Pathophysiological Mechanisms of Focal Cortical Dysplasia: A Critical Review of Human Tissue Studies and Animal Models

Abstract

Summary:  Cortical dysplasia (CD, also known as malformations of cortical development) are the pathological substrates in a large percentage of patients with pharmacoresistant epilepsy who may be amenable to surgical treatment. Therefore, research on the mechanisms of dysplastic lesion formation and epileptogenicity is of paramount importance for the prevention, detection, and treatment of CD-induced epilepsy. The purpose of this review is to discuss and critically evaluate the current state and results of human tissue experimentation (focusing on reported results of studies done on neocortical dysplastic tissue resected from patients with pharmacoresistant epilepsy), and to discuss some of the concerns related to research that uses surgically resected epileptic human tissue. The use of better animal models of CD as a tool toward the better understanding of the mechanisms of pathogenesis, epileptogenesis, and epileptogenicity of dysplastic lesions will be reviewed from the perspective of their usefulness in a model of translational research that should ultimately result in better diagnostic and therapeutic techniques of CD.

There are different types of human histopathological subtypes and animal models of cortical dysplasia (CD) with various mechanisms of epileptogenicity. Beyond the rather descriptive characterization of these substrates, there is a need for the validation of the potential mechanisms at three levels: in vitro electrophysiological studies (acute or chronic organotypic human tissue slice preparations), in vivo recording methods using well characterized and improved animal models of CD, and in human patients with focal epilepsies due to CD. 3) The development of target-specific diagnostic and therapeutic approaches based on the understanding of the molecular mechanisms, e.g., the development of ligand-specific PET/SPECT imaging techniques for the visualization of epileptogenic substrate(s) or drug resistance protein(s). These techniques may more accurately localize the epileptogenic area(s) and predict pharmacological response.

RESEARCH ON HUMAN TISSUE OF CORTICAL DYSPLASIA

Pathological characteristics of CD and issues of nomenclature/classifications

In 1971, Taylor et al. (1971) described CD as a pathological substrate for neocortical epilepsy. CDs are characterized by a disruption of the normal lamination of the cortex that can vary in severity, ranging from a mild disruption of lamination with normal-appearing neurons to a strong loss of laminar organization, usually accompanied by the appearance of dysmorphic and misoriented neurons, neuronal clustering, giant neurons, and/or balloon cells (BCs) (Mischel, 1995; Prayson, 1995; Prayson et al., 1996; Palmini et al., 2004) (Figs. 1 and 2). BCs are “malformed cells of uncertain origin with large, sometimes multiple nuclei surrounded by an excess of opalescent, pseudopodic cytoplasm” (Taylor et al., 1971) (Fig. 3). In addition, heterotopic neurons and gliosis are observed in the white matter (Mischel, 1995; Prayson, 1995; Prayson et al., 1996; Palmini et al., 2004). Giant neurons (also called meganeurons) apparently have a normal content of cytoplasmic organelles and, as in normal pyramidal cells, receive numerous symmetric synapses in cell body and proximal dendrites (Taylor et al., 2001).

Figure 1. Cresylecht violet staining of (A) resected nonepileptic area normal positioning of pyramidal cells (arrows) adjacent to (B) an epileptic region characterized by the presence of abnormal positioning of smaller dysmorphic/dysplastic neurons showing high Nissl staining (Scale bar: 20 μm).

Figure 2. Cresylecht violet staining of a resected nonictal region showing the presence of large neurons with high Nissl staining (meganeurons, arrows) intermixed with balloon cells (double arrows) (Scale bar: 50 μm).

Figure 3. Cresylecht violet staining of a resected nonepileptic area showing many large cells with opalescent cytoplasm and eccentric nucleus (nuclei). BC are mainly found in layer VI and the subcortical white matter (Scale bar: 100 μm).

