Abstract
In cerebellum, GABAA receptors containing α6 subunits are expressed exclusively in granule cells. The number of α6 receptor subtypes formed in these cells and their subunit composition presently are not known. Immunoaffinity chromatography on α6 subunit-specific antibodies indicated that 45% of GABAA receptors in cerebellar extracts contained α6 subunits. Western blot analysis demonstrated that α1, β1, β2, β3, γ2, and δ subunits co-purified with α6 subunits, suggesting the existence of multiple α6 receptor subtypes. These subtypes were identified using a new method based on the one-by-one immunochromatographic elimination of receptors containing the co-purifying subunits in parallel or subsequent experiments. By quantification and Western blot analysis of α6 receptors remaining in the extract, the proportion of α6 receptors containing the eliminated subunit could be calculated and the subunit composition of the remaining receptors could be determined. Results obtained indicated that α6 receptors in cerebellum are composed predominantly of α6βxγ2(32%), α1α6βxγ2(37%), α6βxδ (14%), or α1α6βxδ (15%) subunits. Other experiments indicated that 10%, 51%, or 21% of α6 receptors contained homogeneous β1, β2, or β3subunits, respectively, whereas two different β subunits were present in 18% of all α6 receptors. The method presented can be used to resolve the total number, subunit composition, and abundancy of GABAA receptor subtypes in the brain and can also be applied to the investigation of other hetero-oligomeric receptors.
- GABAA receptor
- composition, α6 subunit
- granule cell
- cerebellum
- antibodies
- immunoaffinity chromatography
- immunoprecipitation
- [3H]muscimol
- [3H]Ro 15-4513
- binding studies
GABAA receptors are ligand-gated chloride ion channels and the site of action of various pharmacologically and clinically important drugs, such as benzodiazepines, barbiturates, steroids, anesthetics, and convulsants (Sieghart, 1995). So far six α, four β, three γ, one δ, one ε, and three ρ subunits have been cloned and sequenced from mammalian brain (Sieghart, 1995; Ogurusu and Shingai, 1996; Davies et al., 1997), and it is assumed that five subunits assemble to form functional GABAA receptors (Nayeem et al., 1994; Tretter et al., 1997). Expression studies have indicated that α, β, and γ subunits have to combine to form receptors closely resembling native receptors. Depending on the type of α, β, and γ subunits used for transfection of cells, however, recombinant receptors with different pharmacological properties do arise (Sieghart, 1995). The distinct but overlapping regional and cellular expression of the individual subunits (Persohn et al., 1992; Wisden et al., 1992) raises the possibility of the existence of an extremely large variety of GABAAreceptor subtypes in the brain. So far the actual extent of GABAA receptor heterogeneity is not known.
GABAA receptors containing α6 subunits are expressed in cerebellar granule cells and in the embryologically related granule cells of the cochlear nucleus only (Laurie et al., 1992; Persohn et al., 1992; Wisden et al., 1992; Varecka et al., 1994;Jones et al., 1997). Thus, all α6 receptors from cerebellum are expressed in the same cell type. In addition, receptors consisting of α6βxγ2 subunits have special properties because they exhibit a high affinity for the inverse benzodiazepine agonist Ro 15-4513 but no affinity for the benzodiazepine agonist diazepam (Sieghart, 1995).
Several studies have investigated the subunit composition of GABAA receptors containing α6 subunits. The results obtained, however, were partially conflicting. Whereas in one study (Quirk et al., 1994) α6 subunits were not observed to occur in combination with other α subunits, other studies demonstrated a partial coexistence of α6 and α1 subunits in the same receptor (Pollard et al., 1993,1995; Khan et al., 1994, 1996). Similarly, estimates of the abundancy of individual receptor subtypes differed between authors. Finally, because of the lack of suitable antibodies, not all α6subunit-containing receptors could be investigated.
The present study was performed to resolve these discrepancies. Using 13 highly specific antibodies directed against different GABAA receptor subunits, we demonstrated that only α1, β1, β2, β3, γ2, and δ subunits significantly co-purified with α6 subunits. To determine the identity and quantitative importance of receptors formed from these subunits, a generally applicable method was developed that is based on a one-by-one elimination by immunoaffinity chromatography of receptors containing the co-purifying subunits. Quantification of the remaining α6 receptors allowed us to estimate the proportion of α6 receptors containing the eliminated subunit. Repeating this subtractive purification by eliminating another co-purifying subunit in a parallel or a subsequent experiment finally allowed us to identify the subunit composition of α6 receptors and to determine their quantitative importance.
MATERIALS AND METHODS
Generation and purification of antibodies. The antibodies anti-peptide α1(1–9), anti-peptide α2(416–424), and anti-peptide α3(459–467) (Zezula et al., 1991), anti-peptide α4(379–421) (Ebert et al., 1996), anti-peptide α5(427–433) (Sieghart et al., 1993), anti-peptide β3(345–408) (Slany et al., 1995), anti-peptide β3(1–13), and anti-peptide γ2(319–366) (Tretter et al., 1997), anti-peptide γ1(324–366) (Mossier et al., 1994), anti-peptide γ3(322–372) (Tögel et al., 1994), and anti-peptide δ(1–44) (Jones et al., 1997) were generated and affinity-purified as described previously. Polyclonal anti-peptide β1(350–404) and anti-peptide β2(351–405) antibodies were generated in a way similar to that described (Mossier et al., 1994).
The N-terminal peptide α6(1–15) (sequence QLEDEGNFYSENVSR-) or the C-terminal peptide α6(429–434) (sequence -VSSTVE) were custom-synthesized with an additional C- or N-terminal cysteine, respectively (piChem, Graz, Austria) and were coupled to keyhole limpet hemocyanin. These adducts were then used for the immunization of rabbits. The antibodies were isolated from the sera of the immunized rabbits by affinity chromatography on thiopropyl-Sepharose 6B coupled to the cysteine residue of the respective peptide according to the recommendations of Pharmacia LKB Biotechnology.
