Chloride channel expression with the tandem construct of alpha 6-beta 2 GABAA receptor subunit requires a monomeric subunit of alpha 6 or gamma 2.

Despite the presence of the multiple subunits (α, β, γ, and δ) and their isoforms for γ-aminobutyric acid, type A (GABAA) receptors in mammalian brains, the αxβ2γ2 subtypes appear to be the prototype GABAA receptors sharing many properties with native neuronal receptors. In order to gain insight into their subunit stoichiometry and orientation, we prepared a tandem construct of the α6 and β2 subunit cDNAs where the carboxyl-terminal of α6 is linked to the amino-terminal of β2 via a linker encoding 10 glutamine residues. Transfection of human embryonic kidney 293 cells with the tandem construct alone failed to induce GABA-dependent Cl− currents, but its cotransfection with the cDNA for α6 or γ2, but not β2, led to the appearance of GABA currents which were picrotoxin-sensitive and, in the case of γ2 containing receptors, responded to a benzodiazepine agonist, U-92330. The high affinity GABA site, however, was detected with [3H]muscimol binding in all combinations of the receptor subunits, including the tandem construct alone or with the β2. No appreciable differences were found in their Kd (2.5 nM) and Bmax values (1.4 pmol/mg of protein). These data are consistent with the view that the polypeptides arising from the tandem construct were expressed with the high affinity GABA site, but unable to form GABA channels. The requirement of a specific monomeric subunit (α6 or γ2) for the tandem construct to express Cl− currents supports a pentameric structure of GABAA receptors consisting of two α6, two β2, and one γ2 for the α6β2γ2 and three α6 and two β2 for the α6β2 subtype.

GABA A 1 receptors, responsible for inhibitory neurotransmission in mammalian brains, are ligand-gated Cl Ϫ channels made of various subunits (␣, ␤, ␥, and ␦) (1)(2)(3). Each subunit consists of several isoforms and contains four transmembranespanning segments (M1 to M4) (1)(2)(3)(4)(5). Despite the existence of the multiple subunits and their isoforms, combinations of ␣x, ␤2, and ␥2 subunits produced Cl Ϫ channels sharing many functional characteristics with native neuronal receptors and displaying the ability to respond to all the GABA A receptor ligands known up-to-date (1,3,6,7). Such cloned GABA A receptors have been proposed to be of pentameric structure with M2 lining the pore in analogy with another member of the four transmembrane ligand-gated channel family, acetylcholine receptors (1)(2)(3). Recent studies, including immunoprecipitation with subunit specific antibodies, have shown the presence of two ␣ subunits per GABA A receptor (8 -11). Further experimental evidence is needed, however, about the stoichiometry of the recombinant GABA A receptors of ␣x␤2␥2 and their modes of association. One way to gain insight into this structural issue is to predetermine the alignment of subunits via gene fusion and to study such fused gene products. Similar approaches have been successful with potassium channels made of their subunits in concatameric or tandem linkages (12,13). In this study we prepared a tandem construct of ␣6 and ␤2 subunit cDNAs of the GABA A receptor where the carboxylterminal of the ␣6 cDNA is linked to the amino-terminal of the ␤2 cDNA via a synthetic oligonucleotide encoding 10 glutamine residues. In order to study their properties, the tandem construct alone or in combination of the monomeric ␣6, ␤2, and ␥2 subunit cDNA was expressed in human embryonic kidney cells (HEK293 cells). Also, recombinant baculovirus carrying the tandem construct or monomeric subunits was prepared for high level of expression in Sf-9 cells.

