Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

GABA_A receptors are a family of ligand-gated ion channels that are the major target of the endogenous inhibitory neurotransmitter (GABA) and maintain the majority of fast inhibitory ion currents in the CNS. They are pentamers composed of subunits (1-6, 1-3, 1-3, , , , and ) that are encoded by a gene family with diverse expression patterns (Sieghart et al., 1999). GABA_A receptors are the targets of several classes of drugs, including benzodiazepines (BZDs), barbiturates, alcohols, neurosteroids, and inhalation anesthetics (Sieghart, 1995). Additionally, GABA_A receptors have been shown to be involved in epilepsy (DeLorey et al., 1998), various behavioral states such as depression and anxiety (Benson et al., 1998; Crestani et al., 1999), and learning and memory (Flood et al., 1992; DeLorey et al., 1998).

Prevailing theory suggests that the subunit composition of an individual GABA_A receptor confers a unique pharmacology that dictates the binding characteristics, functional capacity, and role of the receptor in maintaining the inhibitory tone of the CNS. The GABA_A receptor 1 subunit is the most abundant  subunit in adult brain, highly expressed throughout most brain regions, and is a component of ~50% of GABA_A receptors (Duggan and Stephenson, 1990; McKernan et al., 1991). Recombinant expression studies have indicated that 1 subunit expression confers specific pharmacological properties to the receptor, including GABA sensitivity (Levitan et al., 1988) and maximal efficacy of benzodiazepines (Puia et al., 1991). Furthermore, the expression of 1 versus 2, 3, and 5 in 2 receptors results in differential affinity for several benzodiazepine site ligands (Pritchett et al., 1989; Pritchett and Seeburg, 1990).

Recently, several studies have begun to describe the in vivo role of the 1 subunit in GABA_A receptor pharmacology, function, and related behavior. An association between heightened CNS excitability and reduced 1 subunit expression has been observed during ontogenic development, alcohol dependence and withdrawal, and in animal models of temporal lobe epilepsy (Aicardi and Chevrie, 1970; Mecarelli et al., 1988; Morrow et al., 1990; Devaud et al., 1997; Brooks-Kayal et al., 1998; Poulter et al., 1999). Previous studies have suggested various roles for  subunit isoforms in specific BZD-related behaviors (Mohler et al., 1996; Rudolph et al., 1999; Crestani et al., 2000; Low et al., 2000). Recently, the production of two independent 1 knockout mouse lines has been described (Sur et al., 2001; Vicini et al., 2001). Global deletion of the 1 subunit gene results in viable mice that are surprisingly normal. Although initial electrophysiological studies revealed diminished mIPSCs and loss of zolpidem-induced prolongation of mIPSC decay rates in cerebellar stellate cells (Vicini et al., 2001), we have now analyzed GABA_A receptor subunit expression, ligand binding, and muscimol-stimulated Cl uptake in cerebral cortex as well as ligand binding autoradiography throughout brain. The results reveal interesting stable receptor adaptations that differ in some respects from adaptations observed in 1^/ mice reported by Sur et al. (2001). The goals of the present studies were to identify in vivo relationships between GABA_A receptor subunit expression, receptor adaptations, function, and seizure susceptibility.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects. Male and female wild-type (1^+/+), heterozygous GABA_A receptor 1 subunit knockout (1^+/), and homozygous GABA_A receptor 1 subunit knockout (1^/) mice (Vicini et al., 2001) were derived from 1^+/ breeding pairs at the University of North Carolina (Chapel Hill, NC) or the University of Pittsburgh (Pittsburgh, PA). The wild-type allele consisted of a floxed allele in which the exon encoding nucleotides 1307 to 1509 of the 1 subunit was flanked by loxP sites that lacked a selectable marker gene. The knockout 1 allele consisted of the floxed allele after cre-mediated recombination. Briefly, the floxed allele was produced in Strain 129/Sv/SvJ embryonic stem cells. Chimeric offspring derived from these cells were mated to C57BL/6J mice and subsequently interbred for one generation. These mice were crossed with an actin-cre general deleter mouse line (FVB/N genetic background) to produce the recombined allele (Lewandoski et al., 1997). The cre transgene was subsequently eliminated from the pedigree. Mice that were heterozygous for the wild-type floxed allele and the recombined allele were interbred to produce the mice for experimental analysis. Thus, all mice were of the same mixed genetic background consisting of C57BL/6J (~37.5%), 129/Sv/SvJ (~37.5%), and FVB/N (~25%). All animals were genotyped by Southern blot analysis, as described previously (Vicini et al., 2001). After weaning, mice were group housed with same sex littermates, given free access to standard rodent chow and water, and maintained on a 12-h alternating light/dark schedule with lights on at 7:00 AM. All studies were conducted with mice derived from F5 to F7 generations and were 8 to 13 weeks of age. All studies were carried out in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by each institution's Animal Care and Use committees.

