Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The inhibitory neurotransmitter GABA plays a major role throughout mammalian central nervous system via activation of GABA_A and GABA_B receptors. GABA_A receptors are multisubunit oligomers with an integral chloride channel belonging to the ligand-gated ion channel receptor family. To date, cloning studies have revealed the gene family encompasses a multitude of different subunits, falling into seven families (1-6, 1-3, 1-3, , , , and ) (Whiting et al., 1999). Studies on subunit assembly and stoichiometry suggest the subunits coassemble as pentamers to form a variety of receptor subtypes that are differentially expressed in mammalian brain (Wisden et al., 1992; Fritschy and Mohler, 1995; McKernan and Whiting, 1996). In vitro, an , a , and a  subunit are required to form fully functional GABA_A receptors resembling those in native tissues (Pritchett et al., 1989) where 2:2:1 stoichiometry is often observed (Im et al., 1995; Chang et al., 1996), although ()₂()₁()₂ may also be possible (Sieghart et al., 1999). The presence of different subunits confers different pharmacological, functional, and modulatory properties on receptor subtypes. The largest population of GABA_A receptors in rat brain has subunit composition of 122, whereas 232 and 32/3 together constitute the next most prevalent subtypes (McKernan and Whiting, 1996). There is also a variety of other combinations that form more minor receptor populations (Sieghart et al., 1999), although the precise physiological roles of many of these are not fully defined.

GABA_A ion channel function can be modulated via a variety of allosteric sites (Sieghart, 1992) that are the target for a number of therapeutic agents, such as benzodiazepines, barbiturates, steroids, convulsants, and anesthetics (for review, see Sieghart, 1995; Rabow et al., 1995). Affinity and efficacy at the benzodiazepine site, one of the better characterized, are influenced by  and  (McKernan et al., 1995; Buhr et al., 1996; Wingrove et al., 1997) but not  subunits (Hadingham et al., 1993). The benzodiazepine site is the target for commonly prescribed anxiolytic drugs, occurring at the interface of an (1-6) and a (1-3) subunit, with residues in both influencing modulation (Smith and Olsen, 1995; Wingrove et al., 1997). The presence of 2 confers a classical benzodiazepine pharmacology to GABA_A receptors (Pritchett et al., 1989), whereas receptors containing the less abundant 1 subunit (Benke et al., 1996) display atypical benzodiazepine pharmacology, including a much lower affinity for Ro15-1788 (Wafford et al., 1993). Introduction of a 3 subunit affects the affinity and efficacy of some but not all benzodiazepine ligands (Hadingham et al., 1995). Of the  subunits, the presence of 4 or 6 results in low affinity for classic benzodiazepines such as diazepam but high affinity for the imidazobenzodiazepine Ro15-4513 (Knoflach et al., 1996; Wafford et al., 1996).

