Interaction between GABA(A) receptor beta subunits and the multifunctional protein gC1q-R.

gamma-Aminobutyric acid type A (GABA(A)) receptors were immunopurified from bovine brain using a monoclonal antibody directed against the alpha1 subunit. Of the several proteins that copurified, a 34-kDa protein was analyzed further. After enrichment and tryptic proteolysis, the resulting fragments were sequenced, and the protein was identified as gC1q-R. Using anti-gC1q-R and anti-GABA(A) receptor antibodies, mutual coimmunoprecipitation could be demonstrated from solubilized rat brain membranes. The stability of this interaction was estimated to be very high. Using the yeast two-hybrid system, various GABA(A) receptor subunit intracellular loop constructs were tested for an interaction with gC1q-R. All beta subunits, but not alpha 1 and gamma 2 subunits, were found to bind to gC1q-R. NH(2)- and COOH-terminally truncated beta 2 subunit loops were used to find the region responsible for the interaction with gC1q-R. A stretch of 15 amino acids containing 7 positively charged residues was identified (amino acids 399--413). This region contains residue Ser-410, which is a protein kinase substrate, and it is known that phosphorylation of this residue leads to an alteration in receptor activity. Localization studies suggested a predominantly intracellular localization. Our observations therefore suggest a tight interaction between gC1q-R and the GABA(A) receptor which might be involved in receptor biosynthesis or modulation of the mature function.

As other ion channels promoting fast synaptic transmission, GABA A receptors are clustered at postsynaptic membranes (11)(12)(13). It is presumed that clustering at synapses is achieved by anchoring the receptors to the cytoskeleton. Tubulin and actin as well as a number of additional proteins were found to coimmunoprecipitate with the GABA A receptor from bovine brain (14). Recently, the GABA A receptor-interacting protein (GABARAP) has been described. Apart from binding to the ␥2 subunit of GABA A receptors, this protein also interacts with tubulin and gephyrin (15) and promotes receptor clustering in QT-6 cells (16). However, immunofluorescence experiments in retinas showed no colocalization between GABARAP and the GABA A receptor ␥2 subunit at synapses. In cultured cortical neurons GABARAP was found to be localized to intracellular compartments, implicating a function in receptor sorting and targeting processes that precede and/or initiate receptor clustering (15).
GABA A receptors can be functionally modulated by phosphorylation of specific residues on the intracellular loops of ␤ and ␥ subunits by serine/threonine kinases such as PKC, PKA, Ca 2ϩ /calmodulin-dependent protein kinase, and cGMPdependent protein kinase (PKG) as well as the tyrosine kinase Src (17)(18)(19)(20)(21)(22)(23)(24). PKC␤II as well as the receptor for activated protein kinase C (RACK-1) were shown to coimmunoprecipitate with ␤ subunits from rat brain (25). Also a yet unidentified serine/threonine kinase with a presumable molecular mass of 34 kDa was coimmunoprecipitated with ␣1 containing GABA A receptors from bovine brain (26). In this study we attempted to identify and clone this protein. A protein of the same molecular mass was biochemically purified, digested with trypsin, and the resulting fragments were sequenced by Edman degradation. Our results identify this protein as the multifunctional protein gC1q-R (27). We show that gC1q-R interacts with the GABA A receptor with very high affinity. The interaction site on the GABA A receptor was located to a basic region within the intracellular loop of ␤ subunits which also contains functionally relevant phosphorylation sites.

EXPERIMENTAL PROCEDURES
Immunoprecipitation of GABA A Receptors from Bovine Brain Membranes and Purification of the Associated 34-kDa Protein-GABA A receptors were immunoprecipitated from Zwittergent 3-14-solubilized calf brain membranes as described previously (14). For precipitation, the monoclonal antibody bd24 (28,29) covalently coupled to protein A-Sepharose (bd24 beads) was used. This method coimmunoprecipitates a 34-kDa protein with the GABA A receptor. 6 g of bovine brain membrane protein was used for the immunoprecipitation, which yielded about 3 g of the 34-kDa protein. The 34-kDa protein was separated from the GABA A receptor-bd24 beads complex by extraction with immunoprecipitation buffer containing 0.6 M NaCl (14).
Tryptic Digest and Peptide Sequencing of the 34-kDa Protein-After chloroform-methanol precipitation (30) of the high salt extracted 34-kDa protein, the pellet was resuspended in 10 l of 8 M urea, 10 mM dithiothreitol and incubated at 50°C for 30 min. After cooling to room temperature, 2 l of 100 mM iodoacetamide was added. The solution was incubated for 15 min followed by the addition of 87 l of 50 mM Tris (pH 8.0), 1 mM CaCl 2 , and 1 l trypsin (1 g/l). The protein was trypsin treated for 12 h at 37°C. Peptides were separated by reverse phase high performance liquid chromatography on a C-18 column (Vydac, 1-mm inner diameter ϫ 250 mm) using the linear gradient between solvent A (0.1% v/v trifluoroacetic acid) and solvent B (0.09% v/v trifluoroacetic acid, 80% v/v acetonitrile) at a flow rate of 50 l/min. The following gradient was used: 5% solvent B for 5 min, 5-60% solvent B for 60 min. Peptide masses were determined on-line by mass spectrometry (Sciex, API IIIϩ), using ϳ10% percent of the effluent. The remaining 90% of the effluent was collected manually for sequence determination.
