gamma-Aminobutyric acid and benzodiazepine receptors in the kindling model of epilepsy: a quantitative radiohistochemical study

Quantitative radiohistochemistry was utilized to study alterations of gamma-aminobutyric acid (GABA) and benzodiazepine receptors in the kindling model of epilepsy. The radioligands used for GABA and benzodiazepine receptors were [3H] muscimol and [3H]flunitrazepam, respectively. GABA receptor binding was increased by 22% in fascia dentata of the hippocampal formation but not in neocortex or substantia nigra of kindled rats. Within fascia dentata, GABA receptor binding was increased to an equivalent extent in stratum granulosum and throughout stratum moleculare; no increase was found in dentate hilus or stratum lacunosummoleculare or stratum radiatum of CA1. The increased binding was present at 24 hr but not at 28 days after the last kindled seizure. The direction, anatomic distribution, and time course of the increased GABA receptor binding were paralleled by increased benzodiazepine receptor binding. Unexpectedly, GABA receptor-mediated enhancement of benzodiazepine receptor binding was slightly attenuated in fascia dentata of kindled compared to control rats. The anatomic distribution of the increased GABA receptor binding is consistent with a localization to somata and dendritic trees of dentate granule cells. We suggest that increased GABA and benzodiazepine receptor binding may contribute to enhanced inhibition of dentate granule cells demonstrated electrophysiologically in kindled animals.

GABA receptor binding was increased by 22% in fascia dentata of the hippocampal formation but not in neocortex or substantia nigra of kindled rats. Within fascia dentata, GABA receptor binding was increased to an equivalent extent in stratum granulosum and throughout stratum moleculare; no increase was found in dentate hilus or stratum lacunosummoleculare or stratum radiatum of CAl. The increased binding was present at 24 hr but not at 28 days after the last kindled seizure.
Previous work from our laboratory demonstrated an increase in benzodiazepine receptor binding in the hippocampal membranes from kindled animals (McNamara et al., 1980). This increase was further localized to the dentate granule cells (Valdes et al., 1982). Therefore, we hypothesized that benzodiazepine receptor alterations would be paralleled by GABA receptor alterations in kindled animals.
To test this hypothesis, we utilized quantitative radiohistochemical methods and examined benzodiazepine and GABA receptor binding in sections obtained from the same animals. We compared benzodiazepine and GABA receptor binding with regard to brain region specificity, time course, and cellular localization. The direction, anatomic distribution, and time course of the increased GABA receptor binding were paralleled by increased benzodiazepine receptor binding. Unexpectedly, GABA receptor-mediated enhancement of benzodiazepine receptor binding was slightly attenuated in fascia dentata of kindled compared to control rats.

Materials and Methods
The anatomic distribution of the increased GABA receptor binding is consistent with a localization to somata and dendritic trees of dentate granule cells. We suggest that increased GABA and benzodiazepine receptor binding may contribute to enhanced inhibition of dentate granule cells demonstrated electrophysiologically in kindled animals.
Preparation of animals for radiohistochemistry. Adult male Sprague-Dawley rats (Charles River Breeding Laboratories), weighing between 300 and 425 gm, were stereotaxically implanted with a bipolar electrode in the right amygdala or right entorhinal cortex under pentobarbital anesthesia as previously described (McNamara, 1978). After a postoperative recovery period of at least 7 days, electrical stimulations were administered at daily intervals and kindling developed as previously reported (McNamara et al., 1980). Unless specified otherwise, experimental animals exhibited three kindled seizures (class 4 or 5 (Racine, 1972)), the clonic motor component of which was at least 20 set in duration. These animals will be referred to as "kindled." These animals were sacrificed 24 hr or 28 days after the last kindled seizure. Control animals underwent electrode implantation but were not stimulated.
Kindling, an animal model of temporal lobe epilepsy, is a phenomenon in which repeated subconvulsive electrical stimulation results in progressively more intense seizures and eventually generalized motor convulsions (Goddard et al., 1969). Once established, the kindling effect is permanent.