Cloning and stable expression of selected GABA_A subunits permits detailed analysis of pharmacological profiles associated with specific receptor subtypes and this, together with the knowledge of which combinations predominate in vivo, their localization, and relative abundance (McKernan and Whiting, 1996), may help to dissect the physiological roles of these inhibitory receptors. We have examined the influence of subunit composition on activity at human recombinant GABA_A receptors stably expressed in mammalian cells, with particular emphasis on compounds that modulate ion channel function. The methodology involves measuring ³⁶Cl influx through the GABA_A receptor ion channel pore and, in 96-well format, offers the potential for multiple manipulations simultaneously within the same experiment without the need for washout and recovery associated with electrophysiological experiments where a single cell or oocyte is patch-clamped. Many previous studies used primary cultures, synaptoneurosomes, or microsac preparations from native tissues to examine GABA-activated transmembrane ion flux (Kardos, 1993) but the precise array of subunits is generally unknown and the measurable response may arise from a mixed receptor population. The other primary techniques used to examine GABA_A receptor function are electrophysiological, with much of the evaluation done in transient expression systems such as Xenopus laevis oocytes (Rabow et al., 1995). Comparisons of GABA_A receptor function in stably transfected Ltk cells and transiently transfected oocytes reveal some differences, primarily with regard to the actions of ethanol (Harris et al., 1997). Stable expression of recombinant receptors, however, offers the advantage of a homogeneous clonal population of cells where expression can be carefully monitored, reducing the likelihood of misassembly and overcoming some of the variability seen with transient expression such as multiple transient assembly (Ebert et al., 1996), poor efficiency of transfection, and change of receptor expression with time. In this study, the pharmacological consequences of stably expressing different GABA_A receptor  subunits in combination with (3) and (2s) in mouse Ltk cells are examined.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Tissue Culture and Cell Preparation. Human GABA_A receptor subunits were stably expressed in mouse Ltk fibroblast cells by transfection of the appropriate subunit cDNAs in vector pMSGneo using standard calcium phosphate transfection techniques (Hadingham et al., 1992). Cells were maintained in stock trays in Dulbecco's modified Eagle's medium (DMEM) with 580 mg/l glutamine (Life Technologies Europe, Paisley, Scotland, UK), supplemented with 10% (v/v) FetalClone II serum (Hyclone, Logan, UT) and 1 mg/ml geneticin. Stocks were split 1:8 into fresh trays once a week. For seeding, cells (passage 5-25) were removed from stock plates with 0.05% trypsin/0.53 mM EDTA solution (Life Technologies, Gaithersburg, MD) and resuspended in DMEM supplemented with 10% FetalClone II serum but without geneticin. Cells with subunit composition n32 were seeded into Costar 96-well plates at densities of between 3 and 8 × 10⁴ cells/ml in a volume of 200 µl/well and grown in the presence of 10% serum for 5 to 8 days in an incubator at 37°C. Receptor expression, which is under the control of a dexamethasone-sensitive promoter, was induced 24 h before experiment in the confluent cell monolayers using serum-containing DMEM supplemented with 1 µM dexamethasone (Sigma Chemical, Poole, Dorset, UK).

Early Optimization of Flux Conditions. The assay was developed initially in cells grown on 25-mm coverslips contained in six-well plates. Coverslips were immersed in HEPES/Krebs' buffer containing ³⁶Cl and appropriate drugs for 5 s, and then washed in ice-cold stop buffer (HEPES/Krebs' buffer containing 100 µM picrotoxin, pH 7.4 at 4°C using 1 M Tris) and influx determined by digestion in NaOH and scintillation counting. Although agonist responses and benzodiazepine modulation could be measured, the signal was small and the methodology time-consuming. Assay conditions were optimized, including examining the effects of assay temperature, extracellular chloride concentration, optimizing agonist exposure times, and using chloride transport inhibitors. The assay was then transferred into 96-well format to improve throughput. Cells were initially seeded at densities of 2 × 10⁴cells/ml in a volume of 200 µl/well with 7 days of induction. Following cell washing, ligand, GABA, and test compounds were added simultaneously for 10 s to each well in a 50-µl volume (n = 6 wells/treatment), followed by aspiration and washing with stop buffer. Effects of inducing receptor expression with dexamethasone for different periods were investigated. A progressive improvement in the signal was observed as days of induction were reduced from 7 days to 1 day, with benzodiazepine modulation by flunitrazepam being optimal at 1 day. The current protocol was therefore adopted in which cells are seeded and allowed to grow for 5 to 8 days before being induced 1 day before use. The assay procedure was also amended at this stage to include a 30-s preincubation step with test compounds to allow equilibration before addition of GABA and ³⁶Cl for a further 7 s.