One issue that faces researchers in the analysis of data on CDs is the variable nomenclature used by different groups (Mischel, 1995; Palmini, 2000; Tassi et al., 2001; Crino et al., 2002; Cepeda, 2003; Colombo et al., 2003a, 2003b; Fauser et al., 2004b; Palmini et al., 2004). This led to the lack of agreement on the nomenclature of some constitutive cells and thus to different classification schemes of various pathological subtypes. The first problem resides in the histological definition of the cellular abnormalities that differ between groups. For example, one group may use the term “cytomegalic neuron” to describe a specific subtype of abnormal neurons (also called giant neuron in the classification as proposed by Palmini et al. (2004) under type IB focal CD, while another group equates this term with a subtype of dysmorphic neurons (André, 2005). In addition, it is unclear if all types of malformed neurons (so-called dysplastic or dysmorphic neurons) should be grouped together since these cells may be of various shapes and sizes, have different Nissl staining characteristics, and/or their apical dendrites may be pointing to various directions. This has led to some groups reporting upregulation of specific NMDA receptor subtypes in dysplastic neurons but not in giant cells (meganeurons) and a resulting increased in vitro epileptic activity (Ying et al., 1999; Najm et al., 2000; Crino et al., 2001; Ying et al., 2004), while others reported a decrease in the activity and in the expression of these subunits in large pediatric hemispheric malformations (André, 2005). Upon further analysis of all the data, it becomes clear that the apparent discrepancies in the results on the expression of selected proteins (such as the NR2B subunit of the NMDA glutamate receptor) are most likely due to the use of various pathological subtypes and a lack of consistent histological nomenclature.

These discrepancies highlight the need for a molecular characterization of the various cellular elements in CD that would lead to a more objective cell-specific genetic classification. This could be achieved through the isolation of various cells using laser microdissection (LMD) and single cell cDNA microarray technology.

There are also significant discrepancies in the architectural classifications of CDs: some proposed classifications included a rather heterogeneous variety of malformations of brain development (Raymond et al., 1995), while others were based on histological characteristics (Mischel, 1995; Palmini et al., 2004). The Cleveland Panel recommended a classification system that included the following: Type 1A; cortical architectural abnormalities, type 1B; architectural abnormalities with giant cells (meganeurons), but no dysmorphic neurons or BCs, type 2A; dysmorphic neurons in the setting of architectural organization, and type 2B; the presence of BCs intermixed with dysmorphic neurons in the setting of architectural organization (Palmini et al., 2004). This division is based on histopathological characteristics (Prayson, 1995) with likely distinct electroclinical, functional, and imaging correlates (Marusic et al., 2002; Boonyapisit et al., 2003; Widdess-Walsh et al., 2005). Previous studies showed that types 2A and 2B have a higher degree of epileptogenicity (Palmini, 2000) and are more likely to exhibit various degrees of MRI abnormalities (Lee, 1998). However, types 2A and 2B are different in terms of their ability to retain eloquent function (Boonyapisit et al., 2003). Earlier age of onset, earlier age at surgical resection (Rosenow, 1998), low incidence of seizure risk factors and a lack of in situ epileptogenicity (Marusic et al., 2002) have been previously described in BC-containing CD (Type 2B).

Electroclinical, imaging, and pathological correlations in focal CD

The immediate impact of the discrepancies in nomenclature and histopathological CD classifications is on the electroclinical correlations. Recent studies suggested significant differences in the clinical natural history, MRI characteristics, outcome, and expression of intrinsic epileptogenicity in various histopathological subtypes (Mischel, 1995; Marusic et al., 2002; Boonyapisit et al., 2003; Widdess-Walsh et al., 2005). The more severe histopathological forms of dysplasia (types 2A and 2B), which are more often located in the frontal lobe, expressed epilepsy and pharmacoresistance at an earlier age, showed a higher seizure frequency, showed higher degree of cognitive decline but tended to exhibit better postsurgical seizure outcome.

