Introduction

Top Abstract Introduction Methods Results Discussion References

The GABA_A receptor, a ligand-gated anion channel, carries sites of action for several clinically important compounds, such as benzodiazepines and barbiturates. The receptors consist of several glycoprotein subunits, most of which exist in several isoforms, encoded by different genes (e.g., alpha-1-6, beta-1-3, gamma-1-3) (Macdonald and Olsen, 1994). The distribution of these subunits in the brain shows regional and cellular differences (Wisden et al., 1992). It is thought that there are several GABA_A receptor subtypes in brain due to the differential assembly of subunits. GABA_A receptors can be investigated using radioligands recognizing different binding sites, including the benzodiazepine binding site, GABA binding site and TBPS site (associated with the ion channel domain of the receptor). Benzodiazepines act via the benzodiazepine binding site, which is a modulatory site located on the GABA_A receptor. The presence of a gamma subunit in the GABA_A receptor is necessary for formation of a benzodiazepine recognition site and for potentiation of the GABA response by benzodiazepines (Pritchett et al., 1989b; Günther et al., 1995). This binding site may be located at the interface between alpha and gamma subunits (Sigel and Buhr, 1997; Stephenson, 1995). In contrast, the GABA recognition site may be at the interface of alpha and beta subunits (Sigel and Buhr, 1997). In the rat brain, the gamma-2 isoform is the predominant gamma isoform, is widely expressed throughout most regions of the rat brain and most often is associated with alpha-1 and beta-2 subunits (Benke et al., 1994).

The physiological and pharmacological properties of GABA_A receptors are largely determined by the receptor subunit composition (Macdonald and Olsen, 1994; Sieghart, 1995). Compounds that are selective for receptors incorporating particular subunit combinations might then be useful tools for studying the contribution of individual subunits to the properties of receptors in adult brain cells. So far, only a few such ligands are known; these include the benzodiazepine site ligands zolpidem, which has a relative preference for binding to GABA_A receptor containing the alpha-1 subunit (Arbilla et al., 1986; Wu et al., 1994a), and RY-80, which is selective for receptors containing the alpha-5 subunit (Skolnick et al., 1997). Several studies using various radioligands have shown that the GABA_A receptor population in adult brain is plastic. For example, chronic flurazepam treatment was associated with reduced brain benzodiazepine binding (Rosenberg and Chiu, 1981a, 1981b; Tietz et al., 1986). Using the selective ligand zolpidem, an even greater loss of zolpidem-sensitive sites was demonstrated, suggesting a change in receptor composition (Wu et al., 1994a). This demonstrated the usefulness of such selective ligands as tools. Chronic benzodiazepine treatment was also associated with decreased levels of mRNAs for several GABA_A receptor subunits, including gamma-2 (Zhao et al., 1994; Wu et al., 1994b).

Another possibility for studying the role of GABA_A receptor subunits in adult brain is the use of ASO technology to reduce the expression of selected proteins. Several studies have shown this to be a useful approach for investigating the function of individual proteins in the central nervous system. ASOs directed against the alpha-1, alpha-2, alpha-6 or gamma-2 subunits of the GABA_A receptor have been used to study receptors in cell culture (Brussaard and Baker, 1995; Zhu et al., 1996). More recently, Smith et al. (1998) showed that pretreating rats with an ASO against the alpha-4 subunit of the GABA_A receptor prevented the increased seizure susceptibility observed after progesterone withdrawal, which was apparently related to an increase in alpha-4 expression. In other studies, an ASO for the GABA_A receptor gamma-2 subunit was infused into rat brain and shown to reduce benzodiazepine binding but with some indication of toxic effects (Karle et al., 1997a, 1997b). In a previous study from our laboratory, a similar ASO directed against rat gamma-2 subunit was also used to study GABA_A receptor regulation in adult rat brain. It was found that the convulsive threshold dose for -CCM, a benzodiazepine "inverse agonist," was increased 87% in rats infused by the i.c.v. route with gamma-2 ASO (every 12 hr for 3 days) but was not affected by the sense (control) oligodeoxynucleotide (Zhao et al., 1996). In contrast, there was no change in picrotoxin seizure threshold and no difference in strychnine threshold between sense and ASO-treated rats. These results suggested that ASO treatment had interfered with the actions of the benzodiazepine binding site in vivo and that this may have been selective for the benzodiazepine recognition site of GABA_A receptors. In the present study, the same ASO treatment method against gamma-2 subunit was used to study the effect of such treatment on the binding of several ligands to the GABA_A receptor. It was hypothesized that gamma-2 ASO treatment would decrease benzodiazepine binding without changing the number of GABA_A receptor complexes.