Peptide sequences were determined by Edman degradation, using a model G1005A protein sequencer (Agilent Technologies).
cDNA Cloning and Overexpression of gC1q-R-Recombinant rat gC1q-R (mature form, starting at amino acid residue 72) was expressed as a His-tagged fusion protein in Escherichia coli using plasmid pQE-31 (Qiagen, Hilden, Germany). Two yeast expression libraries from rat hippocampus/cortex (oligo(dT) and randomly primed, kindly provided by Dr. P. R. Worley (Johns Hopkins University, Baltimore)) were used as a template for polymerase chain reaction using primer Ma32f (5Ј-CGGGATCCTCTGCACACGGAAGGAGACAAGG-3Ј) designed to allow in-frame cloning into pQE-31 and primer Ma32r (5Ј-GAGAGCTGGC-CGTTTATATTAGAG-3Ј). The polymerase chain reaction product was digested with BamHI and subcloned into pQE-31 cut with BamHI and SmaI. Expression was performed according to the manufacturer's protocol. Recombinant protein was purified using a nickel-nitrilotriacetic acid matrix (Qiagen). Full-length and mature gC1q-R (starting at residue 72) were also cloned into the mammalian expression vector pCMV (31) for expression of gC1q-R in HEK 293 cells.
Production of a Polyclonal Antibody against gC1q-R-Three Burgunder rabbits were injected with 1 mg of recombinant gC1q-R dissolved in 1 ml TiterMax TM (CytRx, Norcross, GA). Two subsequent immunizations were made at intervals of 2 weeks before the rabbits were bled. Blood was incubated at 37°C for 60 min to let it clot and incubated at 4°C overnight to let it contract. Serum was centrifuged at 10,000 ϫ g for 10 min at 4°C, and the supernatant was collected and stored at Ϫ80°C.
Immunoprecipitation of gC1q-R and the GABA A Receptor from Rat Brain Membranes-Demyelinated rat brain membranes were prepared from fresh tissue essentially after the procedure of Bennett et al. (32). Membranes were solubilized in 100 mM KCl, 2 mM MgCl 2 , 1 mM EGTA, 10 mM HEPES at pH 7.5, and 1% (w/w) Triton X-100 (Pierce). The protease inhibitors pepstatin, antipain, and leupeptin were used at concentrations of 5 g/l, and 1 mM phenylmethanesulfonyl fluoride was included. The protein concentration of the solubilisate was 5 mg/ ml. gC1q-R and GABA A receptors were immunoprecipitated by adding 5 l of anti-gC1q-R serum or 12 l of anti-␤3 (1-13), an antiserum that recognizes ␤2 and ␤3 subunits (33), to 2 ml solubilisate. Incubation was carried out at 4°C overnight. The antibodies were recaptured with 5 l of protein A beads and washed five times with 1 ml of solubilization buffer containing 1% (w/w) ␤-octyl glucoside instead of Triton X-100.

Quantification of Coimmunoprecipitated GABA A Receptors by Radioligand Binding Assay Using [ 3 H]Ro 15-1788 -Washed protein A beads
containing bound anti-gC1q-R antibody, gC1q-R, and the GABA A receptor were resuspended in buffer containing 10 mM potassium phosphate, 100 mM KCl, and 0.1 mM K-EDTA (pH 7.4). Resuspended beads (0.5 ml) were incubated for 90 min on ice in the presence of 3 nM [ 3 H]Ro 15-1788 (87 Ci/mmol) (NEN Life Science Products). Nonspecific binding was determined in the presence of 25 M unlabeled Ro 15-1788. Beads were collected through rapid filtration on GF/C filters presoaked in 0.3% polyethyleneimine. After three washing steps with 4 ml of buffer each, the filter-retained radioactivity was determined by liquid scintillation counting. The specific binding at 3 nM is close to the maximal binding and therefore allowed an estimation of the absolute amount of GABA A receptor.