High-resolution MRI showed abnormalities in the majority of patients reported from various surgical series (Colombo et al., 2003a, 2003b; Zhang et al., 2003; Widdess-Walsh et al., 2005). This rather high incidence of signal and morphologic variations may be due in part to selection bias in patients with epilepsy, whose MRIs showed no lesion, are less likely to undergo surgical resection. CD lesions with BCs tend to show more frequent and more severe MRI FLAIR signal abnormalities (Colombo et al., 2003a, 2003b; Widdess-Walsh et al., 2005). The reason for these clinical differences is unknown but a role for the increased number of glial fibrillary acidic protein (GFAP) positive cells in type 2B is suggested. Recent reports by Colombo et al. (2003a, 2003b) described focal cortical thickening, blurring of the gray white matter junction, and hyperintensity (on T₂-weighted images) of subcortical white matter often tapering toward the ventricle in type 2 CDs. While the focal brain hypoplasia with shrinkage and moderate signal intensity alterations in the white matter core are consistent with type 1 CD (Colombo et al., 2003a, 2003b). Most but not all dysplastic lesions are intrinsically epileptogenic as recent studies showed that most of the ictal patterns reside outside the BC-containing regions where the histopathological changes are less severe (Marusic et al., 2002; Boonyapisit et al., 2003). These findings raise the question about the role of BCs in CD that we shall discuss later.

Pathogenesis of focal CD

Various hypotheses have been advanced to explain the pathogenesis of CD. Most of the hypotheses focused on the role of embryonic/prenatal/developmental insults as the main causes for the majority of dysplastic lesions.

CD as a malformation due to abnormal embryonic brain development

The most dominant hypothesis posited that CD may be a consequence of erroneous migration, maturation and/or cell death during ontogenesis (Crino and Eberwine, 1997; Spreafico et al., 1998a, 1998b; Cotter et al., 1999a, 1999b; Andres et al., 2005). These changes were attributed to defects in late phases of development (Andres et al., 2005). This is supported by the persistence of neuronal elements in the white matter lying between the periventricular region and the cortical mantle. These neurons were shown to exhibit various degrees of neuronal maturation as some cells showed characteristics of neuronal progenitors, such as nestin, undifferentiated neurons or glial cells exhibiting the presence of ß tubulin III (TUJ-1), or vimentin, respectively, and others showing mixed neuronal and glial characteristics such as the coexpression of GFAP and neuronal nuclear (NeuN) and neuronal cytoskeletal proteins (microtubule associated proteins, MAP) (Garbelli et al., 1999; Taylor et al., 2001; Aronica et al., 2003; Fauser et al., 2004a; Ying, 2005). In addition, the persistence of neuronal elements (Cajal Retzius cells) in layer 1 of the neocortex was attributed to a failure of maturation of the cortical mantle through a problem in the apoptosis of these cells (Garbelli et al., 2001; Marin-Padilla et al., 2002; Yun et al., 2003; Thom et al., 2005). Areas where CD is worst present an abnormally increased expression of neurofilaments (Tassi et al., 2001). Proteins and mRNA that are usually expressed during brain development and are downregulated in mature neurons, such as MAP1B, vimentin, nestin, neurofilament, peripherin, α-internexin, and SMI311 (marker for nonphosphorylated filaments), are commonly found in BCs from CD, tuberous sclerosis, and hemimegalencephaly (Yamanouchi et al., 1996; Crino et al., 1997; Yamanouchi et al., 1998; Cotter et al., 1999a, 1999b; Yamanouchi et al., 2000; Tassi et al., 2001; Fauser et al., 2004a; Yamanouchi, 2005), thus indicating certain retention of immaturity characteristics in some of these cells.

