CHAPTER 50 Animal Models of Epilepsy
Epilepsy affects a sizable proportion of the population worldwide and is responsible for a heavy social and economic burden.1 Given that epileptic disorders frequently result in uncontrolled and often medically intractable seizures, further experimental studies offer hope for a deeper understanding of the underlying pathophysiology and, ultimately, better treatments. To achieve these goals, human studies, albeit of great value, may not be sufficient because of both ethical and practical limitations.2 Accordingly, it is not surprising that animal models were developed and have significantly contributed to the epilepsy research literature. A common concern with all animal models is how reliable they are in mimicking the human condition. Are the anatomic and electrophysiologic similarities with human epileptic disorders real or superficial? To what extent can findings in animal research be extended to humans? Furthermore and perhaps most important, what new insights can be gained from animal model studies that cannot be predicted from clinical studies? In the design and interpretation of animal models, it is important to remember that epilepsy is not a single disease, a syndrome, or a homogeneous entity. Although a common feature of epilepsy is the tendency to have spontaneous epileptic seizures,3 the many ways in which seizures are generated (and manifested) are quite varied. Seizures can be motor, sensory, or autonomic and are caused by excessive and abnormal neuronal discharge. Causes of epilepsy range from genetic to acquired to “of unknown origin” (i.e., idiopathic).4 Its symptoms are also many because they depend on the brain areas that are involved in a certain type of seizure and the stage of brain maturation. The type (or types) of seizures that one individual has and other symptoms that are also present can be used to define a specific epileptic syndrome. Because epilepsy involves many levels of structures and activity in the brain, be it from molecules to networks, with causes that range from genes to environmental insults, it is no surprise that many epilepsy models are needed.
A good model of epilepsy should reproduce as many salient aspects as possible for a specific type of human epilepsy.5,6 Ideally, there should be evidence in a model for spontaneous behavioral seizures of the kind (e.g., partial, absence, tonic-clonic) that its human counterpart exhibits, as well as electroencephalographic (EEG; ictal and interictal discharges, focal or generalized) and structural (if any) abnormalities similar to the ones seen in the human epilepsy that it mirrors. The etiology (e.g., cortical malformation [CM], genetic predisposition, focal gliosis) should, if possible, be the same in the animal model as in the human condition. If the human epilepsy has a specific age at onset, the model should do the same. The animal model should have behavioral characteristics (such as memory deficits or developmental retardation) that parallel the human condition. Finally, the model in question should respond to antiepileptic drugs (AEDs) in a manner similar to the human condition after which it is modeled. Even though these goals are scientifically sound, few (if any) animal models actually fulfill all these criteria. Here we review animal models designed to mimic three broad classes of epileptic disorders: (1) temporal lobe epilepsy (TLE), (2) epilepsy associated with a brain malformation, and (3) other focal epilepsies.
Temporal Lobe Epilepsy
Seizures that arise from the temporal lobe are very common in humans and are often resistant to pharmacologic treatment. The most common type of TLE is that associated with hippocampal sclerosis.7 Kindling, pilocarpine, and kainic acid (kainate) are the most common animal models of TLE.
Stimulus-Induced Model of Temporal Lobe Epilepsy
Kindling
Kindling has been used as a model of seizures and epilepsy for more than 3 decades. Initially, these animals were not used for epilepsy, but rather as a physiologic model of learning and memory.8 This model is based on the kindling phenomenon, which involves the progressive development of EEG and behavioral seizures evoked by repeated electrical stimulation of certain brain structures. The first studies from the 1960s showed that repetitive electrical brain stimulation produces an increase in convulsive behavior and eventually generalized motor seizures.8–10 The progression of motor seizures proceeds through several stages as defined by Racine for the amygdala kindling protocol.11 With initial stimulation, there is an initial ictus with orofacial movements (stage 1) and head nodding (stage 2), which is associated with a focal EEG seizure. Further stimulation results in contralateral forelimb clonus (stage 3) and rearing (stage 4). Additional loss of balance, rearing, and falling (stage 5) are accompanied by tonic-clonic seizures involving all four limbs.12 If the kindling stimulus continues to be applied to the focus after a stage 5 seizure is reached, the behavioral sequence evolves further to more severe seizures. Thus, the more advanced the animal in this process, the more intense the seizures and the more widespread and prolonged the electrical seizures or afterdischarges (i.e., the most sensitive parameter of focal seizure activity in kindled rats). Kindling occurs best in neuroplastic areas of the brain; the amygdala is the brain region most susceptible to kindling.13
Although other, but not all encephalic structures can be kindled, there are some differences between them regarding the speed of the kindling process and seizure manifestation.14–16 Stimulation of the cerebellum and superior colliculus, for example, cannot start kindling. Kindling from the hippocampus produces seizures quite similar to but slower to develop than those induced by amygdala kindling. The profile of afterdischarges is different from amygdala kindling in that they recur daily during the kindling process until the seizures recruit other temporal areas.14,17 One peculiar early behavioral response seen during hippocampal kindling in some rats is the so-called wet dog shakes, which start and progressively disappear during the kindling process before the expression of any Racine motor stages.
