Interaction of GABA Receptor/Channel ρ₁ and γ₂ Subunit

Abstract

purpose. To determine whether protein–protein and functional interactions can occur between γ-aminobutyric acid (GABA)_A receptor/channels γ₂ subunit and the retina-specific GABA_C ρ₁ subunit.

methods. Protein–protein interaction was characterized by immunocoprecipitation of these subunits in brain and spinal cord with anti-γ₂ subunit antibody and by Western blot analysis with anti-ρ₁ subunit antibody. The ρ₁ and γ₂ subunits were detected in the adult rat brain and spinal cord lysates that had been previously precipitated with the specific antibodies against the ρ₁ and γ₂ subunits, respectively. A two-microelectrode voltage clamp was used to measure GABA-induced currents in oocytes. In addition, a yeast two-hybrid system was used to detect the interactions of these subunits in vivo.

results. Based on yeast transformed with the N-terminal fragment of the γ₂ subunit (γ₂-N′), the N-terminal fragment of the ρ₁ subunit (ρ₁-N′), and the full-length ρ₁ subunit, the protein–protein interaction of the GABA_A γ₂ subunit and the GABA_C ρ₁ subunit was found in yeast grown in triple-dropout medium (deficient in Leu, Trp, and His) and expressing the LacZ reporter gene. Interaction of the ρ₁ and γ₂ subunits was investigated by functional studies in which γ₂ (γ₂-N′ with 837 bp) and ρ₁ cRNAs were coinjected in Xenopus oocytes. In studies of the functional interaction, after injection of the γ₂ subunit mutant cRNA containing a N-terminal fragment, GABA-induced ρ₁ originated currents declined to 16% of the control level of homooligomeric ρ₁ current. This inhibitory effect of coexpressing γ₂ subunit mutants with ρ₁ subunit on the ρ₁-originated current in oocytes was dose dependent. In addition, coexpression of the GABA ρ₁ and γ₂ subunits in oocytes altered pharmacologic properties of the homooligomeric receptor/channel formed by ρ₁ or γ₂ subunits. Further evidence was provided by results obtained with specific antibodies showing that the ρ₁ subunit was coimmunoprecipitated with the γ₂ subunit from the retina, brain, and spinal cord.

conclusions. The results indicate that protein–protein and functional interactions can occur between the GABA_A γ₂ subunit and the GABA_C ρ₁ subunit. Therefore, the functional role of GABA receptor/channels in the brain, retina, and spinal cord is more diversified because of the possible assembly between the GABA_A γ₂ subunit and GABA_C ρ₁ subunit.

A major inhibitory neurotransmitter in the vertebrate central nervous system (CNS) is γ-aminobutyric acid (GABA). Its cognate receptors have been classified into three distinct subtypes based on their pharmacologic properties. The GABA_A receptor/channel is bicuculline sensitive, the GABA_B receptor is baclofen sensitive, and the GABA_C receptor/channel is insensitive to either bicuculline or baclofen. However, the current classification of GABA receptors does not completely describe GABA receptor/channel heterogeneity, based on pharmacologic profiles of diseases of the GABAergic system, GABA receptor/channel subunit distributions in the CNS, and electrophysiology of the GABA receptor/channel in vivo. ^{1 2 3 4 5 6 7 8} GABA receptor/channel heterogeneity in the mammalian CNS is due to interactions among at least 21 different subunits, including α1 to α6, β1 to β4, γ1 to γ4, δ, ω, π, ζ, and ρ1 to ρ3. ^{9 10 11 12 13 14 15 16 17 18 19 20} Composition of different subunits in the receptor/channel determines the pharmacologic and electrophysiological properties of GABA receptors-channels. ^{10 12 20 21 22} Analysis of molecular structure of GABA receptors-channels reveals that subunits assemble in a putative pentameric heterooligomeric combination to form the receptor/channel. ²³ The GABA_C ρ₁ subunit, cloned from the human retina, forms functional homooligomeric receptor channels when expressed in Xenopus oocytes. ¹¹ The ρ₁ subunit was initially reported to be a retina-specific GABA receptor/channel, exclusively localized in the retina. ²⁴ However, distribution of the ρ₁ subunit is not limited to the retina; it has also been reported in the CNS ²⁵ in the hippocampus, ^{26 27} cerebellum, ^{28 29} anterior pituitary, ¹ dorsal root ganglia, ³⁰ and optic tectum. ³¹  

