Abstract
Direct evidence is lacking to show whether the γ-aminobutyric acid (GABA)B gb1-gb2 heterodimer is the signaling form of the receptor. In this study, we tested whether gb1a or gb2 subunits when coexpressed with truncated receptors or metabotropic glutamate receptor mGluR4 could form functional GABA receptors. Coexpression of the ligand binding N-terminal domain of gb1a or the C-terminal portion of gb1a composing the seven-transmembrane segments and intracellular loops with gb2 could not reconstitute functional receptors. We next examined whether mGluR4, which forms homodimers and is structurally related to GABAB, could act as a surrogate coreceptor for gb1 or gb2. The coexpression of mGluR4 and gb1a led to the expression of gb1a monomers on cell surface membranes as determined by immunoblot analysis and flow cytometry. However, mGluR4-gb1a heterodimers were not formed, and membrane-expressed gb1a monomers were not functionally coupled to adenylyl cyclase in human embryonic kidney 293 cells or activated inwardly rectifying potassium (Kir) channels in Xenopusoocytes. Similarly, the coexpression of mGluR4 and gb2 led to nonfunctional GABA receptors. GABA-activated distal signaling events resulted only after the coexpression and heterodimerization of gb1 and gb2. Taken together with the truncated receptor studies, the data suggest that a high degree of structural specificity is required to form the functional GABAB receptor that is a gb1-gb2 heterodimer.
γ-Aminobutyric acid (GABA) is the most widely distributed inhibitory amino acid neurotransmitter in the vertebrate central nervous system. GABA activities are mediated by three types of GABA receptors, which are classified according to biochemical and pharmacological criteria into ionotropic GABAA/GABACreceptors and metabotropic GABAB receptors (seeMody et al., 1994, for a review).
GABAB receptors, which were first pharmacologically distinguished by Hill and Bowery (1981), act presynaptically and postsynaptically based on their anatomical location and physiological functions. GABAB receptors couple through G proteins to neuronal K+ and Ca2+ channels and lead to the inhibition of adenylyl cyclase. Physiologically, GABABreceptors have been implicated in synaptic inhibition, hippocampal long-term potentiation, short-wave sleep, muscle relaxation, and antinociception, which makes them attractive therapeutic targets (seeBowery and Enna, 2000, for a review).
To date, molecular cloning has identified two main GABAB subtypes termed gb1 (GABABR1 receptor) and gb2 (GABABR2 receptor), which arise from distinct genes (Kaupmann et al., 1997, 1998a; Jones et al., 1998; White et al., 1998; Kuner et al., 1999; Ng et al., 1999a,b; Martin et al., 1999). The human gb1 receptor gene encodes at least three forms of the receptor, termed gb1a, gb1b, and gb1c (gb1c reported only as GenBank accession number AJ012187 and is structurally distinct from the rat gb1c; Isomoto et al., 1998; Pfaff et al., 1999). The human gb2 receptor gene encodes a single form of the receptor. Human gb1 receptor isoforms differ in their extracellular N-terminal domains that are suggested to be responsible for ligand binding (Kaupmann et al., 1998b). GABAB receptors are structurally related to metabotropic glutamate receptors (mGluRs) and together with calcium-sensing receptors belong to the family 3 (C) G protein-coupled receptors (GPCRs).
Unlike other GPCRs, recombinant gb1 and gb2 receptors are functionally inactive when expressed individually (Jones et al., 1998; White et al., 1998; Ng et al., 1999b). It is generally accepted that the functional GABAB receptor results from the coexpression and translocation of the gb1 subunit to the cell surface by the gb2 subunit as a heterodimer (Jones et al., 1998; White et al., 1998; Kaupmann et al., 1998a; Kuner et al., 1999; Ng et al., 1999b). Yeast two-hybrid screening showed that a coiled-coil motif in the carboxyl tails of gb1 and gb2 receptors likely mediate gb1-gb2 heterodimerization (White et al., 1998; Kuner et al., 1999). However, several studies have reported that gb1 (Kaupmann et al., 1997, 1998b) and gb2 (Kuner et al., 1999;Martin et al., 1999) expressed alone can activate inwardly rectifying potassium channel (Kir) channels and/or inhibit cAMP production. To clarify the structural requirements necessary for the expression of the functional GABAB receptor, we expressed full-length and truncated gb1a and gb2 receptors, alone and together, and examined their ability to form functional receptors. To further test the specificity of GABAB receptor heterodimerization, we also examined whether coexpression of gb1a or gb2 with the structurally related mGluR4 receptor was sufficient to form a functional GABA receptor. Our data show that 1) coexpression of gb1a and gb2 leads to GABA-mediated activation of K+ currents and inhibition of cAMP production, 2) coexpression of truncated gb1a receptors with gb2 does not reconstitute functional GABA receptors, 3) coexpression of gb1a and mGluR4 leads to expression of gb1a monomers on the cell surface, 4) coexpression of gb1a and mGluR4 does not result in mGluR4-gb1a heterodimers, and 5) coexpression of gb1a or gb2 with mGluR4 does not result in functional GABA receptors. We conclude that the gb1-gb2 heterodimer is the functional GABAB receptor species.