The neurological alterations associated with the types of kindling behavior seem to be permanent. Kindled animals that have not been stimulated for many months after stage 5 kindling often respond with full seizure behavior on repeat exposure to the original kindling stimulus.8,18 Moreover, kindling of a second structure is usually achieved with fewer stimuli than needed for the primary kindled site.8,19,20 This “transfer phenomenon” might be the physiologic basis for secondary epileptogenesis.21–23 In contrast, if two different nonhomologous areas receive concurrent and alternate stimulation, only one of them undergoes a typical kindling progression, and seizure genesis in the other site is suppressed. This “kindling antagonism” may involve inhibitory mechanisms recruited by the seizure activity from the more dominant site24,25 and could reflect the types of endogenous compensatory mechanisms that the brain has evolved to combat seizure activity.
When the amygdala or other brain areas such as the frontal and posterior cortex, entorhinal cortex, and perforant pathway are kindled, spontaneous seizures can develop in the animals after a period. However, just a few kindled animals reach this point.26,27 Recent evidence from the laboratory of Dan McIntyre suggests that there may be a genetic explanation for this interesting observation. Over time, they isolated strains of “fast”- and “slow”-kindling rats, with the former being highly susceptible to kindling treatment and the latter more resistant.28–30 These rats exhibit different seizure predispositions and behavior that might be attributed to differences in the expression of GABAA receptor subunits and in the kinetics of miniature inhibitory postsynaptic currents in different limbic structures.28,31–33 Behaviorally, the fast-kindling rat strains are more prone than the slow-kindling strains to the development of seizures and appear much like humans with attention-deficit/hyperactivity disorder in that they show easy distraction, hyperactivity, and impulsivity.32,34,35 Importantly, spontaneous seizures in kindled animals are not just a rat phenomenon inasmuch as epilepsy has developed in every animal tested thus far with this stimulation protocol: frogs, reptiles, mice, rats, rabbits, cats, dogs, rhesus monkeys, and baboons.36–38 A persistent question with this model is whether a kindling-like phenomenon occurs in patients with epilepsy. This discussion was initiated by Gowers, who affirmed that “seizures do beget seizures.” In other words, epilepsy could be a progressive condition wherein the chronic recurrence of seizures would increase the likelihood of new seizures, and they could come from new areas (e.g., homologous contralateral sites). There are some hints in epilepsy associated with brain tumors that secondary epileptogenesis does occur in humans.23 Although deliberately causing kindling in humans would be ethically suspicious, to say the least, some reports of such do exist in the literature.39
As a model of epilepsy, the kindling model has been successfully used as a screening tool for AEDs. Lamotrigine, used as an AED in patients with partial and generalized seizures, has long-acting anticonvulsant effects on both amygdala- and hippocampal-kindled seizures in rats.40 GABA uptake inhibitors, such as tiagabine, were also found to be highly effective in the kindling model,41 consistent with the GABA receptor involvement described earlier. Another example is levetiracetam, a relatively new AED that is effective in the kindling, pilocarpine, and kainate models of TLE and in human TLE.42–45 Interestingly, these three drugs are inactive against seizures induced by maximum electroshock (MES), a model of acute seizures in nonepileptic animals and additionally used to screen for potential AEDs.41 More recently, amygdala kindling has also been proposed as an animal model for drug-resistant epilepsy.46 Studies have shown that kindled seizures are less sensitive than primarily generalized seizures to anticonvulsant treatment, as demonstrated with the MES test.41 The response to phenytoin, for example, differs in kindled rats (i.e., in some animals phenytoin has an anticonvulsant effect consistently and in others never).47 Unfortunately, the preparation of kindled rats and drug testing in this model are both labor and time intensive, which limits its use in high-throughput drug discovery and development programs. Some less laborious models are discussed in this respect, including corneal kindling, acute models of focal seizures (e.g., electrical induction of afterdischarges), and genetic models. However, when compared with conventional amygdala kindling, none of these models have been characterized sufficiently as yet to judge their usefulness for drug development.48