Measurement of ³⁶Cl Influx. Following the 24-h induction period with dexamethasone, 96-well plates were removed from the incubator. Cells were initially washed at room temperature with buffer of the following ionic composition 162 mM NaCl, 6.4 mM KCl, 1.7 mM MgCl₂, 1.2 mM CaCl₂, 1.2 mM NaH₂PO₄, 1.2 mM HEPES, pH 7.4, using a Dynatech Ultrawash Plus 96-well plate washer. Induction medium was aspirated and, after a prewash, cells were washed twice for 2-min periods, aspirating after the final wash. Measurement of ³⁶Cl influx was performed in assay buffer where chloride ions were absent to favor uptake of radioactive chloride. Chloride salts were replaced by acetate salts and the chloride-free assay buffer, pH 7.4, at 4°C using 1 M Tris) also contained 11 mM glucose and the chloride transport inhibitors DIDS (10 µM) and furosemide (100 µM). Modulatory compounds or assay buffer was added to the wells in a volume of 40 µl for a 30-s preincubation using a Robbins Hydra 96 dispenser. At t = 30 s a subsequent 40-µl addition of a 2× (³⁶Cl ligand solution ± GABA in chloride-free assay buffer) was then made to each well using a Sagian Beckman Multipette 96-well dispenser, incubating for a further 7 s. At t = 37 s ³⁶Cl influx was terminated by aspiration of incubation buffer and immediate washing with 5× 200 µl of ice-cold picrotoxin stop buffer using a second Dynatech plate washer. Stock ³⁶Cl ligand was supplied by Amersham Pharmacia Biotech (Buckinghamshire, UK) as NaCl in aqueous solution (122 mg Cl/ml) at a concentration of 1 mCi/ml. The final concentration of radioligand in the assay was 4.44 µCi/ml. Scintillation fluid (200 µl) was finally added to each well, the plate sealed, and influx of ³⁶Cl was determined by scintillation counting on a Packard TopCount.

Data Handling and Analysis. ³⁶Cl influx was measured on each plate in replicates of six wells per treatment. Each 96-well plate included the following controls: six wells for basal or background influx in the absence of GABA mimetics (chloride-free buffer plus ³⁶Cl only); six wells treated with a maximal concentration of GABA (buffer plus ³⁶Cl solution containing GABA, 100 µM final concentration); and for modulation of GABA-stimulated flux, six wells treated with ³⁶Cl plus an EC₂₀ concentration of GABA. For benzodiazepine modulation, each plate also contained a full benzodiazepine agonist in the presence of an EC₂₀ concentration of GABA as an internal standard, relative to which everything else was expressed, to control for any minor variations in cell responsiveness from day to day. For 1, 2, 3, and 5, this was the nonselective benzodiazepine agonist chlordiazepoxide (30 µM). Modulation at other sites on the GABA_A receptor is shown as the amount of potentiation of the EC₂₀.