Benzodiazepines are potent anticonvulsants which suppress kindled seizures and retard kindling development (for review, see McNamara et al., 1985). These anticonvulsant effects are presumably mediated at least in part through benzodiazepine receptors which are coupled to y-aminobutyric acid (GABA) receptors in a macromolecular protein complex (Costa et al., 1979;Gavish and Snyder, 1981;Sigel and Barnard, 1984). GABA neurofransmission is enhanced by benzodiazepines (Choi et al., 1977). GABA agonist A paired control was sacrificed with each experimental animal. Brains were rapidly removed and frozen by slow immersion into isopentane chilled in a methanol-dry ice bath. Coronal frozen sections (10 pm) were cut in a cryostat microtome (Harris Mfg. Co.) and thaw-mounted on subbed glass slides. Slide-mounted sections were stored at -70°C until incubation. Electrode placements were verified by methyl green pyronine Y stain of frozen coronal sections.
Upon removal from the freezer, slide mounted sections were allowed to rehydrate for 15 min at room temperature and were then fixed in 0.1% formalln in 50 mM Tris-citrate buffer (pH 7.0) for 10 min at 23°C. To promote removal of endogenous GABA, the sections were subsequently preincubated in 50 mM Tris-citrate buffer at 4°C for 15 min. the standard incubation was carried out for 30 min at 4°C in a solution containing 10 nM [3H]muscimol and 50 mM Tris-citrate ( Fig. 1 Tritium-sensitive film (LKB Ultrofilm) was apposed to the dried sections along with tntrum standards in x-ray cassettes for 10 days to 6 weeks at 4'C. Film was developed in Kodak D-19, fixed in Kodak Rapid FIX at 18°C and mounted on clean glass slides for densrtometnc measurements.
Sections were again apposed t&emulsion-coated coverslips (Kodak NTB2 emulsion, 1:l dilution) and exposed at 4'C for 4 to 6 weeks and processed in Kodak D-19 and regular fixer at 18'C for darkfield mrcroscopy and silver grain counting.
Quantrtatron of the optical density measurements was accomplished as described previously from thus laboratory (Savage et al., 1985) with minor modrfrcatrons.
[3H]Ornrthrne was used for tntrum standards. Radioactivity of the standards was expressed as counts per minute per milligram of protein. The standard curve was linear over the range of optical density measurements made in this study.
B/ochem/ca/ vahdatlon of rad/ohistochem/ca/ methods. Experiments were performed to demonstrate that binding to slide-mounted sections was equrvalent to brndrng to brain membranes.
Frozen sections (10 pm) were obtained from naive male Sprague-Dawley rats. Slide-mounted sections were processed as described for the radrohrstochemrstry above untrl completron of the postincubation rinses. At thus point the wet sections were wiped onto Whatman GF/B filters and the radioactivity of the filters was determined with a liquid scrntrllatton counter. Valrdatron

Results
GABA receptor binding.
[3H]Muscimol binding was increased by 22% (p < 0.001) in the fascia dentata of kindled animals sacrificed 24 hr after the last kindled seizure (Table I) The section representing nonspecific binding (not shown) IS not readily differentiated from the background of the film. Densrtometrrc data presented In Tables I and II were taken from stratum moleculare of the infrapyramrdal blade of fascia dentata (arrows). Structures identified include hrppocampal areas CA3 and CA1 , subrculum (S), and thalamus (7). Magnification X 40.  in the CA1 area of hippocampal. formation, lamina IV of left neocortex, or pars reticulata of substantia nigra (data not shown). The increase in binding was transient, since no significant increases were detectable at 28 days after the last seizure (Table I).
To determine whether there was an increase in the number of [3H]muscimol-binding sites or an increase in affinity, binding isotherms were performed with radiohistochemical methods in two pairs of kindled and control animals. An increase of [3H]muscimol binding was present in the fascia dentata of the kindled animals at all concentrations used (data not shown). The limited number of data points (eight concentrations between 0.5 and 100 nM) did not permit adequate separation of high and low affinity sites, thereby precluding definitive statements regarding a selective alteration of number or affinity of either of the putative sites. In addition, the possibility that the increased [3H]muscimol binding reflects increased availability of GABA receptor-binding sites due to a decrease in bound endogenous GABA in kindled animals cannot be excluded by the present experiments.