Western Blot Analysis. After decapitation, cerebral cortex was rapidly dissected over ice, frozen on dry ice, and stored at 80°C. P2 membrane fractions from cerebral cortex were prepared by homogenization in phosphate-buffered saline buffer (150 mM NaCl and 10 mM Na₂HPO₄/Na₂H₂PO₄, pH 7.4). Aliquots of 25 µg protein/lane were separated by SDS-polyacrylamide gel electrophoresis under reducing conditions using an Xcell II minicell apparatus (Novex, San Diego, CA). Proteins were transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore, Bedford, MA). Blots were probed with GABA_A receptor anti-peptide 1, 2, 3, 5, and 2 (Fritschy and Mohler, 1995), 4 (Kern and Sieghart, 1994), 2/3 (bd17; BMB, Indianapolis, IN), and  (Quirk et al., 1995) antibodies. Antibodies were kind gifts from Drs. Jean-Marc Fritschy (University of Zurich, Zurich, Switzerland) and Werner Sieghart (University of Vienna, Vienna, Austria). Blots were then probed with horseradish peroxidase-conjugated anti-guinea pig (1, 2, 3, and 5 and 2), anti-rabbit (4 and ), or anti-mouse (2/3, actin) antibodies. Specific peptide labeling was detected by enhanced chemiluminescence (Pierce Chemical, Rockford, IL). Blots were apposed to X-ray film (Eastman Kodak, Rochester, NY) under nonsaturating conditions and analyzed by densitometric measurements (NIH Image 1.47). All Western blots were conducted under conditions in which densitometric signals were linear with protein concentration as determined in preliminary experiments. Blots were reprobed with actin and normalized to verify equivalent protein loading.

Radioligand Binding. After decapitation, brains were immediately removed and placed in ice-cold saline from which cerebral cortices were rapidly dissected over ice and either used immediately or frozen on dry ice and stored at 80°C. Membranes were prepared by homogenization of cerebral cortices from eight mice per genotype in 50 volumes of assay buffer (50 mM Tris-citrate, pH 7.4) with a Polytron homogenizer. Samples were centrifuged at 20,000g for 20 min and resuspended in wash buffer five times before freezing the pellets at 80°C overnight. Pellets were washed twice more to remove endogenous GABA and used at a final concentration of 1 mg/ml. High-affinity [³H]muscimol (specific activity 30 Ci/mmol) (PerkinElmer Life Sciences, Boston, MA) binding was conducted over a concentration range of 0.5 to 100 nM in a final assay volume of 500 µl and incubated for 90 min at 0-4°C. Nonspecific binding was determined using 100 µM GABA. The reaction was terminated by rapid filtration under vacuum (<25 in. Hg) using GF/B filter strips (Whatman, Maidstone, UK) pretreated with 0.03% polyethylenimine. Samples were washed twice with 3-ml aliquots of assay buffer at 0-4°C. Filters were dried, added to liquid scintillation cocktail, and counted in a liquid scintillation counter. Saturation binding curves were evaluated using Prism (GraphPad Software, San Diego, CA) to obtain the K_D and B_max values and compared between genotypes by one-way ANOVA.