Whole-Cell Patch-Clamp of Ltk Cells. Responses to GABA were also determined electrophysiologically in Ltk cells stably expressing 1-532 GABA_A receptors after 24-h induction with dexamethasone. Glass coverslips containing a cell monolayer were placed in a chamber on the stage of a Nikon Diaphot inverted microscope and perfused continuously with artificial cerebral spinal fluid containing 149 mM NaCl, 3.25 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 10 mM HEPES, 11 mM D-glucose, D-(+)-sucrose, pH 7.4, and observed with phase-contrast optics. Fire-polished patch pipettes were pulled using conventional 120TF-10 electrode glass and tip diameter was approximately 1.5 to 2.5 µM, with resistances around 4 M. The intracellular solution contained 130 mM CsCl, 10 mM HEPES, 10 mM BAPTA.Cs, 5 mM ATP.Mg, 0.1 mM leupeptin, 1 mM MgCl₂, 100 µM NaVO₃, pH adjusted to 7.3 with CsOH and 320 to 340 mOsm. Cells were voltage-clamped at 60 mV via an Axon 200B amplifier (Axon Instruments., Foster City, CA) and drugs applied via a multibarrel drug delivery system, which could pivot the barrels into place using a stepping motor, ensuring rapid application and washout of drug. GABA was applied to the cell for 5 s with a 30-s washout period between applications. Noncumulative concentration-response curves to GABA were constructed and curves fitted using a nonlinear least-squares-fitting program to the equation f(x) = B_max/[1 = (EC₅₀/x)^n_H], where x is the drug concentration, EC₅₀ is the concentration of drug eliciting a half-maximal response, and n_H is the Hill coefficient.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Assay Optimization. In initial experiments on 132 cells grown on 25-mm coverslips, progressively decreasing the extracellular chloride ion concentration by substituting acetate salts resulted in a progressive increase in the magnitude of stimulation with 10 µM muscimol, whereas agonist-independent basal uptake remained constant (Fig. 1a). Time course experiments were performed in 1 cells under different conditions to select an optimal assay duration. At room temperature in the presence of 8 mM extracellular chloride, a specific signal was obtained at early time points with 10 s being optimal, followed by a progressive decline with no specific signal remaining in excess of 60 s. The temperature was reduced to 4°C to minimize nonspecific membrane transport of chloride and this extended the period during which a specific signal could be observed. However, there was still a progressive rise in basal ³⁶Cl accumulation with increasing incubation time, such that no specific signal remained at 5 min (Fig. 1b). Inclusion of the chloride transport inhibitors DIDS (10 µM) and furosemide (100 µM) further extended the period of time over which a specific window could be observed, with the magnitude of agonist-stimulated accumulation being maintained at a plateau and basal accumulation increasing at a slower rate (Fig. 1c). In 96-well format, a high enough seeding density was critical for optimal responses and density is amended occasionally according to cell growth and receptor expression, currently between 3 and 8 × 10⁴ cells/ml. The number of days of induction was important in maximizing observation of benzodiazepine modulation, with 1 day being optimal. Additionally, preincubation with benzodiazepines for 30 s before addition of GABA and ³⁶Cl was important to accurately measure compounds with slow kinetics such as abecarnil. The effect of DIDS and furosemide on the magnitude of agonist stimulation in 132 cells, both individually and together, is shown in the final 96-well format (Fig. 1d) with a 30-s preincubation and a 10-s flux incubation.

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Optimization of flux assay using 132 cells. a-c, performed in coverslip format with stimulation by 10 µM muscimol. a, effect of varying [Cl]ext. b, time course at 4°C in the presence of 8 mM [Cl]ext. c, effect of DIDS and furosemide on 4°C time course. d, effects of DIDS and furosemide on signal in final 96-well format [3 × 104cells/ml, induced 1 day, [Cl]ext = 0, 4°C, 30-s preincubation, 10-s (36Cl ± GABA) addition].

³⁶Cl Influx: GABA Agonists and Antagonists. The effect of expressing different  subunits in combination with 32s subunits was examined on the response to GABA_A receptor agonists. GABA (Fig. 2) and muscimol both elicited concentration-dependent increases in ³⁶Cl influx, an effect not observed in noninduced controls. Both agonists elicited similar maximal increases in ³⁶Cl influx, although muscimol exhibited higher potency than GABA at each of the subunit combinations. EC₅₀ values (µM) for GABA and muscimol are shown in Table 1, together with a comparison with patch-clamp data. The GABA affinities of diazepam-sensitive recombinant receptors showed little discrimination between subtypes, with the exception of 3 subunit-containing receptors, where both GABA and muscimol were approximately an order of magnitude less potent. Both agonists also displayed slightly lower potency at receptors of subunit composition 432 in flux.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   GABA elicited concentration-dependent increases in 36Cl influx in Ltk cells stably expressing GABAA receptors of composition 1-532s. Each curve represents mean data from at least 12 independent experiments (mean ± standard error of mean).

View this table: [in this window] [in a new window]   TABLE 1 EC50 values (µM) for GABA- and muscimol-stimulated 36Cl influx are shown for GABAA receptors of composition 1-532s, stably expressed in Ltk cells Electrophysiological responses to GABA in Ltk cells, measured using a whole-cell patch-clamp technique, are also shown for comparison. Data are presented as arithmetic mean ± S.E.M. with the number of determinations in parentheses.