Quantification of Coimmunoprecipitated gC1q-R by Immunoblotting-After immunoprecipitation and washing of the beads (5 l), gC1q-R was extracted with 20 l of washing buffer containing 0.6 M NaCl as described before (14). After SDS-polyacrylamide gel electrophoresis on 10% gel, proteins were transferred to nitrocellulose sheets (Hybond-ECL; Amersham Pharmacia Biotech) and immunodecorated as described previously (14). Anti-gC1q-R serum was used at 1:1,000. After incubation with a peroxidase-conjugated anti-rabbit secondary antibody (1:1000, DAKO), positive bands were detected with the ECL system (Amersham Pharmacia Biotech). Coimmunoprecipitation effi-ciency was determined by comparing the intensity of the gC1q-R band from the immunoprecipitate with standard amounts of rat brain membranes.
Determination of the Dissociation Kinetics of the GABA A Receptor from gC1q-R-gC1q-R was immunoprecipitated from Triton X-100-solubilized rat brain extract with anti-gC1q-R antibody. After centrifugation and washing of protein A beads, one aliquot of protein A beads was used in a radioligand binding assay to determine the amount of bound GABA A receptor directly after the immunoprecipitation. Two other aliquots of protein A beads were first transferred to tubes containing 50 ml of solubilization buffer containing 1% (w/w) ␤-octyl glucoside instead of Triton X-100 to allow associated GABA A receptor to dissociate. The tubes were incubated at 4°C for 60 min and 360 min, respectively, with constant agitation. Thereafter the protein A beads were washed, and the residual amount of GABA A receptor bound to the gC1q-R complex was determined by radioligand binding assay. The experiment was carried out twice as indicated above and twice using the same dissociation buffer supplemented with 5 mM dithiothreitol to prevent an S-S double bond formation. Dithiothreitol did not influence dissociation behavior.
Yeast Two-hybrid Assay-The Matchmaker TM two-hybrid system (CLONTECH) was used to assay for an interaction between gC1q-R and the GABA A receptor subunit intracellular loops. Mature rat gC1q-R (starting from residue 72) as well as COOH-and NH 2 -terminally truncated gC1q-R-constructs were cloned into pGAD424 (CLONTECH). The ␣1, ␤1, ␤2, ␤3, and ␥2 subunit loops as well as NH 2 -and COOHterminally truncated ␤2 loop constructs were cloned into pGBT9 (CLONTECH). Expression of the two fusion proteins was done in the yeast strain SFY526. A ␤-galactosidase filter assay was used to test for interactions between gC1q-R and GABA A receptor subunit loop constructs. The assay was made twice with n ϭ 5 for each interaction tested.
Transfection of Recombinant GABA A Receptors and gC1q-R in Cultured Cells-HEK 293 (American Type Culture Collection, Rockville, MD) cells were maintained in minimum essential medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum, 2 mM glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin through standard cell culture techniques. ␣1, ␤2, ␥2, or ␣1, ␤3, ␥2 subunits (34 -37) alone or in conjunction with the mature or full-length gC1q-R were transfected into HEK 293 cells using the calcium phosphate precipitation method (38). Equal amounts (total of 20 g of DNA/90-mm dish) coding for GABA A receptor subunits and/or gC1q-R were used.
Confocal Laser Scanning Microscopy-Rat hippocampal cell cultures were prepared as described (39). Hippocampal cell cultures used for mitochondrial staining were incubated with 200 nM Mito Tracker Red (Molecular Probes) for 30 min at 37°C. For fixation of hippocampal cultures or transfected HEK 293 cells, medium was washed with a saline solution containing 135 mM NaCl, 5 mM KCl, 2 mM MgCl 2 , 2 mM CaCl 2 , 10 mM HEPES, and 30 mM glucose (pH 7.4). The saline was then replaced by 0.1 M phosphate buffer (pH 7.4), and eventually by fixative solution containing 0.1% glutaraldehyde and 4% paraformaldehyde in the same buffer. Fixation was made overnight at 4°C. Prior to immunocytochemistry, cells were incubated with 0.2% Triton X-100 for 10 min and preincubated with 5% normal goat serum for 1 h. The monoclonal antibody bd17 against ␤2/␤3 subunits was used at a dilution of 1:200 and the anti-gC1q-R serum at 1:1000, both in 50 mM Tris-buffered saline. Rhodamine-coupled anti-rabbit IgGs (Jackson Immunoresearch) and FITC-linked anti-mouse IgGs were diluted 1:50 in 13 mM phosphate-buffered saline and incubated either alone or simultaneously for 2 h. Coverslips were mounted on slides with 80% (v/v) glycerol and 0.2% paraphenyldiamine in 0.1 M phosphate (pH 8.6). The cells were visualized using a Zeiss laser scanning microscope (LSM 510). Rhodamine was excited by an HeNe Laser (543 nm), FITC with an argon laser (488 nm).
Construction of Receptor Subunits-The cDNAs coding for the ␣1, ␤2, and ␥2S subunits of the rat GABA A receptor channel have been described elsewhere (34 -36). For cell transfection, the cDNAs were subcloned into the polylinker of pBC/CMV. This expression vector allows high level expression of a foreign gene under control of the cytomegalovirus promoter and a SP6 promoter for in vitro transcription. The ␣ subunit was cloned into the EcoRI, and the ␤ and ␥ subunits were subcloned into the SmaI site of the polylinker by standard techniques.