	Methods

Top Abstract Introduction Methods Results Discussion References

Oligodeoxynucleotides. A 17-base ASO for the region starting at position 2 after the initiation codon was based on the sequence for the rat GABA_A receptor gamma-2 subunit mRNA (Shivers et al., 1989). Antisense (5'-CATGTATTTGGCGAACT-3') and sense (5'-AGTTCGCCAAATACATG-3') phosphorothioate-modified oligodeoxynucleotides were synthesized by Oligos Etc., Inc. (Wilsonville, OR). These were dissolved in sterile, filtered saline for i.c.v. injection.

Animals. Male Sprague-Dawley rats (Harlan, Indianapolis, IN) weighing 250 to 300 g were kept under standard conditions of a 12-hr light/dark cycle with free access to standard rat food and water. After an acclimation period of 4 days, surgery was performed to implant the guide cannula for i.c.v. injections.

Cannula implantation and oligodeoxynucleotide administration. Rats were implanted with unilateral stainless steel guide cannulae aimed at a point 2.0 mm above the lateral ventricle. Under sodium pentobarbital anesthesia (45 mg/kg i.p.), animals were placed in a Kopf rat stereotaxic device. Using sterile technique, the dorsal surface of the skull was exposed, and a hole was drilled to yield an implantation site corresponding to 0.5 mm (caudal) and 1.5 mm lateral to bregma according to the atlas of Paxinos and Watson (1986). A sterilized guide cannula was lowered into the brain tissue 2 mm below skull surface and fixed with dental cement and a screw. A close-fitting stainless steel obturator was used to occlude the cannula. The animals were allowed to recover for 10 days before beginning treatment. Oligodeoxynucleotide solution was administered into the right i.c.v. space of the conscious animals via an injection cannula that extended 2.0 mm beyond the tip of the guide cannula. The rats received the ASO injection (18 µg in 2 µl saline) every 12 hr for 3 days, beginning in the evening. The control group received the corresponding sense oligodeoxynucleotide. Solutions were slowly infused over 1 min using a Harvard infusion pump, and the injection cannula was left in place for an additional 1 min before it was slowly withdrawn and replaced with the obturator. The treatment duration was limited to 3 days by the obvious weight loss of ASO-treated rats (Zhao et al., 1996).

Brain slice preparation. Rats were sacrificed by decapitation 6 hr after the final i.c.v. injection. Brains were removed quickly and immersed in isopentane cooled in an acetone-dry ice bath. Sagittal sections (10 µm) were cut from the right side of each brain in a cryostat microtome (14°C) and were thaw-mounted onto slides previously coated with 0.5% gelatin and 0.05% chrome alum. Slide-mounted tissue sections were transferred to ice-cold slide boxes and stored at 70°C.

[³H]Flunitrazepam binding. Slices were preincubated in 170 mM Tris-HCl buffer (pH 7.4) at 4°C for 30 min and rapidly dried with a cold stream of air. Incubation was performed for 60 min at 4°C in the same buffer in the presence of 5 nM [³H]flunitrazepam (86.0 Ci/mmol, Amersham, Arlington Heights, IL). Nonspecific binding was determined by incubating an adjacent section in the presence of radioligand plus 1 µM clonazepam. Incubation was terminated by rinsing sections twice (30 sec each time) in the Tris-HCl buffer. The slides were then dipped in deionized water and finally dried with a stream of cool air. The dried sections were placed next to tritium-sensitive film (Hyperfilm-[³H], Amersham), which was exposed at 4°C for 12 days.

