest that this EP1 action is mediated through Gβγ subunits and AC7, a Gβγ subunit-sensitive AC isoform. Strikingly, cAMP production induced by D1 receptors alone is mediated by other AC isoforms including AC6, but not AC7. Therefore, the present study demonstrates that distinct AC isoforms mediate cAMP production induced by D1 receptors alone and the facilitative action of EP1 on D1-induced cAMP production (Fig. 9).
A Role for Gβγ-AC7 in EP1-Mediated Facilitation of D1-Induced cAMP Production.
The involvement of Gβγ subunits in the EP1 action on D1 signaling is supported by two findings. First, overexpression of the Gtα subunit as a Gβγ scavenger blocked the facilitative action of EP1 on D1-induced cAMP production (Fig. 5, A and B). Second, knockdown of AC7, a Gβγ-stimulated AC isoform, also abolished this EP1 action (Fig. 8). Although a combination of Gβl and Gγ2 subunits is known to activate a Gβγ-sensitive AC isoform in vitro (Diel et al., 2006), overexpression of Gβ1 and Gγ2 subunits failed to increase D1-induced cAMP response (Fig. 5D). One possibility is that AC activity might be already saturated by GPCR stimulation alone, since it was reported that overexpression of AC6, but not of Gαs, proportionally increases cAMP response induced by β-adrenergic receptor in rat cardiac myocytes (Gao et al., 1998). However, Gβγ overexpression still failed to facilitate cAMP response to SKF-83822 at a nonsaturating dose of this drug (10 nM). Since Gβγ overexpression actually suppressed D1-induced cAMP response in our study, this inhibition could mask a facilitative action of Gβγ, if it exists. It is known that the action of these Gβγ subunits is conditional upon the AC isoform to be coupled. Yoshimura et al. (1996) reported that overexpression of these Gβγ subunits facilitated D1-induced cAMP response with overexpression of AC7, a Gβγ-sensitive AC isoform, but not with overexpression of AC5. Therefore, EP1 activation could facilitate the coupling of D1 receptors to the Gβγ-sensitive AC isoform over the others.
Whereas D1 activation alone could activate both Gαs and Gβγ subunits in theory, AC7 is involved in D1-induced cAMP production only in the presence of EP1 activation. It is known that not all Gβγ dimers can regulate a given effector molecule, as exemplified by inhibition of voltage-gated calcium channels and activation of G-protein-coupled inwardly rectifying potassium channels (Albert and Robillard, 2002). Furthermore, several GPCRs have been shown to use specific Gβγ isoforms for their respective signaling pathways (Albert and Robillard, 2002). Therefore, only Gβγ isoforms that are activated by EP1 receptors in complex with D1 receptors, rather than by D1 receptors alone, could facilitate the activity of AC7, though this possibility remains to be tested.
The D1-EP1 Complex as Signaling Machinery Distinct from the D1 Receptor Alone.
Coprecipitation experiments in this study showed that EP1 and D1 receptors form a complex in HEK-293T cells (Fig. 3, B and C). Such a molecular complex could provide an environment in which Gβγ subunits are activated to a sufficient local concentration to stimulate AC7 activity. The formation of the D1-EP1 complex is also consistent with the fact that EP1 stimulation specifically facilitates D1 receptor signaling, but not those of other GPCRs tested, EP2 or ADRB2 (Fig. 2). Our observation that not all D1 receptors are colocalized with EP1 (Fig. 3A) suggests the existence of both the D1-EP1 complex and the D1 receptor that is not associated with EP1. Since AC6 and AC7 are used differently for cAMP production by D1 stimulation alone and its facilitation by EP1 stimulation, respectively, it is plausible that these AC isoforms selectively bind to the D1 receptor alone and the D1-EP1 complex, respectively. This notion lends support to the concept that a GPCR heteromer acts as a distinct entity from its respective protomer (Ferré et al., 2009).
However, complex formation of EP1 and D1 receptors is not sufficient to cause the facilitative action of EP1 on D1 signaling. For example, EP1 activation suppresses cAMP response induced by ADRB2, which forms a heteromer with EP1, in airway smooth muscle cells (McGraw et al., 2006), although we failed to observe this EP1 action on ADRB2 signaling in HEK-293T cells (Fig. 2B). Thus, whereas EP1 forms a complex with either D1 or ADRB2, EP1 activation can exert opposite actions on cAMP production induced by D1 and ADRB2. Since Gβγ subunits can either stimulate or inhibit AC activity, depending on the type of AC isoforms, heteromeric EP1 receptors could direct distinct AC isoforms to regulate cAMP production induced by D1 and ADRB2. It has been shown that several GPCRs use a specific AC isoform for cAMP production. For example, in mouse smooth muscle cells and HEK-293 cells, AC2 and AC6 mediate cAMP increase induced by EP2 and ADRB2, respectively (Bogard et al., 2012). Consistent with specific GPCR-AC coupling as such, subcellular fractionation showed that ADRB2 and AC6 are localized in lipid rafts, from which EP2, AC2, and AC7 are excluded (Crossthwaite et al., 2005; Bogard et al., 2012). Likewise, D1 receptors that are coupled to AC6 without EP1 and those that are coupled to AC7 with EP1 could be localized in different membrane domains, lipid raft domains, and nonlipid raft domains, respectively. In subcellular fractionation using sucrose density gradients, we have found that D1 receptors and AC5/6 are mainly localized in lipid raft domains, as previously reported (Ostrom et al., 2002; Yu et al., 2004), but that this subcellular distribution of D1 was apparently not altered by simultaneous expression of EP1 (Supplemental Fig. 4). Whereas AC5/6 is distributed similarly to flotillin-1, a marker for lipid rafts, D1 receptors are distributed much more broadly, suggesting a large heterogeneity of D1 receptor complexes. It was reported that AC can form a signaling complex with a GPCR and a Gβγ subunit during their biosynthesis in the endoplasmic reticulum in the heterologous system (Dupré et al., 2009). Whether the D1-AC6 complex and the D1-EP1-AC7 complex are formed separately during their biosynthesis, or alternatively, whether these complexes are dynamically assembled upon GPCR activation, warrants future investigation.
