G protein-coupled receptors form stable complexes with inwardly rectifying potassium channels and adenylyl cyclase.

A large number of studies have demonstrated co-purification or co-immunoprecipitation of receptors with G proteins. We have begun to look for the presence of effector molecules in these receptor complexes. Co-expression of different channel and receptor permutations in COS-7 and HEK 293 cells in combination with co-immunoprecipitation experiments established that the dopamine D(2) and D(4), and beta(2)-adrenergic receptors (beta(2)-AR) form stable complexes with Kir3 channels. The D(4)/Kir3 and D(2) receptor/Kir3 interaction does not occur when the channel and receptor are expressed separately and mixed prior to immunoprecipitation, indicating that the interaction is not an artifact of the experimental protocol and reflects a biosynthetic event. The observed complexes are stable in that they are not disrupted by receptor activation or modulation of G protein alpha subunit function. However, using a peptide that binds Gbetagamma (betaARKct), we show that Gbetagamma is critical for dopamine receptor-Kir3 complex formation, but not for maintenance of the complex. We also provide evidence that Kir3 channels and another effector, adenylyl cyclase, are stably associated with the beta(2)-adrenergic receptor and can be co-immunoprecipitated by anti-receptor antibodies. Using bioluminescence resonance energy transfer, we have shown that in living cells under physiological conditions, beta(2)AR interacts directly with Kir3.1/3.4 and Kir3.1/3.2c heterotetramers as well as with adenylyl cyclase. All of these interactions are stable in the presence of receptor agonists, suggesting that these signaling complexes persist during signal transduction. In addition, we provide evidence that the receptor-effector complexes are also found in vivo. The observation that several G protein-coupled receptors form stable complexes with their effectors suggests that this arrangement might be a general feature of G protein-coupled signal transduction.

Heterotrimeric (␣␤␥) guanine nucleotide-binding proteins (G proteins) are the transducers that convey information from agonist-occupied receptors to a variety of effector proteins. Activation of G proteins in detergent-containing solutions results in dissociation of the ␣ subunit (G␣) from the ␤␥ heterodimer (G␤␥). Depending upon the effector that is being regulated the information is conveyed by G␣ and/or G␤␥. Receptors that couple to G proteins (GPCRs) 1 number in the hundreds, and have in common seven ␣-helical transmembrane domains. Most cells have an assortment of GPCRs, G proteins, and effectors, and in general, signal transduction has been thought of as occurring by a process involving random collisions between these different signaling components as they move about independently of one another in the lipid bilayer. Thus, GPCR-mediated activation of a G protein would be expected to produce G␣ and G␤␥ with the potential to regulate the activity of multiple effectors. However, signal transduction in vivo most often results in the regulation of a select effector. Recent data, as well as many reports from the early literature on G protein-mediated signaling offer a reason for why expectation does not agree with observation. A number of studies have demonstrated stable interactions between GPCRs and G proteins (see Ref. 1 and 2, for review), between G proteins and effectors (3)(4)(5), and between G␣ and G␤␥ of activated G proteins (4,6,7). These data raise the possibility that GPCRs, G proteins, and effectors can form stable complexes that persist during signal transduction in vivo. Such metastable signaling complexes would explain the rapidity and specificity observed during signaling in the intact cell. Indeed, the possibility that GPCRs can form a complex with their effectors was recently shown for the ␤ 2 -adrenergic receptors (␤ 2 AR) and L-type Ca 2ϩ channels (Ca v 1.2) and adenylyl cyclase (8). This receptor-channel complex also contained heterotrimeric G proteins and the modulatory proteins, protein kinase A and protein phosphatase 2A. Whether GPCRs and other effectors, such as Kir3 channels, are part of stable or transient complexes in which the different components of signaling pathways interact with each other is as of yet unknown. Size exclusion chromatography of purified cardiac I KACh channels (Kir3.1-Kir3.4) identified a complex of 480 -520 kDa (9), which would allow for a tetrameric Kir3.1-Kir3.4 structure associated with an equivalent number of muscarinic M2 receptors. Assembly of the GPCR into effector com-plexes may be important to restrict surface expression to fully and correctly assembled effector complexes, as has been shown for the sulfonurea receptor-Kir6 channel assembly (10), GABA B 1 and -2 receptor heterodimerization (11), and Kir3.1 channel heterotetramerization with Kir3.2 or Kir3.4 (12). In all these examples correct assembly results from the presence of endoplasmic reticulum retention sequences or post-endoplasmic reticulum targeting motifs. Thus, complex formation may also form a means to regulate GPCR/effector signaling. Furthermore, evidence for the inclusion of many additional proteins involved in downstream signaling events has led to the suggestion that complexes as large as 2000 kDa may represent functional units (13). Using a combination of co-purification strategies, Western blotting, and mass spectrometry, at least 77 different proteins were demonstrated to be associated with the N-methyl-D-aspartate receptor in the postsynaptic density of neurons including protein kinase As, protein kinase Cs, mitogen-activated protein kinases, tyrosine kinases, phosphatases, PSD-95, and other PDZ domain containing proteins, small G proteins and their modulators, and the G proteincoupled metabotropic glutamate receptor (14). Evidence for large GPCR signaling complexes has been recently reviewed (2).
Here we have further explored the occurrence and nature of complex formation between GPCRs and Kir3 channels and adenylyl cyclase. Catecholaminergic receptor signaling, particularly the ␤ 2 AR and D 2 -dopamine receptors, have been used as prototypic models for GPCR signaling. Most commonly, signaling via the ␤ 2 AR is described through its activation of the stimulatory G protein (G␣ s ) and adenylyl cyclase, whereas that of the D 2 dopamine receptor is exemplified by its inhibitory action on the same effector via G␣ i/o . However, both receptors are also recognized to be physiological regulators of the activity of G protein-activated inwardly rectifying potassium channels (GIRK or Kir3) (15)(16)(17)(18)(19). Here we show that both expressed GPCRs and effectors are part of stable complexes by demonstrating that Kir3 channels and adenylyl cyclase co-precipitate with dopamine D 2 -like receptors and the ␤ 2 AR. Furthermore, by using bioluminescence resonance energy transfer (BRET) we show that both adenylyl cyclase and Kir3 channels are directly associated with the ␤ 2 AR in living cells. The stability of these complexes is not altered by receptor activation or by inactivation of G␣. However, whereas maintenance of the complex is not mediated by G␤␥, complex formation is in some cases G␤␥-dependent. These observations have important implications with regard to G protein-mediated signaling efficiency and specificity.