Expression and Functional Characterization-Xenopus laevis oocytes were prepared, injected, defolliculated, and currents recorded as described (40). Briefly, oocytes were injected with 50 nl of cRNA dissolved in 5 mM K-HEPES (pH 6.8). This solution contained the transcripts coding for the different subunits at a concentration of 10 nM for ␣1, 10 nM for ␤2, 50 nM for ␥2 together or without 150 nM of either full-length or mature gC1q-R. RNA transcripts were synthesized from linearized plasmids encoding the desired protein using the message machine kit (Ambion) according to the recommendation of the manufacturer. A poly(A) tail of about 300 residues was added to the transcripts using yeast poly(A) polymerase (United States Biochemical or Amersham Pharmacia Biotech). The cRNA combinations were coprecipitated in ethanol and stored at Ϫ20°C. Transcripts were quantified on agarose gels after staining with Radiant Red RNA stain (Bio-Rad) by comparing staining intensities with various amounts of molecular mass markers (RNA-Ladder, Life Technologies, Inc.). Electrophysiological experiments were performed by the two-electrode voltage clamp method at a holding potential of Ϫ80 mV. The medium contained 90 mM NaCl, 1 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM Na-HEPES (pH 7.4). GABA was applied for 20 s, and a washout period of 4 min was allowed to ensure full recovery from desensitization. Concentration-response curves for GABA were fitted with the equation where c is the concentration of GABA, EC 50 is the concentration of GABA eliciting half-maximal current amplitude, I max is the maximal current amplitude, I is the current amplitude, and n is the Hill coefficient.

RESULTS
Purification and Identification of gC1q-R-The strategy used for the identification of gC1q-R and characterization of its association with the GABA A receptor is shown in Fig. 1. Calf brain membranes were solubilized with Zwittergent 3-14 and subjected to immunoprecipitation with the monoclonal antibody bd24 covalently bound to Sepharose beads. This procedure immunoprecipitates the GABA A receptor and leads to the copurification of several additional proteins, five of which show prominent bands on a silver-stained gel (14). Incubation of the bd24 precipitate with 0.6 M NaCl extracted a 34-kDa protein (14). About 3 g of the 34-kDa protein was extracted from bd24 immunoprecipitates and concentrated by methanol-chloroform precipitation (30). Peptides resulting from a tryptic digest of the 34-kDa protein were separated and subjected to Edman sequencing. Thereby the partial sequence EVSFQATGESD was obtained which matched the human gC1q-R, a protein of molecular mass 33 kDa (Swiss Protein Q07021). Nine out of 11 residues were identical to the human protein. Of the other 2 residues the alanine residue is also found in the mouse protein, and the aspartic acid residue is conservatively substituted in the human protein by a glutamic acid residue.
An additional partial sequence (A/P)(Q/E)(D/E)QNPVLTSX-PNF also showed high homology to the human gC1q-R and therefore confirmed the identity of the 34-kDa protein with gC1q-R. Further evidence came from a Western blot in which a monoclonal antibody against the human gC1q-R (a generous gift from Dr. B. Ghebrehiwet, SUNY at Stony Brook) gave a weak signal for the bovine 34-kDa protein (not shown).
Three independent clones of the gene for gC1q-R were obtained by polymerase chain reaction from two different rat brain cDNA libraries. All three clones had the same DNA sequence, which differed in several positions from the published sequence of the original clone (41). On the amino acid level, 6 amino acid residues were different, and in addition, one amino acid residue was introduced. Interestingly, all 7 of these amino acid residues are present at the same homologous locations in the mouse gC1q-R gene. Furthermore an expressed sequence tag (GenBank BE128085) from a rat cDNA library was found that contained five of the same mismatching codons as our cloned sequence. The other two differing codons are at a location not encoded by the expressed sequence tag. Differences in our clones compared with the data base entry of the original clone are thus suggestive of a polymorphism in the gC1q-R gene.
The open reading frame of the gC1q-R encodes an immature protein of 279 amino acid residues of which the first 71 amino acids are cleaved off during maturation. It is believed that this sequence is involved in cellular translocation and contains a mitochondrial targeting sequence (41)(42)(43). The DNA coding for the mature form of the protein was subcloned into the bacterial expression vector pQE-31 allowing overexpression of Histagged gC1q-R. The mature and the immature form of gC1q-R were also subcloned into the mammalian expression vector pCMV allowing transfection of HEK 293 cells. Overexpressed mature gC1q-R was used for the production of a rabbit polyclonal antibody.