Finally, the kindling model offers some distinct advantages over systemically drug-induced models of TLE. With kindling, one knows precisely where the epileptogenesis begins (i.e., the site and intensity of the stimuli applied) and can track by numerous methods the progressive changes that occur during a very prolonged latent period, which ends with spontaneous seizures. Additionally, gross brain damage does not appear to be a prerequisite for the initial manifestation of spontaneous seizures.26 After repeated secondary generalized seizures evoked by kindling, cumulative neuronal loss can be detected in the CA1 and CA3 subfields of the hippocampus and in the hilus of the dentate gyrus (DG), similar to classic human hippocampal sclerosis, which is associated with long-term seizure-induced memory dysfunction.49–51 Although kindling does not result in a large loss of neurons, there is clear reorganization of the neuronal circuitry. This reorganization is based on synaptogenesis52 or sprouting of mossy fibers (or both) in the inner molecular layer of the DG53 and in the stratum oriens of the CA3 hippocampal area.54 In addition, kindling-induced seizures result in changes in the morphology of astrocytes that (1) appear to be dependent on seizure intensity, (2) occur early in the kindling process, and (3) persist for weeks after the last seizure. An increase in astrocytic proliferation in the hippocampus, amygdala, and piriform cortex was also observed after kindling.55 Some of these structural alterations induced by kindling are similar to those in the human epileptic temporal lobe, thus raising the possibility that mechanisms operative in kindling may play a role in the pathogenesis of human TLE associated with hippocampal sclerosis.56 As one might expect, the seizure progression and structural changes are accompanied by progressive functional alterations in the kindled animal. These changes have been particularly well studied in the DG. When the first seizure is induced, there is an increase in excitability through N-methyl-D-aspartate (NMDA)-mediated synaptic transmission. Even though this increment is not permanent,57 it appears to play an important long-term role by initiating structural and functional modifications that contribute to long-lasting seizure susceptibility in the hippocampal circuitry.58 Neurochemical and neurophysiologic studies have indicated that the activity of the GABAergic projection from the striatum to the substantia nigra pars reticulata is reduced and the density of GABAergic striatal interneurons is increased in kindled rats.59–62 In addition, GABAergic inhibition is altered in the piriform cortex of kindled rats.63
Chemical-Induced Models of Temporal Lobe Epilepsy
Pilocarpine and Kainate
Pilocarpine and kainate models replicate several phenomenologic features of human TLE, and both can be used as animal preparations to understand the basic mechanisms of epileptogenesis.64–66 Pilocarpine is a cholinergic muscarinic agonist originally isolated from the leaflets of South American shrubs (Pilocarpus jaborandi). The first evidence that the brain damage produced by pilocarpine-induced status epilepticus (SE) could lead to spontaneous recurrent seizures (SRSs) in rats over a long-term period occurred in 1983.67 Kainate is an excitatory amino acid extracted from seaweed (Digenea simplex) and has been used as a model of epilepsy since the late 1970s.68,69 Local or systemic administration of either pilocarpine or kainate to rodents leads to a pattern of repetitive limbic seizures and SE that can last for several hours.67,70,71 Systemic administration of a single dose of pilocarpine or repetitive low doses of kainate to rodents leads to a sequence of seizures that build progressively into a limbic SE lasting up to 24 hours (acute period). In these models, induction of SE is critical for the later development of SRSs. By managing the severity of SE one can decrease animal mortality without decreasing the development of recurrent epilepsy. For pilocarpine, this can be accomplished by using thionembutal or diazepam after the beginning of SE. Pretreatment with scopolamine also minimizes the peripheral cholinergic side effects of pilocarpine.72,73 In some protocols, lithium is applied before pilocarpine to increase its effect and reduce mortality.74 For kainate, a multiple low-dose protocol may decrease mortality and produce animals with robust SRSs.75,76 SE induced by kainate can also be abolished by diazepam. However, some animals may die before the benzodiazepine injection, and this approach may lead to a lower percentage of animals in which robust and long-lasting convulsive SE and subsequent SRSs develop. Seizures during the acute period are similar with both drugs. After injection of pilocarpine, the animal begins with staring and facial automatisms, followed by motor limbic seizures with rearing, forelimb clonus, salivation, intense masticatory movements, and falls. This progression follows the Racine stages.77 For kainate, the seizures are characterized by wet dog shakes, staring, searching, and gnawing, subsequently leading to forelimb clonus (stage 3 of Racine) and followed by Racine stages 4 and 5.78 The electrographic pattern during the acute period is characterized by theta rhythm activity in the hippocampus and low-voltage fast activity in the cortex with progression to high-voltage fast activity with spikes in the hippocampus. Sequentially, spiking activity spreads to the cortex and evolves into