Materials and Methods
Receptor Expression Constructs.
The recombinant murine gb1a (herein referred to as gb1a) receptor (GenBank accession numberAF114168) exhibits 98.5% amino acid identity to the human gb1a receptor (GenBank accession number AJ225028) and was used as a model for gb1a receptors. The gb1a cDNA was constructed from two expressed sequence tags (IMAGE Consortium clone identification numbers 472408 and 319196). The partial cDNAs were assembled by polymerase chain reaction (PCR) using the following oligonucleotides: 472408 sense, 5′-GC GAATTC GGTACC ATG CTG CTG CTG CTG CTG GTG CCT-3′; 472408 antisense, 5′-GG GAATTC TGG ATA TAA CGA GCG TGG GAG TTG TAG ATG TTA AA-3′; 319196 sense, 5′-CCA GAATTC CCA GCC CAA CCT GAA CAA TC-3′; and 319196 antisense, 5′-CG GCGGCCGC TCA CTT GTA AAG CAA ATG TA-3′, which amplified two fragments corresponding to the 5′ 2100 bp and 3′ 1000 bp of the gb1a receptor cDNA coding region. PCR products were cloned into the TA-Cloning vector pCRII-TOPO vector (InVitrogen, San Diego, CA) according to the manufacturer's directions. The EcoRI fragment from PCR cloning using 472408 primers and theEcoRI/NotI product from PCR cloning using 319196 primers were ligated as a 2903-bp open reading frame into pCINeo (Stratagene, La Jolla, CA) or pcDNA3.1 (InVitrogen) vector (Stratagene).
The human gb2 receptor DNA receptor (GenBank accession number AF069755) was subcloned into vector pIRES-puromycin (Clontech, Palo Alto, CA) and used to transfect human embryonic kidney (HEK) 293 cells (Aurora Bioscience, La Jolla, CA). To monitor the transient expression of the gb2 receptor, an N-terminal FLAG-tagged gb2/pcDNA3.1 construct encoding a modified influenza hemagglutinin signal sequence (MKTIIALSYIFCLVFA) followed by an antigenic FLAG (DYKDDDDK) epitope was generated by PCR methods (Ng et al., 1999a).
The N-terminal fragment of the gb1a receptor (N-gb1a), composed of amino acid positions 1 to 625, was generated by PCR. The coding sequence of the N-terminal fragment was amplified by using primer pairs NFP-CJ7843F139 (5′-ACC ACT GCT AGC ACC GCC ATG CTG CTG CTG CTG CTT CTG C-3′) and NRP-CJ7844 (3′-GG GTG CGA GCA ATA TAG GTC TTA AGG GTC GGC CGC CGG CGT CAC CA-5′). Similarly, the C-terminal fragment of the gb1a receptor (C-gb1a), composed of an initiating methionine and amino acid positions 588 to 942, was generated by PCR using primer pairs CFP-CJ7845 (5′-ACC ACT GCT AGC ACC GCC ATG CAG AAA CTC TTT ATC TCC GTC TCA GTT CTC TCC AGC-3′) and CRP-CJ7846 (3′-CAG CTC ATG TAA ACG AAA TGT TCA CTC GCC GGC CGC CGG CGT CAC CA-5′). The N-gb1a and C-gb1a PCR products, flanked by NheI and NotI sites, were then digested and subcloned into the NheI/NotI site of pcDNA3.1, and the DNA sequences were confirmed.