gC1q-R Is Coimmunoprecipitated with the GABA A Receptor from Rat Brain-Initially gC1q-R had been coimmunoprecipitated with the GABA A receptor from bovine brain membranes. After the production of an antibody against the rat gC1q-R we tried to purify gC1q-R from rat brain by coimmunoprecipitation with the GABA A receptor, using a different immunoprecipitation technique. The GABA A receptor was immunoprecipitated from Triton X-100-solubilized rat brain membranes with an antibody against the ␤2/␤3 subunits. As negative controls the same procedure was done in parallel without membranes or alternatively without antibody. After the addition of protein A beads, the immunoprecipitates were extracted with buffer containing 0.6 M NaCl. This separates the gC1q-R protein from the insoluble GABA A receptor complex. This procedure was chosen because of the high background signal that arose from the presence of protein A beads. High salt extracts of the immunoprecipitation, the negative controls, and a standard amount of rat brain membranes (8 g of protein) were loaded on a gel for Western blotting with anti-gC1q-R antibody (Fig. 2). The signal resulting from the coimmunoprecipitated gC1q-R (Fig. 2, lane 2) was about as strong as the signal resulting from 8 g of membrane protein (Fig. 2, lane 1). The negative controls gave no signal (Fig. 2, lanes 3 and 4). The experiment was carried out twice with similar results.
The GABA A Receptor Is Coimmunoprecipitated with gC1q-R-Rabbit polyclonal anti-gC1q-R antibody was used to immunoprecipitate gC1q-R from Triton X-100-solubilized rat brain homogenate (Fig. 3A). Coimmunoprecipitated GABA A receptor was detected by benzodiazepine binding using [ 3 H]Ro 15-1788 (Fig. 3B). This method was chosen because the heavy chain of the precipitating antibody comigrates with the GABA A receptor on SDS-polyacrylamide gel electrophoresis. As negative controls the immunoprecipitation was made either without protein or without anti-gC1q-R antibody (Fig. 3B). A second analogous experiment gave comparable results.
Stability of the GABA A Receptor-gC1q-R Complex-During immunoisolation experiments from bovine brain it became evident that the affinity between gC1q-R and the GABA A receptor must be very high. Washing the protein A beads usually took between 15 and 20 min. It was estimated that during this time most of the GABA A receptor would have dissociated from gC1q-R if the K d characterizing this interaction would have been higher than about 1 nM. To obtain more information on the stability of the GABA A receptor-gC1q-R complex, we carried out a simplified dissociation rate experiment. gC1q-R was immunoprecipitated from Triton X-100-solubilized rat brain extract with anti-gC1q-R antibody and protein A. After centrifugation and washing of the precipitated immune complex, one aliquot of the precipitate was used in a radioligand binding assay to determine the amount of bound GABA A receptor directly after the immunoprecipitation. Two other aliquots of the precipitate were first transferred to tubes containing 50 ml of buffer to allow bound GABA A receptor to dissociate. The tubes were incubated at 4°C for 60 min and 360 min, respectively. After washing, the residual amount of GABA A receptor bound to the gC1q-R complex was determined by radioligand binding assay (Table I). After 360 min only about 40% of the initially bound GABA A receptor had dissociated from gC1q-R, indicating a high stability of the protein complex. Assuming a monoexponential decay a k off value of about 1.4*10 Ϫ3 min Ϫ1 is obtained. The free GABA A receptor concentration upon full dissociation was estimated to be about 1 pM and is unlikely to limit dissociation.
gC1q-R Interacts with ␤ Subunits of the GABA A Receptor-The yeast two-hybrid system was used to test for an interaction between gC1q-R and the large intracellular loops of different GABA A receptor subunits. We chose ␣1, ␤2, ␤3, and ␥2 subunits because gC1q-R was found to copurify with GABA A receptors that were precipitated with antibodies against ␣1   FIG. 3. The GABA A receptor coimmunoprecipitates with gC1q-R. gC1q-R was immunoprecipitated from solubilized rat brain membranes (6 mg of protein) using serum against gC1q-R. Panel A, 50 g of solubilized rat brain membranes (lane 1) and 2% of the gC1q-Rimmunoprecipitate (lane 2) were separated on 10% SDS-polyacrylamide gel electrophoresis and immunoblotted with anti-gC1q-R serum. Panel B, coimmunoprecipitated GABA A receptor was detected by radioligand binding assay using [ 3 H]Ro 15-1788, which labels the benzodiazepine binding site. The starting material for the immunoprecipitation contained 3 pmol of benzodiazepine binding sites.

FIG. 2. gC1q-R coimmunoprecipitates with the GABA A receptor.
The GABA A receptor was immunoprecipitated from solubilized rat brain membranes (2.5 mg of protein) using the polyclonal antibody ␤3-(1-13) against ␤2/␤3 followed by the addition of protein A beads. The same immunobeads were extracted with 0.6 M NaCl and immunoblotted with anti-gC1q-R serum (lane 2). As negative controls, high salt extracts from immunoprecipitations lacking either solubilisate (lane 3) or anti-GABA A receptor-serum (lane 4) were also loaded onto the gel. Lane 1 shows the signal resulting from solubilized rat brain membranes (8 g of protein).