EXPERIMENTAL PROCEDURES
Materials COS-7 cells were purchased from American Type Culture Collection (Rockville, MD). ␣-MEM was purchased from Central Media Preparation Service (University of Toronto, Toronto, ON). Fetal bovine serum, Dulbecco's modified Eagle's medium, LipofectAMINE reagent, and Opti-MEM I, geneticin (G418), T4 DNA ligase, T4 polynucleotide kinase, and oligonucleotides were bought from Invitrogen (Burlington, ON, Canada). All other DNA-modifying enzymes and restriction endonucleases were obtained from New England BioLabs (Beverly, MA). Isopropyl-␤-D-thiogalactopyranoside was purchased from Molecular Probes (Eugene, OR). Low range molecular weight protein markers and nitrocellulose were from Bio-Rad. The protein solubilization reagent B-PER and BCA reagent for protein measurement were obtained from Pierce. Protein G-Sepharose, anti-His tag mouse monoclonal antibody, and glutathione-Sepharose 4B beads were from Amersham Biosciences. In some cases, pre-cast 10 and 8 -16% gradient Tris glycine polyacrylamide gels were used, these and polyvinylidene difluoride membranes were purchased from Invitrogen. All other chemicals for SDS-PAGE were purchased from Bio-Rad. Dopamine, quinpirole, pertussis toxin (PTX), glutathione, aprotinin, leupeptin, phenylmethylsulfonyl fluoride, benzamidine, and soybean trypsin inhibitor were purchased from Sigma. Anti-FLAG (M2) mouse monoclonal antibody was purchased from Sigma and anti-FLAG rabbit polyclonal antiserum was purchased from Santa Cruz (Santa Cruz, CA). Digitonin and rat monoclonal antihemagglutinin (HA) (3F10) antibodies were obtained from Roche Molecular Biochemicals. Mouse monoclonal anti-HA (Y11), anti-Myc (A14; 9E10) antibodies, and rabbit polyclonal anti-G␤ antibodies were acquired from Santa Cruz Biotechnology. Mouse monoclonal anti-HA.11 (16B12) antibody was from Berkeley Antibody Co. (Berkeley, CA). Monoclonal mouse anti-Kir3.1 (NH 2 -terminal) was from Upstate Biotechnology (Lake Placid, NY), and anti-Kir3.1 (COOH-terminal), anti-Kir3.2, and anti-Kv1.5 were from Alomone Labs (Jerusalem, Israel). Chemiluminescent detection of peroxidase-conjugated secondary antimouse or anti-rabbit antibodies (1:2000; Sigma) was performed using ECL Plus (Amersham Biosciences) or Renaissance (PerkinElmer Life Sciences). Plasmid vectors for the production of glutathione S-transferase (GST) fusion proteins were from the pGEX-series of Amersham Biosciences. The mammalian expression vector pCDNA3 was from Invitrogen. All other chemicals used were of reagent grade or higher and were obtained from Sigma.

DNA Constructs
The vector pcDNA3 containing the dopamine D 2L or D 4 receptor cDNA with an amino-terminal, cleavable signal sequence immediately followed by the HA and FLAG epitopes (HA-D 2L , FLAG-D 4 ) have been described by us previously (20,21). Epitope tags were engineered into human Kir3.1, Kir3.2c, Kir3.3, and Kir3.4 (22) by using standard molecular approaches employing the polymerase chain reaction. Oligonucleotides were designed to delete the existing termination codon in the channels and to add the required carboxyl-terminal epitope tag sequence followed by a termination codon. In this fashion we engineered Kir3.1 and Kir3.3 with a carboxyl-terminal HA tag (5Ј-TACCCGTAC-GACGTCCCGGACTACGCC-3Ј), Kir3.2c with a carboxyl-terminal Myc tag (5Ј-GAACAAAAACTCATCTCAGAAGAGGATCTG-3Ј), and Kir3.4 with a carboxyl-terminal FLAG tag (5Ј-GACTACAAGGACGACGAT-GACAAG-3Ј). These constructs are denoted as Kir3.1-HA, Kir3.3-HA, Kir3.2-Myc, and Kir3.4-FLAG, respectively. For expression studies the tagged channels were cloned into the expression vector (23) pCDNA3. The dominant negative G protein-␣ subunits described by Gilchrist et al. (23) were created as oligonucleotide sequences and subcloned into the vector pCDNA3. The GST fusion protein for ␤ARKct and the ␤ARKct-minigene were a kind gift from Dr. R. J. Lefkowitz and Dr. W. J. Koch (Duke University). pcDNA 3.1, pRL-CMV, pGFP10, and pEGFP were obtained from Invitrogen, Promega (Madison, WI), Bio-Signal Packard (Montréal, QC, Canada), and Clontech, respectively. pHAD4.2-GFP10 was made by insertion into pGFP 2 -N3 (BioSignal Packard). p␤ 2 AR-GFP10 and p␤ 2 AR-RLuc code for proteins consisting of GFP10 (a GFP variant described below) and Renilla reniformis luciferase (RLuc) fused to the COOH terminus of the human ␤ 2 -adrenergic receptor, respectively. ␤ 2 AR-GFP10 was constructed as follows: the GFP10 AgeI/BsrgI fragment was subcloned into the AgeI/BsrgI site of pGFP-N1-His␤ 2 AR-YFP (24). pcDNA3-␤ 2 AR-EGFP was a gift from J. Benovic (Thomas Jefferson University) as described (25). A plasmid containing the cDNA for the ␣-subunit of the stimulatory G protein (G␣ s ) was obtained from the American Type Culture Collection (ATCC number 63315). Plasmids containing the cDNA for the G protein ␥2 subunit (G␥2) were subcloned into pcDNA3. A plasmid containing the cDNA for the bovine G protein ␤1 subunit (G␤ 1 , a gift from Arieh Katz (California Institute of Technology)) was subcloned into pcDNA3.1. A plasmid containing the cDNA for type II adenylyl cyclase was a gift from Dr. Randall Reed (26). pCDNA3-␤ 2 AR-GFP was mutated with the QuikChange kit from Stratagene (La Jolla, CA) to produce p␤ 2 AR-EGFP, which coded for the enhanced variant of GFP. pAC-RLuc was prepared by subcloning the cDNAs for type II adenylyl cyclase and RLuc into pcDNA3.1. pAC-RLuc was designed to code for adenylyl cyclase with RLuc fused to its COOH terminus (AC-RLuc). The integrity of all engineered constructs was verified by nucleotide sequencing using the dideoxy chain termination method.