To localize more precisely the increased [3H]muscimol binding within the fascia dentata, a detailed analysis was performed of the distribution of silver grains developed with the emulsion-coated coverslip method. The laminar organization of the principal neurons and their afferents in the fasica dentata permit correlation of the distribution of the silver grains with the distribution of known cellular elements. These analyses disclosed similar percentage increases of silver grain counts over stratum granulosum and the inner, middle, and outer thirds of stratum moleculare (Fig. 3). The increase of silver grain density stopped abruptly at the hippocampal fissure and below stratum granulosum as evidenced by the absence of significant change in subregions of CA1 and in the hilus of fascia dentata. The distribution of the increased number of silver grains coincides with the somata and dendritic trees of the dentate granule cells, a distribution identical to that previously described with benzodiazepine receptors (Valdes et al., 1982).
To further explore the molecular properties of the seizure-induced increase of benzodiazepine receptors, we determined whether the increase selectively involved type 1 or type 2 benzodiazepine receptors (Lippa et al., 1979). To address this question, a relatively selective type 1 benzodiazepine ligand, a triazolopyridazine, CL21 8,872, was utilized. Based on the results of the binding studies (Table II), utilizing the formula described under "Materials and Methods," the proportions of type 1 benzodiazepine receptor were calculated. In the control animals, only 3% of all benzodiazepine receptors in the fascia dentata were type 1, whereas 52% in the substantia nigra pars reticulata were type 1, These values are in agreement with others' results regarding regional distribution of these subtypes of benzodiazepine receptors (Young et al., 1981;Unnerstall et al., 1982;Lo et al., 1983). Importantly, percentages remained unchanged with kindling. Since fascia dentata contains almost exclusively type 2 receptors, the increase in [3H]FLU binding was mostly due to an increase of type 2 receptors, although a concomitant increase of type 1 receptors cannot be ruled out. In the substantia nigra, proportions of type 1 and 2 receptors remained unaltered with kindling.
GABA and benzodiazepine receptor coupling.
The parallel increases of [3H]muscimol and [3H]FLU binding and the similar time course and cellular localization of these alterations raised the possibility that the coupling of GABA and benzodiazepine receptors was also increased in the kindled animals. To examine this possibility, the enhancement of [3H]FLU binding by muscimol was investigated in the fascia dentata. The pars reticulata of substantia nigra was examined as a control, since no kindling-related changes in [3H]FLU or [3H]muscimol were found in this region (Table II).
Contrary to our expectations, no increase in coupling was found. Indeed, in the fascia dentata, the muscimol enhancement of [3H] FLU binding in kindled animals (45%) was less than that in control (54%) a difference which actually reached statistical significance (p c 0.05, Student's two-tailed t test). In the substantia nigra, the enhancement was slightly more in kindled (57%) than in control animals (51%) (p > 0.1, Student's two-tailed t test). No subtypespecific alterations in receptor coupling could be found with the addition of CL21 8,872.
In fact, in substantia nigra, in the presence of CL218,872, musci-

Discussion
The principal findings of this study are: (I) GABA receptor binding is increased in the fascia dentata but not in several other areas of the kindled rat brain; (2) the anatomic distribution of the increased GABA receptor binding in fascia dentata is consistent with a localization to the somata and dendritic trees of the granule cells; (3) the direction, anatomic distribution, and time course of the increased GABA receptor binding are paralleled by increases of benzodiazepine receptor binding; (4) despite the increased GABA receptor binding, GABA receptor-mediated enhancement of benzodiazepine receptor binding is slightly decreased in the fascia dentata of kindled rats; and (5) the kindled seizure-induced increase of benzodiazepine receptor binding in fascia dentata is due mainly to an increase in type 2 receptors; no change in the percentage of subtypes was detected in substantia nigra.
