Proteomic Characterization of Inhibitory Synapses Using a Novel pHluorin-tagged γ-Aminobutyric Acid Receptor, Type A (GABAA), α2 Subunit Knock-in Mouse*

The accumulation of γ-aminobutyric acid receptors (GABAARs) at the appropriate postsynaptic sites is critical for determining the efficacy of fast inhibitory neurotransmission. Although we know that the majority of synaptic GABAAR subtypes are assembled from α1–3, β, and γ2 subunits, our understanding of how neurons facilitate their targeting to and stabilization at inhibitory synapses is rudimentary. To address these issues, we have created knock-in mice in which the pH-sensitive green fluorescent protein (GFP) and the Myc epitope were introduced to the extracellular domain of the mature receptor α2 subunit (pHα2). Using immunoaffinity purification and mass spectroscopy, we identified a stable complex of 174 proteins that were associated with pHα2, including other GABAAR subunits, and previously identified receptor-associated proteins such as gephyrin and collybistin. 149 of these proteins were novel GABAAR binding partners and included G-protein-coupled receptors and ion channel subunits, proteins that regulate trafficking and degradation, regulators of protein phosphorylation, GTPases, and a number of proteins that regulate their activity. Notably, members of the postsynaptic density family of proteins that are critical components of excitatory synapses were not associated with GABAARs. Crucially, we demonstrated for a subset of these novel proteins (including cullin1, ephexin, potassium channel tetramerization domain containing protein 12, mitofusin2, metabotropic glutamate receptor 5, p21-activated kinase 7, and Ras-related protein 5A) bind directly to the intracellular domains of GABAARs, validating our proteomic analysis. Thus, our experiments illustrate the complexity of the GABAAR proteome and enhance our understanding of the mechanisms neurons use to construct inhibitory synapses.

tral nervous system (CNS) (1,2). They are also of therapeutic significance as they are the sites of action for barbiturates, benzodiazepines, general anesthetics, and neuroactive steroids (3). Consistent with their critical roles in regulating neuronal excitability, deficits in the activity of GABA A Rs contribute to a plethora of neurological disorders ranging from anxiety to schizophrenia (4).
Structurally, GABA A Rs can be assembled from 19 different subunits (␣1-6, ␤1-3, ␥1-3, ␦, ⑀, , , and 1-3). The majority of GABA A Rs are believed to be heteropentamers composed of two copies of a single ␣ subunit, two copies of a single ␤ subunit, and one copy of either ␥ or ␦ subunits (5,6). GABA A Rs containing ␣1-3 and ␥ are enriched at inhibitory synapses and mediate phasic inhibition, whereas those containing ␣4 -6 and ␦ are found at extrasynaptic locales and mediate tonic inhibition (1,2). Notably, subunit composition impacts the pharmacological and physiological properties of these varying receptor subtypes (1,7,8). Moreover, GABA A Rs containing unique subunit combinations are selectively targeted to distinct types of inhibitory synapses. However, our understanding of the cellular mechanisms that neurons utilize to regulate GABA A R accumulation at inhibitory synapses is rudimentary. Importantly, the processes that regulate inhibitory synaptogenesis are distinct to those used to build excitatory synapses, which are largely dependent upon PDZ domain-mediated protein-protein interactions (9).
To identify proteins that are relevant for inhibitory synaptogenesis and maintenance, we created knock-in mice in which the pH-sensitive green fluorescent protein (GFP) and the Myc epitope were introduced between amino acids 4 and 5 of the mature GABA A R ␣2 subunit (pH␣2). Following purification on Myc and/or GFP matrices, GABA A R complexes were analyzed by mass spectrometry, and a stable complex of 174 interacting proteins was identified. Importantly, these included the GABA A R ␣1-5, ␤1-3, ␥1-3, and ␦ subunits in addition to the previously identified GABA A R-associated proteins gephyrin (Gphn) and collybistin (Arhgef9). However, 149 of these proteins were novel GABA A R binding partners G-protein-coupled receptors (GPCRs); ion channel subunits; regulators of membrane trafficking and protein stability; modulators of protein phosphorylation; GTPases; and related exchange factors. Sig-and transferred to 30% sucrose solution. Brains were subsequently sliced into 40-m sections and stored in cryoprotectant (30% sucrose, 30% ethylene glycol, 1% polyvinylpyrrolidone in PBS) at Ϫ20°C until use. Sections were washed with PBS before processing. Slide-mounted sections were sequentially washed in 100% ethanol, 95% ethanol, distilled H 2 O and stained with cresyl violet (0.3% glacial acetic acid, 0.5% cresyl violet acetate). This was followed by further rinses in 95% ethanol, 100% ethanol and xylene. Images were acquired with Nikon E800 microscope at 1600 ϫ 1200 resolution using a ϫ4 objective. Twelve sections and three animals per genotype were imaged.