View larger version (11K): [in this window] [in a new window]   Fig. 3.   Agonist-stimulated 36Cl influx was blocked in a concentration-dependent manner by the competitive GABAA antagonist bicuculline (a) and the chloride channel blocker picrotoxin (b). Mean effects of 30 and 100 µM bicuculline and picrotoxin are shown on the response to GABA in 332-containing cells (mean of n = 5 independent experiments). a, , GABA alone; , +30 µM bicuculline; , +100 µM bicuculline. b, , GABA alone; , +30 µM picrotoxin; , +100 µM picrotoxin.

³⁶Cl Influx: Modulation at the Benzodiazepine Site. Next, the role of varying  subunits in mediating different pharmacologies at the benzodiazepine site was examined. With respect to benzodiazepine binding, the nature of the  subunit can exert a marked influence on the observed pharmacology (Table 2) (representative concentration-response curves for selected compounds shown in Fig. 4). From the GABA concentration-response curve at each receptor isoform, a concentration of GABA was selected that produced an increase in ³⁶Cl influx approximately 20% of maximum and the modulatory effects of benzodiazepine site ligands were examined. The net effect of benzodiazepine agonists was to shift the GABA concentration-response curve to the left, whereas compounds exhibiting inverse agonist activity caused a rightwards shift. The nonselective benzodiazepine full agonist chlordiazepoxide was used as a standard, relative to which the effects of other benzodiazepine ligands were expressed. The response to an EC₂₀-equivalent concentration of GABA was significantly potentiated by chlordiazepoxide (30 µM) at 1, 2, 3, and 5 subunit-containing receptors (P < 0.001) and typical data are illustrated in Fig. 5. Chlordiazepoxide efficacy was not significantly different between 1, 2, and 5, although the potentiation observed at 3-containing receptors was significantly larger (P < 0.001).

View this table: [in this window] [in a new window]   TABLE 2 Binding affinity (nanomolar) of benzodiazepine site ligands at 1-532s GABAA receptors Results are mean ± S.E.M. (n = 3-6) (or mean % inhibition of binding at 10 µM). Data for 1, 2, 3, and 5 represent displacement of specific [3H]Ro15-1788 binding, whereas 4 data represent displacement of specific [3H]Ro15-4513 binding, measured for 1 hr at 4°C.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Representative concentration-response curves showing inhibition of binding to the benzodiazepine site for selected benzodiazepine site ligands at 1-532s GABAA receptors. The radioligand for 1, 2, 3, and 5 is [3H]Ro15-1788 and for 4 subunit-containing receptors, [3H]Ro15-4513. a, Ro15-4513; b, zolpidem; and c, flunitrazepam. , 1; , 2; , 3; , 4; , 5.

View larger version (41K): [in this window] [in a new window]   Fig. 5.   Typical flux responses for maximal agonist and benzodiazepine modulation in cells expressing 1, 2, 3, 4, or 5 subunits with 32s. Maximum agonist effects were defined with 100 µM GABA (*P < 0.001 with respect to basal); maximal potentiation of the GABA EC20 response by benzodiazepine was with 30 µM chlordiazepoxide for 1, 2, 3, and 5 and 30 µM bretazenil for 4-containing receptors (+P < 0.001 with respect to EC20). Seeding densities: 1, 8 × 104 ml1; 2, 5 × 104 ml1; 3, 6 × 104 ml1; 5, 6 × 104 ml1; and 4, 6 × 104 ml1.

View this table: [in this window] [in a new window]   TABLE 3 Allosteric modulation of GABA-stimulated flux by benzodiazepine ligands in cells expressing different  subunits with 32s, expressed as percentage of maximum benzodiazepine potentiation (mean ± S.E.M.) from at least eight independent experiments Chlordiazepoxide is 100% by definition for 1, 2, 3, and 5 and bretazenil 100% at 4.

