Epilepsy in mice deficient in the 65-kDa isoform of glutamic acid decarboxylase

Abstract

γ-Aminobutyric acid (GABA), the major inhibitory neurotransmitter in the mammalian brain, is synthesized by two glutamate decarboxylase isoforms, GAD65 and GAD67. The separate role of the two isoforms is unknown, but differences in saturation with cofactor and subcellular localization suggest that GAD65 may provide reserve pools of GABA for regulation of inhibitory neurotransmission. We have disrupted the gene encoding GAD65 and backcrossed the mutation into the C57BL/6 strain of mice. In contrast to GAD67−/− animals, which are born with developmental abnormalities and die shortly after birth, GAD65−/− mice appear normal at birth. Basal GABA levels and holo-GAD activity are normal, but the pyridoxal 5′ phosphate-inducible apo-enzyme reservoir is significantly decreased. GAD65−/− mice develop spontaneous seizures that result in increased mortality. Seizures can be precipitated by fear or mild stress. Seizure susceptibility is dramatically increased in GAD65−/− mice backcrossed into a second genetic background, the nonobese diabetic (NOD/LtJ) strain of mice enabling electroencephalogram analysis of the seizures. The generally higher basal brain GABA levels in this backcross are significantly decreased by the GAD65−/− mutation, suggesting that the relative contribution of GABA synthesized by GAD65 to total brain GABA levels is genetically determined. Seizure-associated c-fos-like immunoreactivity reveals the involvement of limbic regions of the brain. These data suggest that GABA synthesized by GAD65 is important in the dynamic regulation of neural network excitability, implicate at least one modifier locus in the NOD/LtJ strain, and present GAD65−/− animals as a model of epilepsy involving GABA-ergic pathways.

The two isoforms of glutamic acid decarboxylase, GAD65 and GAD67, are coexpressed in γ-aminobutyric acid (GABA)-ergic neurons (1, 2). Whereas GAD67 is a cytosolic enzyme and distributed throughout the cell body, GAD65 is preferentially located in nerve termini, and is reversibly anchored to the membrane of synaptic vesicles (3–6). At least 50% of GAD in brain is present as the pyridoxal 5′ phosphate (PLP)-free apo-enzyme (7, 8); GAD65 has been suggested to constitute the majority of this enzyme reservoir (3, 8), which can be activated by influx of cofactor. In contrast to GAD65, the majority of GAD67 has been reported to be saturated with PLP (1, 3, 8). The functional implication of these differences for GABA-ergic mechanisms has remained elusive.

The major sites of GAD expression outside the central nervous system are ovaries, testis, and the insulin-producing β cells in the islets of Langerhans (9–11), but the function of GABA in these tissues is unknown. Mouse β cells and testis predominantly express GAD67, whereas GAD65 is the major GAD form in mouse ovaries (12–14).

It has been suggested that GAD67 provides basal levels of GABA, whereas GAD65 may supply GABA in situations of a sudden demand (15). GAD67-deficient animals are born with a cleft palate and die within the first day of life, apparently from a respiratory failure (16, 40). GABA levels are decreased approximately 15-fold (16). To study the role of GABA generated by GAD65, we have generated GAD65-deficient mice by gene targeting. GAD65-deficient animals appear normal with regard to GABA levels, behavior, locomotion, reproduction, and glucose homeostasis. GAD65-deficient mice generated in a separate study were reported to have an increased sensitivity to chemical convulsants but did not have epilepsy (17). Here we show that GAD65−/− mice develop epilepsy that involves limbic regions of the brain, results in an increased mortality, and is dramatically influenced by the genetic background of the mice.

MATERIALS AND METHODS

Generation of the Targeting Construct.