(bd24) or ␤2 and ␤3 (␤3-(1-13)). The ␥2 subunit was chosen because the major subunit combination in mammalian cortex probably consists of ␣1␤2␥2 subunits (44). Assuming that in neurons an interaction of gC1q-R and GABA A receptors would take place at the cytoplasmic side, the large intracellular loops of the GABA A receptor subunits mentioned above together with the mature form of gC1q-R (gC1q-R⌬71) were used in the two-hybrid assay. Whereas the intracellular loops of the ␣1 and ␥2 subunit did not indicate an interaction, the ␤1, ␤2, and ␤3 subunits interacted with gC1q-R⌬71 (Table II). The ␤3 subunit, however, seemed to bind to gC1q-R⌬71 to a much lesser degree than ␤1 or ␤2, judging from the intensity of the blue stain.
The Region on the ␤2 Subunit Recognized by gC1q-R Is Highly Positively Charged-To characterize the binding site on the ␤2 intracellular loop for gC1q-R⌬71, two NH 2 -terminally and seven COOH-terminally truncated ␤2 loop constructs were assayed for an interaction with gC1q-R⌬71 using the yeast two-hybrid system. The two NH 2 -terminal (n1, n2) as well as the first COOH-terminal (c1) truncations retained their ability to bind gC1q-R⌬71. All of the other COOH-terminally truncated constructs (c2-c7) lost their ability to bind to gC1q-R⌬71 (Fig. 4). This locates the site of interaction between the COOHterminal ends of c1 and c2, which corresponds to the amino acids 399 -413 (Fig. 4). A closer look at this region reveals that 7 out of 15 amino acid residues are positively charged. Furthermore, this region contains several phosphorylation sites (see "Discussion").
In an attempt to determine the region on gC1q-R⌬71 responsible for binding to ␤ subunits, two NH 2 -terminally and two COOH-terminally truncated versions of the mature gC1q-R⌬71 were constructed. Cutting points were chosen to lie NH 2 -and COOH-terminal to the centrally located ␤ sheet domain (45). None of the truncated constructs showed interaction with ␤2 subunits (Fig. 5). This indicates that both NH 2 -and COOHterminal regions are essential for the binding of gC1q-R to ␤2 subunits. Alternatively, removal of either of the termini could lead to improper folding of gC1q-R, which may cause the observed loss of interaction.
Localization of gC1q-R in Transfected HEK 293 Cells and Rat Hippocampal Neurons-Previous reports have shown that in various cell lines gC1q-R localizes predominantly in the mitochondrial matrix (43, 46 -48). To see whether gC1q-R is also present in the mitochondria of neurons expressing the GABA A receptor, cultured rat hippocampal neurons (26) were chosen for immunocytochemistry and confocal laser scanning microscopy. Fixed and permeabilized cells were incubated with a fluorescent mitochondrial marker (mitotracker red) and anti-gC1q-R antibody followed by the addition of FITC-coupled antirabbit antibody. The stain for gC1q-R colocalized with the stain obtained by the mitochondrial marker, demonstrating a predominant mitochondrial localization of gC1q-R in rat hippocampal neurons (Fig. 7B). Next, HEK 293 cells were tran-siently transfected with full-length gC1q-R and gC1q-R⌬71. To check whether full-length gC1q-R is processed by the cell to mature gC1q-R, membranes from cells transfected with fulllength gC1q-R were subjected to Western blot analysis using anti-gC1q-R serum. Two bands were visible, the upper corresponding to the immature form and the lower one to the mature protein (Fig. 6). The lower one migrated at 34 kDa and was much stronger than the upper one, indicating that the cellular machinery is capable of converting most of the overexpressed precursor protein into mature protein. The rather slow migration of the mature gC1q-R, a protein of 208 amino acid residues, may be because of its highly acidic nature and has been reported also by others (43). Using confocal laser scanning microscopy the intracellular distribution of gC1q-R was then analyzed in fixed and permeabilized HEK 293 cells decorated with anti-gC1q-R antibody followed by the addition of rhodamine-coupled anti-rabbit antibody. HEK cells showed an endogenous signal for gC1q-R (Fig. 7E, red channel) which was granular and tubular, indicative of a mitochondrial localization. In cells transfected with full-length gC1q-R the morphology of the stain remained similar but the intensity was drastically increased (Fig. 7C). In cells transfected with gC1q-R⌬71 the immunoreactivity was retained in the cytoplasm (Fig. 7D), probably arising from the lack of the mitochondrial targeting sequence within the first 71 amino acid residues. ␣1, ␤3, and ␥2 subunits of the GABA A receptor were expressed in HEK 293 cells either alone (Fig. 7E) or together with gC1q-R (Fig. 7F) or gC1q-R⌬71 (not shown). GABA A receptors were detected using bd17 and FITC-coupled anti-mouse antibody. No significant colocalization was found in cells expressing full-length or mature gC1q-R in combination with the GABA A receptor. Also the distribution of GABA A receptors remained unaffected by the presence of gC1q-R. The same results were obtained with GABA A receptors containing ␤2 instead of ␤3 subunits (not shown).