Western Blot Analysis-Proteins separated by SDS-PAGE (8 -10% gel) were transferred to PVDF membranes and blocked in 6% milk in PBST for 1 h. Membranes were further incubated with the appropriate primary antibody (5% milk in PBST), and after extensive washes, they were probed with HRP-conjugated secondary antibodies for 1 h. Western blots were developed using an enhanced chemiluminescence system as per the manufacturer's instructions (Amresco). Membranes were imaged (ChemiDoc MP, Bio-Rad) and analyzed using ImageJ (National Institutes of Health). Two-tailed unpaired t test or analyses of variance with Games-Howell post hoc test (for multiple comparisons with unequal variances) were performed to analyze data (GraphPad, SPSS). Graphs presented show means Ϯ S.E. of the mean (S.E.).
Immunocytochemistry-Hippocampal neurons were prepared from E18 to E19 pH␣2 mice and were used for experiments at 18 days in vitro. For immunocytochemistry experiments, cultures were fixed in 4% paraformaldehyde, 5% sucrose, permeabilized, and probed for the GABA A R ␣2 subunit and gephyrin and were subsequently stained with Alexa Fluor secondary antibodies. 3-5 neurons were imaged from three independent cultures.
Fixed-cell images were acquired using a Nikon Eclipse Ti confocal microscope. Images were taken at 1024 ϫ 1024 resolution with a ϫ60 objective. Calculation of the Pearson's coefficient was performed with the JaCOP (16) plugin for ImageJ software (17).
Coimmunoprecipitation (coIP)-To detect bound gephyrin and collybistin, brains were removed from isoflurane-anesthetized mice (8 -10 weeks). Hippocampi from WT and pH␣2 mice were lysed in lysis buffer containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2% Triton X-100, 5 mM EDTA, 10 mM NaF, 2 mM Na 3 VO 4 , 10 mM Na 4 P 2 O 7 , plus protease inhibitors. These samples were spun at 16,100 ϫ g for 15 min at 4°C, and the supernatant (or lysate) was incubated with 3 g of Myc antibody overnight in lysis buffer (modified to 1% Triton X-100). The addition of protein G-Sepharose beads (GE Healthcare) for 4 h was followed by four quick washes (400 ϫ g, 2 min, 4°C) in 1.5 ml of lysis buffer. For GFP IPs, GFP-Trap beads (Chromotek, catalog no. Gta-200) were incubated with hippocampal lysate overnight. Bound proteins were detected by Western blotting. To detect bound mGluR5, KCTD12, and ephexin, hippocampal/cortical lysates prepared as above were pre-cleared overnight with agarose beads conjugated to IgG. These samples were incubated with GFP-Trap for 2 h and subsequently washed three times for 10 min in 1.5 ml of lysis buffer (modified to 0.2% Triton X-100 and centrifuged at 2500 ϫ g, 2 min, 4°C). Bound proteins were detected by Western blotting. For experiments using HEK293 cells, pre-cleared lysates were incubated with anti-FLAG conjugated beads (Sigma, catalog no. F3165) or GFP-Trap for 2 h and subsequently washed four times in lysis buffer. Bound proteins were detected by Western blotting. A minimum of three independent experiments were performed for all coIP experiments.
Electrophysiology Recordings-After recovery, slices were transferred to a submerged recording chamber on the stage of an upright microscope (Nikon FN-1) with a ϫ40 water immersion objective equipped with DIC/IR optics. Slices were gravitysuperfused with ACSF solution throughout experimentation and perfused at a rate of 2 ml/min with oxygenated (O 2 /CO 2 95:5%) ACSF at 32°C. Adequate O 2 tension and pH 7.3-7.4 values were maintained by continuously bubbling the media with 95% O 2 , 5% CO 2 . Currents were recorded from the dentate gyrus granule cells (DGGCs) in coronal hippocampal slices. Patch pipettes (5-7 megohms) were pulled from borosilicate glass (World Precision Instruments) and filled with intracellular solution (140 mM CsCl, 1 mM MgCl 2 , 0.1 mM EGTA, 10 mM HEPES, 2 mM Mg-ATP, 4 mM NaCl, and 0.3 mM Na-GTP, pH 7.25). A 5-min period for stabilization after obtaining the whole-cell recording configuration was allowed before currents were recorded using an Axopatch 200B amplifier (Molecular Devices), low pass-filtered at 2 kHz, digitized at 20 kHz (Digidata 1440A; Molecular Devices), and stored for off-line analysis. The holding potential was Ϫ60 mV for all recordings.