The molecular structure of GABA_A receptor/channel complex is thought to be a heteropentomeric glycoprotein of approximately 275 kDa composed of combinations of multiple peptide subunits. ³² This has been deduced from Western blot, immunoprecipitation, immunoaffinity chromatography, and in situ hybridization. Topologic structure of GABA receptor/channel is predicted from the structure of another ligand-gated receptor/channel. ³³ The functional heterogeneity of GABA_A receptors in neurons arises, not only from the multiple GABA receptor/channel subunit genes and splice variants, but also from the combinatorial mixing of different GABA receptor/channel polypeptides to form heterooligomeric receptor/channels. ^{16 34 35 36 37 38} Various GABA_A receptor/channel subunits are differentially expressed during specific developmental stages and under varying physiological conditions in different regions of the body. ^{11 39 40} These considerations may contribute to the diversity seen in GABA receptor/channel pharmacology and function. If heterooligomeric receptors were formed by random combinations, it would create numerous GABA receptor/channel isoforms. In fact, functional GABA_A receptor/channels always exist in preferred subunit combinations, as mentioned earlier. ³³ GABA α₁β_2/3γ₂ and α₂β_2/3γ₂ receptor/channels represent two major GABA_A receptor subtypes contributing to 75% to 85% of the diazepine-sensitive GABA_A receptors. ¹⁰ It has been shown that different subunits do not have the same assembly behavior. ^{35 41} GABA_C ρ₁ or ρ₂ subunits can form homooligomeric receptors and express robust currents in oocytes. ^{11 42} However, sole expression of a single GABA_A receptor subunit in oocytes results in a much smaller GABA-induced current compared with heterooligomeric expression, suggesting a low efficiency for GABA_A subunits to form homooligomeric channels in oocytes. ^{35 41} GABA receptor/channel assembly involves several posttranslation maturation steps and is driven by noncovalent protein–protein interaction. ⁴³ There are specific recognition sites on the individual subunit that guides their proper assembly into a functional receptor/channel. However, structural elements that determine the compatibility for the subunit assembly are still unknown. 

Recently, electrophysiological studies of recombinant GABA receptor/channels in Xenopus oocytes suggest that GABA γ₂ and ρ₁ subunits may form a functional receptor/channel in oocytes. ^{44 45 46} However, there is no direct evidence that demonstrates the assembly of a heterooligomeric receptor/channel between GABA γ₂ and ρ₁ subunits at protein–protein interaction level in vitro and in vivo. In the present study, we report that both protein–protein and functional interactions can occur between the retinal specific GABA_C ρ₁ subunit and GABA_Aγ₂ subunit. This determination is based on our results in studies in which immunoprecipitation, the yeast two-hybrid system, and electrophysiological measurements in the oocyte expression system were used. The first approach indicates that ρ₁ and γ₂ subunits can interact, because it was possible to immunoprecipitate the ρ₁ subunit from the retina, brain, and spinal cord lysates with an antibody selective for the γ₂ subunit. Further evidence for protein–protein interaction is that in the yeast two-hybrid system ρ₁ and γ₂ interaction occurred based on elicited gene expression. Such an interaction is reflective of a very specific assembly process occurring in vivo. Furthermore, coexpression of the ρ₁ subunit along with an N-terminal fragment of the γ₂ subunit occurred in Xenopus oocytes, because increases in γ₂ subunit expression inhibited homooligomeric ρ₁-originated currents in a dose-dependent pattern. 

Materials and Methods

Immunoprecipitation and Western Blot Analysis

The protocol of the study adhered to the provisions of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Whole brain, retina, and spinal cord from adult albino rats were lysed with lysis buffer containing 50 mM HEPES (pH 7.5), 150 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, 10 mM sodium pyrophosphate, 10% glycerol, 1% Triton X-100, 1 mM NaF, 1 mM Na-orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 250 μM p-nitrophenylphosphate, 10 μg/mL aprotinin, and 10 μg/mL leupeptin. Cell lysates were centrifuged at 5000 rpm for 5 minutes, and the supernatant was transferred to a new tube containing anti-γ₂ antibody (polyclonal IgG from rabbit against the N terminus of the γ₂ subunit) plus protein a affinity medium (Protein A Sepharose beads; Sigma, St. Louis, MO) and incubated overnight at 4°C. ⁴⁷ Final dilution of the anti-γ₂ antibody was 1:15,000. Cell lysates were centrifuged at 5000 rpm for 5 minutes, and the precipitate was boiled in an equal volume of SDS-PAGE sample buffer for 5 minutes. Aliquots were loaded on SDS-PAGE gels followed by Western blot analysis, using a 1:10,000 dilution of anti-ρ₁ antibody (polyclonal IgG from guinea pig against the N terminus of the ρ₁ subunit) for the detection. ⁴⁷ Briefly, an equal volume of 2× Laemmli buffer was mixed with the immune complex and boiled for 5 minutes. After resolution by SDS-PAGE, proteins were transferred to polyvinylidene difluoride (PVDF) membrane (Millipore, Bedford, MA) and probed with antibodies against specific subunits. Membranes were then incubated with secondary antibodies conjugated with alkaline phosphatase. Secondary antibodies were detected with a Western blot detection kit (Phototope-Star; New England Biolabs, Beverly, MA). 