Physiologic Implications for the D1-EP1 Complex.
Since EP1 activation is critical for augmentation of D1 receptor signaling in striatal slices and hyperlocomotion induced by D1 receptors, the D1/EP1-Gβγ-AC pathway identified in the present study appears to be of physiologic relevance. Thus, our study indicates that striatal neurons express two kinds of D1 receptors: one in the D1-EP1 heteromer and one that is not associated with EP1. The former of these is activated proportionally according to the PGE2 content in the brain. Since exposure to stressors, such as social defeat, increases the PGE2 content in the mouse brain (Tanaka et al., 2012), the D1-EP1 complex may provide the mechanism for stress exposure to modulate D1 receptor signaling through PGE2-EP1 signaling.
However, whether the same molecules identified in the present study mediate the D1-EP1 crosstalk in vivo remains to be tested. For example, whereas AC6, but not AC5 or AC7, mediates cAMP production induced by D1 activation in HEK-293T cells, cAMP production induced by either forskolin or D1 agonists was greatly reduced in the striatum of AC5-deficient mice (Lee et al., 2002). Therefore, HEK-293T cells used in this study may lack a component that makes D1 receptors preferentially coupled to AC5, as seen in striatal neurons. It has been shown that an AC isoform that is not critical for cAMP production by D1 receptors alone can still play a role in dopamine-related behaviors. For example, genetic deletion of the Ca2+/calmodulin-stimulated AC isoforms AC1 and AC8 abolishes locomotor sensitization induced by chronic cocaine treatment (DiRocco et al., 2009). In contrast, little is known about physiologic functions of the Gβγ-sensitive AC isoforms, such as AC2, AC4, and AC7, in the brain. Given the isoform-specific AC functions identified in our study, it is important to choose appropriate stimuli or contexts to reveal the physiologic functions of the Gβγ-sensitive AC isoforms. According to our findings, the D1-EP1 complex offers an excellent platform to analyze functions of these AC isoforms in the brain.
The authors thank Atsushi Mizutani and Nodoka Asamoto for animal care, Tae Arai and Akiko Washimi for secretarial help, and Kimiko Nonomura for technical assistance. The authors also thank Drs. Toshimasa Ishizaki and Satoko Sakamoto for critical advice on biochemical experiments, and Dr. Akiko Maekawa for critical advice on Ca2+ imaging experiments in this study.
Participated in research design: Ehrlich, Furuyashiki, Kitaoka, Narumiya.
Conducted experiments: Ehrlich.
Performed data analysis: Ehrlich, Furuyashiki, Kakizuka.
Wrote or contributed to the writing of the manuscript: Ehrlich, Furuyashiki, Narumiya.
- Received May 10, 2013.
- Accepted July 10, 2013.
This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan; and by a Core Research for Evolutionary Science and Technology grant from Japan Science and Technology Agency.
- adenylyl cyclase
- β2-adrenergic receptor
- 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid/acetoxymethyl ester
- Dulbecco’s modified phosphate-buffered saline
- enhanced green fluorescent protein
- G protein–coupled receptor
- human embryonic kidney
- 16s-9-deoxy-9β-chloro-15-deoxy-16-hydroxy-17,17-propano-19,20-didehydroprostaglandin E2
- polymerase chain reaction
- prostaglandin E2
- reverse transcription
- (R)-(+)-7-chloro-8-hydroxy-3-methyl–1-phenyl-2,3,4,5-tetrahydro-1H-3-benzazepine hydrochloride
- small interfering RNA
- (±)-6-chloro-2,3,4,5-tetrahydro-1-phenyl-1H-3-benzazepine hydrobromide
- 6-chloro-2,3,4,5-tetrahydro-1-(3-methylphenyl)-3-(2-propenyl)-1H-3-benzazepine-7,8-diol hydrobromide
- signal sequence
- Tris-buffered saline
- Copyright © 2013 by The American Society for Pharmacology and Experimental Therapeutics