Cloning of α1, β1, β2, β3, or γ2subunits of GABAA receptors. A rat brain cDNA library was constructed in λZAP (Stratagene, La Jolla, CA) from poly A+ mRNA isolated from the brains of 8- to 10-d-old rats as described in the protocol from Stratagene. α1, β1, β2, β3, and γ2subunits of GABAA receptors were cloned from this cDNA library (Fuchs et al., 1995; Slany et al., 1995), and their sequence proved to be identical to that of the respective sequence published previously.
Culture of human embryonic kidney (HEK) 293 cells and cDNA transfection. Transformed HEK 293 cells (CRL 1573; American Type Culture Collection, Rockville, MD) were grown in DMEM (Life Technologies, Grand Island, NY) supplemented with 10% fetal calf serum (JRH Biosciences, Lenexa, KS), 2 mm glutamine, 50 μm β-mercaptoethanol, 100 U/ml penicillin G, and 100 μg/ml streptomycin in 75 cm2 petri dishes using standard cell culture techniques.
HEK 293 cells were transfected with rat α1, β1, β2, β3, or γ2 subunit cDNAs subcloned individually into the expression vector pCDM8 (Invitrogen, San Diego, CA), using the calcium phosphate precipitation method (Chen and Okayama, 1988). The ratio for the α, β, and γ subunits used for transfection of 3 × 106 cells was 12:6:6 μg cDNA (Zezula et al., 1996). The cells were harvested 48 hr after transfection.
Extraction of GABAA receptors. Membranes from the cerebellum of adult rats or membranes from HEK 293 cells transfected with various GABAA receptor subunit cDNAs were extracted with a deoxycholate buffer containing 0.5% deoxycholate, 0.05% phosphatidylcholine, 10 mm Tris-chloride, pH 8.5, 150 mm NaCl, 500 μm benzamidine, 200 μg/ml bacitracin, and 300 μm phenylmethylsulfonylfluoride (PMSF).
Immunoaffinity chromatography of GABAA receptors and Western blot analysis. The anti-peptide α6(429–434) antibody was the first α6 antibody available in our laboratory and therefore was used to prepare an immunoaffinity column. Immunoaffinity columns were prepared by coupling 3–5 mg of the purified antibodies to 1 ml of protein A–agarose using the ImmunoPure IgG Orientation Kit (Pierce Europe, Oud-Beijerland, The Netherlands) as described previously (Mossier et al., 1994).
The immunoaffinity columns were equilibrated in the deoxycholate extraction buffer. Deoxycholate extracts of cerebellar membranes were applied to the immunoaffinity column at a rate of 2 ml/h. To completely eliminate the respective subunit and its associated receptors from the extract, the extract was cycled three times through the respective immunoaffinity column. The column was washed twice with 4 ml of deoxycholate extraction buffer, twice with 4 ml of IP–high buffer (0.5% Triton X-100, 50 mm Tris-chloride, pH 8.3, 600 mm NaCl, 1 mm EDTA, 500 μmbenzamidine, 200 μg/ml bacitracin, and 300 μm PMSF) and then twice with 4 ml of IP–low buffer (0.2% Triton X-100, 50 mm Tris-chloride, pH 8.3, 150 mm NaCl, 1 mm EDTA, 500 μm benzamidine, 200 μg/ml bacitracin, and 300 μm PMSF). Proteins bound to the column were eluted with a buffer containing 0.1 mglycine-HCl, pH 2.45, 150 mm NaCl, and 0.1% Triton X-100. The eluted proteins were precipitated with methanol/chloroform (Wessel and Flügge, 1984) and subjected to Western blot analysis (Fuchs and Sieghart, 1989). Proteins transferred to polyvinylidene difluoride (PVDF) membranes were detected with digoxigenated primary antibodies (Tögel et al., 1994), as indicated, and the anti-digoxigenin-alkaline phosphatase Fab fragments (Boehringer Mannheim, Mannheim, Germany) and the chemiluminescence substrate CSPD (Tropix, Bedford, MA) according to the instructions of the manufacturer.
Immunoprecipitation and receptor binding assay. For immunoprecipitation, 300 μl of the clear deoxycholate membrane extract were mixed with 30 μl of antibody solution (0–20 μg of antibody), and the mixture was incubated under gentle shaking at 4°C overnight. Then 50 μl of immunoprecipitin (Life Technologies, Gaithersburg, MD) plus 150 μl of an IP–low buffer containing 5% dry milk powder were added, and incubation was continued for 2 hr at 4°C. The precipitate was centrifuged for 10 min at 10,000 ×g, and the pellet was washed twice with 500 μl IP–high and once with 500 μl IP–low buffer.
For [3H]Ro 15-4513 binding assays the precipitated receptors were suspended in 1 ml of a solution containing 0.1% Triton X-100, 50 mm Tris-citrate buffer, pH 7.1, 150 mm NaCl, and 10 or 20 nm[3H]Ro 15–4513 (20.9 Ci/mmol; DuPont NEN, Dreieich, Germany) in the absence or presence of 100 μmRo 15-1788 or various concentrations of diazepam, and were incubated for 90 min at 4°C. For [3H]muscimol binding assays the precipitated receptors were suspended in 1 ml of a solution containing 0.1% Triton X-100, 50 mm Tris-citrate buffer, and 20 nm [3H]muscimol (17.1 Ci/mmol; DuPont NEN) in the absence or presence of 10 μm GABA, and were incubated for 60 min at 4°C (Zezula and Sieghart, 1991). The suspensions were then filtered through Whatman GF/B filters, and the filters were washed twice with 5 ml ([3H]Ro 15-4513 assay) or 3.5 ml ([3H]muscimol assay) of a 50 mm Tris-citrate buffer, pH 7.1. When the percentage of α6 receptors retained by an immunoaffinity column had to be determined, immunoprecipitation with the α6(1–15) antibody and the subsequent [3H]muscimol binding assays were performed in the same experiment with the original extract and the immunoaffinity column efflux.