Yeast Two-Hybrid cDNA Construction and Transformation

cDNAs of the N termini (γN′ and ρN′, see the Results section) were subcloned into the yeast expression vector (pACT2; Clontech, Palo Alto, CA) encoding a GAL4 transcriptional activation domain (AD) of the yeast Saccharomyces cerevisiae. Full-length ρ₁ cDNA was subcloned into a yeast expression vector (pAS2-1; Clontech) encoding the GAL4 DNA-binding domain (BD). The orientation and alignment of the reading frame for the subcloned cDNAs were confirmed by DNA sequencing. The yeast two-hybrid system (Matchmaker; Clontech) was used to determine interaction of the GABA_A receptor/channel γ₂ subunit with the GABA_C receptor/channel ρ₁ subunit. Reporter yeast (strain Y190; Clontech)-competent cells were prepared by the lithium acetate method. Yeast cells were cultured overnight at 30°C with shaking at 250 rpm in 50 mL of medium (YPD; Clontech) and diluted in 300 mL of the medium to an optical density at 600 nm (OD₆₀₀) of 0.2. Yeast cells were centrifuged and rinsed in water at room temperature. They were resuspended in 1.5 mL of TE (Tris-EDTA) buffer/lithium acetate (TE/LiAC; Clontech). Full-length ρ₁ (2.05 kbp) in pAS2-1 was simultaneously transformed into the Y190 cells with each of the N termini γ₂ (γN′, 0.84 Kbp) and ρ₁ (ρN′, 0.80 kbp) in the pACT2 vector. Each plasmid DNA (100 μg) was added to 100 μg herring testes carrier DNA. Yeast cells were heat shocked for 15 minutes in a 42°C water bath, spun down, and resuspended in 0.5 mL TE buffer. Transformation mixture was spread on 100-mm plates containing a synthetic dropout (SD) selection medium without leucine, tryptophan, and histidine (−Leu/−Trp/−His), supplemented with 25 mM 3-amino-1,2,4-triazole (+3-AT; Sigma, St. Louis, MO). Plates were incubated at 30°C for 2 to 4 days. For LacZ selection, only fresh colonies (3 days in culture) on SD/−Leu/−Trp/−His/+25 mM 3-AT agar plates were assayed. Assayed colonies were transferred onto filter paper and then submerged in liquid nitrogen for 10 seconds each time, followed by 1 minute of thawing. Each filter was placed, colony side up, on a presoaked filter and incubated at 30°C. 

Construction of γ₂ and ρ₁ N-Terminal Domains

Construction of N-terminal γ₂ and ρ₁ subunits was based on analysis of the full-length DNA subunit sequences. ^{11 18 48 49} N-terminal fragments were amplified by PCR from full-length sequences inserted into a vector (pBluescript SK(−; Stratagene, La Jolla, CA) vector. PCR primers were designed as follows: 5′-GGA TCCGGCGAGAGGAAAAAAAAGCG-3′ with an introduced BamHI restriction enzyme site (γ₂ sense); 5′-GATCTGAGCAGAAGAATGGGGCTCGAG-3′ with an introduced XhoI restriction enzyme site (γ₂ antisense); 5′-GGATCCCCATGTTGGCTGTCCCA-3′ with an introduced BamHI restriction enzyme site (ρ₁ sense); 5′-ccccgctaccctgatggtcatgGGATTC-3′ with an introduced EcoRI restriction enzyme site (ρ₁ antisense). Each N-terminal fragment was subcloned into the multicloning site (MCS) of a PCR3 vector. 

In Vitro Transcription and Microinjection of Oocytes

Fragments of subunit cDNAs for in vitro transcription were subcloned into the multiple cloning site of a PCR3 expression vector in an orientation of T7 to SP6. Plasmid constructs with inserts were linearized with PstI (New England Biolabs). A T7 in vitro transcription kit (Invitrogen, San Diego, CA) was used in all transcription reactions. Transcription reactions were performed in the presence of the cap analogue diguanosine triphosphate, and were catalyzed by T7 RNA polymerase. The cRNA was quantified by agarose gel electrophoresis and photospectroscopy (Pharmacia, Piscataway, NJ), and then dissolved in sterile RNase-free water to a final concentration of 5 μg/μL. Xenopus oocytes were prepared from frogs by using a method described previously. ⁴⁷ Stage 5 to 6 oocytes were selected and incubated at 18°C in MBS (Modified Barth’s Solution) for 24 hours, followed by injection of cRNA. Microinjection of cRNA into oocytes was performed by positive displacement with a 10-μL micropipette (Drummond Scientific Inc., Broomall, PA). 