Electrophysiology Analysis-Tonic current measurements were measured from an all-points histogram that was plotted for a 10-s period before and during picrotoxin application. A Gaussian fit to these points gave the mean current amplitude, and the difference between these two values was considered to be the tonic current and normalized to cell capacitance (pA/ pF). Throughout the course of the experiment, series resistance and whole-cell capacitance were continually monitored and compensated. If series resistance increased by Ͼ20%, recordings were eliminated from the data analysis. Statistical significance was determined using Student's t test. Spontaneous IPSCs (sIPSCs) were analyzed using the mini-analysis software (version 5.6.4; Synaptosoft, Decatur, GA). sIPSCs were recorded for a minimum of 5 min. To detect sIPSCs, the minimum threshold detection was set to three times the value of baseline noise signal. The recording trace was visually inspected, and only sIPSC events with a stable baseline, sharp rising phase, and single peak were used to negate artifacts due to event summation. Only recordings with a minimum of 100 events fitting these criteria were analyzed. 8 -10 cells were recorded from three animals of each genotype. Amplitude, decay, and frequency distributions of sIPSCs were examined by constructing all-point cumulative probability distributions and compared using the Mann-Whitney test and Kolmogorov-Smirnov test. Values of p Ͻ 0.05 were considered significant.
Large Scale Immunoprecipitation for Mass Spectrometry Analysis-Hippocampus and cortex of age-matched (8 -10 weeks) and sex-matched WT and pH␣2 mice (seven animals each) were prepared as above. Lysates were filtered and precleared with agarose beads conjugated to IgG overnight. For tandem IPs, pre-cleared lysates were incubated with Myc antibody overnight. Sepharose beads were added and incubated at 4°C for 4 h. These beads were washed (three times at 400 ϫ g, 2 min, 4°C), and the proteins were eluted off beads with 200 g/ml c-Myc peptide (Alpha Diagnostics) in lysis buffer. The eluate was incubated with GFP-Trap for 1 h, followed by four washes (2500 ϫ g, 2 min, 4°C) in lysis buffer. Gels were run and stained with silver stain (Sigma), and gel bands of interest from pH␣2 and the corresponding regions from WT mice were excised. For single IPs, pre-cleared lysates were incubated with GFP-Trap for 2 h, followed by four washes in lysis buffer (2500 ϫ g, 2 min, 4°C). Gels were run and stained with colloidal Coomassie (18). Each gel lane (for pH␣2 or WT IP) was cut into five pieces and sent to Taplin Mass Spectrometry Facility (Harvard University) for proteomic analysis.
Mass Spectrometry Analysis-Trypsin digestion, liquid chromatography-tandem mass spectrometry (LC-MS/MS), and MS/MS spectra search in a mouse database (Uniprot) using the Sequest 28 analysis program was performed by Taplin Mass Spectrometry Facility (Harvard University). Peptide matches were considered true matches for ⌬CN scores (⌬ correlation) Ͼ0.2 and XCorr values (cross-correlation) of greater than 2, 2, 3, 4 for ϩ1, ϩ2, ϩ3, ϩ4 charged peptides, respectively (supplemental Tables 1 and 2). A particular protein would only be considered present if at least five such high quality peptides were detected. Three independent mass spectrometry experiments were performed. Proteins identified in pH␣2 mice were compared with those found in WT animals to control for nonspecific binding of proteins. Proteins found at similar levels to a list of nonspecific binding proteins often found in mass spectrometry experiments were removed (19). For tandem IP experiments, proteins were identified by a minimum of seven peptides. Peptides found in WT control IPs were removed from the final list of proteins displayed in Table 1. For GFP-Trap IPs, proteins listed in Tables 2-7 have been identified by a minimum of five peptide, or were at least 3-fold enriched in the pH␣2 compared with WT IPs. Furthermore, these peptides were present in all three experiments. Proteins in Tables 2-7 were manually organized into broad functions through information from GeneCards, HUGO gene nomenclature committee, and the literature.