Autoradiography. t-Butylbicyclophosphorothionate ([³⁵S]TBPS) and tritium-labeled ethyl-8-azido-5,6-dihydro-5-methyl-6-oxo-4H-imidazo[1,5-a][1,4]benzodiazepine-3-carboxylate ([³H]Ro15-4513) were purchased from PerkinElmer Life Sciences. Flumazenil (Ro15-1788) was donated by F. Hoffmann-La Roche (Basel, Switzerland). Picrotoxin was purchased from Sigma-Aldrich (St. Louis, MO). For autoradiography, 14-µm horizontal or frontal serial sections were cut from three to five frozen mouse brains of each genotype using a cryostat (Microm), thaw-mounted onto gelatin-coated object glasses, and stored frozen under desiccant at 20°C. All experiments were carried out in parallel manner with respective genotype, eliminating any day-to-day variation in receptor assays between genotype. The autoradiographic procedures for regional localization of [³H]Ro15-4513 and [³⁵S]TBPS binding were as described previously (Makela et al., 1997). Briefly, sections were preincubated in an ice-water bath for 15 min in 50 mM Tris-HCl, pH 7.4, supplemented with 120 mM. The final incubation in the same buffer was performed with 6 nM [³⁵S]TBPS at room temperature for 90 min and assays with 10 nM [³H]Ro15-4513 at 0-4°C for 60 min. After incubation, sections in both assays were washed 3 × 15 s in ice-cold incubation buffer. Sections were then dipped into distilled water, air-dried under a fan at room temperature, and exposed with plastic [³H]methacrylate or [¹⁴C]methacrylate standards to Biomax MR films (Eastman Kodak) for 1 to 8 weeks. Nonspecific binding was determined with 10 µM Ro15-1788 and 100 µM picrotoxin in [³H]Ro15-4513 and [³⁵S]TBPS assays, respectively. Images from representative autoradiography films were scanned, processed with Adobe Photoshop (version 3.0; Adobe Systems, Mountain View, CA) and Corel Draw 5.0 programs, and printed for figures. The concentration of [³H]Ro15-4513 (10 nM) was greater than or equal to the dissociation constants for a range of recombinant and native GABA_A receptors (Pritchett et al., 1989; Luddens et al., 1990; Pritchett and Seeburg, 1990; Wisden et al., 1991). Therefore, the autoradiographic images should represent the density rather than affinity of binding sites. Autoradiography films were quantified using AIS image analysis system (Imaging Research, St. Catherines, Ontario, Canada) as described previously (Makela et al., 1997). Binding densities for each brain area were averaged from measurements of one to three sections per brain. The standards exposed simultaneously with brain sections were used as reference with the resulting binding values given as radioactivity levels estimated for gray matter areas (nCi/mg for ³H and nCi/g for ¹⁴C). Significance between the mouse lines in different brain regions was assessed by two-way ANOVA followed by Bonferroni's post hoc test using Prism.

Chloride Uptake Assay. After decapitation, brains were immediately removed and placed in ice-cold saline. Cerebral cortices of seven mice per genotype were pooled for each experiment. Synaptoneurosomes were prepared and Cl uptake was conducted as described previously (Morrow et al., 1988). The synaptoneurosomal pellet was resuspended in 6.6 volumes of ice-cold assay buffer (20 mM HEPES, 118 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO₄, and 2.5 mM CaCl₂, pH 7.4) for a final protein concentration of approximately 5 mg/ml. The homogenate was aliquoted 200 µl/assay tube and preincubated at 30°C for 12 min. Muscimol-stimulated Cl uptake was initiated by addition of 0.2 µCi of ³⁶Cl (PerkinElmer Life Sciences) in the presence of various concentrations of muscimol (1-200 µM). The solution was vortexed and uptake terminated after 5 s by addition of 4 ml of ice cold assay buffer containing 100 µM picrotoxin with rapid vacuum filtration over G6 filters (Fisher Scientific, Pittsburgh, PA) using a single manifold filter apparatus (Hoeffer, San Francisco, CA). After two more washes, filters were allowed to dry and radioactive counts determined by liquid scintillation spectroscopy. Basal chloride uptake was measured in the absence of muscimol and subtracted from all tubes to determine muscimol-stimulated chloride uptake. Concentration-response curves were evaluated using computerized nonlinear regression (Prism; GraphPad Software) to obtain the EC₅₀ and E_max values and compared between genotype by one-way ANOVA.

Bicuculline-Induced Seizure Threshold Test. Seizure thresholds were determined at the beginning of the light cycle as described previously (Devaud et al., 1995). Mice were restrained in a Plexiglas plunger-style mouse restraint (Braintree Scientific, Braintree, MA). Threshold determination was made by constant lateral tail vein infusion of bicuculline (Sigma-Aldrich) dissolved in 0.1 N HCl, and diluted with isotonic saline to a final concentration of 0.05 mg/ml, pH 7. The solution was infused at a constant rate of 0.5 ml/min; the endpoint was taken as the first myoclonic jerk of the head and neck. This time point precedes forepaw clonus and generalized tonic/clonic convulsions. Each animal was tested once. Seizure thresholds were determined by experienced observers who were blind to the experimental conditions. Seizure thresholds were calculated from the time of infusion × dose of bicuculline per body weight and presented as milligrams per kilogram of bicuculline. Data were analyzed by one-way ANOVA using Prism.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