³⁶Cl Influx: Effect of  Subunit on Modulation at Other Sites. The influence of the  subunit was also examined on potentiation of the GABA response by a variety of allosteric modulators that act at other sites on the GABA_A receptor, including the barbiturate, steroid, and loreclezole sites. Again, an equieffective concentration of GABA (EC₂₀) was used at each receptor subtype to examine the modulatory effects. The amount of potentiation in each case has been expressed as percentage increase in the EC₂₀ GABA response.

View larger version (36K): [in this window] [in a new window]   Fig. 6.   Effect of expressing different  subunits with 32s on modulation of GABA-stimulated flux at the steroid, barbiturate, and loreclezole sites on the GABAA receptor. Results are expressed as percentage of increase in the response to GABA at the EC20 (mean ± S.E.M.) from at least seven independent experiments.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

A functional ion flux assay using ³⁶Cl has been established in 96-well plate format to measure efficacy at human recombinant GABA_A receptors and examine the influence that the  subunit exerts on receptor pharmacology. The pharmacology of flux responses was consistent with data described previously using electrophysiological techniques. Muscimol and GABA stimulated ³⁶Cl influx in a concentration-dependent manner and were specifically blocked by picrotoxin and bicuculline. The magnitude of maximal agonist stimulation varied across subtypes, with 332 receptors having the largest GABA response, possibly reflecting different levels of receptor expression. Picrotoxin caused a rightward shift of the curve and significant, concentration-dependent depression of the maximum, as previously reported. In contrast, bicuculline, a competitive antagonist that should not depress the maximum, although it has some potential to act as an allosteric inhibitor of channel opening at the GABA site (Ueno et al., 1997), did produce some depression in the current study in addition to a rightward shift. However, there was no significant difference between 30 and 100 µM, in contrast to picrotoxin. The reasons for this depression are unclear, although a slow dissociation rate or incomplete equilibrium with agonist may make an antagonist seem pseudo-irreversible.

In time course experiments during assay development, stimulation with muscimol peaked and then declined to a plateau at longer assay times, possibly due to removal of accumulated cellular chloride or receptor desensitization (Fig. 1b). In the presence of chloride pump inhibitors furosemide and DIDS, this decline was attenuated (Fig. 1c) and agonist-induced accumulation of chloride was enhanced (Fig. 1d), which may imply that the chloride pumps in these cells act to remove accumulated chloride ions. Theoretically, agonist-independent basal accumulation is likely to be the net result of a variety of different processes, including passive diffusion, chloride pump activity, or entry via any spontaneous opening of the GABA_A receptor ion channel. The importance of the 30-s modulator preincubation highlights kinetic considerations. Optimal flux responses are a fine balance between on-rate, which controls initiation, and possible desensitization, which controls termination. GABA_A receptors desensitize rapidly with agonist exposure and clearly an important consideration with the flux assay is to what extent receptors are in a desensitized state during ion flux measurements. However, when GABA EC₅₀ values were determined electrophysiologically in the same Ltk cells by patch-clamp measurement (Table 1), flux data seemed to be in reasonably good agreement. Electrophysiology is the most direct and widely used method of measuring receptor function and the temporal resolution clearly offers more refinement in terms of examining rate of onset and measuring peak responses. Although one cannot determine the true peak of flux by backward extrapolation and the overall flux measurement is potentially more sensitive to desensitization measuring total net ³⁶Cl accumulation over 7 s (i.e., area under the curve rather than peak), the similarity between EC₅₀ values determined by the two methods suggests that desensitization is not a major limiting issue with the current protocol. The current methodology offers a noninvasive means of quantitatively examining large homogeneous populations of GABA_A ion channels and constitutes a slightly different measure to electrophysiology, measuring an integrated response over time in the whole cell.