The construction and characterization of the c-myc-mGluR4 receptor DNA expression vector were reported previously (Han and Hampson, 1999), and the vector was a generous gift from Dr. David Hampson (University of Toronto).
In Vitro Receptor Expression.
In vitro coupled transcription/translation reactions were performed in the presence of [35S]methionine in the TNT Coupled Reticulocyte Lysate system (Promega, Madison, WI) using pcDNA3.1 plasmids containing the gb1a, N-gb1a, or C-gb1a DNAs. Translation products were analyzed by electrophoresis on 8 to 16% Tris-glycine SDS gradient gels (Novex precast gel system, San Diego, CA) under denaturing and reducing conditions. Gels were fixed, dried, and exposed to Kodak X-AR film at −70°C for 4 to 24 h.
Receptor Expression in Whole Cells.
Receptor DNAs (6 μg total DNA/1.2 × 106 cells) were transiently transfected into COS-7 (American Type Culture Collection, Rockville, MD) or HEK 293 (Aurora Bioscience) cells using 36 μl of LipoFECTAMINE reagent (Life Technologies, Ontario, Canada) according to the manufacturer's instructions. Transient gb1 and gb2 coexpression studies were performed using 1:1 ratio of receptor DNAs.
Stable gb2 receptor-expressing cells were selected by growth in puromycin (5 μg/ml) containing medium and limiting dilution. The gb2 receptor RNA expression levels in clones were determined by dot-blot analysis. Briefly, RNA was prepared using TRIZOL reagent (Life Technologies) and 10 μg of total RNA spotted by vacuum with a dot-blot apparatus onto BrightStar-Plus nylon membranes (Ambion, Austin, TX). The blot was hybridized with a32P-labeled DNA fragment encoding the full-length gb2 receptor (106 cpm/ml) in Zip-Hyb solution (Ambion) for 10 h at 50°C and washed at 55°C for 90 min in high-stringency wash buffer. The gb2 receptor-expressing clones were analyzed for cell surface receptor staining by flow cytometry as described later.
Membranes and Immunoprecipitation.
Cells were washed twice with cold PBS, collected by centrifugation at 100g for 7 min, and resuspended in 10 ml of Buffer A [5 mM Tris-HCl, 2 mM EDTA containing 1× protease inhibitor cocktail Complete tablets (Boehringer-Mannheim, Indianapolis, IN), pH 7.4, at 4°C]. Cells were disrupted by Polytron homogenization and centrifuged at 100gfor 7 min to pellet unbroken cells and nuclei, and the supernatant (S1) was collected. The S1 supernatant was centrifuged at 40,000gfor 20 min to recover the crude membrane (P2) fraction. Membranes were then washed once with Buffer A, centrifuged (27,000g for 20 min) and resuspended in Buffer A, and stored at −80°C. The supernatant S1 was centrifuged at 100,000g for 30 min to recover total cellular membranes that were washed and stored in Buffer A. Protein content was determined using the Bio-Rad Protein Assay Kit (Ontario, Canada) according to the manufacturer's instructions.
For the immunoprecipitation experiments, 100,000g membranes were solubilized with digitonin, and samples were immunoprecipitated with a mouse anti-FLAG M2 (Kodak IBI, New Haven, CT) or anti-c-myc 9E10 antibody (Santa Cruz Biochemicals, Santa Cruz, CA) affinity resin essentially as previously described (Ng et al., 1994). Immunoprecipitates were washed and subjected to immunoblot analysis as described later.
Immunoblot Analysis.