A 7.5-kb DNA fragment containing exons 1–3 of the GAD65 gene was isolated from a 129/SvJ genomic DNA library by using a cDNA probe containing exons 3–6 (a kind gift of Roland Tisch, University of North Carolina at Chapel Hill) and used to construct the pmGAD65 targeting vector. A cassette containing the neomycin resistance gene under the control of the phosphoglycerol kinase promoter was introduced into the BamHI site in exon 1 just downstream of the transcription and translation start sites. 129/SvJ-derived JM1 embryonic stem (ES) cells (a kind gift of R. Pedersen, University of California at San Francisco) were electroporated with linearized pmGAD65, and properly targeted G418-resistant ES cell clones were obtained at an overall frequency of 1 in 35. Chimeric animals produced by injection of these cells into C57BL/6 blastocysts were bred with C57BL/6 mice, and germ-line transmission of the mutation was achieved. Experimental knock-out and control mice on a 129/SvJ × C57BL/6 genetic background were derived by intercrossing mice from the N₃-N₄ backcross generation. Experimental knock-out and control mice on a 129/SvJ × NOD/LtJ genetic background were similarly derived from the N₄ backcross generation into the nonobese diabetic (NOD/LtJ) strain of mice (18). These mice are referred to in the text as {129 × B6} and {129 × NOD}, respectively. At this backcross generation, the {129 × NOD} animals do not possess a full complement of diabetes-susceptibility loci, and blood glucose levels were within the normal range at the ages used. Wild-type and mutant littermates were used for each analysis. For Southern blot analysis, 5–10 μg DNA from ES cells or tail biopsies was digested with EcoRI and probed with a 0.5-kb genomic DNA probe, which was external to the targeting construct, by using standard procedures (19) and a neomycin probe to ensure that there was only one copy of the construct in ES cells (data not shown).

Western Blot Analysis.

Brain extracts were prepared in the presence of cofactor as described previously (15) with slight modification. Protein levels were quantitated (Pierce, bicinchoninic acid protein assay reagent kit) and equal amounts were loaded in each lane. The primary antibodies used were as follows: the GAD65-specific antibody is a collection of mouse monoclonal antibodies that recognize the N-terminal and middle portion of GAD65, and the GAD65/67 antibody is a rabbit antibody made to the C-terminal 19 aa of GAD65 (20). Reactivity was detected with an alkaline-phosphatase-conjugated secondary antibody.

GAD Activity.

Mice were deeply anesthetized, and brains were quickly dissected out and homogenized in ice-cold buffer (15) with or without 0.1 M PLP (Sigma). Homogenates were cleared at 55,000 × g, and GAD enzyme activity was measured in supernatants as described previously (15). Protein levels were determined with the bicinchoninic acid protein assay (Pierce). GAD activity was determined in brains of wild-type (n = 4) and homozygous (n = 4) mutant age-matched littermates by using equal amounts of extract prepared in the presence or absence of 0.1 M PLP.

GABA Analyses.

Mice were deeply anesthetized, and brains were quickly dissected out. Samples from different areas of the brain were sonicated in 0.1 M perchloric acid and centrifuged at 27,000 × g for 30 min. Protein levels were determined with the bicinchoninic acid protein assay (Pierce) before centrifugation. GABA content was determined in the supernatants in triplicate by high-performance liquid chromatography/electrochemical detection by using standard methods (21).

Stress-Induced Seizures.

Wild-type and homozygous mutant animals were loosely held in a Plexiglas restrainer for 30 sec at an elevation of 1.5 feet above the table with the experimenter blind to genotype. Animals were observed for the ensuing 10–15 min for seizure activity.

EEG Recordings.

Silver-wire electrodes (0.005“ diameter) soldered to a microminiature connector were implanted bilaterally into the subdural space over frontal and parietal cortex of anesthetized GAD65−/− {129 × NOD} mice at least 24 hr before the experiment. Cortical activity was recorded by using a digital electroencephalograph. Mice moving freely in the test cage were placed on an elevated platform to induce a seizure.

c-fos Immunohistochemistry.

{129 × B6} GAD65−/− animals were loosely held in an elevated restrainer for 30 sec. Two hours later, mice were anesthetized and perfused intracardially with a freshly prepared fixative solution. Immunostaining was performed by using standard protocols (22) on free-floating sections (40-μm thickness) with an anti-c-fos primary antibody (1:15,000, Oncogene Sciences) and detected with an avidin-biotinylated peroxidase complex (Vectastain Elite ABC, Vector Laboratories). Adjacent sections were stained with cresyl violet for comparative localization between animals.

RESULTS

Generation of GAD65−/− Mice.

To study the role of GAD65, the gene for this isoform was disrupted by insertion of a neomycin-resistance cassette into exon 1 (23) downstream of the transcription and translation start sites (Fig. 1a). Mice heterozygous for the targeted allele were backcrossed separately three to four generations into the C57BL/6 and NOD/LtJ backgrounds (18). Heterozygotes were then intercrossed to create animals of all possible GAD65 genotypes (Fig. 1b). Results below refer to {129 × B6} animals, except where the {129 × NOD} background is specifically mentioned. The distribution of gender and genotypes in crosses between heterozygotes approximated the expected Mendelian frequency. Thus, in contrast to GAD67−/− mice (16, 40), GAD65−/− animals are viable past weaning. Western blot analysis of brain extracts revealed approximately wild-type levels of GAD65 protein in heterozygous animals and no detectable protein in the homozygous mutant animals (GAD65−/−; Fig. 1c). GAD67 levels were similar in wild-type and mutant mice. Similar results were obtained for GAD65−/− mice on the {129 × NOD} background. Thus, neither GAD65−/− mice nor GAD67−/− animals (16, 40) show a compensatory expression of the other isoform.