Evidence suggests desensitization and current activation/deactivation kinetics, which contribute to the time course of native GABA_A receptor-mediated inhibitory postsynaptic currents, and synaptic function vary with GABA_A receptor subunit composition (Tia et al., 1996). The  subunit isoform is at least partially responsible for determining the gating kinetics of recombinant GABA_A receptors (Lavoie et al., 1997). Clearly, if comparisons of efficacy are to be made between different receptor isoforms, it is important to compare the actions of modulators at the same effective GABA concentrations at each receptor subtype. This could also apply to electrophysiology. Accordingly, GABA concentration-response curves were constructed regularly to determine accurate EC₂₀ values before modulation studies. Flux efficacies generally compare well to those reported in a variety of electrophysiological assays and, as additional evidence to support this, a correlation is shown between efficacies determined using electrophysiology and flux for 132s receptors (Fig. 7). The fact that there is a good correlation between data from two different expression systems and methodologies (X. laevis oocytes and Ltk cells, respectively) lends further weight to the validity of the flux assay.

View larger version (13K): [in this window] [in a new window]   Fig. 7.   Correlation is shown for benzodiazepine site ligands at 132s GABAA receptors determined using two different methodologies. 36Cl influx was measured in Ltk cells as described herein and electrophysiological responses were determined in X. laevis oocytes expressing 132s receptors, impaled with two 1 to 3 M electrodes containing 2 M KCl and voltage-clamped at 70 mV (further details in Wafford et al., 1993). In each case, effects of benzodiazepines were examined on control GABA EC20 responses with a preapplication time of 30 s. Dotted lines represent 95% confidence intervals of the linear regression.

Benzodiazepine ligands displayed a wide range of intrinsic efficacies at different receptor subtypes. Broadly, the 1,4-benzodiazepines and -carbolines, methyl 6,7-dimethoxy-4-ethyl--carboline-3-carboxylate, abecarnil, and 3-carbomethoxy--carboline were relatively consistent in their action at 1-, 2-, 3-, and 5-containing receptors, being either agonist or inverse agonist at all. The exception was Ro15-1788, which was a pure antagonist at 5 but exhibited a small degree of partial agonist activity at 3. Compounds of different structural types (CL218,872, zopiclone, zolpidem, CGS9895, and CGS8216) exhibited the most subtype selectivity, mainly with regard to 5-containing receptors. Zolpidem, a full agonist at 1, 2 and 3, was without activity at 5, although it should be noted that the 5 affinity is very low. The 1 binding-selective compound CL218,872 was an 5 antagonist, whereas at 1, 2, and 3 receptors it was a partial agonist. CGS9895 also had different efficacy at 532 receptors, being a partial inverse agonist. However, 432 receptors were the most distinct pharmacologically. The imidazobenzodiazepine Ro15-4513 has the highest affinity at 432 receptors but many other ligands have much lower 4 affinity and the concentrations tested in flux were increased accordingly, although the highest concentration that could be tested (30 µM) is unlikely to have been maximal for chlodiazepoxide, flunitrazepam, and zolpidem. The compound with the highest 4 efficacy was FG8205, which was equivalent to bretazenil. The partial inverse agonist Ro15-4513 and the antagonist/partial agonist Ro15-1788 both exhibited high 4 efficacy (87 ± 5 and 53 ± 12% of bretazenil, respectively). Other compounds of note were CGS9895 and the partial inverse agonist CGS8216, which both caused significant 4 potentiation.

All subtypes were sensitive to potentiation by the anticonvulsant loreclezole, in keeping with the presence of a 2 or 3 subunit (Wafford et al., 1994). Although sensitivity to loreclezole does not require an  or  subunit (both 12 and 22 receptors were sensitive), their presence may influence the level of potentiation When different  subunits were expressed in oocytes with 22, the rank order of efficacy was 1 = 2 = 3, 5 was slightly less efficacious, 6 produced significantly less potentiation, and 4 was not determined. This is the same as the rank order observed here with 3 subunit-containing receptors in Ltk cells (1 = 2 = 3 > 5 > 4). Whittemore et al. (1996), describing the pharmacology of the human 4 subunit, also found greater potentiation with loreclezole at 122L receptors than 422L.