[³H]Muscimol binding. [³H]Muscimol binding was performed according to the method of Johnson et al. (1994). Briefly, slide-mounted sections were preincubated twice in 50 mM Tris-acetate buffer, pH 7.1, at room temperature for 15 min. Sections were then transferred to buffer containing 5 nM [³H]muscimol (19.1 Ci/mmol, New England Nuclear, Boston, MA) for 60 min. To determine nonspecific binding, unlabeled 100 µM GABA was added. After the incubation, the slides were rinsed twice (30 sec each time), followed by a final rinse in glutaraldehyde/acetone (2.5% v/v) at 4°C. The dried sections were placed in x-ray cassettes and exposed to tritium-sensitive film at 4°C for 6 weeks.

[³H]GABA binding. The method for [³H]GABA binding was adopted from Bristow and Martin (1988). Slide-mounted tissue sections were incubated in the presence of 50 nM [³H]GABA (92.0 Ci/mmol, Amersham) for 20 min at room temperature in buffer (pH 7.4) including 50 mM Tris-HCl, 190 mM sucrose and 100 µM baclofen to displace binding to GABA_B receptors. Nonspecific binding was assayed in the presence of 100 µM isoguvacine along with 100 µM baclofen. Incubation was terminated by rinsing sections twice (3 sec each time) in 50 mM Tris-HCl buffer at 4°C. Then the slides were dipped in deionized water and finally dried with a stream of cool air. The dried sections were placed in film cassettes and exposed to tritium-sensitive film at 4°C for 14 days.

[³H]EBOB binding. The picrotoxin site on the GABA_A receptor complex, thought to be associated with the anion channel, was examined by [³H]EBOB binding. The assay was performed according to the method of Kume and Albin (1994). Tissue sections were prewashed three times (10 min each time) in 50 mM Tris-HCl buffer (pH 7.4) containing 1 mM EDTA at 4°C and dried under a stream of cool air. Binding of 3 nM [³H]EBOB (44.6 Ci/mmol, New England Nuclear) was carried out for 120 min in 50 mM Tris-HCl and 120 mM NaCl buffer (pH 7.4) at room temperature. Nonspecific binding was assessed in the presence of 20 µM picrotoxin. Incubation was terminated by washing twice (60 min each time) in 50 mM Tris-HCl, pH 7.4, at 4°C. Then the slides were dipped in deionized water and finally dried with a stream of cool air. The dried sections were placed in x-ray cassettes with tritium-sensitive film at 4°C for 3 weeks.

[³H]DPCPX binding. The method for [³H]DPCPX binding was as previously described (Fastbom and Fredholm, 1990). The slide-mounted brain slices were preincubated in 170 mM Tris-HCl buffer (pH 7.4), containing 2 IU/ml adenosine deaminase (Boehringer-Mannheim, Indianapolis, IN), for 120 min at room temperature, and then those slides were incubated in the presence of 0.8 nM [³H]DPCPX (120 Ci/mmol, New England Nuclear) in the above-mentioned buffer for 120 min. Nonspecific binding was studied by adding 30 µM ()-N⁶-phenylisopropyladenosine. After the incubation, the slides were washed twice (2 min each time) at 4°C in 170 mM Tris-HCl buffer and then dried under a stream of cold air. The dried sections were placed in x-ray cassettes along with tritium-sensitive film at 4°C for 8 days.

Quantitative autoradiography. Autoradiograms were generated by exposing the ligand-labeled tissue slides to tritium-sensitive film in cassettes along with standards containing known amounts of radioactivity. Ligand binding was quantified with computer-assisted densitometry using the NIH Image software. To quantify ligand binding density, the optical density of coexposed standards were determined, and a standard curve was generated. The tritium standards (courtesy of Dr. E. I. Tietz) were 10 disks of rat brain paste containing known amounts of [³H]thymidine, mounted on a single slide and fixed with paraformaldehyde vapor. The amount of each ligand bound in the rat brain regions was determined by converting optical density measurement to pmol/mg protein (Tietz et al., 1986). For each brain, 14 regions were measured on 4 sagittal sections, and the mean of the 4 values for each region was used. Specific binding was the difference between total and nonspecific binding.