Figure 1.

Targeted disruption of the GAD65 gene in ES cells and mice. (a) Schematic of the wild-type locus, the targeting construct pmGAD65, and the expected targeted allele. Solid and open boxes represent coding and untranslated exons, respectively. Arrowed lines indicate the EcoRI restriction fragment sizes for the wild-type and targeted alleles. Restriction sites are: E, EcoRI; X, XbaI; Xh, XhoI; B, BamHI. (b) Genomic Southern blot analyses of ES cell DNA and tail biopsies. Genomic DNA from the parental and targeted ES cells along with genomic DNA from offspring obtained after breeding of heterozygote mice was digested with EcoRI and probed with a 5′ 500-bp genomic fragment (thick, black line). Wild-type and targeted restriction fragments are 6 kb and 4 kb, respectively. (c) Western blot analysis of GAD expression in GAD65+/+, GAD65+/−, and GAD65−/− brain extracts with an antibody recognizing GAD65 (1) or GAD65 and GAD67 (2).

Loss of GAD65 Affects GAD apo-Enzyme but Not Holo-Enzyme Levels.

GAD activity in brain extracts was measured in the presence and absence of PLP to estimate apo- vs. holo-enzyme levels. In the absence of exogenous PLP, similar GAD activity levels were measured in wild-type and mutant brains, demonstrating that brain GAD holo-enzyme activity remains the same in the absence of GAD65 (Fig. 2). In contrast, brain extracts prepared and analyzed in the presence of cofactor revealed a 50% reduction in total GAD activity in GAD65-deficient mice (P < 0.05, Student’s t test). Addition of PLP resulted in a 3-fold increase in GAD activity in GAD65−/− mice, compared with a 10-fold increase in wild-type mice. Similar results were obtained for GAD65−/− mice on the {129 × NOD} background. Thus, although GAD65 provides the majority of PLP-inducible enzyme activity, the mutant mice reveal an unexpected ability of GAD67 to contribute to the apo-GAD reservoir in the brain.

Figure 2.

Loss of GAD65 affects apo- but not holo-GAD enzyme activity. GAD enzyme activity was measured in brain extracts of GAD65+/+ (n = 4, solid bars) and GAD65−/− (n = 4, open bars) mice in the presence or absence of PLP. Data were converted to pmol CO₂ evolved per min/μg protein and expressed as percent of wild-type levels in the presence of PLP (2.69 ± 0.23 pmol per min/μg; mean ± SEM).

Normal Development, Locomotion, Reproduction, and Glucose Homeostasis in GAD65−/− Mice.

Homozygous GAD65−/− mice are largely indistinguishable from their littermates. Adult brain weights are unaffected, and immunohistochemical analysis of GAD65+/+ and GAD65−/− brain sections with a GAD65/GAD67 antibody reveal normal cytoarchitecture and no apparent loss of neuronal populations (data not shown). In contrast to GAD67−/− mice, there is no cleft palate. Thus, GABA generated by GAD65 is not a crucial trophic factor during development.

A rare neurological disorder in humans, stiff-man syndrome (SMS), seems to involve a dysfunction of the GABA-ergic system and is characterized by a high titer of autoantibodies to GAD65, stiffness of the axial muscles, and painful spasms (20, 24, 25). GAD65−/− animals are not stiff, suggesting that a functional impairment of GAD65 by autoantibody binding is not an underlying cause of SMS. GAD65−/− mice are also not hyperactive and do not easily startle. There is no evidence of tremor, spasticity, or movement disorder. The mice are not ataxic, as judged by performance on a Roto-Rod (data not shown). GAD65 therefore is not required for gross control of locomotion.

Because significant GAD and GABA expression is detected in ovaries, testis, and the insulin-producing beta cells in the islets of Langerhans (9–11), GABA has been suggested to function as a paracrine-signaling molecule in those tissues. Both male and female GAD65−/− mice show normal fertility, glucose tolerance, and body weight (data not shown). Thus, loss of GAD65 does not markedly affect the neuroendocrine systems regulating reproduction or glucose homeostasis.