MATERIALS AND METHODS
Construction of the ␣6-␤2 Tandem Subunit-The coding regions of the ␣6 and ␤2 subunits were isolated upon digestion of the mammalian expression vectors containing the specific GABA A receptor subunit cDNA (14). For ␣6, ClaI cut the vector at the molecular cloning site near the 5Ј end of the insert, and ScaI cut the vector at a site near the 3Ј end of the insert, which is 9 bases upstream from the stop codon. This restriction digestion yielded a fragment of expected size, 1377 bp. For ␤2, Kpn2I cut the vector at a site near the 5Ј end, which is 11 bases downstream from the start codon, and EcoRI cut the vector at the 3Ј ends of the insert, including the stop codon. A double-stranded oligonucleotide (61-mer) was purchased (Genosys) with a sequence of 5Ј-ACTGTCGAGCAGCAGCAACAGCAGCAACAGCAGCAACAGATGTG-GAGAGTCCGGAAAAGGG-3Ј. In this linker, the first 9 bp represent the ␣6 sequence before the stop codon beginning with the distal half of the ScaI recognition site, the next 30 bp represent 10 glutamine residues, and the last 22 bp represent the ␤2 sequence, including the start codon (ATG) and the recognition site for Kpn2I with an overhang at the 3Ј end. The fragment representing the rest of the expression vector was isolated by digestion with ClaI and EcoRI. Ligation products were formed from all the four fragments with proper sticky ends and were used to transform the competent Epicurean coli SCS1 strain (Stratagene). The correct plasmid was selected with polymerase chain reaction (PCR) using a primer pair, one of which anneals to the glutamine linker, and was verified with sequencing. Human embryonic kidney cells (HEK293 cells, ATCC CRL 1573) were transfected with the vector carrying the tandem cDNA alone or with ␣6, ␤2, or ␥2 cDNA in a 2:1 ratio (4 to 2 g/36-mm culture dish), in the presence of a transfection reagent, DOTAP (Boehringer Mannheim Gmbh). For infection of insect Sf-9 cells, recombinant baculoviruses were prepared using a vector, PVL 1393 (Pharmingen), carrying the tandem ␣6-␤2 construct or the cDNA for ␣6, ␤2, or ␥2. All other procedures for DNA analysis, construction, and purification were described elsewhere (15).
Northern Blotting and 3Ј-RACE Assays-mRNAs were prepared in the presence of guanidium thiocyanate from HEK293 cells transfected with the carriers of indicated GABA A receptor subunit cDNAs. Northern blotting was carried out with nylon transfer membranes, following the procedures provided by Schleicher and Schuell. The 32 P Probes for * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  ␣6 and ␤2 were prepared from PCR which were carried out in a 50-l volume using 1 unit of Amplitaq DNA polymerase (Perkin-Elmer) in the vender-supplied buffer, 20 M dNTP ([␣-32 P]dCTP), the respective cDNA as a template, and selective primer pairs annealing to the 5Ј and 3Ј end of the coding region of the template cDNA. The cycle parameters were at 94°C for 1 min, 55°C for 30 s, and 72°C for 1-3 min, depending on the expected product size, with a final extension at 72°C for 10 min after 30 cycles. The radioactive probes were purified using a Qiagen column and used within a few days. Fig. 1 shows Northern analysis of the mRNA preparations from HEK293 cells which were transfected with the tandem construct of ␣6 and ␤2 subunits (␣6-␤2) alone or in combination with ␥2. For the sake of comparison, authentic mRNAs for individual subunits were prepared using a transcription in vitro system with sp6 RNA polymerase (Promega) and linearized vectors at the 3Ј end of the coding region. In the analysis, stringent washing conditions were employed under which the 32 P-labeled probe for the ␣6 subunit interacted only with ␣6 mRNA, but not with ␤2 or ␥2 mRNA (Fig. 1A).
In the mRNA preparations from HEK293 cells transfected with ␣6-␤2 alone or in combination with ␥2, the 32 P-labeled ␣6 probe interacted with a single band of an estimated molecular mass of about 3.5 kDa (judging from RNA molecular mass standards and as compared with the ␣6 mRNA of 1.7 kDa) (Fig. 1A). The 32 P-labeled ␤2 probe detected the same band as expected from the tandem construct (data not shown). The mRNA species for the tandem construct was further confirmed by the 3Ј-RACE following the procedures provided by Life Technologies, Inc. (Fig. 1B). Briefly, following reverse transcription of the mRNA preparations using 3Ј-RACE adaptor primer ((dT) 17 CATGATR-CAGCTGCGCACCGG), PCR was carried out with a primer specific for the ␣6 or ␤2 and the universal amplification primer (the vender provided). The primer for ␣6 was a 23-mer beginning at the base 400 of ␣6 (5Ј-CGGACACATTTTTCCACAATGGG) and that for ␤2 was a 21-mer beginning at the base 1051 of ␤2 (5Ј-GGTCAACTACATCTTCTTTGG).
In HEK293 cells transfected with ␣6, ␤2, and ␥2 subunits, PCR with the ␣6 primer and the universal primer produced one product with an expected size of 1491 bp and with the ␤2 primer also produced a single product with an expected size of 1743 bp. In the cells transfected with the tandem construct, PCR with the ␣6 primer and the universal primer produced one product matching with the expected size of 3663 bp for the ␣6-␤2, and similar PCR with the ␤2-specific primer and the universal primer produced the product of the same size (1743 bp) as the one found in the cells transfected with ␤2 monomers. These data confirmed the proper construction of ␣6 and ␤2 subunits in tandem as described above and have established its in vivo transcription.  (17). GABA and drugs were dissolved in the external solution and were applied through a U-tube placed within 100 m of the target cell.