Total [3H]Ro 15-4513 binding in the extract before or after immunoprecipitation of α6 subunit-containing GABAA receptors was measured using a polyethyleneglycol (PEG) precipitation assay as described (Zezula and Sieghart, 1991). For this, 100 μl of the deoxycholate extract (or of the supernatant from the immunoprecipitation with anti-α6antibodies) was incubated for 90 min at 4°C in a total volume of 1 ml with a buffer containing 50 mm Tris-citrate, pH 7.1, 150 mm NaCl, 50 μg γ-globulin, 15% (wt/vol) PEG, and 10 or 20 nm [3H]Ro 15-4513 in the absence or presence of 100 μm Ro 15-1788. The suspension was then filtered through Whatman GF/B filters, and the filters were washed twice with 3.5 ml of an 8% PEG solution.
For the determination of total [3H]muscimol binding in the extract, the PEG precipitation assay could not be used. This was attributable to the relatively high viscosity of the PEG solutions, prolonging the time needed for filtration of the samples, and the rapid dissociation of [3H]muscimol from its binding site. Total [3H]muscimol binding therefore was determined after all GABAA receptors present in the extract were precipitated with an antibody mixture containing 8 μg β1(350–404), plus 8 μg β2(351–407), plus 10 μg β3(1–13) antibody, using the same assay as described above. The validity of this approach was demonstrated by the observation that [3H]Ro 15-4513 binding data were identical whether receptors were precipitated with PEG or with this antibody mixture.
RESULTS
Anti-α6 antibodies
The N- or C-terminal amino acid sequences α6(1–15) or α6(429–434) are unique for the α6subunit of GABAA receptors (Lüddens et al., 1990). Antibodies generated against these sequences were able to immunoprecipitate native GABAA receptors solubilized from rat cerebellar membranes in a dose-dependent manner (Fig.1). Whereas anti-peptide α6(1–15) antibodies precipitated up to 15 ± 4% (mean ± SD; n = 4) of all [3H]Ro 15-4513 binding sites present in the extract, anti-peptide α6(429–434) antibodies precipitated only 5 ± 1% (mean ± SD; n = 4) of these sites. Of the [3H]Ro 15-4513 binding sites precipitated by these antibodies, 23 ± 2% were diazepam sensitive, whereas 77 ± 2% of these sites were diazepam insensitive.
Immunoprecipitation of GABAA receptors solubilized from rat cerebellum. Solubilized receptors (470 fmol of [3H]Ro 15-4513 binding sites) were incubated with increasing amounts of α6(1–15) or α6(429–434) antibodies in a final volume of 350 μl. Receptors present in the pellets (solid symbols) or the supernatant (open symbols) were determined by specific [3H]Ro 15-4513 binding. Identical results were obtained when the supernatant from the α6(1–15) or α6(429–434) immunoprecipitation was investigated. The values are mean ± SD of four separate experiments performed in triplicates. SD bars that were smaller than the diameter of the symbols are not shown.
Interestingly, however, it was demonstrated that in the same experiment the percentage of total [3H]Ro 15-4513 binding sites eliminated from the supernatant was higher than that actually found in the precipitate. Thus, whether 15–20 μg of α6(1–15) or α6(429–434) antibodies was used for immunoprecipitation, the amount of [3H]Ro 15-4513 binding sites in the supernatant was reduced by 30 ± 3% (mean ± SD; n = 4) (Fig. 1). These results seem to indicate that each of these antibodies had a similar ability to bind to α6 subunit-containing GABAA receptors, but that part of the receptors presumably were lost during washing of the precipitate.
In other experiments the ability of α6(1–15) antibodies to immunoprecipitate [3H]muscimol binding sites was investigated. As with [3H]Ro 15-4513 binding, the amount of [3H]muscimol binding sites retained in the pellet was smaller than that removed from the supernatant. Thus, α6(1–15) antibodies precipitated 22 ± 3% (mean ± SD; n = 3) of all [3H]muscimol binding sites present in cerebellar extracts but eliminated 42 ± 3% of these binding sites from the supernatant. Overall, these data indicated that more [3H]muscimol than [3H]Ro 15-4513 binding sites were precipitated from cerebellar extracts (or removed from the supernatant) by α6(1–15) antibodies.
Proteins precipitated from cerebellar extracts by α6(1–15) antibodies were then subjected to SDS-PAGE and Western blot analysis. The α6(429–434) antibody as well as two α6(1–15) antibodies purified from the sera of different rabbits were able to identify a single protein band with an apparent molecular mass of 56–57 kDa (Fig.2). The observation that three antibodies directed against two distinct epitopes of the α6 subunit specifically identified the same protein supports the conclusion that the protein with apparent molecular mass of 56–57 kDa was the α6 subunit of GABAA receptors. This conclusion is in agreement with previous reports indicating that the α6 subunit exhibits an apparent molecular mass of 56–57 kDa (Lüddens et al., 1990; Pollard et al., 1993; Quirk et al., 1994). The different signal intensity of the antibodies reacting with identical amounts of the immunoprecipitate indicates that these antibodies exhibited a differential affinity for α6subunits under Western blot conditions.