Voltage Clamp

Two-microelectrode voltage-clamp recordings were performed at room temperature with continuous perfusion (10 mL/min). The glass microelectrodes (resistance ∼2.0 MΩ) were made with a horizontal puller (PD-5; Narishige USA, Greenvale, NY), and were backfilled with 3 M KCl. The perfusion bath was connected to the voltage-recording amplifier (Axoclamp 2A; Axon Instruments, Burlingame, CA) by an Ag-AgCl-Agar-3 M KCl bridge. Data were filtered with a four-pole Bessel filter at 500 Hz. Data acquisition was performed with the pCLAMP software (Axon Instruments). 

Statistical Analysis

Data were presented as original values, or as the mean ± SE. Significant differences were determined by using the paired t-test or ANOVA and the Tukey post-hoc range test at the confidence interval indicated. 

Results

Construction of N-Terminal Domain of the ρ₁ and γ₂ Subunits

It has been suggested that the N-terminal part of ligand-gated channels has a large extracellular domain containing ligand-binding sites. ^{50 51} Its C-terminal part is believed to traverse the cell membrane and to form four transmembrane segments with a large intracellular loop between the third and fourth transmembrane segments. The second transmembrane region is likely to form the lining of the channel pore. The intracellular loop contains specific sites for channel regulation by protein kinase-induced phosphorylation. In the present study, N-terminal constructs of the ρ₁ and γ₂ subunits were composed of 292 and 271 amino acids, respectively (Fig. 1) . These N-terminal domain peptides were also used to produce fusion proteins in bacteria, by using a glutathione S-transferase (GST) fusion system. N-terminal domain fusion proteins were then used to raise specific antibodies against the ρ₁ and γ₂ subunits that demonstrate tissue and functional specificity. ⁴⁷  

Immunoprecipitation of the Retinal ρ₁ Subunit from the Rat Brain and Spinal Cord

Immunoprecipitation experiments were performed to determine a protein–protein interaction between GABA γ₂ and ρ₁ subunits in the mammalian CNS. Brain and spinal cord lysates from adult rats were obtained and incubated with specific antibodies against the GABA γ₂ subunit. Brain and spinal cord proteins that were pulled down by the anti-γ₂ subunit antibody were analyzed in a polyacrylamide gel. A 50-kDa band was recognized by the anti-ρ₁ subunit antibody in the brain and spinal cord, but not in the control lanes in which brain and spinal cord proteins were immunoprecipitated with an anti-Erk (mitogen-activating protein [MAP] kinase) antibody (Fig. 2A) . Western blot analysis was used to detect γ₂ subunits in brain and spinal cord lysates that had been immunoprecipitated with the selective γ₂ antibody. A 48-kDa band was detected by the anti-γ₂ antibody in the brain and spinal cord tissues, but not in lysates immunoprecipitated in the control experiment with the anti-Erk antibody (Fig. 2B) . In addition, the retinal lysate from adult rats was used to incubate with the antibody against the GABA γ₂ subunit. Western blot analysis demonstrated that there was a 50-kDa band in the retina identified by the anti-ρ₁ antibody (Fig. 2C) . Protein bands with molecular weights of 48 and 50 kDa identified in Western blot by anti-γ₂ and anti-ρ₁ antibodies are consistent with the sizes of GABA γ₂ and ρ₁ subunits, respectively. These results suggest that there is indeed interaction and assembly of the γ₂ and ρ₁ subunits in the CNS. 

Next, we tested whether an interaction between the γ₂ and ρ₁ subunits occurs as a posttranslational subunit–subunit interaction. To probe this possibility, GST-ρ₁ fusion protein containing the N-terminal part of the ρ₁ subunit was purified using a glutathione-Sepharose matrix. Lysates of the brain and spinal cord from adult albino rats were incubated with GST-ρ₁ fusion proteins that were bound to GST affinity matrix. Oocytes expressing ρ₁ subunit were also incubated with GST-ρ₁ fusion proteins. After the fusion protein and affinity matrix were rinsed with cold PBS, GST-ρ₁ fusion proteins were eluted from the affinity matrix with a glutathione elution buffer and subjected to Western blot analysis. The anti-ρ₁ antibody was used to probe for interaction of the GST-ρ₁ fusion protein and native ρ₁ subunit from the brain and spinal cord (Fig. 3A) and the GST-ρ₁ fusion protein and expressed ρ₁ subunit from cRNA-injected oocytes (Fig. 3B) . In addition, lysates from brain and spinal cord tissues, and from cRNA-injected oocytes were immunoprecipitated with anti-Erk antibodies and probed with the anti-ρ₁ antibody for the control, respectively. Western blot analysis demonstrated that GST-ρ₁ fusion proteins used in these precipitation experiments were detectable, but native ρ₁ subunit peptides were not detected. These results imply that γ₂ and ρ₁ subunit interaction or assembly in the CNS is unlikely to be a posttranslational event. 