Hippocampal and cortical lysates (prepared as above) from male WT mice were pre-cleared with GST alone. These samples were then incubated with GST tagged to various GABA A R subunits immobilized on glutathione-agarose beads overnight. Beads were washed three times (400 ϫ g, 2 min, 4°C), and bound proteins were detected by immunoblotting. A minimum of three independent GST experiments was performed for each protein studied.

Creation of a pHluorin/Myc-tagged GABA A R ␣2 Subunit
Knock-in Mouse-To date, our understanding of the mechanisms responsible for the formation and maintenance of inhibitory synapses has been limited. These issues are confounded by the structural diversity of GABA A Rs and technical limitations such as the paucity of high affinity subunit-selective antibodies. To overcome these limitations, mice were created in which pHluorin, a pH-sensitive GFP, and the Myc epitope (EQKLISEEDL, Fig. 1, A and E) were introduced into the GABA A R ␣2 subunit. These reporters were introduced into exon 3 of the GABA A R ␣2 subunit gene between the codons encoding amino acids 4 and 5 of the mature protein (pH␣2). This location was chosen because studies in expression systems suggest that the respective modifications are functionally silent (14,20). pH␣2 mice were created using homologous recombination in ES cells, blastocyst injection, and Cre-mediated excision of the neomycin selection marker (Fig. 1B). Mice were genotyped by PCR using primers that detect the presence of pHluorin insertion (Fig. 1C), and the respective mice were backcrossed on the C57BL/6J background in excess of 10 generations. The presence of the pHluorin and Myc reporters was confirmed by DNA sequencing (Fig. 1E).
pH␣2 Subunit Is Associated with Endogenous GABA A R Subunits and Known Receptor-associated Proteins-pH␣2 homozygotes were viable, bred normally, and did not exhibit any overt phenotypes. In addition, Nissl staining did not reveal any gross abnormalities in the structure of the hippocampus between WT and pH␣2 mice (Fig. 1D). To confirm the expres- sion of the pH␣2 subunit, immunoblotting was utilized with ␣2 subunit antibodies. In accordance with the addition of pHluorin, the molecular mass of the ␣2 subunit was increased by ϳ30 kDa in extracts prepared from pH␣2 homozygotes compared with WT ( Fig. 2A). However, there were no significant differences in the total expression levels of the GABA A Rs ␣4 and ␤3   subunit, GAPDH, gephyrin, NMDA receptor NR1 subunit, and tubulin in pH␣2 mice compared with wild type animals ( Fig. 2B; Plasma membrane accumulation of the ␣2 subunit is dependent upon oligomerization with receptor ␤ subunits (1,2,21). To test whether pH␣2 subunits are associated with endogenous receptor ␤ subunits, detergent-solubilized brain extracts were subjected to immunoprecipitation with Myc or GFP antibodies. As measured by immunoblotting, the ␣2 and ␤3 subunits were detected to immunoprecipitate with Myc or GFP antibodies from pH␣2 but not WT brains (Fig. 3, A and B). Molecular, genetic, and biochemical approaches suggest that the multifunctional protein gephyrin and the GDP-GTP exchange factor collybistin play important roles in determining the synaptic accumulation of GABA A Rs (1,2,12,22). Consistent with this, both of these proteins were detected to immunoprecipitate with Myc/GFP antibodies from pH␣2 but not WT brain extracts. Thus, in mouse brain pH␣2 assembles with endogenous GABA A R subunits and is associated with gephyrin and collybistin.
pH␣2 Subunits Are Targeted to Functional Inhibitory Synapses-In the brain, GABA A Rs containing ␣2 subunits are highly concentrated at inhibitory synapses (1,2,23,24). To assess whether this synaptic targeting also occurs in pH␣2 mice, 18 days in vitro hippocampal cultures produced from these mice were stained with ␣2 and gephyrin antibodies and imaged by confocal microscopy. Endogenous green fluorescence colocalized with GABA A R ␣2 subunit immunoreactivity ( Fig. 3C; p Ͻ 0.001) at gephyrin-positive postsynaptic inhibitory specializations ( Fig. 3C; p Ͻ 0.005).