Binding Studies-Sf-9 cells were infected with the baculovirus carrying the recombinant cDNAs as described previously (18). Binding of radioactive ligands was measured in the Sf-9 cell membranes, using filtration techniques as described elsewhere (19,31). Briefly, [ 3 H]muscimol or [ 3 H]Ro 15-4513 binding was measured in the medium containing 118 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 20 mM Hepes/ Tris, pH 7.3, the radioactive ligand at various concentrations (1-40 nM), 30 g of membrane proteins in a total volume of 500 l at 4°C for 60 min. Nonspecific binding was estimated in the presence of excess unlabeled ligands and was subtracted to compute specific binding as described earlier (19,31). RESULTS We examined whether GABA induces Cl Ϫ currents in HEK293 cells transfected with the GABA A receptor cDNAs, using the whole cell patch clamp technique. In the cells transfected with the tandem construct alone, no currents were detected upon application of GABA at 1-20 M (with more than 30 patches). If GABA A receptors are pentameric as proposed, functional expression of the tandem construct of ␣6-␤2 would require the presence of monomeric subunits. Therefore, HEK293 cells were transfected with the ␣6-␤2 and one of the monomeric subunits. Addition of ␣6 or ␥2, but not ␤2, led to the appearance of GABA-induced whole cell currents, which were sensitive to picrotoxin, a specific inhibitor of GABA A receptor Cl Ϫ currents (Fig. 2). Moreover, a benzodiazepine agonist, U-92330 (19,31), enhanced GABA currents by 136 Ϯ 9% in the receptor made of ␣6-␤2 plus ␥2, but had no effect on that made of ␣6-␤2 plus ␣6 (0 Ϯ 5%) (Fig. 2). The picrotoxin sensitivity and the ␥2-dependent interaction with the benzodiazepine site ligand have already been known in the ␣6␤2 and ␣6␤2␥2 subtypes of GABA A receptors (16,19,31). It should be also noted that no GABA-induced Cl Ϫ currents were detected in the whole cell patch with HEK293 cells (or Sf-9 cells) transfected (or infected) with the subunit of ␣6, ␤2, or ␥2 alone (19,31). GABA dose-dependently increased Cl Ϫ currents in the cells transfected with ␣6-␤2 plus ␣6 or plus ␥2 (Fig. 3). Analysis of the data with a logistic equation of E/E max ϭ [GABA] n /(K 0.5 n ϩ [GABA] n ) yielded a half-maximal GABA concentration (K 0.5 ) of FIG. 1. Northern analysis and 3-RACE assays with mRNAs from HEK293 cells transfected with the tandem construct of ␣6 and ␤2 GABA A receptor subunit cDNAs. The cells were grown to about 70% confluence in a 75-cm 2 culture flask and incubated in the presence of the tandem construct of ␣6-␤2 (8 g) or with the cDNA for ␥2 (4 g) and DOTAP (60 l) for 24 h. The cells were harvested 24 h later, and mRNAs were prepared in the presence of guanidium thiocyanate following the standard procedures. A, Northern blotting was carried out with nylon membranes containing 10 g of extracted mRNAs after resolution on a 1% agarose gel, following the venderprovided procedures. The 32 P-labeled probe for the ␣6 was prepared using PCR in the presence of [␣-32 P]dCTP. The ␣6, ␤2, and ␥2 mRNAs were prepared in vitro using a Promega transcription kit. B, 3Ј-RACE assays were carried out using the kit from Life Technologies, Inc., the universal primer and the ␣6 specific primer (a) or the ␤2-specific primer (b). The detailed information for the primers and PCR reaction were described under "Materials and Methods." Earlier electrophysiological and binding studies have shown that properties of the receptors expressed in HEK293 cells or Sf-9 cells are indistinguishable, but the latter, a high expression eucaryotic system, is more reliable for radioactive ligand binding experiments (18,20). Thus, Sf-9 cells were infected with recombinant baculovirus carrying the tandem construct alone or in combination with that for ␣6, ␤2, or ␥2 cDNA. Binding experiments were carried out with [ 3 H]muscimol and [ 3 H]Ro 15-4513, the ligands for high affinity GABA site and the benzodiazepine site, respectively. The data were analyzed with Scatchard analysis (Table I, Fig. 4). [ 3 H]Muscimol binding was observed in all the cell membranes, including the cells infected with ␣6-␤2 alone or in combination with one of the monomers. The K d value for [ 3 H]muscimol (the high affinity GABA site) was 2.5 Ϯ 0.2, 2.6 Ϯ 0.3, 2.7 Ϯ 0.2, and 2.1 Ϯ 0.2 nM for the receptors from ␣6-␤2 alone or with ␣6, ␤2, or ␥2, respectively. Also, the maximal binding level was not noticeably variable among those receptors, ranging from 1.6 to 1.2 pmol/mg of protein. Despite the high expression of the high affinity GABA site, the Sf-9 cells expressing the ␣6-␤2 alone again failed to produce Cl Ϫ currents upon GABA (5 M) application.