Cell Culture and Transfection
COS-7 cells were grown in ␣-MEM supplemented with 10% fetal bovine serum, whereas HEK-293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin, 2 mM L-glutamine in a humidified incubator at 37°C and 5% CO 2 . For transfection, COS-7 cells were grown to 60 -80% confluency. DNA constructs were transfected into cells by lipofection using LipofectAMINE and Opti-MEM I. The transfection protocol was employed and optimized as recommended by the supplier of the transfection reagents (Invitrogen). Transfection efficiency determined by using pCDNA3 expressing lacZ was about 60 -80%. The cells were assayed or harvested for further experiments 2 days after transfection. For transient transfection of HEK 293 cells were seeded at a density of 10 5 cells/cm 2 in polylysine-coated plates and transiently transfected ϳ24 h later with one or more plasmids using LipofectAMINE Plus (Invitrogen). Within experiments the concentration of DNA was kept constant by adding vector pCDNA3.1 or pCDNA3.1-luciferase. Sham transfections were also performed with these vectors.
Membrane Solubilization and Immunoprecipitation D 2 and D 4 Dopamine Receptor Expressing COS-7 Cells-Prior to harvesting, cells were washed three times with ice-cold phosphatebuffered saline (PBS; pH 7.2) and subsequently collected by scraping in 10 ml of ice-cold PBS, spun at 1,200 rpm (5 min), and stored at Ϫ80°C. The pellets were quickly thawed at room temperature and resuspended in 10 ml of TE buffer (10 mM Tris, pH 7.4, 10 mM EDTA, pH 8.0, 5 g/ml aprotinin, 2 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride). Next, the cells were homogenized on ice using a Polytron (2 ϫ 10 s, maximum speed). The homogenate was spun for 3 min at 2,000 ϫ g. The resulting supernatant was spun for 20 min at 34,000 ϫ g. The pellet was solubilized in RIPA buffer (150 mM NaCl, 50 mM Tris, pH 7.5, 1 mM EDTA, pH 8.0, 1% Nonidet P-40, 0.5% sodium deoxycholate, 5 g/ml aprotinin, 2 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride), and incubated at 4°C for 1 h with gentle rocking. The resulting homogenate was spun for 10 min at 14,000 ϫ g and the supernatant stored at Ϫ10°C until assayed. For immunoprecipitation assays we used about 300 g of total protein from the solubilized protein preparation as determined using the BCA protein assay (Pierce). The solubilized protein (300 g) was aliquotted into microcentrifuge tubes and adjusted to a volume of 500 l using RIPA buffer. Next we added 1 g of the primary antibody to the samples and incubated the samples for 4 h at 4°C with gentle rocking. To precipitate the antibody complexes in the mixture we added 40 l of 50% protein G-agarose slurry in RIPA buffer. This mixture was incubated for 3 h at 4°C with gentle rocking. The beads were pelleted, washed three times in RIPA buffer, and finally taken up in 50 l of SDS loading buffer and stored at Ϫ20°C until electrophoresis.
For dopamine D 2 receptor-Kir3.2 co-immunoprecipitation experiments from brain, rat striatal tissue (500 mg) was homogenized in buffer containing 50 mM Tris-Cl (pH 7.6), 150 mM NaCl, 1% Igepal Ca630, 1% Triton X-100, 0.5% sodium deoxycholate, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and proteinase inhibitor mixture (Sigma, 5 l/100 mg of tissue) for 1 h on ice. Next, the solubilized tissue was spun at 12,000 ϫ g for 20 min. The supernatant was collected and protein concentrations were determined. For co-immunoprecipitation experiments, striatal lysates (500 -700 g of protein) were incubated with anti-Kir3.2 (2 g) or as negative control nonspecific normal rabbit serum (2 g, Sigma, R9133) for 4 h at 4°C, followed by the addition of 20 l of protein A/G-Sepharose (Sigma) for 12 h. Pellets were washed four times in buffer as described above, boiled for 5 min in SDS sample buffer, and subjected to SDS-PAGE. The D 2 receptor was detected on Western blots with an anti-D 2 antibody as recommended by the manufacturer (D2R14-A, BioTrend Chemikalien, Germany). ␤ 2 AR Containing Constructs Expressed in HEK 293 Cells-Two days after transfection HEK 293 cells co-expressing the ␤ 2 AR-EGFP and either RLuc or AC-RLuc were collected by centrifugation, and samples containing 0.5 mg of cell protein were dissolved in 0.5 ml of a 1% digitonin solution. Insoluble material was removed by centrifugation at 100,000 ϫ g for 1 h and anti-GFP (Clontech) was added to the supernatant. After incubating the supernatant at room temperature for 1 h, anti-mouse IgG-agarose (Sigma) was added and the incubation continued for an additional 1 h. Both the immunoprecipitable and nonprecipitable material were assayed for RLuc activity (Promega Dual Luciferase assay kit) with a Turner Design TD 20/20 Luminometer.
HEK 293 cells expressing ␤ 2 AR and various Kir3 constructs were washed with ice-cold PBS and resuspended in 5 mM Tris (pH 7.4), 2 mM EDTA, 5 g/ml leupeptin, 10 g/ml benzamidine, and 5 g/ml soybean trypsin inhibitor. Cells were homogenized by Polytron and solubilized with RIPA buffer with the same protease inhibitor mixture. The solubilized cells were centrifuged to remove debris and stored at 80°C until use. 0.2 g of primary antibody was added to 450 l of clarified protein extract and incubated with shaking for 16 h at 4°C. Antibody complexes were precipitated with 100 l of a 50% slurry of protein Gagarose, washed in RIPA buffer, and resuspended in SDS-PAGE sample buffer.
In some experiments, ␤ 2 AR was immunoprecipitated from extracts of adult mouse heart or brain tissue. Briefly, 10 -12-week mice were sacrificed, their hearts and brains were removed, and tissue was minced and washed with ice-cold phosphate-buffered saline. The cells were then disrupted by homogenization with a Polytron (1 ϫ 20 s burst) in 15 ml of ice-cold lysis buffer containing 5 mM Tris-HCl (pH 7.4), 2 mM EDTA (plus a protease inhibitor mixture consisting of 5 g/ml leupeptin, 10 g/ml benzamidine, and 5 g/ml soybean trypsin inhibitor). The homogenate was centrifuged at 500 ϫ g for 15 min at 4°C. The pellets were then homogenized as before, spun again, and the two supernatants were pooled. The supernatant was centrifuged at 45,000 ϫ g for 15 min and the pellets were washed twice in the same buffer. Membranes were solubilized in 100 mM NaCl, 10 mM Tris (pH 7.4), 5 mM EDTA, 0.3% n-dodecyl-␤-D-maltoside plus the protease inhibitor mixture. Removal of detergent and concentration of the solubilized receptor was performed by dialysis using Centriprep cartridges (Amicon) against an ice-cold solution (Buffer A) containing 100 mM NaCl, 10 mM Tris-HCl (pH 7.4), 2 mM EDTA (plus protease inhibitors described above) until the detergent concentration was reduced below 0.05%. Affinity purified polyclonal anti-␤ 2 AR (1:1000 dilution) was added to the concentrate and gently agitated for 2 h at 4°C. Anti-rabbit IgG-agarose (Sigma; at an 11:1 secondary to primary antibody molar ratio) and protease inhibitor mixture were then added. The precipitation was allowed to proceed overnight at 4°C with gentle agitation. The immunoprecipitate was centrifuged at 12,000 rpm in a microcentrifuge for 10 min at 4°C. The pellet was washed three times in buffer A and finally resuspended in 100 l of reducing SDS-PAGE sample buffer for 30 min, sonicated, and centrifuged at 12,000 rpm. The supernatant was then subjected to SDS-PAGE and Western blotting as described below.