Next, we compared the properties of phasic and tonic inhibition in the dentate gyrus granule cells (DGGCs) of WT and pH␣2 mice (Fig. 4). Examination of sIPSCs revealed that there was no significant difference in the amplitude ( Collectively, these data suggest that GABA A Rs containing pH␣2 subunits are targeted to inhibitory synapses, and their incorporation at these subcellular specializations does not have an impact on GABAergic inhibition. Isolation of GABA A Rs from the Brains of pH␣2 Mice Using Two-step Tandem Affinity Purification-To assess which proteins associate with GABA A R subunits in the brain, a two-step immunoaffinity purification protocol was performed. First, hippocampi and cortices from age/sex-matched WT and pH␣2 mice were solubilized and exposed to Myc antibody followed by binding to G-Sepharose beads. After extensive washes, bound material was eluted with Myc peptide and exposed to immobilized GFP-Trap beads. Bound material was subsequently eluted using 2% SDS and subjected to SDS-PAGE followed by silver staining. Bands that were present in the pH␣2 lane and the adjacent lane from WT mice were excised and subjected to LC-MS/MS (Fig. 5). Three independent purifications were performed for both WT controls and pH␣2 animals. Table 1 shows a list of the proteins identified by MS analysis that associate with pH␣2. Proteins listed were identified by a minimum of seven peptides. Furthermore, proteins that bound non-specifically (in WT controls) were removed. Using these criteria, the GABA A R ␣1, ␣3, ␣4, ␣5, ␤1, ␤2, ␤3, and ␥2 subunits in addition to the ␣1 subunit of the Na ϩ /K ϩ -ATPase subunit copurified with the pH␣2 (Table 1 and supplemental Table 1). Although there was some contamination between bands, the majority of GFP and ␣2 subunit peptides were identified in the major silver-stained product at ϳ80 kDa. Atp1a1 was found at the 100-kDa region, ␣4 subunit at the 65-kDa region, and the rest were found in the 50 -55-kDa region of the gel. Collectively, these results suggest that pH␣2 is capable of assembling with the ␥2 and multiple ␣ and ␤ subunit isoforms in the brain.
GFP-Trap Purification of GABA A Rs Reveals Their Association with Known Binding Partners-To increase the probability of identifying proteins that are associated with the ␣2-containing GABA A Rs, a single purification with GFP-Trap was used. Lysates from hippocampi and cortices of age-and sex-matched WT and pH␣2 mice were incubated with GFP-Trap beads. These samples were then subjected to SDS-PAGE followed by Coomassie staining (Fig. 6). The single step purification method led to an increased yield of protein compared with the tandem purification as indicated by the increased number of peptides identified and greater protein coverage for GABA A R ␣2 (Figs. 5B and 6B; GFP/myc IP 8.4%, GFP IP 43%). Three independent purifications were performed, and proteins iden-tified by LC-MS/MS in all three experiments and found to be at least 3-fold enriched in the pH␣2 samples are listed in Tables  2-7 and supplemental Table 2.
Identification of Novel Components of the GABA A R Proteome Using GFP-Trap Purification-In addition to known interacting proteins as detailed in Table 2, 149 novel binding partners for GABA A Rs were identified in material purified from pH␣2 animals. For brevity, these proteins were divided into five FIGURE 6. Single-step purification to isolate pH␣2 complexes. Detergent-solubilized hippocampal and cortical lysates from WT and pH␣2 mice were immunoprecipitated with GFP antibodies and subjected to SDS-PAGE and colloidal Coomassie staining (A). Each gel lane was cut into five pieces and pooled for mass spectrometry analysis. Protein coverage of GABA A R ␣2 subunit (blue, underline) identified by MS analysis (B). Example of MS/MS spectrum for tryptic peptide identified as GABA A R ␣2 is shown (C). The sequence of the identified peptide is indicated.

TABLE 1 Proteins identified with pH␣2 identified using tandem myc/GFP-Trap purification
Summary of MS/MS analysis results of proteins associated with pH␣2 after purification using Myc and pHluorin tag from three independent experiments. Age-and sex-matched WT mice were used as controls for non-specific binding of proteins. Total peptides indicate the sum of peptides found in all experiments.