Immunoblotting and quantitative densitometry. The left side of the cerebellum was rapidly dissected and stored at 70°C. The cerebellar tissues were homogenized in 3.0 ml of ice-cold medium containing 0.32 M sucrose, 10 mM Tris-HCl (pH 7.0), 1 mM sodium EDTA and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 µg/ml aprotinin and 1 µg/ml leupeptin). The homogenates were centrifuged at 1000 × g for 10 min. The supernatants were collected and centrifuged at 10,000 × g for 20 min. The resulting pellets were washed once with 5 ml of 10 mM Tris-HCl, pH 7.0, to remove the protease inhibitors. Aliquots of protein, 10 µg from each sample, were loaded onto the gel, and subjected to SDS-polyacrylamide gel electrophoresis (10% acrylamide) then transferred to nitrocellulose membrane overnight at 4°C. The nitrocellulose membrane was treated for 1 hr at room temperature with blocking reagent, consisting of TBS buffer (20 mM Tris-HCl, pH 7.6, and 137 mM NaCl), 0.1% Tween 20 and 5% nonfat dry milk. The membrane was then exposed to anti-GABA_A receptor beta-2/beta-3 subunit monoclonal antibody (bd-17; Boehringer-Mannheim) 1.0 µg/ml, overnight at 4°C. The membrane was then exposed to anti-mouse IgG (dilution 1:1000; Boehringer-Mannheim) for 2 hr at room temperature. The membrane was washed and placed in ECL reagent solution (Boehringer-Mannheim) for 1 min. The blot was then placed in an autoradiography cassette along with light-sensitive autoradiography film (Hyperfilm ECL, Amersham) for varying periods as needed to achieve adequate band density but staying well below the saturation level of the film. Autoradiographs were analyzed with a BioRad imaging densitometer (model GS-670). For each set of data on an autoradiograph, the mean density of the control samples was set as 1.0, and then the relative density of each band was taken as a fraction of this value.

Statistical analyses. Each set of the antisense and sense groups were treated in parallel throughout. For analysis of radioligand binding, the analyses for the antisense and sense groups were carried out using analysis of variance (ANOVA), with treatment and brain region as variables. In the case of a significant treatment effect, planned comparisons were further performed for each brain region. Results of the immunoblot analysis were evaluated by Student's t test. In all cases, P < .05 was considered statistically significant.

	Results

Top Abstract Introduction Methods Results Discussion References

[³H]Flunitrazepam binding. Figure 1A shows a representative autoradiogram of 5 nM [³H]flunitrazepam binding in a brain section obtained from a rat after treatment with the sense oligodeoxynucleotide. [³H]Flunitrazepam binding was unevenly distributed throughout the rat brain, with a pattern like that in previous reports (Tietz et al., 1986; Suzuki et al., 1996). Nonspecific binding ranged from 2% to 20% of total binding in those areas examined. As shown in figure 2, there was a 15% to 25% decrease of specific [³H]flunitrazepam binding in most brain areas in rats that had received gamma-2 ASO, compared with rats treated with a sense oligodeoxynucleotide. Analysis of these results by ANOVA revealed a significant difference among areas (F = 69.1, df = 14, P < .0001) and a significant ASO treatment effect (F = 34.2, df = 1, P < .0001) but no significant interaction between brain region and treatment (F = 0.9, df = 14). The post-hoc analysis showed a significant decrease of [³H]flunitrazepam binding in several brain regions, including frontal cortex, cerebellar molecular layer, zona reticulata of substantia nigra and CA3 of hippocampus.