Crude membranes (40,000g) were solubilized in SDS sample buffer consisting of 50 mM Tris-HCl, pH 6.5, 10% SDS, 10% glycerol, and 0.003% bromophenol blue with 10% 2-mercaptoethanol and separated on 8 to 16% Tris-glycine SDS gradient gels. The full-length gb1a receptor and N-gb1a truncated receptor were detected using affinity-purified rabbit polyclonal antibodies 1713.1 raised against the peptide:acetyl-DVNSRRDILPDYELKLC-amide and antibody 1713.2 raised against the peptide:acetyl-CATLHNPTRVKLFEK-amide in the N-terminal tail of the gb1 receptor. gb2 receptors were detected using affinity-purified rabbit polyclonal antibody 1630.1 raised against the peptide:acetyl-CSGKTPQQYEREYNNK-amide and antibody 1630.2 raised against the peptide:acetyl-QDVQRFSEVRNDLTC-amide of the gb2 receptor. The characterization of gb1 and gb2 antibodies have been reported elsewhere (Belley et al., 1999; Ng et al., 1999b). Specific immunoreactivity was revealed by secondary antibody coupled to horseradish peroxidase and chemiluminescence detection using the Renaissance Western Blot Chemiluminescence Reagent Plus kit (New England Nuclear, Boston, MA). The whole-cell expression of the C-gb1a truncated receptor was detected using a GABABreceptor antibody AB1531 (Chemicon Int, Ontario, Canada) raised against the peptide:acetyl-PSEPPDRLSCDGSRVHLLYK-amide in the C-terminal tail of the gb1 receptor. Specific C-gb1a immunoreactivity was revealed by a secondary antibody coupled to alkaline phosphatase and detected using the Immuno-Blot Alkaline Phosphatase Assay Kit (Bio-Rad).
Densitometry.
Determinations of immunoreactive band intensity were made by scanning on a GS-719 calibrated imaging densitometer (Bio-Rad) and analyzed using ImageQuant 5.0 software (Molecular Devices, Sunnyvale, CA). In the immunoblot shown (see Fig.3), a rectangular region was defined around the ∼130-kDa band in the gb1a/mGluR4 and gb1a/FLAG-gb2 lanes and the corresponding region in gb1 and pcDNA3.1 lanes. The pixel/density of the defined region was determined, and the background, as defined by pcDNA3.1, was subtracted from all subsequent band determinations. To ensure analysis in the linear range, X-ray films were exposed to immunoblots of 25- to 50 μg of protein for various times.
Receptor Binding Assays.
The synthesis of the [125I]CGP71872 ([125I]3-(1-(R)-(3-((4-azido-2-hydroxy-5-iodobenzoylamino)pentyl) hydroxyphosphoryl)-2-(S)-hydroxypropylamino)ethyl)benzoic acid) photoaffinity label and conditions for receptor binding have been described elsewhere (Belley et al., 1999).
Functional Assays.
cAMP determinations were made using a scintillation proximity assay kit (Amersham, Ontario, Canada). Briefly, HEK 293 cells were washed and detached, and 77,000 to 100,000 cells/well were resuspended in Hanks' balanced salt solution containing 25 mM HEPES, pH 7.4, 100 μM 4-(3-butoxy-4-methoxybenzyl)-2-imadazolidinone (Ro 20-1724; BIOMOL Research Laboratories, Plymouth Meeting, PA) and incubated for 20 min at 37°C. Then, 2 μM forskolin and ligands (10−9 to 10−3 M) were added and incubated for 30 min at 37°C. Cells were lysed by boiling, and cAMP levels were determined by scintillation proximity assay according to the manufacturer's instructions.
Pigment aggregation assays in Xenopus laevis melanophores were performed as previously described (Ng et al., 1999b). To monitor the efficiency of transfection, two internal control GPCRs were used (pcDNA1amp-cannabinoid 2 and pcDNA3-thromboxane A2). For Gi-coupled responses (pigment aggregation), cells were preincubated with fibroblast-conditioned growth medium to induce pigment dispersion, and drugs were added. Absorbance readings at 600 nm were measured using a Bio-Tek Elx800 Microplate reader (ESBE Scientific) before (Ai) and after (Af) a 1.5-h incubation with the ligands.
Data from the cAMP and melanophore assays were analyzed by nonlinear least-squares regression using the computer-fitting program Prism version 2.01 (GraphPad, San Diego, CA).
Voltage-clamp experiments in X. laevis oocytes were performed as described in detail elsewhere (Ng et al., 1999b). X. laevis oocytes were isolated and injected with 5 to 10 ng (in 25–50 nl) of Kir3.1 and Kir3.2 and different combinations of receptor mRNAs. The standard recording solution was KD-98 (98 mM KCl, 1 mM MgCl2, 5 mM K-HEPES, pH 7.5) unless otherwise stated. Data collection and analysis were performed using pCLAMP v6.0 (Axon Instruments, Burlingame, CA) and Origin version 4.0 (MicroCal Software, Northampton, MA) software. For subtraction of endogenous and leak currents, records were obtained in ND-96 (96 mM NaCl, 2 mM KCl, 1 mM MgCl2, 5 mM Na-HEPES, pH 7.5), and these were subtracted from recordings in KD-98 before further analysis.