CD as a malformation due to perinatal and early postnatal brain injury

In addition to the prenatal hypotheses of CD formation, recent pathological studies on surgical or autoptic human tissue suggested a role for perinatal insults such as neonatal periventricular hemorrhage and early postnatal traumatic lesions in the formation of dysplastic lesions (Sarnat, 1987, 1991; Marin-Padilla, 1995, 1996, 1997, 1999; Lombroso, 2000; Marin-Padilla, 2000). Reports from surgical series suggest that up to one-third of patients with CDs have a history of a perinatal complication (Palmini, 1994; Widdess-Walsh et al., 2005). The development of CD may not be restricted to the perinatal period: Marin-Padilla et al., 2003 reported pathological changes consistent with CD in the brains of two patients who survived 7 and 9 years an episode of violent shaking (shaken infant syndrome) early in their lives. These changes were similar to those found rat receiving in utero radiation (Marin-Padilla et al., 2003). The authors suggested that in the two children undamaged and/or partially damaged cortical regions survived the original insult and underwent postinjury reorganization that transformed the residual cortex structural and presumably functional organizations. The results of the postinjury reorganization included progressive CD with cytoarchitectural disorganization, laminar obliteration, morphologic, and functional (synaptic reorganization) transformation of some neurons, preservation of layer 1 intrinsic fibers and Cajal-Retzius cells, and the presence of large (hypertrophic) intrinsic neurons with intense neurofilament immunoreactivity (Marin-Padilla et al., 2002). This then was an example of a rat model confirming the human data.

CDs may not be static abnormalities

Previous reports have suggested that BC or BC-like cells exhibit phenotypic characteristics of immature or stem cells (Crino et al., 1997; Cepeda C, 2003; Ying, 2005). These studies showed that BCs are heterogeneous cell populations expressing cell surface markers for pluripotent stem cells and proteins for multipotent progenitors, or immature neurons/glia. The presence of stem cell/progenitor markers in the BCs could be due to a persistent postnatal neurogenesis or early embryonic insult that resulted in arrest of proliferation/differentiation at their early stages (Ying, 2005). Based on these data, it may be hypothesized that the presence of spontaneous neurogenesis associated with some forms of human CD may represent an erroneous and maladaptative mechanism for neuronal circuitry repair. However, since controlled longitudinal studies in patients with CD are not feasible, there is a need for new animal models of true CD to test these hypotheses and to examine the various underlying cellular and molecular mechanisms.

Mechanisms of epileptogenicity in CD

Although previous studies linked dysplastic lesions in patients with pharmacoresistant epilepsy with in situ epileptogenicity (Morioka et al., 1999; Boonyapisit et al., 2003; Matsumoto et al., 2005; Otsubo et al., 2005), much evidence has suggested to many that not all dysplastic lesions are epileptic. Only parts of some dysplastic lesions show in situ epileptogenicity (Marusic et al., 2002; Boonyapisit et al., 2003). Moreover, data from various animal models of CD show that the majority of these models are not epileptic despite in vitro evidence of increased excitability and/or decreased inhibition. These data lead to the following questions:

• 1
  What are the mechanisms of epileptogenicity of CD?
• 2
  Why is epileptogenicity expressed in some lesions but not others?
• 3
  Why is epileptogenicity expressed at various postnatal ages despite the assumed presence of CD since birth? Is there a role for a “second hit” in the expression of epileptogenicity in CD?

Mechanisms of epileptogenicity in CD

There is an extensive modification of neurotransmitters receptors in CD: For example, NR2A/B, NR2B and NR1-1a, 1b, 2a, 2b and GluR2/3 are increased in FCD tissue, more specifically in dysplastic neurons (Ying et al., 1999; Najm et al., 2000; Crino et al., 2001; Ying and Najm, 2002; Avoli et al., 2003; Crino, 2004; Ying et al., 2004; Moddel et al., 2005; Yamanouchi, 2005) (Fig. 4). A potential functional role of the NMDA receptor NR2B subunit in the expression of epileptogenicity in CD was recently validated in vitro using freshly resected brain slices (Moddel et al., 2005). In contrast, Andre et al. (2005) describe that NR2B expression is decreased in some dysplastic (cytomegalic) neurons dissociated from CD tissue of young children. Other glutamate receptor subunits are distributed equally both in CD and normal cortex, such as NR1-3a, 3b, and 4a (Ying and Najm, 2002). Increased or decreased expressions of glutamate and GABA receptors subunits mRNA have also been described (Crino et al., 2001). This abnormal expression of glutamate receptors in CDs may be part of the mechanism of epileptogenesis in this tissue.