View larger version (91K): [in this window] [in a new window]   Fig. 1.   Sample autoradiographs of radioligand binding to rat brain tissue after 3-day i.c.v. sense (control) oligodeoxynucleotide treatment. Results are shown for (A) 5 nM [3H]flunitrazepam binding, (B) 5 nM [3H]muscimol binding, (C) 50 nM [3H]GABA binding in the presence of 100 µM baclofen, (D) 3 nM [3H]EBOB binding and (E) 0.8 nM [3H]DPCPX binding in the presence of adenosine deaminase. The regional distribution of binding density for each radioligand was similar to that expected from published studies (see text for references).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Specific binding of 5 nM [3H]flunitrazepam to rat brain tissue after i.c.v. infusion of gamma-2 ASO or sense oligodeoxynucleotide for 3 days. The bars represent the mean ± S.E. (n = 4-5). ANOVA revealed a significant difference among areas and a significant ASO treatment effect but no significant interaction. *P < .05 vs. sense oligodeoxynucleotide. CTX-FR, frontal cortex; CTX-OCC, occipital cortex; SNR, zona reticulata of substantia nigra; SC, superior colliculus; IC, inferior colliculus; thal, thalamus; CPu, caudate-putamen; CA1 SO, hippocampal CA1 region, stratum oriens; CA1 SR, CA1 stratum radiatum; CA3 SO, CA3 stratum oriens; CA3 SL, CA3 stratum lucidum; DG MOL, molecular layer of dentate gyrus; CB-GRA, granular layer of cerebellum; CB-MOL, molecular layer of cerebellum.

[³H]Muscimol binding. As reported by Olsen et al.(1990), the abundance of [³H]muscimol binding was the highest in cerebellar granule cell layer, whereas the molecular layer displayed much lower binding. Intermediate levels were seen in the thalamus and frontal cortex, with lower levels in hippocampus and substantia nigra (fig. 1B). Nonspecific binding in the presence of 100 µM GABA was at the level of film background. ASO treatment had no effect on [³H]muscimol binding (fig. 3a). Analysis of the results showed a significant difference among areas (F = 178.3, df = 13, P < .0001) but no significant ASO treatment effect (F = 0.1, df = 1) and no significant interaction between brain region and treatment (F = 0.01, df = 13).

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Specific binding of GABA recognition site ligands. a, Binding of 5 nM [3H]muscimol to rat brain tissue after i.c.v. infusion of gamma-2 ASO or sense oligodeoxynucleotide for 3 days. The bars represent the mean ± S.E. (n = 4-5). ANOVA revealed a significant difference among areas but no significant ASO treatment effect and no significant interaction. Abbreviations as in figure 2. b, Specific binding of 50 nM [3H]GABA, in the presence of 100 µM baclofen, after i.c.v. infusion of gamma-2 ASO or sense oligodeoxynucleotide for 3 days. The bars represent the mean ± S.E. (n = 4-5). ANOVA revealed a significant difference among areas and a significant ASO treatment effect but no significant interaction. *P < .05 vs. sense oligodeoxynucleotide. Abbreviations as in figure 2.

[³H]GABA binding. As seen in a sample autoradiogram (fig. 1C) [³H]GABA binding, in the presence of baclofen, was greatest in the granule cell layer of the cerebellum, with a lower level in the molecular layer, in agreement with a previous study (Bristow and Martin, 1988). The location of GABA_A sites labeled with [³H]GABA correlated well with the [³H]muscimol binding. Nonspecific binding ranged from 10% to 45% of total [³H]GABA binding in those areas examined. There appeared to be a modest effect of ASO treatment on GABA binding (fig. 3b). Data analysis by ANOVA showed a significant difference among areas (F = 140.90, df = 13, P < .0001) and a significant ASO treatment effect (F = 6.26, df = 1, P < .02) but no significant interaction. The post-hoc analysis showed a significant change in only the cerebellar granule cell layer, where [³H]GABA binding was decreased 10%.

[³H]EBOB binding. The results showed that the regional distribution of [³H]EBOB binding was similar to that reported for TBOB (Olsen et al., 1990) and a previous [³H]EBOB study (Kume and Albin, 1994). Nonspecific binding was 3% to 6% of total binding at 3 nM [³H]EBOB. High levels of binding were found in cortex, inferior colliculus, superior colliculus and substantia nigra, whereas low levels of binding were found in cerebellar granule cell layer (fig. 1D). ASO treatment appeared to have the greatest effect on [³H]EBOB binding in hippocampal regions (fig. 4). There was a significant difference among brain areas (F = 45.80, df = 13, P < .0001) and a significant ASO treatment effect (F = 47.14, df = 1, P < .0001) but no significant interaction. The post-hoc analysis showed that a significant decrease of [³H]EBOB binding was limited to frontal cortex and all measured areas of the hippocampal formation, where the binding was reduced by 24% to 45%.