Flow Cytometry.
Analysis was performed using live intact cells, which were incubated with primary antibodies for 1 h in Hanks' balanced salt solution, followed by incubation with secondary antibody conjugates under similar conditions. Goat anti-rabbit antibodies coupled with Alexa-488 (Molecular Probes, OR) were used to detect rabbit gb1 or gb2 antibodies. We analyzed 10,000 cells per condition with a Becton Dickinson (San Jose, CA) FACSVantage SE flow cytometer configured to detect fluorescein isothiocyanate fluorescence.
Results
Coexpression of Ligand Binding N-Terminal Domain of gb1a or Transmembrane Domain (TM) 1 to 7 Segments of gb1a with gb2 Does Not Result in Functional GABA Receptors.
We asked whether coexpression of N-terminal (N-gb1a) and C-terminal (C-gb1a) truncated gb1a receptors with gb2 was sufficient to form functional GABABreceptors. N-gb1a composed the signal peptide and the entire extracellular N-terminal domain, including the putative TM 1 segment, which was retained to anchor and orient the protein in the plasma membrane (Fig. 1A). The C-gb1a composed the receptor from TM 1 to 7 through to the C-terminal tail containing coiled-coil and PDZ (PSD-95, Disc-large, and ZO-1) domains for protein-protein interactions (Fig. 1A).
In vitro transcription/translation studies revealed the expression of a ∼63-kDa N-gb1a monomer and a ∼40-kDa C-gb1a monomer corresponding to the nonglycosylated forms of these receptors. In crude membranes prepared from whole cells, immunoblotting and [125I]CGP71872 photoaffinity labeling revealed the expression of a ligand-binding ∼95-kDa N-gb1a species, presumably representing a glycosylated form of the receptor, and a mature nonglycosylated ∼40-kDa C-gb1a species that did not bind ligand (Fig.1, B–D). Competition studies at N-gb1a revealed a rank order of affinities of CGP71872 > SKF-97541 [3-aminopropyl(methyl)phosphinic acid] ≥ GABA ≥ (+)-baclofen > saclofen similar to gb1a (data not shown), suggesting that N-gb1a retains the pharmacological characteristics of the full-length receptor. The soluble N terminus of gb1a alone was reported previously to be sufficient to specify agonist and antagonist binding, although agonist affinities were higher possibly because this construct lacked the TM 1 segment present in N-gb1a, which may influence agonist affinities (Malitschek et al., 1999).
The ability of gb1a, N-gb1a, and C-gb1a to form functional receptors with gb2 was determined in a stable high-level gb2-expressing HEK 293 cell line selected on mRNA and surface expression (Fig.2A). As expected, GABA mediated a dose-dependent decrease (46–61%) in forskolin-stimulated cAMP levels (IC50 = 49–321 nM) in stable gb2 HEK 293 cell lines transfected with gb1a but not when receptors were expressed alone (n = 4; Fig. 2B), indicating that gb1a and gb2 monomers are not functionally coupled to adenylyl cyclase in HEK 293 cells. GABA-mediated inhibition of cAMP synthesis was not observed after the coexpression of the ligand binding N-gb1a construct or C-gb1a in the clonal gb2 cell line (n = 4; Fig. 2B), indicating that the functional GABAB receptor requires both full-length gb1a and gb2 subunits for signaling via adenylyl cyclase.
Coexpression of mGluR4 Promotes Plasma Membrane Expression of gb1a but Not Function.
mGluR4 can undergo dimerization and exhibits motifs required for protein-protein interactions. Thus, we reasoned that mGluR4 might act as a surrogate coreceptor for gb1a translocation and the functional expression of gb1a. We examined the ability of mGluR4 to promote the plasma membrane expression of gb1a by immunoblot analysis and flow cytometry and the ability to form heterodimers with gb1a by differential immunoprecipitation and blotting.