Identification of α6 subunits in cerebellar membrane extracts. GABAA receptors were immunoprecipitated from cerebellar membrane extracts using α6(1–15) antibodies. Receptors were then subjected to SDS-PAGE and Western blot analysis, using digoxigenated antibodies.Lane 1, α6(429–434) antibodies;lane 2, α6(1–15) antibodies from rabbit 15; lane 3, α6(1–15) antibodies from rabbit 16. Antibodies bound to proteins were detected using anti-digoxigenin-alkaline phosphatase Fab fragments and a sensitive chemiluminescence detection system as described in Materials and Methods. The gel was calibrated with proteins of known molecular mass. The experiment was performed twice with similar results. The protein smear below the 56–57 kDa band was detected by all three antibodies and may have represented differentially glycosylated, partially degraded, or alternatively spliced α6 subunits.
Isolation, subunit composition, and quantitative importance of GABAA receptors containing α6 subunits
After solubilization of GABAA receptors from cerebellar membranes, 67.8% of the [3H]Ro 15-4513 or [3H]muscimol binding sites present in the membranes could be recovered in the extract. This corresponded to 92.5% of the binding sites identified in the extract and in the 100,000 × g pellet after extraction. Because there was no significant difference in the efficiency of solubilization by detergent between [3H]muscimol binding sites or diazepam-sensitive or -insensitive [3H]Ro 15-4513 binding sites, it can be concluded that the extracted receptors were representative of the entire functional α6subunit-containing GABAA receptor population.
To quantitatively isolate GABAA receptors containing α6 subunits, cerebellar extracts were cycled three times through an immunoaffinity column containing anti-peptide α6(429–434) antibodies. In the final efflux of this column, anti-peptide α6(1–15) antibodies no longer were able to precipitate GABAA receptors, and α6subunits no longer could be demonstrated in Western blots, indicating that this procedure eliminated most if not all α6receptors from the extract. In the same efflux, [3H]Ro 15-4513 binding was reduced by 31 ± 1% (mean ± SD; n = 3), and [3H]muscimol binding was reduced by 45 ± 1% (mean ± SD; n = 3). These percentages correspond closely to the 30 ± 3% reduction of [3H]Ro 15-4513 and 42 ± 3% reduction of [3H]muscimol binding sites observed in cerebellar extracts after immunoprecipitation with α6(1–15) antibodies (see above).
To identify GABAA receptor subunits co-purifying with α6 subunits, receptors bound to the α6(429–434) immunoaffinity column were eluted by a change in the pH value of the buffer and were probed with 13 different antibodies, each of which specifically recognized a distinct GABAA receptor subunit. As shown in Figure3A (or Fig.4A), in addition to the α6 subunit, α1, β1, β2, β3, γ2, and δ subunits were present in the α6(429–434) column eluate. Thus, α1(1–9), β1(350–404), β2(351–405), β3(345–408), γ2(319–366), and δ(1–44) antibodies identified proteins with apparent molecular mass of 51 kDa, 51–54 kDa, 50–53 kDa, 51–56 kDa, 41–44 kDa, and 53 kDa, respectively. Proteins with identical apparent molecular mass could be identified by these antibodies in parallel control experiments investigating recombinant GABAA receptors containing the respective subunits (experiments not shown). The β3(345–408) antibody, in addition to the 51–56 kDa protein, identified a second protein with an apparent molecular mass of 42–47 kDa. The protein with lower molecular mass seemed to be a partially degraded β3 subunit, because staining of this protein was variable in different experiments (compare Figs. 3 and 4), and increased with increasing time needed for the isolation of GABAA receptors (compare Fig. 3A–C).
Subunit composition and quantification of α6 receptors before or after elimination of δ or α1 subunit-containing receptors. Cerebellar extracts were chromatographed on an α6(429–434) immunoaffinity column either before (A) or after (B) chromatography on a δ(1–44) immunoaffinity column, or after (C) chromatography on both a δ(1–44) and an α1(1–9) immunoaffinity column. α6(429–434) column eluates were subjected to SDS-PAGE and Western blot analysis using the following digoxigenated antibodies: α1(1–9), α6(1–15), β1(350–404), β2(351–405), β3(345–408), γ2(319–366), and δ(1–44). Western blots are from a typical experiment that was performed three times with similar results. In parallel experiments, the original extract as well as the efflux from the δ(1–44) or the α1(1–9) column were subjected to immunoprecipitation with α6(1–15) antibodies and subsequent [3H]muscimol binding assays. Data are presented as percentage of [3H]muscimol binding sites precipitated in the original extract and are means of a single experiment performed in triplicate. The experiment was repeated three times with similar results.
Subunit composition and quantification of α6 receptors before or after elimination of γ2 and α1 subunit-containing GABAA receptors. Cerebellar extracts were chromatographed on an α6(429–434) immunoaffinity column either before (A) or (B) after chromatography on a γ2(319–366) immunoaffinity column, or after (C) chromatography on both a γ2(319–366) and an α1(1–9) immunoaffinity column. α6(429–434) column eluates were subjected to SDS-PAGE and Western blot analysis using α1, α6, β1, β2, β3, γ2, and δ antibodies as described in Figure 3. Western blots are from a typical experiment that was performed three times with similar results. In parallel experiments, the original extract as well as the efflux from the γ2(319–366) or the α1(1–9) column were subjected to immunoprecipitation with α6(1–15) antibodies and subsequent [3H]muscimol binding assays. Data are presented as percentage of [3H]muscimol binding sites precipitated in the original extract and are means of a single experiment performed in triplicate. The experiment was repeated three times with similar results.