Protein–Protein Interaction of γ₂ and ρ₁ Subunits in the Yeast

To further confirm protein–protein interaction between GABA γ₂ and ρ₁ subunits in vivo, a high-fidelity yeast two-hybrid expression system was used. In the reporter yeast strain Y190, expression of the His-3 and LacZ reporter genes requires the DNA BD and AD of the GAL4 to be colocalized upstream from the activation sequence of GAL4. In addition, we fused full-length ρ₁ to the DNA BD, whereas γ₂N′ and ρ₁N′ terminal fragments were fused to the transcriptional AD of the yeast GAL4 transcriptional activator. The colocalization is mediated in yeast by the specific interaction of these fusion proteins that were connected to the AD and BD. In addition, 3-amino-1,2,4-triazole (3-AT) was added to the triple-dropout medium to prevent leaky His-3 activity, thereby ensuring the fidelity of the system. Thus, activation of transcription of the His-3 and LacZ reporter genes in our two-hybrid assay was due to the specific interaction of the γ₂ or ρ₁ N′-terminal fragment and ρ₁ subunit that were fused to the AD and BD, respectively. The 5-bromo-4-chloro-3-indolyl-β-d-galactoside (X-gal) reaction (detection of LacZ reporter gene expression) was also used as the other indication of the interaction between AD and BD fusion proteins. The X-gal reaction ensures that survival of transformed yeast grown in SD−Leu/−Trp/−His selection medium is due to transcription of the His-3 reporter gene, not to external histidine contamination. 

Y190 cells were simultaneously transformed with ρ₁-BD cDNA and each of the γ₂N′-AD and ρ₁N′-AD cDNA, in turn, and the transformation mixture was plated on SD−Leu/−Trp/−His/+25 mM 3-AT agar medium. Results shown in Table 1A indicate that in both transformations there was significant growth of Y190 primary colonies. The ρ₁N′/ρ₁ transformation was used as the positive control, because their interaction has been extensively characterized. Control experiments were also performed by simultaneously transforming yeast with antisense ρ₁-BD cDNA and γ₂N′-AD or by ρ₁N′-AD cDNA and by plating transformation mixture on SD−Leu/−Trp/−His/+25 mM 3-AT medium (Table 1A) . No growth of Y190 was observed in these plates. To further ensure that neither BD nor AD could independently activate transcription of the reporter genes, the reporter yeast strain Y190 was transformed with either BD or AD vector. The transformation mixture was plated on SD−Leu/−Trp/−His/+25 mM 3-AT agar medium (Table 1A) . There was no growth of Y190, again showing that the activation of transcription of the His-3 gene was due to specific interaction between the ρ₁ and γ₂ subunits. 

The affinity for protein–protein interaction is indicated in the yeast by the number of primary colonies on SD medium agar plates. There was no significant difference in the number of primary colonies from the ρ₁-BD with ρ₁N′-AD transformation and ρ₁-BD with γ₂N′-AD transformation (Table 1B) . This suggests that there is comparable protein–protein affinity between these subunits. The auxotrophic integrity of the Y190 cells was also determined, because the integrity of the reporter yeast strain (Y190) determines the fidelity of the yeast two-hybrid assay. Before each experiment, Y190 cells were transformed with either BD or AD vector, and the transformation mixture was plated on either SD−Trp or SD−Leu, respectively (Table 1C) . Specific growth on these media showed that the reporter yeast strain maintained its auxotrophic integrity. 

Functional Interaction of the γ₂ N Terminus with the ρ₁ Subunit in Xenopus Oocytes

It has been suggested that there are specific peptide motifs in the N-terminal domain that mediate protein–protein interaction for both homooligomeric and heterooligomeric interactions. ⁵² The competitive inhibition studies in Xenopus oocytes were designed by coinjecting cRNAs of full-length subunits and C-terminal, part-truncated subunits. These C-terminal, part-truncated subunits are mutant receptor/channels that do not have the ability to conduct chloride ions. If there is an assembly occurring among full-length and mutant subunits, the population of functional receptor/channel in the membrane is decreased. Coexpression of the C-terminal, part-truncated mutant of the ρ₁ subunit (ρN′) with the full-length ρ₁ subunit in oocytes resulted in a cRNA concentration-dependent reduction in GABA-induced whole-cell current (Fig. 4A) . Whether the γ₂ subunit interacts with the ρ₁ subunit was investigated by coexpression of the C-terminal, part-truncated mutant of the γ₂ subunit (γN′) with the native ρ₁ subunit in Xenopus oocytes. Coinjection of the ρ₁ subunit cRNA with the γN′ mutant in oocytes demonstrated a concentration-dependent inhibition of GABA-induced whole-cell current (Fig. 4B) . The reduction in GABA-induced whole-cell current of the ρ₁ subunit suggests that the N terminus of the γ₂ subunit contains motif(s) for heterooligomeric interaction with the ρ₁ subunit. 