Figure 4. Photomicrographs of cresylecht violet stained section from (A) nonepileptic normal cortex and (B) epileptic dysplastic type 2A cortex with misshapen dysmorphic neurons. Immunocytochemical staining using NR2B antibodies showing (D) immunopositivity in the dysplastic/epileptic region but not in the normal cortex (C) (Scale bar: 100 μm).

Inhibitory neuronal cytoarchitecture is also disrupted in dysplastic tissue, with absence of chandelier-like and hypertrophic basket-like terminals, but results are variable among patients (Alonso-Nanclares et al., 2005). Histological quantification in CD versus non-CD tissues shows that the dysplastic region has larger numbers of asymmetric (excitatory) synapses than nondysplastic regions (Marco and DeFelipe, 1997).

The two studies are different in that only pediatric tissue was used in UCLA group's study compared to tissue resected from a mixed patient population of adults and children in our study (Moddel et al., 2005). In addition, the neurons that were described as showing the lowest expression are the meganeurons but not the dysplastic (dysmorphic) neurons that we studied. These discrepant results point to the importance of common understanding of the phenotypical characteristics of various types of neurons that may be expressed in various types of CD.

In vitro electrophysiological studies of dissociated neurons from freshly resected CD tissue from children shows that normal appearing neurons have characteristics of neurons from control cortex, but dysmorphic neurons have lower capacitance, increased resistance, and decreased time constant (Cepeda, 2003). On the other hand, “giant” neurons are more excitable, as they present increased capacitance, decreased resistance, and longer time constant (Cepeda, 2003). BCs are not excitable, but it is hypothesized that their presence may lead to modification of the structure of surrounding cortex, thus resulting in a net excitability of the tissue (Cepeda, 2003). Initial studies using modern techniques, such as cDNA array and proteomics, report changes in various molecules in CD as compared to normal cortex (Kim et al., 2003; Eun et al., 2004). These molecules are associated with neurogenesis, apoptosis, migration, and antioxidant processes, among others (Kim et al., 2003; Eun et al., 2004). Further studies are needed to validate the roles these pathways in the expression of epileptogenicity in some forms of CD.

Is there a role for glial cells in CD-induced epileptogenicity?

Studies that analyze glia in epileptic tissue point to an important role of those cells in seizure initiation and maintenance (Dudek et al., 1998; Steinhauser and Seifert, 2002; D'Ambrosio, 2004). Gliosis is frequently found in epileptic tissue (Ribak, 1986; Dudek et al., 1998). There is evidence that glia in epileptic tissue is not able to maintain K⁺, Na⁺, and Ca⁺⁺ homeostasis (Grisar et al., 1983b, 1983a; Grisar, 1984; Grisar, 1986; Dudek et al., 1998; Grisar et al., 1999; Steinhauser and Seifert, 2002; D'Ambrosio, 2004). Glial cells may also have their electrophysiological characteristics modified, suggesting inability to maintain ionic balance in epileptic tissue, thus may be contributing to its hyperexcitability (Bordey et al., 2001). In addition, glial dysfunction may be at the basis of decreased glutamate clearance mechanisms as the glial glutamate transporters (GLT1) have been shown to play a role in the expression of in vivo epileptogenicity ion animals: GLT-1 knockout mice show evidence of spontaneous seizures and in vivo epileptogenicity (Rothstein et al., 1996; Tanaka et al., 1997).

The differential expression of epileptogenicity within some dysplastic lesions

As stated above, the center of BC-rich areas does not exhibit ictal EcoG patterns (Marusic et al., 2002; Boonyapisit et al., 2003) (Fig. 5). These findings raise the important question on the role of BCs in the lack of expression of epileptogenicity in some subtypes of CD: Could these cells play a protective role against epileptogenicity? Further characterization of BCs and their role in epileptogenicity in some types of focal CDs is needed. In addition, direct EcoG recordings from MRI-identified dysplastic regions show that epileptogenicity is mainly in the surrounding mostly MRI normal cortex. These findings suggest that a successful surgical treatment would not be possible without the extension of the resection to the surrounding epileptic cortex rather then to be at times limited to a rather small part of the dysplastic region.