Densitometric analysis of immunoblots of crude (40,000g) membranes prepared from COS-7 cells coexpressing gb1a and FLAG-gb2 showed a ∼15-fold increase in the expression of a ∼130-kDa gb1a over the staining in gb1a-expressing cells (Fig.3A). White et al. (1998) showed that the coexpression of gb1a and gb2 resulted in the membrane expression of a ∼130-kDa mature glycosylated gb1a monomer. c-myc-mGluR4 promoted a smaller 4- to 7-fold increase in the membrane expression of a ∼130-kDa gb1a monomer identical in size to the gb1a monomer in cells coexpressing gb1a and gb2. Negligible ∼130-kDa gb1a immunoreactivity was detected in crude membranes prepared from cells expressing gb1a alone or in vector-transfected cells (Fig. 3A). Consistent with these findings, flow cytometry of whole live cells, which reveals only cell surface gb1a expression, showed gb1 antibody labeling in 50 to 75% of cells coexpressing gb1a/FLAG-gb2, 45 to 60% of cells coexpressing gb1a/c-myc-mGluR4, and 5 to 20% in controls (n = 3; Fig. 3B). Thus, mGluR4 appears to promote the expression of a mature glycosylated gb1a monomer in crude membranes, although slightly less efficient than gb2 (Fig. 3A).
Because gb2 undergoes heterodimerization with gb1a for cell surface targeting, we asked whether mGluR4 can form heterodimers with gb1a. We used a differential coimmunoprecipitation and immunoblotting strategy similar to the one used to show gb1-gb2 heterodimers (Ng et al., 1999b). The gb1 receptor antibodies were used to blot gb1a immunoprecipitated with FLAG antibodies directed against FLAG-gb2. Consistent with previous findings, the gb1a-gb2 heterodimer and gb1a monomer were only detected in gb1a/FLAG-gb2-coexpressing cells (Fig.4C). The gb1 receptor antibodies were then used to blot gb1a immunoprecipitated with c-mycantibodies directed against c-myc-mGluR4. In contrast, gb1a immunoreactivity was not detected in immunoprecipitate prepared from cells coexpressing gb1a and c-myc-mGluR4 (Fig. 4C), even though a ∼110-kDa immunoreactive species, likely corresponding to the glycosylated mGluR4 monomer (Han and Hampson, 1999), was detected in this sample with a mGluR4 antibody (Shigemoto et al., 1997; data not shown). This demonstrates that gb1a did not coimmunoprecipitate with c-myc-mGluR4. The gb1a immunoreactivity was not detected in c-myc antibody-immunoprecipitated samples from vector-transfected cells, FLAG-gb2-expressing cells, or gb1a receptor-expressing cells indicating that gb1a forms a specific heterodimer with gb2. Although GABAB receptors undergo heterodimerization and mGluRs undergo homodimerization, these data are the first to demonstrate the structural specificity between these family C GPCRs.
To examine whether the mGluR4-promoted expression of the mature gb1a monomer is sufficient to result in functional GABA receptors, we coexpressed mGluR4 and gb1a in X. laevis oocytes and melanophores. Consistent with our previous findings, coexpression of gb1a and FLAG-gb2 with Kir 3.1/3.2 in X. laevis oocytes resulted in a significant stimulation of Kir current in response to 100 μM GABA (297 ± 30.5% increase over control current,n = 11) measured at −80 mV, whereas modulation of Kir 3.1/3.2 was not seen in oocytes expressing gb1a or FLAG-gb2 individually (n = 11; Fig. 4A). GABA (100 μM) could not stimulate Kir current in oocytes coexpressing gb1a and mGluR4 (n = 11; Fig. 4A).
In melanophores transiently cotransfected with the gb1a and FLAG-gb2 receptors, GABA mediated a dose-dependent pigment aggregation response with an IC50 value of 0.6 to 8 μM (n = 4; Fig. 4B). GABA activity was not detected, testing concentrations up to 1 mM, in melanophores transiently cotransfected with c-myc-mGluR4 and gb1a or c-myc-mGluR4 and FLAG-gb2. Thus, gb1a receptors do not form functional GABA receptors after coexpression with mGluR4. The functional GABA receptor results only from the coexpression of gb1 and gb2 subunits.