The distribution of this seizure-induced increase in GABA receptor binding in stratum granulosum and throughout stratum moleculare supports a localization to the somata and dendritic trees of the dentate granule cells. This pattern does not coincide with the distribution of either of the principal granule cell afferents, namely, the perforant path (outer two-thirds of stratum moleculare) or commissural/associational (inner one-third of stratum moleculare) fibers (Blackstad, 1956;Gottlieb and Cowan, 1973;Steward, 1976;Wyss, 1981). Likewise, this pattern does not conform to the distribution of blood vessels in fascia dentata. Localization to glia seems unlikely because of the distinct laminar distribution of the increase together with the absence of NaYndependent GABA receptor binding in primary glial cultures (Ossola et al., 1980). Localization to interneurons seems less likely, since the number of granule cells vastly exceeds the number of interneurons. The distribution of the seizureinduced increase in GABA receptor binding is identical to the seizureinduced increases of benzodiazepine receptor binding which have also been tentatively localized to the dentate granule cells (Valdes et al., 1982).
The striking parallel in the seizure-induced increases in GABA and benzodiazepine receptor binding is not surprising in view of the extensive evidence linking these two sites to the same macromolecular complex. Paradoxically, the increased GABA receptor binding is not paralleled by increased responsiveness of the benzodiazepine receptors to exogenous GABA; indeed the kindled animals exhibited a slight decrease in magnitude of GABA enhancement of benzodiazepine binding compared to controls. This result confirms our previous finding in well washed membranes prepared from microdissected fascia dentata of kindled animals (Fanelli and McNamara, 1983). Essential to the understanding of this apparent conflict is that radioligand binding with [3H]muscimol interacts with a high affinity GABA receptor, whereas GABA regulation of benzodiazepine receptor binding reflects a low affinity GABA receptor. One possible explanation is that the increased [3H]muscimol binding represents desensitized GABA receptors which are functionally uncoupled. Alternatively, the seizure-induced increase of benzodiazepine receptors may be in a conformational state reflecting the highest affinity for benzodiazepines, and added GABA cannot increase this affinity further.
The tentative localization of seizure-induced increases of GABA and benzodiazepine receptor binding to dentate granule cells raises the possibility that these molecular alterations may underlie alterations of granule cell physiology. Indeed the granule cells receive both feed-forward and recurrent inhibition which appear to be mediated by GABA (Buzsaki and Czeh, 1981;Douglas et al., 1983;Tuff et al., 1983;Fricke and Prince, 1984). To test this idea, G. L. King and FT. Dingledine, in collaboration with this laboratory, employed field potential recordings to examine the responses of granule cells to orthodromic and antidromic activation in isolated hippocampal slices prepared from kindled animals 24 hr after the last seizure (King et al., 1985). Increases were apparent in tests of both feedforward and recurrent inhibition of granule cells in kindled animals as compared to controls. The increased inhibition was almost certainly synaptic and GABA mediated. Likewise, Tuff et al. (1983) found enhanced paired-pulse inhibition of granule cells in vivo following kindling. Liebowitz et al. (1977) have reported increased GABA release from hippocampal slices from kindled animals; it is not clear whether this reflects synaptic release and whether it is from fascia dentata and/or hippocampus proper. Nevertheless, it seems reasonable to speculate that the increased release of GABA, together with the increase of GABA and benzodiazepine receptor binding, may contribute, at least in part, to the enhanced inhibition of dentate granule cells demonstrated electrophysiologically.
A final paradox deserving of comment is that the direction of the biochemical changes together with the electrophysiology point to an increase of inhibition of granule cells in the kindled animals. This would certainly argue against the idea that these biochemical changes constitute the molecular basis of kindling; indeed the transient nature of the GABA receptor increase is more consistent with a response of the brain to repeated seizures. Lesion data support the idea that the hippocampal formation plays a facilitatory role in kindling development (Dasheiff and McNamara, 1982;Savage et al., 1985). The granule cells of fascia dentata, as a gateway and rate-limiting step to the flow of information through the hippocampal trisynaptic circuit (Winson and Abzug, 1978) would then be an appropriate site of endogenous modulation. We propose that the increases in GABA and benzodiazepine receptors represent part of a carefully orchestrated response of the brain to repeated seizures, designed to stabilize granule cell excitability and reduce the likelihood of hippocampal activation and subsequent seizures.