1 Subunit Deletion Produces Selective Loss of 2/3 and 2 Subunit Peptides and Elevations in 2 and 3 Subunit Peptides. Western blot analysis demonstrated the successful elimination of 1 subunit peptide expression in cerebral cortex (Fig. 1) as well as whole brain (Vicini et al., 2001). 1^+/ and 1^/ mice exhibited 39 ± 3% (p < 0.0001) and 100% reductions in 1 subunit peptide expression in the P2 membrane fraction of cerebral cortex, respectively. Because the major GABA_A receptor subtype containing 1 subunits expressed in brain is l22 (Barnard et al., 1998), we measured the effect of the loss of 1 subunit expression on 2/3 and 2 subunit expression (Fig. 1). Reductions in 2/3 subunit expression (35 ± 4%; p < 0.01; 65 ± 7%; p < 0.001) were observed in 1^+/ and 1^/ mice, respectively. Expression of 2 subunit peptide was reduced by 47 ± 9% (p < 0.001) in 1^/ mice. Adaptations in the expression of other GABA_A receptor  subunit variants were observed in 1^/ mice. There were 37 ± 4% (p < 0.001) and 39 ± 6% (p < 0.001) increases in 2 and 3 subunit peptide expression, respectively, in 1^/ mice. No significant differences in 4, 5, or  subunit peptide expression were observed across genotypes. Data were analyzed from generations F5 to F7 and identical results were obtained (data not shown); therefore, data from all generations were collapsed in Fig. 1.

View larger version (79K): [in this window] [in a new window]   Fig. 1.   Western blot analysis of GABAA receptor subunit polypeptide expression of 1+/+, 1+/, and 1/ mice. The membrane fraction (P2) of cerebral cortex was examined using 1, 2, 3, 4, 5, 2/3, 2, and  subunit-specific polyclonal antibodies. A, histogram of subunit expression showing data expressed as the percentage of control after normalization to -actin expression. B, representative blot of subunit expression paired with corresponding -actin expression. Data represent the mean ± S.E.M. of three to six independent experiments of n = 6 mice/genotype/experiment and analyzed using a one-way ANOVA with Tukey's post hoc test. , p < 0.01; , p < 0.001 statistical significance compared with 1+/+ mice.

Pharmacological and Functional Characterization of 1 Subunit Knockout Mice. Inhibition of 1 subunit expression resulted in a reduction of GABA_A receptor density in the membrane fraction of the cerebral cortex. High-affinity [³H]muscimol saturation binding studies revealed a 56 ± 8% reduction (p < 0.005) in binding sites with unaltered affinity (K_D) in 1^/ mice (Fig. 2A). Saturation binding experiments using the direct GABA_A receptor antagonist [³H]SR-95531 were conducted to verify the apparent loss in receptor number observed with [³H]muscimol binding because agonist binding can vary with activation state of the receptor (Colquhoun, 1998). Binding of 25 and 200 nM [³H]SR-95531 were measured because these concentrations correspond to the approximate half-maximal (K_D) and saturating (B_max) concentrations of the ligand (McCabe et al., 1988). In agreement with the reduction in [³H]muscimol binding, 1^/ mice exhibited a 57 ± 14% reduction (p < 0.05) in the binding of 200 nM [³H]SR-95531 (Fig. 2B). At 25 nM [³H]SR-95531, binding was reduced by 38 ± 11% (p < 0.05) and 53 ± 8% (p < 0.01) in 1^+/ and 1^/ mice, respectively.