Discussion
Native GABAB receptors are well known to couple to membrane K+ channels as well as to adenylyl cyclase in neurons (Bowery and Enna, 2000). Therefore, we evaluated the ability of recombinant gb1a and gb2 receptors to modulate Kir channel activity in X. laevis oocytes and to inhibit cAMP levels in X. laevis melanophores and HEK 293 cells. Under our assay conditions, gb1a and gb2 receptors, when expressed alone, are nonfunctional, consistent with the intracellular retention of gb1 in the absence of gb2 (Couve et al., 1998; White et al., 1998), low agonist affinities of gb1 monomers (Kaupmann et al., 1997, 1998b), and the lack of detectable binding sites on gb2 (Jones et al., 1998;Kaupmann et al., 1998; White et al., 1998; Kuner et al., 1999; Ng et al., 1999a). These results are discrepant with studies by Kaupmann et al. (1997), who reported that the rat gb1a receptor, when expressed alone, can inhibit forskolin-stimulated cAMP levels and that human gb1a and gb1b receptors can activate Kir channels in HEK 293 cells (Kaupmann et al., 1998b). In the latter study, however, modulation was only detectable in ∼10% of the cells where current was measured and was significantly attenuated compared with other GPCRs. We have not detected any modulation by gb1a or gb2 alone for Kir 3.1/3.2, Kir 3.1/3.4, Kir 3.2, or Kir 3.4 (data not shown). Two studies have also reported that the gb2 receptor, when expressed alone, can mediate baclofen-inhibited forskolin-stimulated cAMP production (Kuner et al., 1999; Martin et al., 1999). As proposed by Martin et al. (1999), the discrepancy may be due to higher levels of stable gb2 receptor expression in their CHO cells achieved using inducible systems. Our clonal gb2 receptor-expressing HEK 293 cell line showed high RNA and surface gb2 receptor expression, but receptors were nonligand binding (data not shown) and exhibited no functional activity following up to 1 mM GABA or 1 mM (R)-baclofen treatment (data not shown). It will be important to determine the gb2 receptor density conferring functional activity. gb2 may bind a yet-to-be-discovered ligand. gb1 and gb2 monomer activity may also be cell line-dependent. A number of studies have highlighted the importance of an appropriate cell background in the identification of orphan receptors, in particular in the identification of CGRP and adrenomedullin receptors (McLatchie et al., 1998). It is possible that the functional expression of gb1 or gb2 receptors in certain cell lines is due to the coexpression of an endogenous surrogate coreceptor.
We reasoned that if the gb1-gb2 heterodimer were the active form of the GABAB receptor, this would be a specific functional interaction. We coexpressed a truncated N-terminal portion of gb1a, containing the major determinants for ligand binding, with gb2, and a C-terminal portion of gb1a, containing the TM 1 to 7 segments, extracellular and intracellular loops and carboxyl tail, with gb2. GABA did not mediate inhibition of forskolin-stimulated cAMP production in gb2-expressing cells transfected with N-gb1a, suggesting that although the ligand binding N termini of gb1a and gb2 are present, the C-terminal portion of gb2, absent the C-terminal portion of gb1a, is not sufficient alone to promote coupling to effector pathways. Although we determined that N-gb1a exhibits binding characteristics similar to gb1a whereas gb2 does not bind ligands, we did not determine whether high-affinity agonist binding result in cells coexpressing N-gb1a and gb2. The lack of high-affinity binding likely does not explain the lack of function because GABA concentrations were tested up to 1 mM. The data suggest that N-gb1a and gb2 monomers and/or N-gb1a-gb2 heterodimer expressed in these cells are functionally inactive. GABA (up to 1 mM) also did not mediate inhibition of forskolin-stimulated cAMP production in gb2-expressing cells transfected with C-gb1a. This suggests that although C-gb1a contains the intracellular domains for G protein interactions and the coiled-coil domain for heterodimerization with gb2, the extracellular N terminus of gb2 is not sufficient in the absence of the N terminus of gb1a to bind and mediate the intrinsic activity of agonist. The C-gb1a and gb2 monomers and/or C-gb1a-gb2 heterodimers that are expressed in these cells are functionally inactive. The functional GABAB receptor coupled to the inhibition of adenylyl cyclase with a nanomolar potency for GABA results only from the coexpression of the full-length gb1a and gb2 receptors. Thus, the most likely explanation is that the functional GABAB receptor is a preexisting gb1-gb2 heterodimer with the major site for ligand binding and effector coupling conferred by gb1a.