Altered Pharmacologic Properties of the Recombinant GABA_H Receptor/Channel

The plant-derived convulsant bicuculline is a potent antagonist at GABA_A receptor/channels. Application of 100 μM bicuculline resulted in a more than 75% reduction of GABA-induced current in oocytes expressing heterooligomeric receptor/channel of the GABA_A α₁, β₂, and γ₂ subunits, but showed no effect on homooligomeric receptor/channels of the GABA_C ρ₁ subunits (data not shown). Application of 100 μM bicuculline showed no effect on recombinant GABA_H receptor/channels (assembled by the γ₂ and ρ₁ subunits) in Xenopus oocytes (Fig. 5A) . In contrast, I4AA is an antagonist of GABA_C receptor/channels in the retina but is an agonist at recombinant GABA_C receptor/channels in Xenopus oocytes. ³ Application of 100 μM I4AA showed neither agonist nor antagonist effects at recombinant GABA_H receptor/channels in Xenopus oocytes (Fig. 5B) . The pharmacologic studies provided additional evidence of the assembly of the γ₂ subunit with the ρ₁ subunit in Xenopus oocytes indicating the importance of subunit composition in the diversity of the GABAergic system. 

Discussion

Subunit composition determines the pharmacologic and electrophysiological properties of the GABA receptor/channels in the brain. Because the retina is a window of the brain, this study is important as well for understanding the GABA receptor/channels in the retina. The native GABA_A receptor/channel is a heterooligomeric and pentameric protein containing γ subunits. ⁵² The GABA_C receptor/channel ρ₁ subunit has been expressed in Xenopus oocytes and forms a homooligomeric receptor/channel with a high affinity for GABA and a large ionic conductance of current. ^{4 5} In addition, pharmacologic and electrophysiological profiles of the GABA_C receptor/channel are distinct from GABA_A receptor/channels. Recently, pharmacologic and electrophysiological studies on some GABA receptor/channels have revealed characteristics different from other GABA receptor/channels. These differences could be due to novel GABA receptor/channels formed by unidentified subunits, splicing variants of cloned subunits, or subunit compositions forming novel hybrid receptor/channels. Considering that some drugs modify CNS responses through their interaction with specific subunits of GABA receptor/channels, a better understanding of their mode of action depends on additional characterization of GABAergic system diversity. Such an understanding can be obtained through further studies on subunit composition of GABA receptor/channels. We undertook in the present study to determine the protein–protein interaction of the GABA γ₂ and ρ₁ subunits in vivo. 

In vivo expression determined by the yeast two-hybrid and the Xenopus oocyte systems is ideal for studying protein–protein interaction and subunit assembly, because subunits are very likely to be expressed in their native conformations in these systems, as opposed to in vitro translation systems. The other advantage of these systems is that they allow examination of the subunit assembly at the translation level. Because C-terminal, part-truncated mutant subunits can be expressed in these systems, there is an additional advantage of using such systems to determine the specific peptide domain(s) that mediate interaction and assembly of these subunits. The affinity for protein–protein interaction in the yeast two-hybrid system can be determined by counting the number of primary colonies grown in the triple-dropout medium. We found, by using the two-hybrid system, that the γ₂ subunit indeed interacts with the ρ₁ subunit and that the affinity of this interaction is similar to the ρ₁-to-ρ₁ subunit interaction to form a homooligomeric receptor/channel. This suggests in neurons expressing both the γ₂ and ρ₁ subunit mRNAs that there may be equal possibilities for homooligomeric ρ₁ assembly or heterooligomeric ρ₁/γ₂ assembly. 

Despite our evidence for assembly and interaction, it is also possible that the reduction in whole-cell current may be due to disruption of translation of the full-length ρ₁ cRNA by the N-terminal cRNA. If this decrease in whole-cell current were due to disruption of translation, one would expect a similar reduction in whole-cell current on coexpressing full-length ρ₁ cRNA with the cRNA of the N-terminal–truncated mutant of the ρ₁ subunit. Coexpression of the N-terminal–truncated ρ₁ mutant cRNA with the full-length ρ₁ cRNA, however, showed no effect on the whole-cell current, compared with homooligomeric ρ₁ expression (data not shown). In fact, increasing the amount of coexpressed γN′ and ρN′ mutants competitively inhibited GABA-induced currents of the homooligomeric ρ₁ subunit in Xenopus oocytes (Fig. 4) . There are two possible explanations for the competitive inhibition: (1) Coexpression of a truncated mutant with a wild-type subunit in Xenopus oocytes resulted in a population of mutant receptor/channels that may not be translocated to the membrane, or (2) if these mutant receptor/channels were translocated to the membrane, they may not be able to form functional chloride channels because of the absence of channel-forming domains. 