Preliminary results of analysis of GAD65−/− {129 × NOD} mice suggest that they are not markedly different from GAD65−/− {129 × B6} mice with regard to the above parameters.

GAD65−/− Animals Develop Epilepsy Resulting in Increased Mortality.

Although the animals appeared neurologically normal, there was a significant increase in mortality in the GAD65−/− colony (Fig. 3), such that less than 70% of the GAD65−/− animals survived to 7 months of age, as compared with 99% for both wild-type and heterozygous animals. To determine the cause of death, small groups of animals were monitored by constant videotaping. Several spontaneous seizures were observed during the animals’ active cycle in mice as young as 12 weeks. The behavioral manifestations of seizures consisted of clonic movement of forelimbs, loss of postural control, and falling, followed by focal clonic movements of both the forelimbs and hindlimbs. Typically, the animal remained still for several seconds, followed by a righting of posture, swaying of the head, occasionally followed by a wild running phase. After a seizure, the animal remained responsive, but relatively inactive, for several minutes and then recovered fully. Spontaneous seizures were rare, occurring less than once per week in only a small fraction of GAD65−/− animals. However, on occasion, GAD65−/− mutants seized repeatedly over a period of less than 30 min, culminating in the death of the animal. Thus, GAD65−/− mice suffer from an epileptic syndrome resulting in increased mortality.

Figure 3.

GAD65-deficient mice are prone to sudden death. Spontaneous deaths among homozygous mutant animals were observed as early as the fourth postnatal week. Solid, shaded, and open circles represent survival of wild-type, heterozygous, and homozygous mutant animals, respectively. Shown below each time point is the total number of animals represented for each genotype.

Seizures Can Be Induced by Mild Stress, and Genetic Background Influences the Seizure Frequency.

We noticed that removal of the cage lid and other routine handling occasionally induced a seizure in GAD65−/− mice, suggesting that fear and/or mild stress could precipitate seizures. A standard restraint–stress paradigm was applied to quantitate the propensity for stress-induced seizures. When GAD65−/− mice were loosely held in an elevated restrainer for 30 sec, 5% of {129 × B6} homozygous mutant animals exhibited partial motor seizures with incomplete generalization during the ensuing 10 min. The behavioral characteristics were similar as for the spontaneous seizures described above. Notably, GAD65−/− {129 × NOD} homozygotes showed an 11-fold higher incidence of stress-induced seizures than the GAD65−/− {129 × B6} mice (Table 1). These results suggest the presence of at least one modifier locus in the {129 × NOD} background that enhances the phenotypic expression of the GAD65 mutation. Although more frequent, individual seizures may be less severe in GAD65−/− mice on the {129 × NOD} background. These mice have not yet been followed long enough to assess their mortality rate.

Table 1.

Incidence of stress-induced seizures in GAD65-deficient mice crossed into C657Bl/6 and NOD/LtJ backgrounds

Genetic background	{129 × B6}	{129 × NOD}
GAD65+/+	0%	0%
GAD65−/−	5%	56%

Animals crossed into the C57Bl/6 background (GAD65+/+, n = 10; GAD65−/−, n = 19) or the NOD/LtJ background (GAD65+/+, n = 14; GAD65−/−, n = 16) were loosely held in a Plexiglas restrainer for 30 sec at an elevation of 1.5 feet above the table with the experimenter blind to genotype. Animals were observed for the ensuing 10–15 min for seizure activity. At this backcross generation, the {129 × NOD} animals do not possess a full complement of diabetes susceptibility loci, and blood glucose levels were within the normal range at the ages used (data not shown). 

Basal GABA Levels Are Affected in GAD65−/− {129 × NOD} but Not in GAD65−/− {129 × B6} Mice.