electrographic recurrent seizures that build up to continuous epileptiform activity.71,79 This acute period is followed by a “silent period” (seizure-free phase) with progressive normalization of EEG activity and behavior of variable extent that lasts from days to weeks. This sort of latent interval is believed to play an important role in “ripening of the focus,” a process that renders the animal chronically epileptic.80 After this latent period, animals begin to have frequent SRSs (chronic period).67,70,73,81–83 The main features of the SRSs observed during the long-term period resemble those of human complex partial seizures.66,67,79,84 SRSs are characterized by facial automatisms, head nodding, forelimb clonus, rearing, and falling and electrographically by paroxysmal hippocampal discharges that rapidly spread to cortical regions. The ensuing spontaneous seizures show gradual electrographic synchronization of cortical and hippocampal activities and a longer duration of ictal events. Forelimb clonus and rearing with falling (kindling stages 4 or 5) are the clinical hallmarks of this SRS phase.71,79
As models for temporolimbic epilepsy, the pilocarpine and kainate models are important instruments for screening new AEDs. Studies have shown that diazepam, phenobarbital, valproic acid, and trimethadione protect against acute pilocarpine-induced SE whereas phenytoin and carbamazepine are ineffective.66 In contrast, during the chronic period, phenobarbital, carbamazepine, phenytoin, and valproic acid but not ethosuximide are able to control pilocarpine-induced SRSs. Valproic acid was likewise able to abolish SRSs in kainate-treated animals.85 These AEDs can also control complex partial seizures in humans.86 SRSs may thus be more reliable than acute seizure models for finding new AEDs with better efficacy against complex partial seizures. In the kainate model, a ketogenic diet also seems to be effective in reducing SRSs.87
In hippocampal tissue removed from patients with refractory TLE, neuronal loss is most prominent in the hippocampal formation, particularly in the DG and the CA1 and CA3 subfields.88–90 Mossy fiber sprouting,91,92 hippocampal gliosis,88,89 cell dispersion in the granular cell layer of the DG, and local ectopic areas91 are additional neuropathologies. Although these morphologic changes are much more prominent in pilocarpine- and kainate-treated rats than in humans,73 the structural changes in these models of epilepsy during the acute and chronic periods resemble the anatomic changes found in many patients with mesial TLE.64,65,67,69,73,93–95 Morphologic analysis of the rodent brain during the acute period shows that both kainate- and pilocarpine-induced SE invariably leads to cellular loss and injury in the hippocampal subfields CA1 and CA3 and in the DG hilus. Cell loss is also often seen in the septum, olfactory tubercle, amygdala, piriform cortex, neocortex, thalamic nuclei, and substantia nigra.67,96–98 In fact, the thalamus can exhibit severe atrophy in various nuclei.99 Damage to subpopulations of GABAergic inhibitory neurons is also extensive throughout the hippocampus.100 In the following months the cell damage tends to progressively involve other areas. In the chronic period, cell loss is often noted in the DG (mainly attributed to a decrement in GABAergic interneurons) and in the hippocampal subfields CA1 and CA3 (loss of pyramidal cells). The decreased subpopulation of interneurons in the DG is associated with a functional hyperexcitability.65 In the kainate model, the CA3 area seems to be preferentially damaged in comparison to area CA1, which probably reflects the higher distribution of hippocampal kainate receptors in CA3. This differs from the typical hippocampal sclerosis in humans, in which CA1 is usually more damaged than CA3. As in humans, the DG and the CA2 area are relatively spared.69,88,89,93 Cell dispersion is evident in the granule cell layer of the DG101; increased neurogenesis102 and supragranular and intragranular mossy fiber sprouting73,103,104 are also important features in these models during the period of SRSs. Interestingly, SRSs can occur in animals with complete absence of mossy fiber sprouting.105,106 These morphologic changes seem more likely to represent a reactive response rather than a direct consequence of the initial insult. As observed in kainate-treated rats, axonal sprouting and synaptic reorganization take place not only in the DG but also in other hippocampal areas, such as CA1 and CA3.107 These cellular and synaptic modifications are associated with neurochemical and cellular dysfunctions that lead to permanent hyperexcitability. For example, the decrease in GABAergic cells in the DG and entorhinal cortex in the pilocarpine model of epilepsy is associated with decreased inhibitory synaptic input to the granule cells and to the layer II stellate cells, respectively, with a consequent increase in network excitability. Reduced granule cell inhibition in the DG seems to precede the onset of SRSs by days to weeks.108,109 Decreased expression of some postsynaptic GABAA receptor subunits,110–116 loss of GABAergic interneurons