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Specific binding of 3 nM [3H]EBOB after i.c.v. infusion of gamma-2 ASO or sense oligodeoxynucleotide for 3 days. The bars represent the mean ± S.E. (n = 4-5). ANOVA revealed a significant difference among areas and a significant ASO treatment effect but no significant interaction. *P < .05 vs. sense oligodeoxynucleotide. Abbreviations as in figure 2.

[³H]DPCPX binding. As shown in figure 1E, [³H]DPCPX binding showed a wide distribution of A1 receptor in CNS. The distribution of [³H]DPCPX binding was in agreement with a previous study (Fastbom and Fredholm, 1990). High densities of binding were seen in hippocampus, molecular layer of cerebellum and thalamus, whereas moderate levels of binding were seen in other regions. Nonspecific binding of [³H]DPCPX could not be differentiated from background in any brain region. ASO treatment produced only a modest effect [³H]DPCPX binding (fig. 5). Data analysis by ANOVA showed a significant difference among areas (F = 92.6, df = 13,82, P < .0001) and a significant ASO treatment effect (F = 11.3, df = 1, P < .02) but no significant interaction. Despite the significant treatment effect, post-hoc analysis showed that no single area had a significant difference between ASO and sense oligodeoxynucleotide treatments (P  .15 in each area).

View larger version (26K): [in this window] [in a new window]   Fig. 5.   Specific 0.8 nM [3H]DPCPX binding after i.c.v. infusion of gamma-2 ASO or sense oligodeoxynucleotide for 3 days. The bars represent the mean ± S.E. (n = 4-5). ANOVA revealed a significant difference among areas and a significant ASO treatment effect but no significant interaction. Post-hoc analysis found no single region with a significant treatment effect. Abbreviations as in figure 2.

Immunoblot for GABA_A receptor beta-2/beta-3 subunit in cerebellum. Quantitative Western blot analysis was used to determine beta-2/beta-3 receptor subunit level after gamma-2 ASO treatment. The monoclonal antibody bd 17 recognized two different proteins, corresponding to the beta-2 and beta-3 subunits, with molecular weights of 56 and 58 kDa, respectively, in the cerebellar membrane preparation. These results were in agreement with a previous study in which this antibody was used to detect beta-2/beta-3 subunits in a brain membrane homogenate preparation (Mhatre and Ticku, 1994). The ASO treatment had no significant effect on the intensity of bands seen in the immunoblot (fig. 6).

View larger version (35K): [in this window] [in a new window]   Fig. 6.   Top, representative autoradiogram showing  subunit protein immunoreactivity in the cerebellum. The results shown in the autoradiogram depict immunoreactivity levels from sense (lanes 2, 4, 6, 8) and gamma-2 ASO (lanes 1, 3, 5, 7)-treated rats. Bottom, results of densitometry. The mean density of the four controls was set to 1.0 and used to compare each sense and ASO-treated sample (mean ± S.E., n = 4). There was no significant treatment effect (Student's t test).

	Discussion

Top Abstract Introduction Methods Results Discussion References

In this study, it was expected that the ASO treatment would reduce the availability of gamma-2 protein, which should interfere with normal assembly GABA_A benzodiazepine receptors. The ability of benzodiazepines to bind to a GABA_A receptor and to increase the GABA-dependent gating of the intrinsic Cl channel, requires a gamma subunit in the receptor, and the gamma-2 subunit has been most closely associated with typical benzodiazepine actions (Pritchett et al., 1989a; Macdonald and Olsen, 1994). Günther et al. (1995) also addressed the role of the gamma-2 subunit in GABA_A receptor function by producing mutant mice that lacked this subunit. The neonatal mice expressed GABA_A receptors lacking gamma-2 subunits, with almost complete absence of [³H]flumazenil binding (benzodiazepine recognition sites), but a much smaller reduction in [³H]SR-95531 binding (GABA recognition sites). These mice failed to thrive, displayed abnormalities in sensorimotor behavior and survived only a short time, indicating the likelihood of physiological changes in GABA_A receptors resulting from loss of the gamma-2 subunit.