View larger version (29K): [in this window] [in a new window]   Fig. 2.   [3H]Muscimol and [3H]SR-95531 binding profiles in 1+/+, 1+/, and 1/ mice. A, saturation isotherm and Scatchard analysis of [3H]muscimol binding to crude membrane fraction of pooled cerebral cortices of 1+/+ and 1/ mice (n = 8 mice/genotype/experiment). [3H]Muscimol maximal binding (Bmax) was reduced from 2104 ± 154 to 931 ± 70 fmol/mg of protein in 1+/+ and 1/ mice, respectively, with a nonsignificant reduction in affinity (1+/+, KD, 28.9 ± 4.7 nM; 1/, KD, 19.6 ± 1.1 nM). B, binding of 25 nM (~KD) and 200 nM (~Bmax) [3H]SR-95531 to crude membrane fraction of pooled cerebral cortices of 1+/+ and 1/ mice (n = 2 mice/genotype/experiment). Binding of 25 nM [3H]SR-95531 was reduced from 627 ± 42 to 391 ± 45 to 289 ± 14 fmol/mg of protein in 1+/+, 1+/, and 1/ mice, respectively. Binding of 200 nM [3H]SR-95531 was reduced from 1047 ± 128 to 445 ± 44 fmol/mg of protein in 1+/+ and 1/ mice, respectively, whereas binding was reduced nonsignificantly to 743 ± 92 fmol/mg of protein in 1+/ mice. Data represent the mean ± S.E.M. of three independent experiments, each conducted in triplicate and analyzed using a one-way ANOVA with Tukey's post hoc test. , p < 0.05; , p < 0.01 statistical significance with respect to 1+/+ mice.

/

/

/

View larger version (75K): [in this window] [in a new window]   Fig. 3.   Representative autoradiographs of GABAA receptor binding to 6 nM [35S]TBPS and 10 nM [3H]Ro 15-4513 in 1+/+ and 1/ mouse brains. The images were scanned from the films with each pair of them being processed with similar brightness and contrast settings. Ctx, cerebral cortex; Th, thalamus; Gr, cerebellar granule cell layer; Str, striatum; Hi, hippocampus; IC, inferior colliculus.

View this table: [in this window] [in a new window]   TABLE 1 Alterations in brain regional ligand binding in GABAA receptor subunit 1/ mice Results are means ± S.E.M. for three and five brains for the control and knockout groups, respectively. The values for 10 nM [3H]Ro15-4513 binding are in nanocuries per milligram and those for 6 nM [35S]TBPS binding in nanocuries per gram in reference to microscale standard for brain gray matter. Two-way ANOVA detected significant (p < 0.0001) interactions between the genotype and brain region, and significant brain region [F(5,36) > 63, p < 0.0001] and genotype effects [F(1,36) > 376, p < 0.0001] in [35S]TBPS and [3H]Ro15-4513 binding.

+/

/

+/

/

View larger version (26K): [in this window] [in a new window]   Fig. 4.   [3H]Zolpidem and [3H]flunitrazepam binding profiles in 1+/+, 1+/, and 1/ mice. Saturation isotherm and Scatchard analysis of [3H]zolpidem (A) and [3H]flunitrazepam (B) binding to crude membrane fraction of pooled cerebral cortices of 1+/+, 1+/, and 1/ mice (n = 2 mice/genotype/experiment). [3H]Zolpidem binding was reduced from 1283 ± 117 to 737 ± 126 to 36 ± 12 fmol/mg of protein in 1+/+, 1+/, and 1/ mice, respectively, with no significant change in KD between 1+/+ and 1+/ (20.0 ± 1.0 nM versus 14.4 ± 3.4 nM, 1/ KD, not detectable). [3H]Flunitrazepam binding (Bmax) was reduced from 1798 ± 25 to 1289 ± 29 to 971 ± 58 fmol/mg of protein in 1+/+, 1+/, and 1/ mice, respectively, with no significant change in affinity (KD) (1+/+, 6.4 ± 0.8 nM; 1+/, 6.1 ± 0.9 nM; 1/, 5.2 ± 1.1 nM). Data represent the mean ± S.E.M. of three to four independent experiments, each conducted in triplicate and analyzed using a one-way ANOVA with Tukey's post hoc test.

/

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Muscimol-stimulated Cl uptake in pooled cortical synaptoneurosomes of 1+/+, 1+/, and 1/ mice (n = 7-8 mice/genotype/experiment). The maximal response (Emax) to muscimol was significantly reduced from 30.4 ± 1.1 to 21.7 ± 0.7 nmol of Cl/mg of protein in 1+/+ and 1/ mice, respectively, with a slight reduction to 28.8 ± 1.4 nmol of Cl/mg of protein in 1+/ mice. The potency of muscimol was significantly reduced from 3.7 ± 0.6 to 7.9 ± 1.0 µM in 1+/+ and 1/ mice, respectively, with a slight reduction to 5.0 ± 0.7 µM in 1+/ mice. Data represent the mean ± S.E.M. of four independent experiments each conducted in quadruplicate and analyzed using a one-way ANOVA with Tukey's post hoc test.