The co-purification of α1, β1, β2, β3, γ2, and δ subunits together with α6 subunits was not caused by a cross-reactivity of α6(429–434) antibodies with these subunits, because neither α6(429–434) nor α6(1–15) antibodies were able to precipitate [3H]muscimol binding sites or GABAAreceptor subunits from extracts of forebrain membranes, which do contain α1–5, β1–3, γ1–3, and δ, but no α6 subunits (Persohn et al., 1992; Wisden et al., 1992). These data therefore indicate that any one of the α1, β1, β2, β3, γ2, and δ subunits can be colocalized with α6 subunits in the same GABAA receptor.
In contrast, the α2, α3, α4, α5, γ1, and γ3 subunits could not be detected in the eluate of the α6(429–434) immunoaffinity column, although all of these subunits, except the α3 and α5 subunits, could be identified in cerebellar extracts by the respective antibodies (experiments not shown). This indicates that unspecific adsorption of receptors or an exchange of subunits during extraction was not a problem in this study.
Isolation, subunit composition, and quantitative importance of GABAA receptors containing α6 and γ2 subunits
Because GABAA receptors are composed of five subunits, the co-purification of a total of seven different subunits by the α6(429–434) immunoaffinity column indicated that a mixture of GABAA receptor subtypes with different subunit composition was purified. To isolate GABAA receptors containing α6, βx, and γ2 subunits, GABAA receptors containing any one of the other co-purifying subunits were quantitatively removed by immunoaffinity chromatography. In the first step, receptors containing δ subunits were eliminated from cerebellar membrane extracts using a δ(1–44) column (Fig. 3). The δ(1–44) antibody specifically recognized the δ but no other subunits of the GABAAreceptor (Jones et al., 1997). Interestingly, in the pH 2.45 eluate of the δ(1–44) column, α1, α6, β1, β2, β3, δ, and other subunits, but no γ2 subunits, could be identified (R. Pelz, M. Jechlinger, and W. Sieghart, unpublished data).
To determine the composition of the remaining α6receptors, the efflux of the δ(1–44) column subsequently was chromatographed on the α6(429–434) column. As shown in Figure 3B, δ subunits could no longer be identified in the eluate of this column, indicating that these subunits had been completely eliminated by the δ(1–44) column. The presence of six different subunits (α1, α6, β1, β2, β3, γ2) in the eluate of the α6(429–434) column indicates that GABAAreceptors retained by this column were still heterogeneous.
In the efflux of the δ(1–44) column, α6(1–15) antibodies were able to precipitate 70% of the [3H]muscimol binding sites that could be precipitated by these antibodies in the original extract (Fig.3B). This indicates that 30% of the α6subunit-containing GABAA receptors were retained by the δ(1–44) column and contained the δ subunit.
In the next step, the efflux of the δ(1–44) column was chromatographed on an α1(1–9) immunoaffinity column. The α1(1–9) antibody has been demonstrated to selectively identify only α1 but no other GABAA receptor subunits (Nusser et al., 1996; Zezula et al., 1991). The α6 subunit-containing receptors remaining in the efflux of the α1(1–9) column were then collected by the α6(429–434) column. In the pH 2.45 eluate of this column, only α6, β1, β2, β3, and γ2subunits, but no α1 subunits, could be detected (Fig.3C). The five subunits present in this eluate still could have been combined in a variety of different ways, resulting in a multiplicity of pentameric αβ or αβγ receptors with different subunit composition and stoichiometry. At this point, therefore, no conclusion on the identity and composition of the receptors isolated by this procedure could be made.
As expected, the intensity of the individual signals for α6, β1, β2, β3, and γ2subunits was lower in Figure 3C than in 3A orB. In the efflux of the α1(1–9) column, 32 ± 3% (mean ± SD; n = 3) of the α6 subunit-containing receptors present in the original extract could be precipitated by α6(1–15) antibodies (Fig. 3C). Thus, 32% of α6 receptors were composed of α6, β1, β2, β3, and γ2subunits. The observation that 70% of the α6 receptors could be precipitated before and only 32% after the α1(1–9) column additionally indicates that 38% of α6 receptors were removed by the α1(1–9) column and thus contained α1 as well as α6subunits.
All of these percentages were obtained by investigating binding of [3H]muscimol to the precipitated receptors. Because [3H]muscimol binding sites can be demonstrated only on receptors containing α and β, or α, β, and γ subunits (Zezula et al., 1996), these experiments indicate that the 32% of α6 and 38% of α1α6receptors so far discussed must also have contained β subunits. Whether all or only some of these receptors additionally contained γ2 subunits cannot be answered at this time.
Isolation, subunit composition, and quantitative importance of GABAA receptors containing α6 and δ subunits
In another experiment (Fig. 4), GABAA receptors containing γ2 subunits were eliminated from cerebellar membrane extracts using a γ2(319–366) column. The high specificity of this immunoaffinity column has been demonstrated previously (Mossier et al., 1994). In the pH 2.45 eluate of the γ2(319–366) column, α1, α6, β1, β2, β3, γ2, and other subunits, but no δ subunits, could be identified (experiments not shown). This again supports the conclusion that γ2 and δ subunits, at least in the cerebellum, seem not to be present in the same GABAAreceptors.
Receptors remaining in the efflux of the γ2(319–366) column were then chromatographed on the α6(429–434) column. In the eluate of this column, α1, α6, β1, β2, β3, and δ subunits, but no γ2 subunits, could be detected (Fig.4B). Immunoprecipitation with α6(1–15) antibodies in the efflux of the γ2(319–366) column indicated that receptors composed of these subunits represented 30% of the α6 receptors present in the original extract (Fig.4B). All of these receptors contained the δ subunit, because 30% of all α6-containing GABAA receptors could also be bound to the δ(1–44) immunoaffinity column, as discussed above (Fig. 3B).