A recent report on GAD65−/− mice crossed into the C57BL/6 background found no differences in brain GABA levels in wild-type and mutant mice (17). In marked contrast, GABA content in total brain extracts of neonatal GAD67−/− mice on the {129 × B6} background is decreased to 7% of levels in normal mice (16), suggesting that GAD67 synthesizes the majority of basal GABA levels in the brain. In our study, quantitative analysis of GABA extracted from cerebral cortex, cerebellum, and hippocampus of wild-type and GAD65−/− mice confirmed that basal GABA levels are not affected in GAD65−/− mice on the {129 × B6} background (Fig. 4). Analysis of wild-type and mutant {129 × NOD} mice, however, revealed (i) significantly higher basal GABA levels in all three brain areas in wild-type mice of this strain compared with the {129 × B6} strain; and (ii) a 23, 12, and 20% decrease in GABA levels in the cortex, cerebellum, and hippocampus, respectively, of GAD65−/− mice compared with wild-type mice (Fig. 4). These results suggest that wild-type {129 × NOD} mice generally maintain a higher basal GABA level in the brain than {129 × B6} mice and that GAD65 contributes significantly to these elevated levels.

Figure 4.

Decreased basal GABA levels in GAD65−/− {129 × NOD} but not in GAD65−/− {129 × B6} mice. GABA contents were measured in brain extracts of cerebral cortex, cerebellum, and hippocampus of GAD65+/+ (solid bars) and GAD65−/− (open bars) mice (n = 15 for each region for {129 × B6} animals and n = 24 for each region for {129 × NOD} animals) by using standard HPLC methods (21). GABA content values are significantly different between wild-type {129 × NOD} and homozygous mutant {129 × NOD} mice (∗∗, cortex and hippocampus, P < 0.001; ∗, cerebellum, P < 0.02; Student’s t test). GABA content values are also significantly different between wild-type {129 × B6} and wild-type {129 × NOD} mice (cortex, P < 0.01; cerebellum and hippocampus, P < 0.001; Student’s t test).

Analysis of basal GABA levels in brain does not discriminate between neuronal and glial compartments, nor does it selectively detect the releasable pool of neurotransmitter. It therefore is possible that differences in these parameters exist between wild-type and mutant mice that were not detectable in the current analysis.

Seizures in GAD65−/− Animals Involve Limbic Structures.

The ability to reliably precipitate seizures in GAD65−/− {129 × NOD} mice by mild stress enabled an electrophysiological analysis of the excitability phenotype (Fig. 5a). Chronic electroencephalogram (EEG) recordings revealed that seizures began with the bilateral onset of intermittent high-amplitude spike discharges, converting rapidly to fast continuous spiking during the clonic phase. Termination of the seizure was followed by a period of relative electrographic and behavioral depression.

Figure 5.

EEG recording and analysis of c-fos-like expression in GAD65−/− mouse brain during and after stress-induced seizures. (a) Monopolar EEG recording (left hemisphere) showing typical neocortical seizure activity in an adult GAD65−/− {129 × NOD} mouse after mild stress (mouse was placed on a small elevated platform). The seizure lasted approximately 40 sec, beginning with an abrupt transition from rhythmic background activity (1) to bilateral (right hemisphere not shown) continuous high-amplitude spiking during clonic forelimb activity (2), followed by a prolonged (>60 sec) postictal period of low-amplitude synchronous rhythms and behavioral depression (3, only initial stage is shown). Calibration: upper trace, 5 sec, 100 μV; lower traces, 2 sec, 300 μV. (b) Immunohistochemical staining of c-fos in brain sections of GAD65−/− {129 × B6} mice 2 hr after exposure to mild stress that either did (Left) or did not (Right) induce a seizure. c-fos expression (dark staining of neuronal nuclei) is observed in (1) the granule cell layer (GC) of the dentate gyrus and CA3 of the hippocampus, (2) amygdala (Am) and piriform cortex (Pir), and (3) lateral entorhinal cortex (L Ent) of GAD65−/− mice after a seizure, but not in nonseizure GAD65−/− controls. No significant staining was observed in GAD65+/+ controls (data not shown). H, hippocampus. (Bar = 500 μm).

Immediate early genes, such as c-fos, are rapidly and transiently expressed in cells after activation by seizures, and patterns of c-fos expression have been used to map neuronal networks involved in these events (22). Two hours after a stress-induced seizure in GAD65−/− mice, induction of prominent c-fos-like immunoreactivity was observed in the granule cell layer of the dentate gyrus and the pyramidal cells of the CA3 region of the hippocampus (Fig. 5b). Significant staining was seen in other limbic structures, including the amygdala, piriform cortex, and lateral entorhinal cortex. Staining was also observed in the dorsomedial thalamic nuclei and ventromedial hypothalamic nuclei (data not shown).

The pattern of c-fos-like immunoreactivity provides evidence of limbic involvement in the epilepsy syndrome of GAD65-deficient animals.