Western Analysis
Prior to electrophoresis the protein samples taken up in SDS-loading buffer were heated for 2-5 min at 80°C and spun quickly at 14,000 ϫ g. Samples were size separated by gel electrophoresis using 10 or 8 -16% gradient Tris glycine polyacrylamide gels, followed by transfer of the material to nitrocellulose or polyvinylidene difluoride membranes. For immunological detection of material, nonspecific binding to polyvinylidene difluoride or nitrocellulose membranes was blocked by overnight pretreatment of the membrane with blocking buffer containing 5% nonfat dry milk powder in Tris-or phosphate-buffered saline with 0.2% Tween 20 (TBS-T or PBS-T). Next the membranes were incubated for different times (dependent on the primary antibody used) with the appropriate dilution of primary antibody in blocking buffer anti-Kir3.2, 1:1000). After incubation with the primary antibody the membranes were washed extensively with TBS-T or PBS-T, and secondary antibody detection was performed using standard protocols using peroxidase-coupled antisera and enhanced chemiluminescence.

BRET Analysis
BRET was performed as described previously (24) with the following modifications. D4.2-GFP10, ␤ 2 AR-GFP10, or ␤ 2 AR-YFP fusion proteins were constructed and used as the fluorescent acceptor. Donor constructs consisted of fusion proteins between Kir3.1 or adenylyl cyclase and luciferase. To assay for BRET, cells were grown in 6-well tissue culture plates and transiently transfected with the relevant constructs. Forty-eight hours post-transfection, HEK-293T cells were washed twice in PBS, detached with PBS, and resuspended in PBS containing 0.1% glucose (w/v) and 10 Ϫ4 M ascorbic acid and the protease inhibitor mixture (5 g/ml leupeptin, 10 g/ml benzamidine, 5 g/ml soybean trypsin inhibitor). The cell suspension was assayed for protein concentration using the Bio-Rad protein assay with bovine serum albumin as a standard. Cells were then distributed in 96-well microplates (white Optiplate from Packard Bioscience) at a density of ϳ100,000 cells (ϳ20 -100 g of protein) per well. In these experiments, we have used the newly developed BRET 2 technology, which permits a better separation of the two emission peaks when compared with the original BRET partners, RLuc and the YFP variant of GFP. The BRET 2 technology uses a new coelenterazine (Deep Blue C coelenterazine), which after being oxidized by Renilla luciferase, emits light between 390 and 400 nm as compared with 470 nm for the original substrate coelenterazine H. A novel mutant of the GFP (GFP10) is excited by light released from luciferase and re-emits fluorescence at 505-508 nm. The increase in the spectral resolution between the luciferase and GFP emission peaks (105 nm for BRET 2 versus 50 nm for the original BRET system) results in more effective quantification of protein/protein interactions. Similar results were obtained when using BRET 1 (EGFP or YFP) or BRET 2 (GFP10). The BRET signal generated is calculated by the ratio of light emitted by the GFP partner over the light emitted by the RLuc partner. Deep Blue C coelenterazine (for BRET 2 ; BioSignal Packard Bioscience) or coelenterazine H (for BRET1; Molecular Probes) were added at a final concentration of 5 M. Signals were collected on a Packard Fusion instrument using either 410/80-nm (luciferase) and 515/30-nm (GFP) band pass filters for p␤ 2 AR-GFP10 or 470/60 (luciferase) and 550/80-nm band pass filters (GFP) for p␤ 2 AR-YFP. Whether or not BRET occurred was determined by calculating the ratio of the light passed by the 515/30 filter to that passed by the 410/80 or 550/80 to the 470/60 filter. This ratio is referred to as the BRET ratio.

Receptor Modulation
Agonist treatment of dopamine receptor expressing cells was done with 1 M dopamine or quinpirole for 5 min at 37°C, 2 days after transfection. For PTX treatment the cells were washed with serum-free ␣-MEM 1 day after transfection, and subsequently the cells were grown for another day in ␣-MEM containing 100 ng/ml PTX. After treatment, the plates on which the cells were grown were washed three times with ice-cold PBS and cells were harvested by scraping and stored at Ϫ20°C. To block trans-Golgi transport of newly synthesized material 5 h after transfection, growth media was supplemented with 10 g/ml brefeldin A. After incubation for 16 -20 h with brefeldin A cells were harvested as described above. In the case of ␤ 2 AR expressing cells, 10 M isoproterenol was also included in all solutions used for cell harvesting and solubilization.
For the GST-␤ARKct-mediated G␤␥ competition experiment we purified GST and GST-␤ARKct from Escherichia coli strain BL21-Codon-Plus-RIL containing pGEX3 plasmid constructs expressing the GST fusion proteins. The proteins were isolated using B-PER (Pierce) and glutathione-Sepharose 4B beads as recommended by the manufacturer. A solubilized membrane preparation expressing HA-tagged D 4 receptors and c-Myc-tagged Kir3.2cc were incubated with 0.2 M of the GST fusion proteins during the immunoprecipitation protocol.

Receptor Binding and cAMP Assays
Cells grown in 24-well tissue culture plates were assayed as previously described (27) to determine the number of ␤-adrenergic receptors. Binding of the hydrophobic ligand (Ϫ)-[ 125 I]cyanopindolol (CYP) and the hydrophilic ligand (Ϫ)-[ 3 H]CGP12177 (CGP) were used to determine the total number and cell surface receptors, respectively. Nonspecific binding was determined by including 10 M (Ϫ)-propranolol. Cells were also assayed for basal and agonist-stimulated levels of cAMP. Briefly, cells in 24-well tissue culture plates were incubated at 37°C for 10 min in serum-free medium containing 1 mM isobutylmethylxanthine in the absence or presence of 10 M (Ϫ)-isoproterenol, and cellular cAMP levels were measured by radioimmunoassay. Protein concentration was assayed using the Bio-Rad protein assay reagents.