The identification of only 30% of the α6 receptors in the efflux of the γ2(319–366) column indicates that 70% of these receptors were retained by this column and thus contained γ2 subunits. Combined with the above observation (Fig.3B) that 70% of all α6 receptors could be precipitated in the efflux of the δ(1–44) column and were composed of α1, α6, β1, β2, β3, and γ2 subunits, these data suggest that α6receptors contain either γ2 or δ subunits.
In the next step, the efflux of the γ2(319–366) column was chromatographed on the α1(1–9) column, and α6 receptors remaining in the efflux of this column were then either collected by a subsequent α6(429–434) immunoaffinity chromatography or precipitated by α6(1–15) antibodies (Fig. 4C). In the eluate of the α6(429–434) column, α6, β1, β2, β3, and δ subunits, but no α1subunits, could be identified. Immunoprecipitation experiments indicated that 15% of all α6 subunit-containing GABAA receptors could still be precipitated in the efflux of the α1(1–9) immunoaffinity column (Fig.4C) and thus were composed of α6βxδ subunits.
Because 30% of all α6 (and δ) subunit-containing receptors could be precipitated before and only about 15% after chromatography on the α1(1–9) column, these results additionally indicate that 15% of all α6subunit-containing receptors are composed of α1α6βxδ subunits. Thus, the α6 and δ subunit-containing receptors α1α6βxδ and α6βxδ obviously are present in cerebellum at a 1:1 ratio. As expected, the signal strength of the individual protein bands was reduced according to the receptors removed by the various immunoaffinity columns (compare Fig. 4A–C). In this experiment the staining of the β3 subunit was quite prominent. Because staining intensity depends on the individual properties of the digoxigenated antibody batch used, different staining intensities obtained with different antibodies do not necessarily reflect differences in the amount of protein present in the extract.
Results so far presented indicate the existence of at least four α6 subunit-containing GABAA receptor subtypes in cerebellum that are composed of α6βxγ2, α1α6βxγ2, α6βxδ, and α1α6βxδ subunits. The same four α6 receptor subtypes were also identified when the sequence of columns was changed, and an α1(1–9) column was used before the γ2(319–366) column to eliminate receptors containing the respective subunits from cerebellar extracts. In addition, the quantitative data obtained were consistent with each other and not dependent on the sequence of columns used (experiments not shown). These results strongly suggest that none of the antibodies used for immunochromatography exhibited a significant cross-reactivity and that the α6(1–15) or α6(429–434) antibodies were able to recognize or precipitate these four α6 subunit-containing GABAA receptor subtypes with comparable efficiency. The experiments described were repeated several times, and the average proportion of the four GABAA receptor subtypes calculated from the individual experiments is given in Table1. In addition, taking into account that only 45 ± 1% of all GABAA receptors in the cerebellum contained the α6 subunit, the absolute contribution of the various α6 receptors to total GABAA receptors present in cerebellum was calculated (Table 1).
Relative and absolute abundancy of α6receptor subtypes in rat cerebellum
Isolation, subunit composition, and quantitative importance of GABAA receptors containing α6 and distinct β subunits
The low number of α6 receptors remaining in the extract after complete removal of γ2 and α1(α6βxδ, 15% of all α6receptors) or of δ and α1 subunits (α6βxγ2, 32% of all α6 receptors) prevented a direct investigation of the β subunit composition of these receptors, even more so because each immunoaffinity chromatography step is time consuming and enhances degradation and inactivation of receptors. Therefore, the β subunit-composition of α6 receptors was investigated in the original extract from cerebellum only.
For this, cerebellum extracts were first chromatographed on a β1(350–404) immunoaffinity column (Fig.5A). In the efflux of this column, β1 subunits no longer could be demonstrated (experiments not shown), indicating that receptors containing this subunit had been removed completely. Precipitation with α6(1–15) antibodies indicated that 85 ± 1% (mean ± SD; n = 4) of the original α6 receptors were still present after removal of the β1 subunit-containing receptors and suggested that 15% of all α6 receptors contained β1 subunits (Fig. 5A).
Quantification of α6 receptors containing different β subunits. [3H]muscimol binding to GABAA receptors immunoprecipitated with α6(1–15) antibodies was determined in cerebellar membrane extracts before or after chromatography on a β1(350–404), β2(351–405), or β3(345–408) immunoaffinity column as indicated. Data are presented as percentage of [3H]muscimol binding sites precipitated by α6(1–15) antibodies in the original extract and are mean values ± SD of three to four experiments performed in triplicate. The proportion of [3H]muscimol binding sites retained on the initial anti-β immunoaffinity columns (A, C, E) was significantly different (Student’s t test) from that remaining in the extract after the other two β subunits had been removed (efflux, B, D, F): A, efflux D (p = 0.007); C, efflux F (p = 0.002); E, efflux B (p = 0.001). [3H]muscimol binding sites present in the original extract were significantly different (p = 0.0001) from the sum of the [3H]muscimol binding sites retained on the initial anti-β immunoaffinity columns (A + C + E) and were also significantly different (p = 0.007) from the sum of the [3H]muscimol binding sites found in the efflux of B, D, andF.
The efflux of the β1(350–404) column was then chromatographed on a β2(351–405) immunoaffinity column (Fig. 5B). On this second column all receptors containing β2 subunits were adsorbed, as indicated by the absence of β2 subunits in the column efflux (experiments not shown). In the same efflux, however, 21 ± 7% (mean ± SD;n = 3) of the original α6 receptors could be precipitated using α6(1–15) antibodies. Because GABAA receptors containing β1 as well as those containing β2 subunits now had been completely removed from the extract, the remaining 21% of the α6receptors thus contained only β3 subunits.