DISCUSSION

GAD65−/− mice previously were reported to have an increased susceptibility to seizures induced by chemical convulsants (17, 26), a phenotype shared by a variety of knock-out mouse models (27–30). More importantly, however, our study shows that GAD65−/− mice develop spontaneous seizures that are easily induced by mild stress and whose frequency is strongly affected by the genetic background.

Elevated expression patterns of the immediate early gene, c-fos, after a seizure in GAD65−/− animals revealed the involvement of limbic structures. Three genetic models of epilepsy that involve limbic structures have been previously described in the mouse (31–33); seizures induced by vestibular stimulation in the inbred mouse strains SWXL-4 and El, and spontaneous seizures in Ca²⁺/calmodulin-dependent kinase II α-subunit-deficient mice. EEG recordings of seizures and c-fos expression patterns after seizures have been described for the SWXL-4 mice and are similar to the results obtained for GAD65−/− animals. The gene defect(s) for both inbred strains are still unknown, whereas seizures in the kinase mutant mice are rare and cannot be easily evoked. Other seizure models involving defects in GABA-ergic signaling, including the tissue nonspecific alkaline phosphatase-deficient mutant with reduced brain levels of GABA (34) and the GABA receptor β3 subunit^−/− mouse (35), both show generalized convulsions and early lethality in 95% of affected animals, which has precluded further characterization of the seizure phenotype. Thus, EEG recordings and c-fos activation patterns have not yet been described in these models. GAD65−/− animals are the first epilepsy model involving a gene defect within the GABA-ergic system with a significant neonatal survival and the first to show a clear evidence of limbic involvement. The ease with which seizures are induced by stress presents GAD65−/− animals as a model to study epilepsy involving limbic structures, the short- and long-term neuronal dysfunction after seizures, and the efficacy of anticonvulsant drugs.

Distinct qualities of GAD65 suggest that this isoform may have evolved for dynamic control of GABA-ergic synapses. The low saturation with PLP (3, 8), the localization at nerve termini (3), and the reversible anchoring to the membrane of synaptic vesicles that store and secrete GABA (4–6) are features that distinguish GAD65 from the cytosolic GAD67 protein and reveal a potential mechanism for activation of an apo-enzyme reservoir that can deliver newly synthesized GABA directly into vesicles for rapid secretion. The phenotype of the GAD65−/− mice is consistent with GAD67 synthesizing GABA to meet basal functional requirements for inhibitory neurotransmission. GABA synthesized by GAD65 is, however, essential for controlling excitability of limbic networks. Thus, GAD65 appears to synthesize a dynamic buffer of inhibitory neurotransmitter to provide a rapid response to certain stimuli like fear or stress, thereby fine-tuning complex mammalian neural networks.

Interestingly, in contrast to the normal basal GABA levels in GAD65−/− {129 × B6} mice, there is a measurable decrease in the generally higher brain GABA levels in mutant mice on the {129 × NOD} background. It can be speculated that the {129 × NOD} mice, which are unusually active, require a larger basal pool of dynamic GABA synthesized by GAD65 to maintain a balance between stimulatory and inhibitory synapses. Although GABA levels in brain extracts are unaffected in GAD65−/− mice on the {129 × B6} background, GABA release, measured by in vivo microdialysis in the cortex during a prolonged stimulation with K⁺, is significantly decreased, indicative of defects in GABA secretion attributable to the absence of GAD65. (T. K. Hensch, M. Fagiolini, N. Mataga, M. P. Stryker, S.F.K., and S.B., unpublished data).

Stress can induce seizures in epileptic patients (36) and readily precipitates seizures in GAD65−/− mice. Interestingly, long-term restraint stress has been shown to directly affect limbic pathways in rats, as evidenced by c-fos staining, although seizures are not evoked (37). We propose that in the absence of GAD65-generated GABA, inhibitory networks fail to suppress abnormal synchronization within this and synaptically related brain regions, thereby lowering the threshold for seizures.

The propensity of GABA antagonists to induce seizures, and of GABA mimetics to suppress seizures, has long been recognized (38). These correlations implicate GAD and the GABA-receptor genes as candidate genes for familial seizure disorders in humans, although disease-associated mutations in these genes have not yet been identified. Our study specifically identifies GAD65 as an essential gene for controlling epilepsy affecting limbic structures and implicates at least one modifier locus by comparison of the C57BL/6 and NOD/LtJ backgrounds. Thus, GAD65 and the modifier locus (loci) are important candidate genes for familial epilepsy in humans (39).