The interaction of γ₂ and ρ₁ subunits in both the Xenopus oocyte and yeast two-hybrid systems suggests, but does not necessarily imply, that there is assembly of these subunits in the retina and CNS. We therefore used immunoprecipitation to show that there is assembly of these subunits in the retina and CNS. The fidelity of an immunoprecipitation reaction largely depends on whether there are specific and nonspecific posttranslation interactions in the protein clumping. Our data showed that subunit interaction is a cotranslation process rather than a posttranslation process, by using a fusion protein and tissue lysate incubation assay. In addition, we found that the γ₂ subunit was immunoprecipitated with the ρ₁ subunit when specific antibodies were used, providing strong indication of assembly of these two subunits. Incubation of GST-ρ₁ fusion protein with brain and spinal cord lysates and with the homogenate from Xenopus oocytes expressing the ρ₁ subunit showed that there was no posttranslation subunit–subunit interaction. More evidence in favor of cotranslation subunit interaction comes from coexpressing full-length ρ₁ subunit with ρ₁ mutants that do not have the N-terminal signal sequence in Xenopus oocytes, by using the voltage recording of GABA-induced whole-cell currents (data not shown). In addition, if interaction of subunits could be a posttranslation process, one would expect the mediation of chaperone proteins, which have never been shown for the ionotropic GABA receptor channel subunits. Cotranslation interaction of these subunits indicates that the subunits assemble in the ER or Golgi to form the receptor/channel complex, which is then translocated to the membrane as a complete unit. 

In summary, in our study, we showed, by the yeast two-hybrid expression system, that the γ₂ subunit interacted with the ρ₁ subunit, and we also showed, by the Xenopus oocyte expression system, the assembly of these two subunits. Most important, we showed that the ρ₁ and γ₂ subunits assembled in the retina and CNS to form a novel hybrid GABA receptor/channel. It is very likely that this hybrid receptor/channel possesses pharmacologic and electrophysiological properties that are distinct from those of GABA_A and GABAc receptor/channels. Considering that subunit composition defines GABAergic system properties, additional studies are warranted to explore in the CNS the electrophysiological and pharmacologic profiles of these hybrid GABA receptor/channels. 

 Supported by the American Physiological Society, a Porter fellowship (GME), and National Eye Institute (NIH) Grant EY11653 (LL).

 Submitted for publication November 9, 2001; revised March 8, 2002; accepted March 19, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Putative structures of N-terminal domain of the ρ₁ and γ₂ subunits. The cDNA fragments encoding 292 and 271 amino acids (aa) of ρ₁ and γ₂ subunits, respectively, were amplified with PCR, using the full-length cDNA of ρ₁ and γ₂ subunits.

Figure 2.

 

Determination of interaction between GABA γ₂ and ρ₁ subunits in the brain and spinal cord. (A) Immunoprecipitation of the ρ₁ subunit from rat brain (BR) and spinal cord (SC) using anti-γ₂ antibody. Whole-brain and spinal cord lysates from adult albino rats were incubated overnight at 4°C with a 1:15,000 dilution of anti-γ₂ antibody in a protein A–affinity column. The ρ₁ subunit (∼50 kDa) was detected in Western blot analysis of both precipitates, with anti-ρ₁ antibody. (B) The γ₂ subunit (∼48 kDa) was detected in Western blot analysis of the immunoprecipitates from rat brain and spinal cord, with the same anti-γ₂ antibody that was used to pull down the precipitate. (C) Immunoprecipitation of the ρ₁ subunit from rat brain and retina (RT), with anti-γ₂ antibody. Whole-brain and retinal lysates from adult albino rats were incubated overnight at 4°C with a 1:15,000 dilution of anti-γ₂ antibody in a protein A–affinity column. The ρ₁ subunit (∼50 kDa) was detected in Western blot analysis of both precipitates using anti-ρ₁ antibody. In these experiments, an anti-Erk antibody (polyclonal IgG from rabbit) was used for the negative control. This antibody did not pull down any proteins that could be detected by the anti-ρ₁ or the anti-γ₂ antibody.

Figure 3.