Expression in Xenopus Oocytes and Electrophysiological Recording
Kir3 channel cDNAs were linearized and cRNAs were synthesized and expressed in Xenopus oocytes as described previously (16,22). cDNA constructs for various tagged Kir3 (Kir3.1, Kir3.2c, and Kir3.4) channel isoforms and the HA-D 4 , D 4 -GFP10, ␤ 2 AR-EGFP, and ␤ 2 AR-GFP10 constructs were linearized by restriction enzymes and purified using Geneclean (Bio 101, Vista, CA). Capped mRNA was made using T7 RNA polymerase and the mMessage mMachine (Ambion). Individual oocytes were injected with 5-10 ng (in 25-50 nl) of each cRNA. Recordings were made 48 h after injection at room temperature using a Geneclamp 500 amplifier (Axon Instruments). Oocytes were voltage clamped and perfused continuously with different recording solutions. Data was recorded at a holding potential of Ϫ80 mV or from ϩ80 mV to Ϫ140 mV in 20-mV steps. Drugs were added to the bath with a fast perfusion system. Data collection and analysis were performed using pCLAMP v6.0 (Axon Instruments) and Origin v4.0 (MicroCal) software.

Expression and Function of Tagged GPCRs and Effectors-
Previous studies have shown that covalently coupling either a GFP variant or RLuc to the COOH terminus of the ␤ 2 AR does not affect its affinity for ligands or the efficacy with which the receptor mediates hormonal stimulation of AC (24,25,28). The expression of ␤ 2 AR-EGFP resulted in fluorescence that was associated with internal membranes and with the plasma membrane (data not shown). Ligand binding experiments revealed that all of the detectable endogenous ␤ 2 -adrenergic receptors in HEK 293 cells are present on the cell surface. Transfected HEK 293 cells expressed 2.8 Ϯ 0.1 pmol of ␤ 2 AR-EGFP per mg of cell protein, and 43 Ϯ 7% of the expressed receptors were on the cell surface. Thus, the distribution of binding sites reflected the distribution of fluorescence, and indicated that the ␤ 2 AR-EGFP, associated with both the plasma membrane and internal membranes, was capable of binding ligand. Similar results were obtained for ␤ 2 AR-YFP (24) and ␤ 2 AR-GFP10 (data not shown), which were used in BRET assays.
Epitope tagging the NH 2 terminus of D 2L and D 4 receptors did not affect their function as determined by receptor-mediated regulation of downstream effectors (Ref. 21, and data not shown). BRET experiments require that the proteins being studied have either GFP or RLuc fused to them. Because fusing different GFP variants (EGFP, YFP, and GFP10) to the COOH terminus of the ␤ 2 AR did not oblate its function, a similar approach was taken in making a fusion protein of dopaminergic receptors and GFP. However, this rendered the dopaminergic receptor inactive as determined by its inability to activate G proteins or regulate effectors. It seemed unlikely that fusing RLuc to the COOH terminus would result in a functional dopaminergic receptor, and no further attempts were made to intercalate GFP or RLuc into the receptor at a point where it would not disrupt receptor function.
Based on biochemical and structural information, it was decided that attaching RLuc to the COOH terminus of AC would be least likely to disrupt its function. Hormone-induced accumulation of cAMP by cells expressing ␤ 2 AR-EGFP either with or without G s (i.e. G␣ s ␤ 1 ␥ 2 ) was increased more than 6-fold by co-expression of AC-RLuc indicating that the AC-RLuc fusion protein retained its ability to function in G proteinmediated signal transduction (Table I).
To simplify biochemical analysis, four different human Kir3 channel subunits were altered by the addition of epitope tags to their COOH-terminal ends (Kir.3.1-HA, Kir.3.2-Myc, Kir3.3-HA, and Kir3.4-FLAG). When expressed in COS-7 cells the individual channel subunits were readily detectable by Western analysis with antibodies against the different epitope tags (Fig. 1A). The migration pattern of the epitope-tagged channels corresponds closely to the molecular mass predicted from the cloned channels (Kir3.1, 56.1 kDa; Kir3.2c, 48.5 kDa; Kir3.3, 44 kDa; Kir3.4, 47.5 kDa). The migration of Kir3.1 at slightly higher than predicted molecular weight may be because of its reported glycosylation. We and others have found that the addition of epitope tags to Kir3 channel subunits do not affect their ability to assemble into functional channels (Ref. 29, and data not shown). In co-immunoprecipitation experiments we established that the differently tagged human Kir3 channel subunits could be immunoprecipitated with each other in all possible combinations (Fig. 1B, and data not shown). Although Kir3 subunits will associate in all possible subunit combinations ( Fig. 1 and data not shown), several of these combinations do not form functional channels (22). Kir3 channel subunits did not immunoprecipitate with each other when the receptors were expressed in separate cell populations and the two different populations were mixed just prior to solubilization (Fig. 1B). This makes it unlikely that subunit association under our experimental conditions is an experimental artifact. A recombinant expression vector encoding a fusion protein composed of the Kir3.1 channel subunit and RLuc was prepared. In experiments performed with Xenopus oocytes, coexpression of Kir3.1-RLuc with Kir3.4 resulted in the formation of functional inwardly rectifying potassium channels that can be activated by ␤ 2 AR-EGFP, and were essentially indistinguishable from wild-type Kir3.1/Kir3.4 channels under twoelectrode voltage clamp (data not shown). Previous studies have also demonstrated that COOH-terminal fusions of GFP to Kir3.1 did not affect channel function (30). These experiments, which demonstrated that the various receptors, Kir3, and adenylyl cyclase fusion proteins are functional, allowed us to proceed with studies of the physical interactions between these receptors and their effectors.
Co-immunoprecipitation of GPCRs with Their Cognate Effectors-Immunoprecipitation of epitope-tagged dopaminergic receptors or ␤ 2 AR-GFP fusion proteins with the appropriate antibody resulted in the precipitation of co-expressing Kir3 channel subunits (Figs. 2, 3, and 4). Reversing the approach and immunoprecipitating the Kir3 channel subunits resulted in precipitation of the co-expressed receptors (Fig. 2, A and D). The D 4 receptor migrates in the Western analysis close to its predicted molecular weight of 44,000. As expected, the ␤ 2 AR-GFP fusion protein runs at a higher molecular weight (Mw) than the ␤ 2 AR (predicted M r ␤ 2 AR, 46,000), however, it migrates at a lower M r than predicted for the fusion protein (His-␤ 2 AR-GFP, 76,500). Whether this constitutes a gel running artifact because of structural features of the fusion protein is unknown. Considering that both the His tag and GFP portion are still functional for the modified ␤ 2 AR it is unlikely because of proteolytic digestion. The specificity of the receptor/channel interaction is highlighted by the observation that separate expression of the receptors and the Kir3 channels and subsequent solubilization, mixing of membrane preparations, and receptor immunoprecipitation did not result in co-association of Kir3 ( Fig. 2A). Furthermore, when co-expressed, FLAG-D 4 receptors did not immunoprecipitate a Myc epitope-tagged version of the closely related inwardly rectifying potassium channel Kir2.1 (Kir2.1-Myc), suggesting that these receptors do not indiscriminately associate with all Kir channels family members (Fig. 2C).