Binding of [ 3 H]Ro 15-4513 (a benzodiazepine site ligand) was observed only in the membranes from the cells infected with ␣6-␤2 with ␥2 as expected (see above). The B max for the benzodiazepine site ligand was similar to that for [ 3 H]muscimol, suggesting all the receptors containing both the high affinity GABA site and benzodiazepine site. DISCUSSION In this study we have shown that expression of the tandem construct of ␣6-␤2 subunit cDNA alone failed to produce Cl Ϫ currents in response to GABA application, but in the presence of the monomeric subunit of ␣6 or ␥2 subunit, its expression led to the appearance of GABA-and picrotoxin-sensitive Cl Ϫ currents. This could be interpreted to mean that the tandem construct alone forms receptors with only even-numbered subunits, which are not functional (no chloride channels), but in the presence of the monomeric subunit of ␣6 or ␥2, forms a functional pentameric receptor with Cl Ϫ channels. This interpretation led us to propose that the functional receptor/Cl Ϫ channels may consist of two ␣6-␤2 and one ␣6 or ␥2. Since these receptors displayed similar functional and pharmacological properties with the monomeric ␣6␤2␥2 or ␣6␤2 subtypes, we propose that the ␣6␤2␥2 subtype consists of two ␣6, two ␤2, and one ␥2 subunits and that the ␣6␤2 subtype consists of three ␣6 and two ␤2 subunits. This proposal is in agreement with earlier immunoprecipitation studies which indicated the presence of two ␣ subunits in the ␣␤␥ subtypes of cloned GABA A receptors (8 -11). An alternative interpretation of our results is that the functional receptors of ␣6␤2␥2 would be made of one ␣6-␤2 and three monomeric ␥2, leading to one ␣6 subunit per receptor. This is incompatible with the presence of two ␣ subunits per receptor already shown by the immunoprecipitation studies (8 -11).
We also propose the orientation of the subunits in the ␣6␤2␥2 subtype to be ␣6-␤2-␣6-␤2-␥2. Such an arrangement could minimize interactions between homologous subunits and  Binding was measured in the membranes from the Sf-9 cells expressing the ␣6-␤2 alone (E) or with ␥2 subunit cDNA (q) using filtration techniques as described under "Materials and Methods" The data were fit with a one-site binding isotherm and were also shown in a Scatchard analysis (inset). The data represent the average of three measurements, and the binding parameters are shown in Table I. Tandem ␣6-␤2 GABA A Receptor Construct domains and accommodate two ␣6-␤2 tandem subunits per receptor. This arrangement is also compatible with the assignment of the benzodiazepine site at the interface between ␣6 and ␥2 subunits (22)(23)(24), with the N-terminal of ␣6 involved in the formation of the benzodiazepine site (14,25). Future study with the tandem ␥2-␣6, ␤2-␣6, and ␤2-␥2, where the hyphen represents a C-to N-terminal linkage, will be useful to test this proposal.
Of considerable interest is the appearance of the high affinity GABA site without Cl Ϫ channels in the cells expressing the ␣6-␤2 construct alone. Two types of GABA sites are known to be on GABA A receptors, the high affinity site with nanomolar dissociation constants and low affinity sites with micromolar dissociation constants (1,21). The high affinity GABA site has been reported on the ␤ subunit (1,26), and low affinity GABA sites were not localized yet, but their affinity was markedly affected by mutations on the ␤ subunit (27) and different ␣ isoforms (7,28,29). This indicates that low affinity GABA sites could be influenced by secondary, tertiary, and quaternary interactions among the subunits. Furthermore, its occupancy seems to lead to channel openings, because the Hill coefficient of near 2 was observed for GABA currents and no channel openings were observed with GABA at concentrations occupying nearly 70% of the high affinity sites (29,30). It appears that the formation of low affinity GABA sites and GABA-sensitive Cl Ϫ channels require association of proper five receptor subunits, which could not be achieved with the tandem ␣6-␤2 construct alone.