Gene symbol Reference Name
Total peptide

WT pH␣2
GFP-Aequorea Green fluorescent protein 0 groups based on literature searches of their presumed functions: 1) G-protein coupled receptors (GPCRs), ion channels, and transporters (Table 3); 2) regulators of protein trafficking, stability, and cytoskeletal anchoring (Table 4); 3) regulators of GTP exchange and protein phosphorylation (Table 5); 4) miscellaneous enzymes (Table 6); and 5) miscellaneous proteins (Table 7). These various binding partners presumably act sequentially to control receptor assembly, forward trafficking in the secretory pathway, trafficking to and stabilization at inhibitory synapses, receptor endocytosis, and endocytic sorting followed by lysosomal or proteosomal degradation.
Cullin1, Ephexin, KCTD12, Mitofusin2, mGluR5, PAK5/7, and Rab5 Bind to the Intracellular Loop of Specific GABA A R Subunits-To confirm our MS findings, we examined the binding of selected hits to the intracellular domains of GABA A R subunits. Our initial studies focused on the GPCR mGluR5 (Grm5), the kinase PAK5/7 (Pak7), the GTPases mitofusin2 (Mfn2), and Rab5, the Rho guanine nucleotide exchange factor ephexin (Ngef) and regulator of ubiquitination cullin1 (Cul1) (Tables 3-5). These proteins were chosen for their range in the total number of peptides identified by MS analysis as follows: from a lower number of peptides (e.g. mGluR5; 0 peptides WT

Known binding partners of GABA A R subunits and their closely associated proteins identified using GFP-Trap purification
Proteins associated with pH␣2 were purified using pHluorin tag from three independent experiments. Age-and sex-matched WT mice were used as controls for non-specific binding of proteins. Proteins listed have appeared in all three experiments, have been identified by a minimum of five peptides, and there is a 3-fold difference between peptides found in pH␣2 compared with WT IPs.
The potential interaction of ephexin with GABA A Rs was of particular interest because ephexin belongs to the same family of GDP-GTP exchange factors (GEFs) as collybistin, a molecule that plays a key role in determining the formation of hippocampal inhibitory synapses (22,41). To further corroborate our findings in pH␣2 mice, we expressed FLAG-ephexin, pH␣2, and ␤3 in HEK293 cells. Reciprocal immunoprecipitation with FLAG and GFP antibodies revealed the robust association of ephexin with GABA A Rs in HEK293 cells (Fig. 8B).
Together, these studies demonstrate that proteins identified by mass spectroscopy can be validated in the brain and in expression systems.

Discussion
Inhibitory fast synaptic transmission is critically dependent upon the accumulation and stabilization of selected GABA A R subtypes at inhibitory postsynaptic specializations. To further elucidate the processes neurons utilize to regulate the synaptic accumulation of these critical ligand-gated ion channels, we have created mice in which the ␣2 subunit is modified with pHluorin and Myc reporters by targeting the respective gene using homologous recombination. These reporters were introduced between residues 4 and 5 of the mature subunit. pH␣2 homozygotes were viable and did not exhibit any overt phenotypes but exhibited endogenous fluorescence at inhibitory synapses. Moreover, the properties of sIPSCs and tonic currents, the unitary events that underlie phasic and tonic inhibitory synaptic transmission, were similar between genotypes. Importantly, gephyrin and collybistin, which were previously reported to associate with GABA A R ␣2 in HEK293 cells (22), could be shown to coimmunoprecipitate in brain lysates, highlighting the necessity for the tagged protein to enable high affinity purifications.
Consensus opinion suggests that the ␣1-3 subunits are components of synaptic GABA A Rs and that the anxiolytic and sedative properties of benzodiazepines are mediated by specific receptor subtypes containing individual ␣ subunit isoforms. Therefore, we assessed which receptor subunits associate with pH␣2 using tandem purification on Myc and GFP antibodies followed by LC-MS/MS. This approach revealed that the pH␣2 subunit copurified with ␣1, ␣3, ␣4, ␣5, ␤1-3, and ␥2 subunits. Using GFP-Trap alone, we further detected association with the ␥1, ␥3, and ␦ subunits. Although these results are not quantitative and do not discriminate between surface and intracel- lular populations, our results do suggest the existence of multiple receptor subtypes with mixed ␣ and/or ␤ subunits, supporting previous observations of the coexistence of different ␣ subunits in a single receptor complex (42)(43)(44)(45)(46). Consistent with our results, previous studies to identify proteins associated with the GABA A R ␣5 subunit through MS analysis exclusively identified other GABA A R subunits, including ␣1-3, ␣5, ␤1-3, and ␥2 (47). A more recent investigation into the proteins associated with the GABA A R ␣1 subunit isolated 18 associated proteins via MS analysis, more than half of which were other GABA A R subunits (48), further supporting the possibility of a more heterogeneous population of receptors than originally predicted (5,49). It is important to note that some of these subunit interactions may represent "non-productive" or nonfunctional receptor assembly intermediates that are not present on the plasma membrane (1,2,25,50). Because GABA A Rs are a major target for pharmacological agents such as benzodiazepines, barbiturate, neurosteroids, and general anesthetics (3), the heterogeneity of these receptors may have major implications in the design of subunit-selective drugs for therapeutic use.