Experiments were also done to determine whether immunoprecipitation of ␤ 2 AR-GFP would result in the co-precipitation of AC-RLuc. The membranes of cells expressing ␤ 2 AR-GFP together with either soluble RLuc or AC-RLuc were dissolved in various detergent-containing solutions and immunoprecipitates were prepared using antibodies against GFP. RLuc was co-precipitated with ␤ 2 AR-GFP from cells co-expressing AC-RLuc but not from cells co-expressing RLuc if the cell membranes were dissolved in a solution containing 1% digitonin (Table II). When other detergents (i.e. n-dodecyl-␤-D-maltoside, Lubrol PX, sodium cholate, or a mixture of Triton X-100 and sodium cholate) were used to dissolve cell membranes no coprecipitation of AC-RLuc with the ␤ 2 AR-GFP was observed. These results are consistent with previous studies showing that preservation of lipophilic protein complexes is largely dependent on the type of detergent used to dissolve the membranes.
Experiments were then performed to verify that these interactions also occur in native tissue. Dopamine D 2 receptors could be co-immunoprecipitated from brain (striatum) with Kir3.2 (Fig. 3A, right panel). In negative control experiments, using a nonspecific IgG antiserum, D 2 receptors were not immunoprecipitated. The specificity of the anti-D 2 antibody is shown by using Western analysis of HEK 293 cells transfected with HA-tagged D 2 receptors or vector as negative control (Fig. 3A, left panels). ␤ 2 AR were immunoprecipitated from mouse heart or brain, and samples were blotted for ␤ 2 AR (to verify immunoprecipitation, Fig. 3B, left panel), adenylyl cyclase V/VI (expressed ubiquitously), and Kir3.2. In both heart and brain extracts, the ␤ 2 AR could co-immunoprecipitate adenylyl cyclase (Fig. 3B, middle panel). In brain tissue, Kir3.2 was also co-immunoprecipitated with ␤ 2 AR (Fig. 3B, right panel). As a negative control, we attempted to co-immunoprecipitate Kv1.5, another membrane protein highly expressed in both tissues. The anti-␤ 2 AR antibodies could not immunoprecipitate , no matter which partner is immunoprecipitated (not shown). Kir channels that were expressed separately and mixed during protein extraction ( †) showed no co-immunoprecipitation. There was no cross-reactivity between the different antibodies and the epitopes employed in this study (not shown). The figure is representative of at least three separate experiments. The different antisera used for immunoprecipitation (IP) and immunoblotting (IB) are indicated. Kv1.5 from either tissue (data not shown) demonstrating the specificity of the interaction in vivo. In the Western analysis the migration of the native receptors and heterologous expressed D 2 receptors is at slightly higher than predicted molecular weights, which is in concordance with the reported glycosylation of these proteins.
Neither Agonist Treatment nor ADP-ribosylation of G i by PTX Affects Co-precipitation of GPCRs with Their Effectors-To determine whether agonist-mediated activation of signal transduction affects co-precipitation of receptors and effectors, cells expressing Kir3.2-Myc channel subunits and either epitope-tagged dopamine D 2L or D 4 receptors were incubated with 1 M dopamine or quinpirole. Similarly, we co-expressed epitope-tagged ␤ 2 AR with Kir3.1 or Kir3.2c and explored whether stimulation with 10 M isoproterenol would affect receptor-channel co-immunoprecipitation. Treatment with either agonist for 5 min is sufficient for maximal receptor-mediated activation of its effectors. The receptors were subsequently immunoprecipitated from agonist-treated cells and the amount of co-precipitated Kir3.2c was compared with that obtained from unstimulated cells. Coprecipitation of Kir3 channel subunits with epitope-tagged D 2L and D 4 receptors (Fig. 4) or ␤ 2 AR (Fig. 5) was largely unaffected by receptor agonists, although subtle changes cannot be excluded from these experiments. The GFP fusion constructs of ␤ 2 AR that were not His-tagged served as negative controls showing the specificity of the protocol used to demonstrate co-immunoprecipitation of Kir3.1 with Histagged ␤ 2 AR-GFP (Fig. 5). His-␤ 2 AR-GFP could co-immunoprecipitate Kir3.2 (D). In a triple co-transfection protocol in which His-␤ 2 AR-GFP, Kir3.1, and Kir3.4-FLAG were co-expressed we could effectively co-immunoprecipitate Kir3.4-FLAG with His-␤ 2 AR-GFP (D). In reverse co-immunoprecipitation protocols His-␤ 2 AR-GFP could also be immunoprecipitated with the Kir3.4-FLAG and Kir3.1-HA channel subunits, whereas in cells transfected with untagged versions of these channels the receptor could not be immunoprecipitated (D). The figure is representative of at least three separate experiments. Tests using nonspecific antibodies as controls were negative for co-immunoprecipitation (not shown). The different antisera used for immunoprecipitation (IP) and immunoblotting (IB) are indicated.
Because the activation of dopamine D 2L and D 4 receptors and consequent opening of Kir3.2c channels is sensitive to ADP-ribosylation of G␣ i/o by PTX, we analyzed whether treatment with this toxin altered receptor/channel interactions. Although the conditions employed in this study block dopamine D 2L and D 4 receptor mediated-signal transduction (Ref. 21, and data not shown), it did not alter the amount of dopamine D 2L and D 4 receptor that was co-precipitated with Kir3.2c channels compared with nontreated cells (Fig. 4). The lack of PTX effect on co-precipitation of Kir3.2c with dopamine receptors indicates that a functional G␣ i/o is not required for the interaction between dopaminergic receptors and Kir3 channel subunits. This observation is also consistent with the lack of an effect of agonist treatment on receptor-channel co-immunoprecipitation. This is further supported by the observation that coexpression of dominant-negative G␣ i constructs (23), which can block GPCR-mediated activation of Kir3 channels, did not markedly change the dopamine receptor-Kir3.2c complex formation either (Fig. 6A). These dominant-negative G␣ i constructs consist of a short carboxyl-terminal fragment of the G protein that competes with the G protein for interactions with the receptor and thus blocks receptor-mediated activation of Kir3 channels.