Electrophysiological Measurements-Recombinant GABA A receptors with the subunit composition ␣1␤2␥2 were expressed in Xenopus oocytes alone or in combination with full-length or mature gC1q-R. Coexpression with gC1q-R did not significantly alter the expression level of the GABA A receptor, as judged by the current amplitude of the GABA A receptor, its activation characteristics by GABA (EC 50 and Hill coefficient), and apparent desensitization (not shown).

DISCUSSION
In this study we identify a novel GABA A receptor-interacting protein, the previously described multifunctional protein gC1q-R. In an earlier study, it was shown that upon immunoprecipitation of ␣1-containing GABA A receptors from bovine brain membranes, a 34-kDa protein was coimmunoprecipi-  II Yeast two-hybrid assay of gC1q-R with different GABA A receptor subunit intracellular loops GABA A receptor subunit intracellular loop constructs and the mature form of gC1q-R (gC1q-R⌬71) were expressed in the yeast SFY526 strain as fusion proteins with an amino-terminal GAL4 activation domain and GAL4 DNA binding domain, respectively. Interaction was detected by measuring the activity of ␤-galactosidase with 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside as a substrate in a colony-lift filter assay. ϩ, strong interaction; (ϩ), weak interaction; Ϫ, no interaction. The experiments were done twice with n ϭ 5.

Subunit loop Interaction
Ϫ tated. Biochemical purification and sequencing of tryptic fragments led to the identification of this protein as gC1q-R. cDNA cloning and bacterial overexpression of the rat gC1q-R were followed by the production of an antibody. Experiments with solubilized rat brain membranes showed that gC1q-R can be coimmunoprecipitated with the GABA A receptor and vice versa (Figs. 2 and 3B). gC1q-R is thought to be localized predominantly in mitochon-dria where it seems to be involved in maintenance of oxidative phosphorylation (43). Using confocal laser scanning microscopy in rat hippocampal neurons we also found a signal originating from gC1q-R exclusively in mitochondria (Fig. 7B). However, since the discovery of gC1q-R there have been several reports indicating extramitochondrial functions of this protein. gC1q-R was identified originally as a cell surface protein binding to the globular heads of the complement factor C1q (27). Also, a nuclear localization has been postulated (46). Recently, a study using immunogold electron microscopy showed that gC1q-R is also located at other extramitochondrial locations in normal tissue, as e.g. in zymogen granules, condensing vacuoles, and endoplasmic reticulum (49). Although in most of the analyzed tissues, including rat cerebral cortex and cerebellum, immunoreactivity was localized almost exclusively in mitochondria, an extramitochondrial localization of a small percentage of gC1q-R in GABAergic neurons cannot be ruled out. If the amount of extramitochondrial gC1q-R is small enough it will not be visible with immunostaining techniques. Whether gC1q-R is associated with mature GABA A receptors in the surface membrane or with GABA A receptors during biosynthesis or transport to the surface is unknown. In a recent study gC1q-R was implicated in the regulation of the cellular localization and the expression of the G-protein-coupled ␣ 1B -adre-nergic receptor (42). It is intriguing to note that gC1q-R interacts with two different membrane receptors. During experimentation it became evident that the affinity between the GABA A receptor and gC1q-R must be very high. During the course of 6 h only about 40% of initially bound GABA A receptor dissociates from gC1q-R. This indicates that the complex is very stable.
Using the yeast two-hybrid system we identified GABA A receptor ␤ subunits as interaction partners with gC1q-R. Only the intracellular loop portions of GABA A receptor subunits were used in the assay because an interaction would most likely take place on the cytoplasmic side of neurons. By using truncated versions of ␤2 subunit loops a region of 15 amino acids was identified which is required for the interaction (Fig.  4). This region contains 7 positively charged residues, 3 lysines and 4 arginines, and shows no homology to the respective regions on ␣1 and ␥2. On both the ␣1 and ␥2 loops there are also clusters of 7 and 6 positive amino acids within a small stretch of 12 amino acid residues. Despite these positive charges, ␣1 and ␥2 loops did not interact with gC1q-R. Therefore, the interaction is not only based on electrostatic interaction between acidic residues on gC1q-R and basic residues on ␤ subunit loops. Additional evidence supporting this hypothesis came from an experiment in which site-directed mutagenesis of a serine residue lying within the positively charged region on ␤2 to an alanine or glutamine residue led to a reduced interaction between the loop and gC1q-R (not shown). Neither mutation introduced nor removed an electric charge but nevertheless decreased the strength of interaction between ␤2 and gC1q-R.