Behavioral Phenotype of 1 Subunit Knockout Mice. Breeding 1^+/ mice resulted in a Mendelian distribution of genotypes among the offspring of 1:2:1 (1^+/+/1^+//1^/). The distribution of genotypes among offspring did not vary with generation or gender. Furthermore, the number of pups per litter (approximately eight) remained constant between generations. Knockout mice were overtly normal and indistinguishable from littermates except for a 25-Hz tremor when handled and a small, but significant reduction in body weight. In a sample of 60 to 70 mice per genotype, 1^/ mice exhibited 10 ± 3% (p < 0.01) lower body weights compared with wild-type mice between 9 and 12 weeks of age. A few 1^/ mice seemed dwarfed at birth and oftentimes did not survive into adulthood.

+/

/

/

/

View larger version (33K): [in this window] [in a new window]   Fig. 6.   Bicuculline-induced seizure thresholds of 1+/+, 1+/, and 1/ mice. Seizure thresholds were determined as the dose of bicuculline required to reach seizure threshold normalized to the mass of the mouse. Data represent the mean ± S.E.M. of 23 to 26 mice/genotype and were analyzed using one-way ANOVA with Tukey's post hoc test. , p < 0.001 statistical significance with respect to 1+/+ mice.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Production of mice with targeted deletion of individual GABA_A receptor subunits has provided valuable insight into the contribution of specific receptor subtypes to CNS mechanisms controlling inhibitory tone, GABA-related disorders, and actions of drugs that act upon GABA_A receptors. Likewise, the initial characterization of the 1^/ mouse line has revealed several interesting phenomena related to 1 expression. Although the mice exhibit reduced GABA_A receptor number and function and adaptations in GABA_A receptor subunit expression, deletion of the 1 subunits resulted in viable offspring with a handling-induced tremor, but maintained normal fertility, litter size, and body weight across several generations. The overtly normal behavior of the mice is remarkable and suggests that the CNS expresses roughly twice as many GABA_A receptors as required to maintain normal inhibitory tone. However, handling-induced tremor in ^/ mice may suggest reduced CNS inhibition and increased seizure susceptibility in these mice provides evidence that further blockade of GABA_A receptors severely impairs inhibitory neurotransmission.

The 1^/ mice provide an interesting model to study the assembly of GABA_A receptors. Deletion of the 1 subunit resulted in a dramatic reduction in the number of GABA_A receptors in membrane fractions and brain sections presumably due to loss of the 1 subunit-containing receptors. Loss of the majority of cortical GABA_A receptors as measured by [³H]muscimol, [³H]SR-95531, and [³⁵S]TBPS binding in conjunction with similar magnitude losses of 2/3 and 2 subunit expression provides further support that 1 subunits are required for the assembly of most 2/32 receptors in brain, despite compensatory increases in 2 and 3 subunit peptide expression. The observed decrease in [³H]flunitrazepam saturation binding and [³H]Ro15-4513 autoradiographic binding is also consistent with this hypothesis. In a previous study,  subunits were also shown to determine receptor assembly in vivo as deletion of 5 subunits in hippocampal pyramidal cells prevented receptor assembly and reduced receptor binding (Fritschy et al., 1997).

Absolute differences in brain regional binding levels were detected by autoradiographic analysis of [³⁵S]TBPS versus [³H]Ro15-4513 binding as described previously (Olsen et al., 1990; Turner et al., 1991; Makela et al., 1997; Korpi et al., 2002). Furthermore, deletion of the 1 subunit differentially altered [³⁵S]TBPS versus [³H]Ro15-4513 within brain regions. These discrepancies are likely due to several factors, including differences in the receptor subtypes labeled by each ligand (Korpi et al., 2002) and differences in the percentage of receptor occupancy of the ligands (based on the concentrations tested). The reduction in [³⁵S]TBPS binding in the 1^/ mice was greater than the reduction in [³H]Ro15-4513 binding in most brain regions. These data may indicate that [³H]Ro15-4513 preferentially labels receptor populations that are less affected by deletion of 1 subunits or those that are compensatorily increased by elevated 2 and 3 subunit expression. [³⁵S]TBPS exhibits weak binding to some of the 2 and 3 subunit-containing receptors (Lüddens and Korpi, 1995) that would be labeled by [³H]Ro15-4513. Therefore, compensatory increases in 2 or 3 subunit-containing receptors would be expected to diminish the loss of [³H]Ro15-4513 binding more than the loss of [³⁵S]TBPS binding. These data are consistent with the possibility that compensatory changes in 2/3 subunit-containing receptors are present in 1^/ mice. Clearly, both ligands demonstrate dramatic reductions in GABA_A receptor binding sites.