Figure 5. Cartoon showing the differential expression of epileptogenicity in dysplastic regions that are devoid of balloon cells (Type 2A) but not in BC-rich regions (Type 2B).

Why is epileptogenicity expressed at various postnatal ages despite the assumed presence of CD since birth? Is there a role for a “second hit” in the expression of epileptogenicity in CD?

Previous studies on the natural history of CD report that patients who develop epilepsy do so at various ages. Some patients may develop epilepsy during the early postnatal period and others exhibit their first seizures later on in life (Mischel, 1995; Raymond et al., 1995; Sisodiya, 2004; Widdess-Walsh et al., 2005). In one of the retrospective studies, a history of head trauma or some form of CNS infection was reported in a quarter of these patients (Widdess-Walsh et al., 2005). The effect of a “second hit” was recently tested on the Eker rat, which carries a spontaneous germline mutation of the tuberous sclerosis TSC2 gene (TSC2±). Early postnatal irradiation (“second hit” stimulus) led to lower seizure thresholds (latencies to flurothyl-induced seizures) and the expression of dysplastic cytomegalic neurons and giant astrocyte-like cells, similar to cytopathologies observed in TSC lesions of patients (Wenzel et al., 2004).

Assuming that the majority of the dysplastic lesions are formed during embryonic brain development or during the perinatal period, these human and animal model observations lead to the following questions: (1) Is there a need for a “second hit” before the transformation of a preexisting “dormant” lesion into an active/epileptogenic focus?, and (2) what are the mechanisms that underlie the transformation of these dormant lesions into active/epileptic foci?

RESEARCH ON ANIMAL MODELS OF CD

Why do we need animal models of CD?

As discussed above, most of the studies on human tissue resected from patients with pharmacoresistant epilepsy due to CD are rather descriptive in nature, and as new hypotheses are continuously generated, there is a need for an in vivo (and at times in vitro) validation of these results using appropriate animal models. The main questions that need to be answered through the use of animal models of epilepsy include: (1) How does CD develop (pathogenesis of CD)? (2) What are the cellular and molecular changes that lead to the development and expression of epilepsy (mechanisms of epileptogenesis)? and (3) What are the mechanisms of epileptogenicity? In addition, the use of animal models should serve as an essential link between the in vitro results and the patient leading to the introduction of novel diagnostic and therapeutic modalities for the management of patients with epilepsy. This need is highlighted by the failure of almost all antiepileptic drugs to deliver satisfying results of complete seizure control without adverse effects in a significant number of patients with epilepsy (Kwan and Brodie, 2000; Jacobs et al., 2001). Some of the animal models introduced to date have proven to be important for our understanding of the basic mechanisms underlying epileptogenesis, epileptogenicity, and have been valuable in the screening of novel antiepileptic drugs (Sarkisian, 2001).

In order to achieve these goals there is a need for clinicians to clearly state the desired characteristics for new animal models. These may include: (1) Reproduction of the human pathology (i.e., various pathological subtypes of human CD) and, (2) the expression of in vivo epileptogenicity (e.g., spontaneous and predictable epileptic seizures). Animal models should also exhibit electrophysiological patterns that closely resemble the dysplastic pathologies in humans.

Overview of some animal models of cortical malformations

Many brain malformations and abnormalities are caused by genetic and environmental factors. Several animal models of CD have been previously described: genetic, fetal insults, and neonatal lesion models (Sarkisian, 2001).

Genetic models of CD

Although models that clearly mimic human malformations are still limited, some genetic models seem to reproduce some rare dysplastic pathologies. The telencephalic internal structural heterotopia (TISH) model describes a mutant rat exhibiting a forebrain anomaly resembling the rare human neuronal migration disorder of double cortex. The bilateral heterotopia is prominent below the frontal and parietal neocortices, but is rarely observed in temporal neocortex. The cells in the heterotopia exhibit a “rim-to-core” neurogenetic pattern rather than the characteristic “inside-out” pattern observed in normotopic neocortex. Studies have shown that some of the animals with the TISH phenotype exhibit spontaneous recurrent electrographic and behavioral seizures (Lee et al., 1997).