In addition to receptor subunits, we also isolated the known GABA A R binding partners gephyrin, collybistin, PKC, PKA, and GABA B R2. To the best of our knowledge, this is the first time that these respective protein-protein interactions have been simultaneously demonstrated for GABA A Rs in their native environment. The use of a single GFP-Trap protein purification yielded a 174-multiprotein complex comprising 149 novel protein components that copurified with pH␣2 compared with material isolated from WT mice. Novel components of the GABA A R complex include other receptors, proteins required for trafficking, ubiquitination/degradation, GTPases and their regulators, cytoskeletal components, and a host of enzymes. Significantly, the PSD95 family of proteins, which is enriched in excitatory synapses (32), was absent from these purifications.
As an initial means of assessing the significance of our MS experiments, we tested the interaction of selected proteins from brain extracts with GST fusion proteins encoding the intracellular domains of GABA A R subunits. Our studies focused on mGluR5, PAK5/7, mitofusin2, Rab5, ephexin. and cullin1 due to the availability of suitable antibodies. All of the proteins bound to the intracellular domains of the receptor ␣1, ␣2, ␤3, or ␥2 subunits, confirming the veracity of our GFP-Trap purifications.
We further validated some of the MS results by demonstrating that mGluR5, KCTD12, and ephexin coIP with pH␣2 from brain lysates. We are particularly interested in ephexin due to some similarities with collybistin. Collybistin is a member of the Dbl family of GEFs necessary for the proper clustering of gephyrin and gephyrin-dependent GABA A Rs (41). Like collybistin, ephexin also belongs to the Dbl family of GEFs and therefore has a similar domain structure to collybistin. Studies on ephexin have described its role in axon guidance in retina ganglion cells (13) and dispersal of synaptic acetylcholine receptor clusters in the neuromuscular junction through its capacity to activate Rho family GTPases (51). Numerous regulators of the actin cytoskeleton such as the Rho family GTPases have been demonstrated to be critical for synapse remodeling at excitatory synapses (52). In addition, similar roles for the regulation of the actin cytoskeleton at inhibitory GABAergic synapses have only more recently begun to emerge (53). Although how ephexin, other GTPases, and GTPase regulators identified here may affect GABA A Rs remains to be seen, it is tantalizing to speculate that they may have similarly important roles at inhibitory synapses.
Typical contaminants such as highly abundant proteins (e.g. actin, tubulin, and ribosomal proteins) and proteins that bind unfolded proteins (e.g. heat shock proteins) are commonly found in affinity-purified protein preparations (54). Our use of proper WT controls removed many of these contaminants. Furthermore, the requirement for the detection of proteins from three different experiments unveiled protein binding partners that may weakly but stably form a complex with pH␣2. Thus, potential pH␣2-associated proteins cannot readily be discarded due to a low number of total peptides discovered. Indeed, although only six peptides were identified for mGluR5, we demonstrated that it was robustly coimmunoprecipitated with pH␣2 (Fig. 8A).
Previously described GABA A R-associated proteins have been demonstrated to be essential for regulatory processes crucial for GABA A R function (1,2,55). The characterization of the protein components that form the inhibitory synaptic complex described here have wide-ranging ramifications for the understanding of GABA A R activity and trafficking and therefore its role in synaptic transmission and plasticity. The vast majority of proteins purified here are novel putatively GABA A R-associated proteins, indicating that the inhibitory synapse is likely to be far more complex than initially appreciated. Thus, the challenge still remains to elucidate the effects of these associations on GABA A Rs. Considering the crucial role of GABA A R in brain function, it is of fundamental importance to ascertain the underpinning mechanisms that govern these receptors thereby clarifying its role in CNS health and disease.