Benzodiazepine binding sites, located at the interface of  and 2 subunits (for review, see Sigel and Buhr, 1997), were severely affected by deletion of the 1 subunit. Complete loss of high-affinity [³H]zolpidem binding in 1^/ mice confirms the selectivity of this ligand for 1 subunit-containing receptors as described in vitro (Pritchett et al., 1989). Previous studies suggest that up to 70% of BZD binding sites have properties of 1 subunit-containing (type I) receptors (Braestrup and Nielsen, 1981; Niddam et al., 1987; Dennis et al., 1988). However in 1^/ mice, we observed a loss of 46% of [³H]flunitrazepam binding, suggesting that compensation by increased expression of 2 and 3 subunits may have resulted in the increased assembly of 2 and 3 subunit-containing (type II) BZD receptors. This possibility is consistent with the observation that the decrease in the B_max of [³H]zolpidem binding (1247 fmol/mg of protein) was greater than the loss of maximal [³H]flunitrazepam binding (827 fmol/mg of protein) in 1^/ mice. Alterations in BZD pharmacology observed in 1^/ mice would be expected to impact the behavioral actions of BZDs and may aid in defining properties associated with actions at various BZD receptors.

Elegant studies conducted in knockin mice containing a point mutation at the BZD binding site of the  subunit have provided models in which the behavioral responses to an array of BZD site ligands have been selectively attenuated without affecting receptor assembly or sensitivity to GABA (for review, see Rudolph et al., 2001). In 1(H101R) mice, the sedative and amnesic effects of diazepam were ablated, whereas the seizure-protective effect was partially reduced (Rudolph et al., 1999). Similar studies conducted in 2(H101R) and 3(H101R) mice have attributed the anxiolytic effects of diazepam to 2-containing receptors (Low et al., 2000). Together, these studies have begun to dissociate the receptor subtypes responsible for the anxiolytic, sedative, myorelaxant, motor-impairing, hypnotic, amnesic, and ethanol-potentiating effects of BZDs. Furthermore, the pharmacological and behavioral profile of current and future selective BZDs can be examined in these mice (e.g., zolpidem) (Crestani et al., 2000). These results would predict that the 1^/ mice will exhibit a loss of the sedative-hypnotic effects of diazepam and zolpidem with a partial loss in the anticonvulsant effect of diazepam. We have observed a complete loss of the ability of diazepam (1-30 mg/kg) to prevent bicuculline-induced seizures in 1^/ mice (Kralic et al., 2001). Additional studies are underway to further compare these models.

The alterations in muscimol-stimulated ³⁶Cl uptake in cerebral cortex of 1^/ mice indicate a dramatic reduction in GABAergic tone that may underlie the increased seizure susceptibility in 1^/ mice. Studies in stably transfected mouse L(tk) cells reported that the maximal effect of muscimol-stimulated chloride flux varies with expression levels of 1, 5, or 6 subunits in conjunction with the  and  subunits (Harris et al., 1998). Several in vivo studies suggest that the presence of the 1 subunit contributes to the efficacy of GABA agonists. Knockdown of 1 subunits using antisense deoxyoligonucleotides resulted in a decrease in GABA-mediated chloride flux (Malatynska et al., 2000). Patch-clamp recordings measured a reduction in the amplitude of evoked inhibitory postsynaptic currents in slices from visual cortex after treatment with 1 antisense deoxyoligonucleotides (Brussaard and Baker, 1995). A decrease in 1 subunit mRNA and peptide expression has been associated with a decrease in the E_max of muscimol-stimulated chloride flux after chronic ethanol administration (Morrow et al., 1988; Devaud et al., 1997), whereas there is no change in total receptor number (for review, see Grobin et al., 1998). Although these studies support the hypothesis that 1 subunits contribute to the functional capacity of the receptor, a reduction in GABA_A receptor number may also contribute to the loss of agonist efficacy. In addition, the reduced potency of muscimol in 1^/ mice suggests that  subunits modulate the potency of muscimol, a property that should not be affected by receptor number, but is consistent with lower GABA sensitivities of 2 and 3 subunit-containing receptors than 1 subunit-containing receptors (Lüddens and Korpi, 1995).