The truncated receptor studies, however, do not rule out the possibility that once gb1 is expressed on the plasma membrane with gb2, the mature gb1 monomer is rendered functional. Membrane-expressed gb1 monomers may occur under certain cellular environments and could account for the reported ability of gb1 monomers to couple to adenylyl cyclase in HEK 293 cells or Kir channels in X. laevisoocytes (Kaupmann et al., 1997, 1998b). To address this, we asked whether mGluR4 could act as a surrogate coreceptor (translocator protein) for gb1a. The mGluR4 receptor shares many structural features with GABAB receptors, including protein-protein interacting PDZ and SCR domains (Kaupmann et al., 1998b), and forms functional homodimers (Han and Hampson, 1999), making this a candidate coreceptor. It should be noted that mGluR4 does not exhibit a coiled-coil domain present in gb1 and gb2 that mediates the heterodimerization of these receptors (White et al., 1998; Kuner et al., 1999). Coexpression of gb1a with gb2 resulted in the membrane expression of a mature gb1a monomer corresponding to the glycosylated form of the receptor (White et al., 1998). Coexpression of gb1a with mGluR4 also resulted in the membrane expression of a mature gb1a monomer, but mGluR4 was slightly less efficient than gb2 and did not form heterodimers. mGluR trafficking has been reported to involve an interaction with Homer/Vesl family of proteins (Ciruela et al., 1999;Roche et al., 1999), but a Homer/Vesl consensus sequence in gb1a is lacking. Possibly, mGluR4 transiently stabilizes gb1a in the endoplasmic reticulum such that it can fold/mature and traffic to the cell surface, but the mechanism remains unknown at this time. This, however, provided a model system to test whether mature gb1a monomers, in the absence of gb2, are functionally coupled. The coexpression of mGluR4 and gb1a in oocytes and melanophores did not result in the formation of active GABA receptors. Similar results were obtained after the coexpression of mGluR4 and gb2. This indicates that the functional GABAB receptor results only from the coexpression of gb1 and gb2 and that the functional receptor is a gb1-gb2 heterodimer.
The mode of gb1-gb2 heterodimer ligand binding and activation may resemble the insulin tyrosine kinase receptors that exist as preformed dimers that, on ligand binding, undergo transition to an active conformation (Weiss and Schlessinger, 1998). Future experiments planned using fluorescence donor-gb1 and fluorescence acceptor-gb2 pairs in fluorescence resonance energy transfer-based assays will be valuable in confirming the conclusions of this study. Of interest, a growing number of GPCRs have been reported to exist as dimers (Hebert and Bouvier, 1998), but the therapeutic importance is largely unclear. In the case of the dopamine D2 receptor, monomers and homodimers are differentially bound by butyrophenone and benzamide neuroleptic antagonists, which exhibit different side effects profiles (Ng et al., 1996). Coexpressed κ- and δ-opioid receptors result in a new receptor with distinct pharmacology (Jordan and Devi, 1999). gb1a, gb1b, and gb1c isoforms differ in their ligand-binding extracellular N-terminal domains. gb1a and gb1b differ in their extrasynaptic localizations (Billinton et al., 1999; Fritschy et al., 1999) but are colocalized with gb2 (Benke et al., 1999), raising the possibility that coexpression of gb1 isoforms with gb2 could result in pharmacologically and functionally distinct GABABheterodimers.
Acknowledgments
We thank Kevin Clark for the preparation of manuscript figures, Ken MacDonald for technical assistance, and Louise Charlton for administrative assistance.
Footnotes
-
Send reprint requests to: Dr. Gordon Y. K. Ng, Departments of Biochemistry, Molecular Biology and Chemistry, Merck Frosst Center for Therapeutic Research, 16711 TransCanada Hwy., Kirkland, Quebec, H9H 3L1 Canada. E-mail: gordon_ng{at}merck.com
-
↵1 The work in the laboratory of T.E.H. was supported by the Medical Research Council of Canada, the Heart and Stroke Foundation of Canada, and the Fonds de la Recherche en Santé du Québec.
- Abbreviations:
- GABA
- γ-aminobutyric acid
- mGluR
- metabotropic glutamate receptor
- GPCR
- G protein-coupled receptor
- TM
- transmembrane domain
- Kir
- inwardly rectifying potassium channel
- HEK
- human embryonic kidney, PCR, polymerase chain reaction
- bp
- base pair(s)
- Received November 12, 1999.
- Accepted January 27, 2000.
- The American Society for Pharmacology and Experimental Therapeutics