In other experiments, all receptors containing β2subunits were first removed from the cerebellum extract using a β2(351–405) immunoaffinity column (Fig. 5C). In the efflux of this column, only 34 ± 2% (mean ± SD;n = 4) of the original α6 receptors were present. From this it can be concluded that 66% of all α6 receptors contained a β2 subunit. A subsequent chromatography on a β3(345–408) column (Fig.5D) eliminated an additional 24% of the α6receptors. The remaining 10 ± 1% (mean ± SD;n = 3) of receptors thus contained only β1 subunits.
Finally, the cerebellum extract was chromatographed first on a β3(345–408) column. In the efflux of this column, 63 ± 2% (mean ± SD; n = 4) of the α6 receptors were still present (Fig. 5E), indicating that ∼37% of all α6 receptors contained a β3 subunit. A subsequent chromatography on a β1(350–404) column removed an additional 12% of α6 receptors. The remaining 51 ± 8% (mean ± SD; n = 3) of α6 receptors thus contained only β2 subunits.
Interestingly, a comparison of the proportion of α6receptors retained by the β subunit-specific columns from the original extract with that remaining in the extract after removal of the other two β subunits revealed striking and statistically significant differences (see legend to Fig. 5 ). Although 15% of all α6 receptors were removed by the β1 column from the original extract (Fig. 5A), only 10% of α6 receptors were left after elimination of all β2 and β3 subunits (Fig. 5D). Although 66% of all α6 receptors were removed by the β2 column from the original extract (Fig. 5C), only 51% of these receptors were left after removal of β1 and β3 receptors (Fig.5F). Finally, although 37% of all α6receptors were removed by the β3 column from the original extract (Fig. 5E), only 21% of these receptors were left after removal of β1 and β2 subunits (Fig.5B).
In addition, the sum of α6 receptors retained by the β1, β2, and β3columns from the original extract was 118% (Fig. 5A,C,E), whereas the sum of the receptors remaining in the extract after two of the three β subunits had been removed was 82% (Fig.5B,D,F). These differences could not be explained by a cross-reactivity of the antibodies, because β1, β2, or β3 antibodies were unable to precipitate recombinant α1βxγ2 receptors containing the wrong β subunit (experiments not shown). These data therefore suggest that 18% of the α6 receptors in cerebellum contain more than one type of β subunit. Because of the variability of binding data, however, a further calculation of the proportion of receptors containing β1β2, β1β3, or β2β3 subunit combinations does not provide reliable results.
DISCUSSION
Composition and quantitative importance of GABAAreceptors containing α6 subunits
In the present investigation, 13 antibodies, each one highly specific for a different GABAA receptor subunit, were used to investigate the subunit composition and quantitative importance of GABAA receptors containing α6 subunits. Chromatography on an α6(429–434) immunoaffinity column quantitatively removed α6 subunits and 45 ± 1% of all GABAA receptors from cerebellar extracts, supporting previous conclusions (Khan et al., 1996; Jones et al., 1997) that 45% of all GABAA receptors in the cerebellum contain the α6 subunit. In the eluate of this column, in addition to the α6 subunit, only α1, β1, β2, β3, γ2, and δ subunits of GABAA receptors could be demonstrated, suggesting that any one of these subunits can be colocalized with α6 subunits in native GABAA receptors.
In contrast, α2, α3, α4, α5, γ1, or γ3 subunits did not co-purify with α6subunits. This is to be expected for α2, α3, α5, or γ1subunits, which are not expressed in the granule cells of cerebellum (Persohn et al., 1992; Wisden et al., 1992). The existence of minor amounts of receptors containing γ3 and α6subunits has been demonstrated previously after purification of GABAA receptors by a γ3 subunit-specific immunoaffinity column (Tögel et al., 1994). The observation that α4 subunits did not co-purify with α6 subunits, although these subunits are expressed in cerebellar granule cells and could be identified in cerebellar extracts (E. Bencsits, V. Ebert, and W. Sieghart, unpublished data), indicates that receptors containing α4 as well as α6subunits, if they exist at all, are quantitatively not important. Thus, the great majority of α6 subunit-containing GABAA receptors is composed of α6 and α1, β1, β2, β3, γ2, or δ subunits.
A new strategy for the determination of the subunit composition and quantitative importance of hetero-oligomeric receptors
A random assembly of α6 subunits with six other subunits into pentameric receptors (Nayeem et al., 1994; Tretter et al., 1997) would result in a total of 210 GABAA receptor subtypes with distinct subunit composition. It is impossible to isolate a single receptor subtype from an even much less heterogeneous mixture by immunoenrichment. In the present study, therefore, immunodepletion was used to purify and characterize GABAA receptors. Receptors containing one of the co-purifying subunits were eliminated from extracts by chromatography on subunit-specific antibodies. Quantification and Western blot analysis of α6 receptors remaining in the extract then allowed us to estimate the proportion of α6 receptors containing the eliminated subunit and to determine the composition of the remaining receptors. Repeating this procedure by eliminating all co-purifying subunits in parallel or subsequent experiments finally allowed us to identify the subunit composition of α6 receptor subtypes and to determine their quantitative importance.
α1, γ2, or δ subunit-containing α6 receptors
In agreement with previous studies (Khan et al., 1994, 1996;Pollard et al., 1995), 52% of the [3H]muscimol binding sites precipitated by α6(1–15) antibodies could be eliminated from cerebellar extracts by an α1subunit-specific column, indicating that α1α6 receptors are as abundant as receptors containing homogeneous α6 subunits (Table 1). Other experiments indicated that 70% of α6 receptors could be eliminated from cerebellar membrane extracts by a γ2subunit-specific (Fig. 4) and 30% by a δ subunit-specific column (Fig. 3). In addition, it was demonstrated that γ2 and δ subunits did not co-purify with each other, supporting the conclusion that these subunits do not co-exist in the same GABAA receptor (Quirk et al., 1995).