Evidence for Assembly of GPCR-Effector Complexes Prior to Their Incorporation into the Plasma Membrane and a Potential Role for G␤␥ in the Assembly Process-It has been well described that the activation of Kir3 channels is G␤␥-dependent (31). It is also known that expression of the COOH-terminal domain of G protein-coupled receptor kinase (␤ARKct) binds G␤␥ and interferes with channel activation (32). However, it is not known whether the ␤ARKct can interfere with formation of the receptor-channel complex. Co-expression of ␤ARKct with Kir3.2c channels and D 4 receptors results in a dramatic loss of receptor-channel complex formation (Fig. 6A). Considering that modulation and disruption of GPCR signaling by PTX and agonist treatment did not affect markedly the co-immunoprecipitation of the receptor-channel complex we postulated that G␤␥ plays an early role in the complex formation. To determine whether the receptor-channel complex is formed during biosynthesis (rather than at the plasma membrane) we incubated cells with brefeldin A shortly after transfection to block transport of newly synthesized proteins from Golgi to trans-Golgi. This treatment did not block complex formation as accessed by co-precipitation of epitope-tagged Kir3.2c channel subunits with the epitope-tagged D 2L and D 4 receptor, indicating that the receptor-channel complex forms prior to its appearance at the cell surface (Fig. 6A). Furthermore, brefeldin A treatment did not prevent the ␤ARKct-mediated block of the receptorchannel formation (Fig. 6A), suggesting that the G␤␥ subunit plays a role in early complex formation (Fig. 6A). To determine whether the G␤␥ subunits are critical for the maintenance of the dopamine receptor-Kir3.2c complex we added an excess of purified GST-␤ARKct to the solubilized membrane preparation. The addition of the ␤ARKct did not cause the dissociation FIG. 3. Co-immunoprecipitation of catecholamine receptors with Kir3 channel subunits and adenylyl cyclase in native tissue. A, dopamine D 2 receptors were co-immunoprecipitated from brain lysates (striatum) using anti-Kir3.2. Immunoprecipitation with nonspecific IgG immune serum did not result in the detection of D 2 receptors with anti-D 2 antibodies. The selective detection of the D 2 receptors with the anti-D 2 antiserum is shown by using HEK 293 cells transfected with either HA-tagged D 2 receptors (HA-D2) or pCDNA3 vector (-ctr). B, mouse heart and brain lysates were incubated with anti-␤ 2 AR. Immunoprecipitation of ␤ 2 AR is shown in the right panel. The slight shift in molecular weight of the receptor band in the immunoprecipitation was seen consistently. Co-immunoprecipitation of either adenylyl cyclase V/VI (middle panel) or Kir3.2 subunits (right panel) with ␤ 2 AR (as indicated by the closed arrows in the blots) is shown. The anti-␤ 2 AR antibodies could not co-immunoprecipitate Kv1.5, which is expressed in both tissues (not shown). The figure is representative of at least three separate experiments. Open arrows represent antibody bands. The different antisera used for immunoprecipitation (IP) and immunoblotting (IB) are indicated. Direct probing of lysates without immunoprecipitation is indicated as none.

TABLE II
Luciferase activity co-immunoprecipitates with ␤ 2 AR-EGFP when co-expressed with AC-RLuc but not RLuc Membranes from HEK 293 cells expressing the indicated protein were dissolved with the indicated buffers and the soluble proteins were immunoprecipitated with antibodies to GFP as described under "Experimental Procedures." The immunoprecipitates were assayed for RLuc activity. Data are the mean Ϯ S.D. for three independent experiments.

Transfection
Luciferase in pellet % ␤ 2 AR-EGFP/RLuc 2 Ϯ 1 ␤ 2 AR-EGFP/AC-RLuc 19 Ϯ 4 of the dopamine receptor-Kir3.2c complex (Fig. 6B), suggesting that the ␤ARKct cannot destabilize the receptor-effector complex once it is formed. Of relevance in this regard is our observation that D 4 receptors and Kir3 channels co-precipitated under the conditions used for our experiments were not associated with detectable amounts of G␤␥ (Fig. 6B). Taken together these data strongly support the view that G␤␥ plays a role in early complex formation rather than maintenance of the complex.

BRET Provides Evidence for Direct Interaction of GPCRs with Their Cognate Effectors in Vivo That Survives Agonist
Occupancy of the Receptor-BRET occurs if the distance between donor (RLuc) and acceptor (GFP) is less than 100 Å. By and large, the distance between two individual proteins is less that 100 Å only if they are directly associated with each other. Furthermore, BRET can be assayed in living cells making it useful for determining if stable protein/protein interactions occur in vivo. Being able to attach the GFP or RLuc to receptors and effectors without disrupting their ability to engage in G protein-mediated signal transduction paved the way for the use of BRET to determine whether GPCRs and their cognate effectors form stable signaling complexes in vivo.
We have previously demonstrated that human Kir3.1 requires co-expression of the Kir3.2c or Kir3.4 subunits to become a functional channel (22). BRET was observed between Kir3.1-RLuc and the ␤ 2 AR-GFP10 when cells co-expressed either Kir3.2c or Kir3.4 (Fig. 7A) and there were no changes in the BRET ratio in response to agonist, again suggesting that receptor and effector are associated with one another regardless of whether or not signal transduction is activated (Fig. 7B). In the absence of co-expressed Kir3.2c or Kir3.4 BRET between Kir3.1-RLuc and the ␤ 2 AR-GFP10 is markedly reduced, suggesting that the presence of a functional channel improves or stabilizes the interaction with the receptor (Fig. 7A). BRET also occurred when ␤ 2 AR-GFP10 and AC-RLuc were co-expressed regardless of whether or not the cells were treated with agonist (Fig. 7, A and B), suggesting that the receptor and effector interact directly with one another in living cells, and that the interaction is not dependent upon activation of the signal transduction pathway. It has been demonstrated that the ␤ 2 AR forms oligomers in vivo by showing that BRET occurs when cells co-express ␤ 2 AR-YFP and ␤ 2 AR-Rluc (24). We used co-expression of ␤ 2 AR-GFP10 and ␤ 2 AR-RLuc as a positive control for our experiments and observed BRET as expected (Fig. 7A). Finally, there was no BRET when the functionally inactive D 4 -GFP10 fusion protein was co-expressed with Kir3.1-RLuc and Kir3.4, a subunit combination that produces functional heterotetrameric Kir3 channels (data not shown). The BRET ratio was the same for cells co-expressing soluble GFP10 and effector-RLuc fusion proteins or ␤ 2 AR-GFP10 and soluble RLuc, as it was for cells expressing RLuc alone or RLuc fusion proteins alone, indicating the specificity of the assay and that BRET did not occur between donor and acceptor if the proteins did not associate. DISCUSSION Co-immunoprecipitation experiments were used to demonstrate that three different GPCRs (D 2L and D 4 dopaminergic receptors and the ␤ 2 AR) stably interact with their downstream effectors including several inwardly rectifying potassium channel (Kir3) subunits and adenylyl cyclase. Co-precipitation of effectors with their receptors occurred regardless of whether or not signal transduction was activated. Co-precipitation of effectors with receptors suggests that these complexes also exist in living cells, which was confirmed using BRET. As with the immunoprecipitable complexes, the interaction of receptors and effectors in living cells was stable in the absence as well as the presence of agonists. With the exception of D 4 -GFP, all of our fusion proteins were functional and had properties similar to their untagged counterparts. Expressing Kir3 channels and dopaminergic receptors in separate cells and mixing the detergent-soluble material from these cells did not result in the formation of complexes containing receptors and effectors, suggesting that the assembly of these complexes is a biosynthetic event. This observation is further supported by evidence that the trans-Golgi transport blocker brefeldin A is ineffective in preventing receptor-channel complex formation, which supports the view that the association of a receptor with its effector occurs prior to their appearance at the cell surface.