The ASO treatment produced a widespread decrease in binding of the benzodiazepine ligand [³H]flunitrazepam. This was consistent with our previous observation of a significant increase in convulsive threshold for -CCM, a benzodiazepine "inverse agonist," after the same 3-day i.c.v. treatment. The 10% to 25% decrease in [³H]flunitrazepam binding was similar to the results reported by Karle and Nielsen (1995), who noted a 9% to 15% reduction in benzodiazepine binding in cerebral cortex subsequent to a 2-day treatment with i.c.v. infusions of a gamma-2 ASO that differed somewhat from the one used in the present study.

It had been hypothesized that the gamma-2 ASO treatment might result in the selective loss of benzodiazepine binding sites without changing the other binding sites present on GABA_A receptors. Indeed, the GABA recognition site was largely unaffected by the ASO treatment. [³H]Muscimol binding showed no change in any brain region, including cerebellar granule cell layer. This region has the highest level of [³H]muscimol binding and demonstrated a 25% decrease of [³H]flunitrazepam binding after 3-day ASO i.c.v. infusion. [³H]GABA binding was only decreased 10% in this area, with no significant change in other regions, including those that had shown diminished benzodiazepine binding. The effects of gamma-2 ASO treatment suggested that even though GABA_A receptors had been affected, as indicated by the reduced availability of benzodiazepine recognition sites ([³H]flunitrazepam binding), there was little effect on the availability of GABA recognition sites ([³H]muscimol and [³H]GABA binding). These findings suggest that gamma-2-deficient receptors, of unknown subunit composition, were being expressed in ASO-treated rats. Under the experimental conditions used for the autoradiographic binding, [³H]muscimol and [³H]GABA show dramatic differences in relative binding among brain regions (fig. 3; Olsen et al., 1990). In recent studies, it was shown that high affinity [³H]muscimol binding was dramatically reduced in the cerebellar granule cell layer of alpha-6-null/delta-deficient mice (Jones et al., 1997; Mäkelä et al., 1997), indicating that receptors containing these subunits are responsible for much of the high affinity [³H]muscimol binding, particularly in cerebellar granule cells. Moreover, delta subunit protein and mRNA disribution closely approximate the distribution of [³H]muscimol binding (Jones et al., 1997) and, in forebrain, alpha-4 and delta subunit distributions are similar (Wisden et al., 1992; Jones et al., 1997). Such findings indicate that [³H]muscimol especially labels a population of GABA_A receptors containing alpha-4 delta or alpha-6 delta. Thus, the lack of gamma-2 ASO effect on [³H]muscimol binding could also suggest that these receptors do not include a gamma-2 subunit.

It is known that GABA_A receptors can be expressed from only alpha and beta subunits and that such receptors will respond to GABA in a picrotoxin- and bicuculline-sensitive fashion but will be insensitive to benzodiazepine (Verdoorn et al., 1990; Angelotti and Macdonald, 1993). However, the presence or absence of a gamma subunit does affect the GABA response of such receptors (Verdoorn et al., 1990), so that receptors affected by ASO treatment may have had an altered GABA response, resulting in decreased GABA-mediated inhibitory transmission. This was thought to be the basis for some behavioral changes in ASO- but not sense oligonucleotide-treated rats (Zhao et al., 1996). Reduced GABA neurotransmission might also eventually result in cell damage or even loss of neurons. In their initial report, Karle and Nielsen (1995) noted only a modest decrease in [³H]flunitrazepam binding and no loss of [³H]muscimol binding in cerebral cortex after the 2-day ASO treatment. However, when this treatment was continued for up to 6 days, there was a greater decrease in [³H]flunitrazepam binding (21%) and a 12% decrease in [³H]muscimol binding in cortex (Karle et al., 1997a). The decreased GABA binding appeared to be at least partially dependent on the duration of gamma-2 ASO treatment and may be a secondary result of the gamma-2 ASO treatment. Subsequent work by the same group, using intrahippocampal ASO infusion, suggests this to be the case (Karle et al., 1997b). In our own study, behavioral effects of the ASO (but not sense) treatment, which includes serious weight loss (Zhao et al., 1996), limited the treatment to 3 days. The reasons for this are not clear but likely involve less effective GABA function in GABA_A receptors that are deficient in the gamma-2 subunit (Verdoorn et al., 1990), as discussed above.