In contrast to the benzodiazepine site, barbiturate and neurosteroid modulatory sites of GABA_A receptors are less well defined. In particular, studies concerning which subunits influence steroid sensitivity have yielded contradictory results, particularly with regard to the  subunit, depending on the steroid examined and the experimental conditions (Lambert et al., 1995). Although the type of  subunit seems to have little impact, direct and modulatory effects of steroids are preserved even at homomeric 1 receptors. In a recent report (Maitra and Reynolds, 1999), changing or removing the  subunit altered the efficacy of pregnanolone to potentiate GABA (212L > 112L 12L), in agreement with the current findings, and efficacy was also influenced by the  subunit. In the case of the barbiturate site, the EC₅₀ value for pentobarbitone was not dependent on receptor subtype but maximum potentiation differed with the type of  and  subunits present (Thompson et al., 1996). Progress is currently being made in defining residues which influence anesthetic action using site-directed mutagenesis (Carlson et al., 2000; Cestari et al., 2000). The larger potentiation with pentobarbital at 4 compared with 1, observed here, has been reported for 422L (Whittemore et al., 1996) and 412s receptors (Wafford et al., 1996) expressed in oocytes. In the former study, the authors also examined pregnanolone and loreclezole. Pregnanolone produced a smaller potentiation of 4 than pentobarbital but had higher efficacy than loreclezole, as observed here. The efficacy of pregnanolone was greater at 422L than at 122L receptors, in agreement with the present observations, although in the current study the efficacy for pregnanolone at 2 and 5 exceeded that at 4.

Potentiation of the EC₂₀ response by pentobarbital was reported to be of the same magnitude for 412s and 612s expressed in oocytes (Wafford et al., 1996) and, in the same system, potentiation of 6 by pentobarbital was significantly greater than that for 1, 2, 3, or 5 (Thompson et al., 1996). Therefore, from oocyte work, it would be predicted that the magnitude of maximal pentobarbital potentiation at 4 should greatly exceed 1, 2, 3, or 5. Interestingly, in a study where expression in Ltk cells was compared with oocytes (Harris et al., 1997), some interesting differences were noted between the two expression systems. Pentobarbital potentiation of 632s was only about 50% higher than 132s in Ltk cells, whereas in oocytes the potentiation of 632s was 3- to 4-fold greater than other subtypes. In addition, the potentiation of 532s by pentobarbital was much greater in cells than oocytes, consistent with the present observations.

It is well documented that pentobarbital can directly activate GABA_A receptors in the absence of agonist and the magnitude of activation varies with receptor subtype, depending on both  and  subunits. Direct activation typically occurs at concentrations above 100 µM except for 6-containing receptors where direct activation is observed above 10 µM (Thompson et al., 1996). Although the focus of the current study was to compare maximal potentiation of different GABA_A receptor subtypes via a variety of modulatory sites, some direct activation of receptors with 100 µM pentobarbital is likely. Although pentobarbital was tested in the absence of GABA, it was not possible to resolve this component of the response due to some degree of spontaneous opening of the ion channel, which has previously been reported using this assay and confirmed electrophysiologically (Alder et al., 1998). Therefore a submaximal concentration of pentobarbital (10 µM) was also examined, with the largest potentiation being observed at 432 receptors (46 ± 3%).

Although the nature of the  subunit is a critical determinant of binding and efficacy at the benzodiazepine modulatory site on the GABA_A receptor, producing distinct pharmacological profiles, its influence on other modulatory sites examined here seems to be more subtle. Although all receptor subtypes were sensitive to modulation by pentobarbital, loreclezole, and pregnanolone, there were differences in the level of potentiation at different subunit combinations. This 96-well format flux assay should aid further investigation of recombinant GABA_A receptors and the structural features that influence and modulate ion channel function.