Furthermore, the number of [3H]Ro 15-4513 binding sites removed from cerebellar extracts by α6(429–434) or α6(1–15) antibodies was 69% or 71% of the [3H]muscimol binding sites eliminated by these antibodies, respectively. Because [3H]Ro 15-4513 binding sites are present on GABAA receptors containing αγ or αβγ subunits and [3H]muscimol binding sites are present on receptors composed of αβ, αβγ, and αβδ subunits (Quirk et al., 1995; Sieghart, 1995; Zezula et al., 1996), these data agree with the conclusion that 70% of the α6 receptors contained a γ2 subunit. The observation that the [3H]muscimol binding sites of γ2 or δ subunit-containing α6 receptors add up to 100% additionally indicates that all α6receptors contain either a γ2 or a δ subunit. From this it can be concluded that receptors composed of α6βx subunits, and consequently also those composed of α6γ2 subunits, which would contribute to [3H]Ro 15-4513 but not to [3H]muscimol binding sites, are not significantly expressed in cerebellum.
Further fractionation of the 70% α6 receptors containing γ2 subunits using an α1 subunit-specific column indicated that 37 ± 3% of α6 receptors are composed of α1α6βxγ2 and 32 ± 3% of α6βxγ2subunits. α1α6βxγ2receptors have been identified previously (Khan et al., 1994, 1996;Pollard et al., 1995), and quantification of these receptors led to comparable results (Khan et al., 1994).
Recombinant receptor studies have indicated that α6βxγ2 receptors, in contrast to α1βxγ2 receptors, exhibit a high affinity [3H]Ro 15-4513 binding that could not be inhibited by diazepam (Lüddens et al., 1990; Sieghart, 1995). Other studies have indicated that in GABAA receptors containing α6 and α1 (Khan et al., 1996) or α1 and α3 subunits (Araujo et al., 1996), each one of the subunits expressed its characteristic benzodiazepine pharmacology. Because 32% of α6 receptors are composed of α6βxγ2, whereas 37% are composed of α1α6βxγ2subunits, these two receptor subtypes are responsible for 46.4% and 53.6% of all [3H]Ro 15-4513 binding sites precipitated by α6(1–15) antibodies, respectively. Assuming that α6βxγ2receptors contain two α6 subunits (Im et al., 1995), these two receptor subtypes contain a total of 73% α6and 27% α1 subunits. The present observation that 23 ± 2% of [3H]Ro 15-4513 binding precipitated by α6(1–15) antibodies could be inhibited by diazepam is supported by a recent study (Khan et al., 1996) and is in agreement with the conclusion that each one of the subunits expresses its characteristic benzodiazepine pharmacology.
Further fractionation of the 30% α6 receptors containing δ subunits using an α1 subunit-specific column indicated that 15 ± 3% of all α6 receptors were composed of α1α6βxδ and 14 ± 2% of α6βxδ receptors. Although the existence of α1α6βxδ receptors in cerebellum has been implicated previously (Pollard et al., 1995), their abundancy was not determined.
β Subunit composition of α6 receptors
When β1-, β2-, and β3-specific immunoaffinity columns were used to eliminate GABAA receptors from cerebellar extracts in parallel experiments, it was demonstrated that the total percentage of α6 receptors removed was 118%. In the absence of a significant cross-reactivity of the β1, β2, or β3 subunit-specific antibodies, these data suggested the colocalization of different β subunits in 18% of the α6 receptors. This conclusion is supported by recent evidence indicating the colocalization of two different β subunits in native receptors (Li and De Blas, 1997). The proportion of α6 receptors containing homogeneous β subunits was then determined by measuring α6 receptors remaining in the extract after the removal of the other two β subunits. The results obtained indicated that 10, 51, or 21% of all α6 receptors contained homogeneous β1, β2, or β3subunits, respectively. Because of the variability of binding data, a reliable estimation of the β subunit composition of the remaining 18% of α6 receptors was not possible. The observation that β1 and β2 as well as β3subunits are co-purifying with α6 and γ2(Fig. 3C) or α6 and δ subunits (Fig.4C), however, indicates that the α6βxγ2 or α6βxδ receptor subtypes might exist in up to six isoforms containing different β subunit combinations (homogeneous β1, β2, or β3 subunits, β1β2, β1β3, or β2β3). The same might be true for receptors consisting of α1α6βxγ2 or α1α6βxδ subunits. Whether all of the resulting 24 α6 receptors with different subunit composition actually exist cannot be answered by this study.
Subunit stoichiometry of native α6 receptors
The present results, in agreement with studies investigating other receptors, indicate that native α6 receptors can contain two different α (Sieghart, 1995) or two different β subunits (Li and De Blas, 1997), and in addition contain either a γ2or a δ subunit. Overall, these results suggest a subunit stoichiometry of two α, two β, and one γ (or one δ) subunit for native α6 receptors. This is in agreement with studies investigating the subunit stoichiometry of α6β2γ2 (Im et al., 1995) or of other recombinant receptors (Chang et al., 1996; Tretter et al., 1997). The method of subtractive purification of GABAAreceptors developed in the present study can be used to investigate whether all native α6 receptors exhibit this stoichiometry or whether other stoichiometries also exist (Backus et al., 1993). In addition, this method can also be applied to the investigation of other hetero-oligomeric receptors.
Footnotes
This work was supported by Grants P9003-Med and P9828-Med of the Austrian Science Foundation and by the European Commission Biotechnology Programme, Project ERBBIO4CT960585. The participation of W. Kern in the initial experiments of this study is gratefully acknowledged.
Correspondence should be addressed to Dr. Werner Sieghart, Section of Biochemical Psychiatry, University Clinic for Psychiatry, Währinger Gürtel 18-20, A-1090 Vienna, Austria.