No change was found in beta subunit immunoreactivity in the cerebellum after gamma-2 ASO treatment. As the GABA recognition site is thought to be associated with beta subunits (Sigel and Buhr, 1997), this was in keeping with the minimal effects of ASO treatment on [³H]GABA and [³H]muscimol binding. In the cerebellum, beta-2 is the most abundant isoform, followed by the beta-3 subunit, whereas beta-1 is present in a very small proportion of cerebellar GABA_A receptors (Li and de Blas, 1997). The lack of change in the beta subunit immunoreactivity also indicated that the reduction in the gamma-2 subunit was not compensated by altering the levels of the beta subunits and adds further support for the idea that the gamma-2 ASO treatment affected neither the subunits involved in the GABA binding site nor the number of GABA_A receptor complexes.

The apparent effect of treatment on the chloride channel of the GABA_A receptor complex, as expressed by [³H]EBOB binding, differed greatly among brain regions. Although several areas, such as substantia nigra pars reticulata, showed small, statistically insignificant decreases, there was no significant effect of ASO treatment here, or in the cerebellum, where there was a 25% decrease in [³H]flunitrazepam binding. However, all areas of the hippocampal formation showed large decreases in [³H]EBOB binding. In contrast to other regions of the rat brain, the cut surface of the hippocampal tissue of gamma-2 ASO-treated rats appeared edematous and discolored, with a less distinct laminar structure. A similar change in hippocampus after local gamma-2 ASO injection was observed by Karle et al. (1995). In that study, the intrahippocampal ASO treatment also resulted in a 51% decrease in [³⁵S]TBPS binding, and, in contrast to the present study, significant decreases in hippocampal [³H]muscimol and [³H]quinuclidinyl benzilate binding (for the muscarinic acetylcholine receptor), and a decrease in protein content, suggesting loss of hippocampal cells (Karle et al., 1995). In the present study, it is possible that the greater loss of [³H]EBOB binding in the hippocampus may be related to the proximity of this region to the site of ASO infusion, resulting in a higher local ASO concentration. Other possibilities include the sensitivity of hippocampal structures to the loss of inhibitory function, resulting in increased excitatory activity and subsequent neuronal damage. Such damage may be associated with release of endogenous substances, such as fatty acids, which have been shown to inhibit the binding of a similar ligand (TBPS) to the chloride channel portion of the GABA_A receptor (Koenig and Martin, 1992). To further evaluate possible nonspecific effects of the gamma-2 ASO treatment, [³H]DPCPX binding was used as an indicator of possible changes in brain tissue secondary to decreased GABA-ergic neurotransmission. There was a small but significant ASO treatment effect on [³H]DPCPX binding. However, no single area showed a significant change after 3-day gamma-2 ASO treatment. This modest change indicated that the 3-day ASO treatment did cause moderate secondary effects, especially in hippocampus. Because there was no loss of GABA, muscimol or DPCPX binding in hippocampus, it is unlikely that the ASO treatment had caused a loss of neurons, and any secondary changes that might have eventually resulted in neuronal loss with further treatment were still in a reversible stage. This is in accord with the recent findings by Karle et al. (1997b).

In summary, the gamma-2 ASO treatment reduced benzodiazepine binding with little effect on GABA binding. The findings support the hypothesis that, at least up to 3 days, the ASO treatment described can selectively alter the expression of gamma-2 subunits and thereby alter the composition and function of the GABA_A receptor. Reducing the availability of one GABA_A receptor subunit might be expected, through feedback mechanisms, to alter the transcription and/or translation of that subunit, as well as for other subunits. It is anticipated that the pattern of that response may provide insight into the regulation of GABA_A receptor subunit expression and assembly in adult central neurons.