The present study demonstrates that GABA_A receptor subunit expression and function are altered in the cerebral cortex of 1^/ mice. In a previous study using these mice, Vicini et al. (2001) demonstrated that 1 subunits contribute to the developmental shortening of spontaneous inhibitory postsynaptic currents and mIPSCs in stellate cells of the cerebellum. Because GABA_A receptor binding using [³⁵S]TBPS and [³H]Ro15-4513 are reduced in cerebellum and throughout brain, it is likely that GABA_A receptor function, subunit expression, and assembly are also altered in many brain regions. Studies are underway to further investigate this possibility.

The decrease in bicuculline-induced seizure threshold measured in 1^/ mice suggests that 1 subunit-containing receptors influence seizure susceptibility through maintenance of inhibitory tone. As described earlier, a strong association between the expression of the 1 subunit and seizure susceptibility has been drawn from developmental, epileptic, and alcohol dependence and withdrawal models. The developing brain expresses predominately GABA_A receptor 2, 3, and 5 subunits during embryonic and early postnatal stages shifting to high levels of 1 subunit expression in adult brain (Poulter et al., 1992; Laurie et al., 1992). This switch in the expression pattern of  subunits during brain development is associated with a decreased seizure susceptibility in adult brain (Aicardi and Chevrie, 1970; Mecarelli et al., 1988). Moreover, models of seizure-prone and -resistant mice express 50% less and 200% more 1 subunit than controls, respectively (Poulter et al., 1999). Withdrawn ethanol-dependent rats exhibit increased seizure susceptibility and a concomitant reduction in 1 subunit expression (Devaud et al., 1997). These findings support the hypothesis that 1 subunit expression serves a critical role in regulating CNS excitability and seizure susceptibility and a potential target for anticonvulsant drug therapy.

Recently, production of an 1 subunit knockout mouse (MSD mice; Sur et al., 2001) was achieved using a different gene targeting method (Sur et al., 2001; Vicini et al., 2001). Although both mouse lines share mostly similar phenotypes, they also differ in several ways. In addition to the successful deletion of the 1 subunit and production of viable knockout mouse lines, both lines exhibited similar changes in GABA_A receptor pharmacology and an absence of any overt behavioral differences. Furthermore, both lines displayed increased peptide expression of 2 and 3 subunits. Although these adaptations were consistent across generations in our mouse line, the adaptations were gradually lost in successive generations of MSD mice (Sur et al., 2001). This difference is likely due to distinct gene-targeting methods or breeding strategies used. Interbreeding of 1^/ mice may have resulted in the smaller litter sizes observed in MSD mice in earlier generations. Litter sizes may have recovered in size due to selection of knockout mice able to survive 1 gene deletion or those containing adaptations in 2 and 3 subunit expression. Measurement of mRNA revealed no adaptations in 3 and 2 expression in MSD mice (Sur et al., 2001), whereas 2/3 and 2 peptide expression was significantly reduced in our mice, suggesting either another inherent difference between lines or adaptations in subunit expression mediated by translational control or assembly mechanisms. Although functional studies conducted in both mouse lines used different methods and targeted different brain regions, they collectively showed that GABA_A receptor function is diminished in the absence of 1 subunit expression. Differences between these mouse lines may be inherent to the line or due to varying techniques used for breeding and investigation. Together, these independent studies provide a thorough examination of the role of the 1 subunit in GABA_A receptor pharmacology, function, and related behaviors.

In conclusion, we have shown that global deletion of the 1 subunit of the GABA_A receptor results in reduced agonist-induced receptor function, BZD pharmacology, and increased seizure susceptibility without affecting viability or overt behavior. The increased seizure susceptibility of 1^/ mice supports the role of 1 subunit expression in the maintenance of CNS inhibitory tone suggested by models of epilepsy and alcohol withdrawal. Adaptations in GABA_A receptor subunit expression provide further evidence for plasticity in the GABAergic system. Results from the present studies suggest a specific role for 1 subunit-containing receptors in GABAergic neurotransmission and extend previous findings supporting the 1 subunit-containing receptor as a target for the development of GABA_A receptor subtype-specific drugs, e.g., for new antiepileptic/anticonvulsant drugs. Our new mouse model will be useful in the study of mechanisms that regulate GABA_A receptor assembly as well as testing possible new drug entities. Furthermore, important pharmacodynamic studies remain such as determining the response of 1^/ mice to the actions of benzodiazepines, alcohol, volatile anesthetics, barbiturates, and neurosteroids, all of which act upon GABA_A receptors.