in the entorhinal cortex, and changes in neuropeptide Y expression, including ectopic expression in the granule cells and mossy fibers, have also been observed in rats with pilocarpine-induced epilepsy.117 Some of these changes precede the onset of SRSs by weeks and correlate with profound alterations in receptor function that contribute to the process of epileptogenesis. Increased glutamate release in the hippocampus was also observed during the acute period118,119 and could further contribute to the epileptogenic process in this model. For example, constant activation of NMDA-type glutamate receptors, which leads to increased intracellular Ca2+ and consequent activation of lipases, proteases, and nucleases, can kill cells by necrosis, apoptosis, or both. Thus, activation of NMDA receptors plays an important role in SE and brain damage in these models.120 The expression of proteins related to ionotropic NMDA and metabotropic glutamate receptors in the hippocampus is also modified in the pilocarpine model of epilepsy.121,122
As much as the pilocarpine and kainate effects in rats can resemble the human condition, species differences between rodents and primates are frequently used to explain existing discrepancies. To this end, developmental changes are often called on to explain features such as tectonic malformations123 and DG cell dispersion91 in humans even though these findings have been demonstrated to take place in adult epileptic rats.101 As a means to more properly evaluate the extent to which rodent models mimic the human condition, pilocarpine has also been used in a nonhuman primate, the marmoset (Callithrix jachus).124,125
In seminal work by Luiz Mello and colleagues, administration of pilocarpine was used to induce SE in marmosets followed, after a silent period, by SRSs. It was observed from this model that prolonged SE is required to promote injury during the acute period and that the mortality rate is higher during SE in marmosets than in rats. That rodents seem to be more resistant to pilocarpine-induced SE (in wild types, 20% to 30% of animals will not reach full SE after drug administration) may be the reason that more damage is necessary for spontaneous seizures to develop. Different from the rodent pilocarpine and kainate models but similar to human TLE, brain damage is minimal during the chronic period in marmosets and generally limited to the limbic structures (mainly the hippocampus), and neurogenesis, exuberant in rodents,102 is mild in this primate.126 Cellular dispersion in the granular cell layer of the DG; neuronal loss in CA1, CA3, and the DG hilus; and supragranular mossy fiber sprouting were detected in marmosets with SRSs. Regarding morphology, one interesting observation in the marmoset pilocarpine model was the finding of tectonic malformations in the DG characterized by invaginations and lateral displacement of the granular cell layer, mainly in the posterior plates, similar to that described by Sloviter and coworkers in humans with TLE.123 This new model of TLE in marmosets may have a greater resemblance to the alterations seen in human TLE and may provide a new tool to more properly evaluate the causes of and consequences associated with TLE.
Cortical Malformations and Epilepsy
CMs are developmental neuronal disorders in humans characterized by (1) disorganization of cortical architecture or (2) the presence of abnormal, immature, and eventually nondifferentiated neurons (or both).127–130 Genetic or environmental prenatal factors can lead to focal or diffuse CMs, or both, which are often associated with refractory epilepsy.127–129 The anatomic features of CMs associated with epilepsy have been widely studied.131–134 Although no single animal model can reproduce the precise anatomic disorganization and cell abnormalities observed in humans, these models have proved useful in studying the functional characteristics of dysplastic neuronal networks. Some CM models are based on induced injury in the immature brain (e.g., freeze, undercut, irradiation, teratogen exposure); others are based on spontaneous genetic mutations or on trangenesis that gives rise to abnormal cortical architecture. Here we review some of the injury-based and genetic models of CMs.
Injury-Based Models of Cortical Malformations
Brain malformation can be induced in animals by teratogenic compounds or by mechanical treatment. These models include prenatal exposure (embryonic days 14 to 16) to methylazoxymethanol acetate (MAM)135 or to irradiation (γ- or x-rays)136 and the neonatal application (postnatal days 0 to 2) of a freeze lesion in the cortical plate of rodents.137,138 There are similarities in the embryonic development of the human and rat brain, and neurogenesis occurs in a relatively known spatiotemporal manner.139 Deleterious agents applied at selected stages of development can reduce neuronal populations in the injured brain regions,140 similar to those seen in some forms of CM in humans. This reduction and altered architecture subsequently lead to a focal or diffuse structural and functional impairment of the brain that contributes to epileptogenesis.