Receptor-effector complexes were also detected in native tis- sues. This indicates that the receptor/effector interactions are of physiological relevance and not merely a consequence of the conditions of the heterologous expression systems. Specifically, we have provided evidence for the existence of dopamine D 2 receptor-Kir3.2 complexes in mouse brain (striatum), and ␤ 2 AR-Kir3.2 and adenylyl cyclase V/VI complexes in mouse heart and brain tissues.
The observation that the receptors can co-immunoprecipitate with nonfunctional homomeric Kir3.1 or Kir3.3 channels suggests that functionality of the receptor-channel complex may not be a prerequisite for assembly of this complex. This is also supported by data in which D 4 mutants unable to promote G protein GDP-GTP exchange can co-immunoprecipitate Kir3.2 (data not shown). Furthermore, homomeric Kir3.1, which forms a nonfunctional channel, can mediate a modest BRET signal when expressed with ␤ 2 AR. In addition, a functional transducer did not appear to be necessary for complex formation because neither PTX treatment nor expression of dominant negative G␣ i constructs prevented co-precipitation of dopaminergic receptors and Kir3 channels. However, the experiments do not exclude subtle changes in the efficacy of the interaction. Similarly, G␤␥ does not appear to be necessary for maintenance of the mature signaling complex. Note, however, that whereas we were unable to implicate G␤␥ in the stability of a receptor-effector complex once it was formed, we did find evidence of a role for the G protein heterodimer in facilitating an initial interaction between receptor and effector. The apparent exclusive role of G␤␥ in early complex formation, rather than maintenance of the complex, is consistent with our inability to significantly disrupt the receptor-effector complex by  (24), thus cells expressing ␤ 2 AR-RLuc and ␤ 2 AR-GFP were included in experiments as a positive control. Data are presented as mean Ϯ S.E. for three (␤ 2 AR-GFP10/AC-RLuc) and five (␤ 2 AR-GFP10/Kir3-RLuc) experiments conducted in triplicate in all cases. Asterisks denote statistical difference between the indicated conditions and ␤ 2 AR-GFP and RLuc alone (p Ͻ 0.05). B, in none of the cases did stimulation with 1 M isoproterenol have an effect on the BRET ratio. Data are normalized to the signal in the absence of ligand in each case. Cells were harvested 48 h post-transfection, counted, and transferred to 96-well plates (100,000 cells/wells). The energy transfer reaction was initiated by adding 5 M Deep Blue C coelenterazine or coelantrazine H to each well and the BRET assessed in a fusion (Packard) microplate reader with the filter settings described under "Experimental Procedures." Kir3.1-Luc and AC-RLuc are COOH-terminal fusion proteins of Kir3.1 and type II adenylyl cyclase with luciferase. ␤ 2 AR-GFP is a COOH-terminal fusion protein of the ␤ 2 -adrenergic receptor with GFP10. manipulations that affect the functional activity of the mature complex. Nevertheless, some of the data support that functionality may be a contributing factor with regards to the efficacy of complex formation. This is revealed by the observations that D 4 -GFP10, which does not functionally couple to its effectors, does not mediate BRET with Kir3.1-Kir3.4 channel complexes. Furthermore, the co-expression of Kir3.1 with Kir3.4 or Kir3.2, rather than Kir3.1 by itself, significantly enhanced the BRET signal with ␤ 2 AR. Similarly, Kir2.1 channels, which are not direct effectors of dopamine receptors, did not form complexes with D 4 receptors. Whether these differences in receptor-channel complex formation are related to a more efficacious assembly of the complex or merely represent improved expression because of either increased stability or decreased degradation of the constituents of the complex is as of yet unknown.
The co-immunoprecipitation protocols for the receptor-channel complexes using "strong" detergents (RIPA) in combination with the BRET data support the view that the receptor-channel complex is mediated by direct protein/protein interactions. However, it cannot be excluded that the co-immunoprecipitation and BRET of the receptor-effector complexes is a resultant of the co-localization of the signaling partners into lipid microdomains or rafts (33).
How G␤␥ is involved in receptor-channel complex formation during early synthesis is of yet unknown. G proteins have been implicated in the maintenance of the integrity of the Golgi system and regulation of trafficking in this system (34,(35)(36)(37). Because the inhibitory effects of ␤ARKct on complex formation were brefeldin-insensitive an indirect role of G␤␥ via interference with the Golgi system seems unlikely, supporting a more direct role in complex formation. However, it has also been reported that G␤␥ heterodimers are assembled in the cytosol and that their final targeting to the plasma membrane is not affected by brefeldin (38). These different observations are difficult at present to reconcile with a mechanism on how G␤␥ plays a role in receptor-channel complex formation.
The existence of stable receptor/G protein interactions as well as G protein-effector complexes have been reported by many investigators (see Ref. 2 for review). Our findings extend this paradigm to include direct interactions between receptors and their associated effector molecules. The concept of a signaling complex containing many of the components involved in G protein-mediated signal transduction is supported by data demonstrating that the ␤ 2 AR can be co-precipitated from rat brain with one of its effectors, the voltage-gated L-type calcium channel Ca v 1.2 (8), and that the precipitate includes heterotrimeric G proteins, AC, protein kinase A, and phosphatase PP2A. Indeed, recent work has shown that molecular scaffolds such as protein kinase A-binding proteins in addition to scaffolding protein kinase A may also function as anchors for larger signaling complexes (8,39,40). However, our data also suggest that the G proteins may not be essential for maintenance of the complex.
Neurons or cardiomyocytes contain many different receptors, G proteins, and effectors that would have to transiently associate, dissociate, and then subsequently re-associate after each activation cycle. Such a scheme is difficult to reconcile with our current notions of signaling specificity in these cells. We have demonstrated that stable signaling complexes exist between three different GPCRs and their downstream effectors, suggesting that assembly of these signaling complexes is likely to be a biosynthetic event, rather than ones directed by agonist at the cell surface. The exact constituents of each complex may depend on the specific receptor, cell and/or tissue type studied. Preassembly of signaling components into a specific subcellular location and/or macromolecular complex may provide a means of insuring specificity and rapidity of response (41).