Interestingly, the stretch of basic amino acids identified in our study contains one or two functionally relevant serine residues, the number depending on ␤ subunit subtype (␤1S409, ␤2S410, and ␤3S408/S409). Phosphorylation of these residues by either PKC, PKA, Ca 2ϩ /calmodulin-dependent protein kinase, and PKG causes modulation of GABA A receptor function (17)(18)(19)(20)(21)(22)(23). Thus, binding of gC1q-R might interfere with receptor phosphorylation by a competitive mechanism and therefore indirectly modulate channel activity. An alternative role of receptor phosphorylation is suggested below. Initially we thought that gC1q-R itself might possess kinase activity despite lacking a kinase sequence signature. In an earlier work a novel kinase with putative mass of 34 kDa had been described which phosphorylates ␤ subunit loop constructs at the same site (26). However, kinase assays with immunoprecipitated gC1q-R or bacterially overexpressed and purified gC1q-R did not result in phosphorylation of ␤ subunit loop substrates (not shown). This makes an identity of gC1q-R with this unknown kinase unlikely.
Recently gC1q-R was shown to be associated with different PKC subtypes (50). In in vitro kinase assays, addition of gC1q-R blocked PKC-mediated substrate phosphorylation dose-dependently. Although such a modulation has not yet been shown for PKC␤II, which associates to GABA A receptor ␤1 and ␤3 subunits (25), it might be speculated that gC1q-R interferes with PKC-mediated phosphorylation of GABA A receptors.
To see whether gC1q-R can modulate the function of mature GABA A receptors, recombinant ␣1␤2␥2 GABA A receptors were coexpressed with full-length or mature gC1q-R in Xenopus oocytes. No effect on the expression level of GABA A receptors, agonist affinity, or apparent receptor desensitization was found. Thus, we could not demonstrate a direct modulation of mature GABA A receptor function by gC1q-R.
gC1q-R might also be involved in GABA A receptor biosynthesis or trafficking. Because of its many different interaction partners, a chaperon function has recently been ascribed to FIG. 7. Localization of gC1q-R in rat hippocampal neurons and transfected HEK cells. Rat hippocampal neurons (panel A) were fixed and permeabilized for immunostaining. Mitochondria were stained with mitotracker red, and endogenous gC1q-R was stained with anti-gC1q-R antiserum followed by a FITC-labeled anti-rabbit antibody. gC1q-R colocalized with mitochondria as shown in yellow in the twocolor merged image (panel B). HEK 293 cells were transiently transfected with gC1q-R (panel C), gC1q-R⌬71 (panel D), GABA A receptor ␣1␤3␥2 (panel E), and GABA A receptor ␣1␤3␥2/gC1q-R (panel F). Immunostaining was made on fixed, permeabilized cells. gC1q-R and gC1q-R⌬71 were visualized using rabbit anti-gC1q-R serum and rhodamine-labeled anti-rabbit antibody using the red fluorescence channel (panels C-F). The GABA A receptor was stained with the mAb bd17 and FITC-coupled anti-mouse antibody (panels C and D). gC1q-R showed a particulate organization indicative of a mitochondrial localization for both the endogenous or the transfected full-length protein (panels C and F). gC1q-R⌬71 lost its ability to enter mitochondria and was retained in the cytoplasm (panel D).
gC1q-R (50). gC1q-R might support correct folding or assembly of GABA A receptors prior to membrane delivery. In this case gC1q-R would not be expected to be colocalized with surface membrane-bound GABA A receptors. Indeed, proteins interacting with ligand-gated ion channels at locations other than the surface membrane have been described, e.g. GABARAP. GABARAP colocalizes with GABA A receptors in intracellular compartments but not in postsynaptic densities (15). Furthermore, the close homology of GABARAP to p16, an intra-Golgi trafficking factor, suggests a role of this protein in receptor transport. For the nicotinic acetylcholine receptor also, a protein interacting with the receptor during biosynthesis has been described. The immunoglobulin-binding protein, a resident protein of the ER, was shown to bind only to newly synthesized, unassembled ␣ subunits of the nicotinic acetylcholine receptor, suggesting a function in receptor assembly, receptor subunit folding, or degradation of misfolded subunits (51,52). Similar functions could also be conceived for gC1q-R. The phosphorylation state may also be important for the interaction here.
In summary we have identified a new protein interacting with the GABA A receptor, the previously described multifunctional protein gC1q-R. Interaction was specific for ␤ subunits, with the site of interaction located within the large intracellular loop. The presence of the functionally relevant Ser-410 within the interacting site might suggest a modulatory role of gC1q-R either in biosynthesis or the mature receptor.