Methylazoxymethanol Acetate
MAM is an antimitotic methylating agent known since the 1960s to induce brain malformations141–145 in rats when the exposure takes place at the beginning of the third week of intrauterine development. MAM has a short half-life and exerts an antiproliferative action on dividing, but not quiescent neuroepithelial cells via methylation of nucleic acids.135,143 Prenatal injection of MAM into female rats at gestational day 15 selectively affects the proliferation of specific neuronal cell populations by disrupting the sequence of normal brain development. The offspring resulting from prenatal MAM exposure exhibit multifocal brain malformations, microcephaly,141,146 loss of lamination, vascular abnormalities, and neuronal heterotopia in the hippocampal and periventricular locations.147–153 The abnormal cell clusters (heterotopia) first appear postnatally in the hippocampus (postnatal days 1 to 2), and their appearance is preceded by a distinct sequence of perturbations in neocortical development.154 These induced neuronal heterotopias possess many features of CMs associated with epilepsy in humans, such as focal cortical dysplasia, periventricular nodular heterotopia, and tectonic hippocampal malformations in patients with TLE.123,148,155 In addition, heterotopic neurons have an abnormal synaptic network, with neurons communicating directly between the neocortex and hippocampus,156 and they exhibit characteristics of neocortical cells. For example, molecular analysis has revealed that hippocampal heterotopic cells express mRNA markers normally found abundantly in layer II and III supragranular neocortical neurons and exhibit firing properties strikingly similar to those of supragranular cortical neurons.157 MAM-treated rats have increased susceptibility to induced seizures in vivo,158–160 and hyperexcitability in vitro is refractory to the commonly available AEDs.161 This hyperexcitability can be attributed to a combination of different cellular mechanisms.162,163 Functionally, heterotopic hippocampal cells (1) lack Kv4.2 A-type potassium channels,162 which results in abnormal neuronal firing, and (2) exhibit changes in glutamate receptor function and expression.164–166 The inhibitory synaptic system is also altered in hippocampal heterotopia, including changes in the inhibitory drive such as a prolonged duration of inhibitory synaptic events associated with a decrease in transporter-mediated GABA reuptake.131 These alterations in GABAergic inhibition could represent some type of postnatal compensatory response to the intrinsic hyperexcitability of these animals because spontaneous electrographic seizures, although present, are rare in MAM-exposed rats.165 Interestingly, some of these same deficits were found to occur in dysplastic tissue samples from patients with focal cortical dysplasia and epilepsy.132
Irradiation
Different radiation sources (x-rays, γ-rays) have been used to induce CMs in animal models, but the results have been similar in all cases. The histologic abnormalities in irradiated animals are attributed to the initial injury from irradiation associated with continued cortical development in an altered cellular environment. Immature, migrating neurons and radial glia are particularly sensitive to radiation.136,167 Different CMs were produced in rats exposed to x-rays in utero (200 cGy)168 and γ-irradiation (150 to 250 cGy). Pregnant rats exposed to external γ-irradiation on gestational day 16 to 17 produce offspring with various degrees of CMs and architectural abnormalities, depending on the dose of in utero radiation.169 Timing of the exposure is also critical,136 with exposure on gestational day 16 to 17 producing the most severe malformations. These abnormalities include microcephaly; diffuse CM with dyslamination and lack of columnar organization in multiple areas of the neocortex; subcortical, periventricular, and hippocampal neuronal heterotopias; and dysgenesis of the corpus callosum.169–172 Not only do radiation-exposed rats exhibit epileptiform discharges, but spontaneous seizures also develop in vivo.172–174 The epileptogenicity in vivo increased with mild to moderate radiation doses. However, high doses of in utero radiation leading to more severe pathologic changes in the neocortex and hippocampus are not associated with the occurrence of spontaneous seizures,173,175,176 again suggesting potential compensatory mechanisms or widespread brain damage that could preclude the ability to generate network activity. Extracellular recordings from brain slices of in utero irradiated rats showed enhanced epileptiform activity in the dysplastic areas in vitro.177 Anatomic reduction of GABAergic interneurons (parvalbumin- and calbindin-positive cells178