 

Determining interaction of GABA ρ₁ subunit fusion protein with GABA γ₂ subunit from the brain and spinal cord or from exogenous expression in oocytes. (A) Brain and spinal cord lysates from adult albino rats were pooled and incubated with GST-ρ₁ fusion protein that was bound to a GST affinity matrix. After it was raised with PBS, anti-ρ₁ antibody was applied to detect the 63-kDa GST-ρ₁ fusion protein in Western blot, but native ρ₁ protein from the brain and spinal cord was not detected. (B) Homogenates of Xenopus oocytes expressing the GABA ρ₁ subunit were incubated with GST-ρ₁ fusion protein that was bound to a GST affinity matrix. After it was raised with PBS, anti-ρ₁ antibody was used to detect the 63-kDa GST-ρ₁ fusion protein in Western blot, but the ρ₁ subunit protein expressed in Xenopus oocytes was not detected. Immunoprecipitated lysates from brain and spinal cord tissues (A) and from cRNA injected oocytes (B) with anti-Erk antibodies served as the control in Western blot analysis.

Table 1.

 

Growth of Y190 Cells on Synthetic Dropout Media

Construct	−Leu/−Trp/−His+3AT Selection
A.
pAS2-1 ρ1-sense + pACT2 γ N′	+
pAS2-1 ρ1-sense+ pACT2 ρ N′	+
pAS2-1 ρ1-antisense+ pACT2 γN′	−
pAS2-1 ρ1-antisense+ pACT2 ρ N′	−
pAS2-1 ρ1-sense	−
pAS2-1 ρ1-antisense	−
pACT2 γ N′	−
pACT2 ρ N′	−

Construct	Colonies (n)
B.
ρ1-BD/ρ1N′-AD	69 ± 21
ρ1-BD/γ2N′-AD	66 ± 17

Construct	SD −Leu	SD −Trp
C.
pAS2-1 ρ1sense	−	+
pACT2 γ N′	+	−
pACT2 ρ N′	+	−

 (A) Y190 reporter yeast cells were simultaneously transformed with pAS2-1 ρ₁ full-length (BD) and pACT2 γ₂ N-terminus (AD) and plated on a −Leu/−Trp/−His triple-dropout medium. Other Y190 cells were transformed with pAS2-1 ρ₁ full-length and pACT2 ρ₁ N-terminus (positive control), pAS2-1 ρ₁-antisense and pACT2 ρ₁/pACT2 γ₂ to show specific posttranslational interaction. Cells were also transformed with either BD or AD, to show that neither alone could activate the transcription of reporter genes. (B) Number of primary Y190 colonies on the −Leu/−Trp/−His+3AT dropout medium. There is no significant difference in the number of colonies in both homooligomeric and heterooligomeric ρ₁ assembly. (C) Y190 cells were transformed with BD or AD and plated on −Leu or −Trp single-dropout (SD) medium. Growth on the SD medium shows posttransformation auxotrophic integrity of the Y190 cells.

Figure 4.

 

Determination of GABA γ₂ and ρ₁ subunit functional interaction in Xenopus oocytes. (A) Competitive inhibition of GABA-induced ρ₁ subunit activation by C-terminal deletion mutants of the ρ₁ subunit (ρN′). There was a proportionate decrease in whole-cell current when the amount of N-terminal ρ₁ cRNA coexpressed with 25 ng full-length ρ₁ cRNA in Xenopus oocytes increased. Data were analyzed by one-way ANOVA (0.05 significance level, n = 44). Top: Representative traces from whole-cell current recordings in Xenopus oocytes. (B) Competitive inhibition of GABA-induced ρ₁ subunit activation by C-terminal deletion mutants of the γ₂ subunit (γN′). There was a proportionate decrease in whole-cell current when the amount of N-terminal γ₂ cRNA coexpressed with 25 ng full-length ρ₁ cRNA in Xenopus oocytes increased. Data were analyzed by one-way ANOVA (0.05 significance level, n equals; 40).

Figure 5.

 

Effects of bicuculline and I4AA on GABA-induced currents in Xenopus oocytes expressing γ₂ and ρ₁ subunits. (A) The ρ₁ and γ₂ subunits were coexpressed in Xenopus oocytes, and GABA-induced whole-cell currents were recorded by voltage clamp in the presence and absence of the GABA_A receptor/channel antagonist bicuculline. Bicuculline showed no effect on the GABA-induced whole-cell currents in oocytes that heteromerically expressed the γ₂ and ρ₁ subunits. (B) The ρ₁ and γ₂ subunits were coexpressed in Xenopus oocytes, and GABA-induced whole-cell currents were recorded by voltage clamp in the presence and absence of the GABA_C receptor/channel agonist/antagonist I4AA. I4AA (100 μM) showed no agonist or antagonist effect on whole-cell currents in oocytes that heteromerically expressed the γ₂ and ρ₁ subunits.