Other genetic models include the Reeler mutant rats, and Ihara mutant rats (Amano et al., 1995, 1996, 1999; Takahashi et al., 2000; Ross, 2002; Arai et al., 2003; Frotscher et al., 2003).

Difficulties encountered in the studying of such models are the determination of how mutations in diverse sets of genes ultimately lead to alterations in brain excitability (Sarkisian, 2001). Moreover, spontaneous seizures were observed only in genetic models that showed bilateral or diffuse lesions (Lee et al., 1997).

Fetal insult models of CD

These models include those induced by administration of methylazoxymethanol acetate (MAM) and fetal irradiation of pregnant rats. Both these methods induce multifocal CD in newborn rats (Roper et al., 1995, 1997; Roper, 1998; Baraban et al., 2000; Castro et al., 2001; Kondo et al., 2001; Calcagnotto et al., 2002; Castro et al., 2002; Marin-Padilla et al., 2003; Kellinghaus et al., 2004).

In rats, prenatal exposure to methylazoxymethanol (MAM) consistently results in offspring with multifocal brain malformations, microcephaly, cortical thinning, loss of lamination and clusters of displaced neurons in the hippocampus, i.e., hippocampal heterotopia (Colacitti et al., 1999; Baraban et al., 2000). These animals have an increased sensitivity to various proconvulsant agents (de Feo et al., 1995; Baraban and Schwartzkroin, 1996; Germano et al., 1998; Chevassus-au-Louis et al., 1999). Isolated mini-slices containing heterotopia are capable of independent burst generation in vitro (Baraban et al., 2000) and seizure activity induced in hippocampal slices from MAM-exposed rats is resistant to commonly available antiepileptic drugs (Smyth et al., 2002). Although these animals exhibit salient features of the clinical condition, and have played an important role in determining how a malformed brain can generate abnormal electrical discharge, our overall understanding of hippocampal heterotopia remains limited. As for most animal models of CD, there is a lack of definite data on the spontaneous epileptogenicity in MAM-induced CD, although recent data indicate that a small percentage of these animals (<20%) exhibit spontaneous seizures arising from the hippocampus or the neocortex (Harrington et al., in press).

The in utero irradiation rat model of CD: The in utero irradiation of pregnant rats leads to the development of various degrees of cortical malformations and architectural abnormalities in the neocortical areas that are similar to those seen in some forms of CD in humans (Roper et al., 1995; Kondo et al., 2001). Hippocampal formation abnormalities also involve cell dispersion in the CA1 and to a lesser extent in CA3 (Roper et al., 1995; Kondo et al., 2001). The significance of these hippocampal abnormalities remains unclear as a similar pattern in human is yet to be described. Dyslamination and lack of columnar organization are seen in multiple areas of the neocortex as well as clustering of neurons in the molecular layer. Studies also showed that the cortical malformations inflicted in these rats would cause interictal epileptic discharges in a large number of the animals and spontaneous seizures in a much smaller percentage of these rats as compared to normal controls that is estimated around 10–20% (Kondo et al., 2001; Kellinghaus et al., 2004). Moreover, higher doses of in utero radiation leads to more severe pathological changes (microcephaly, absence of corpus callosum, and more diffuse and severe cortical dysplastic abnormalities), but these rats exhibit no spontaneous seizures. In vivo epileptogenicity increased with mild to moderate radiation doses, but occurred less frequently in the high-dose radiation group. There is a positive correlation between the radiation doses the rats received in utero and the severity of the cortical and hippocampal disorganization (Jensh, 1987; Fukui, 1991; Fushiki, 1996; Miki, 1999; Kellinghaus et al., 2004), suggesting a nearly linear dose–response curve regarding the extent of histological abnormalities seen in the brain postnatally. It has been shown that migratory cells are the most radiosensitive (Altman, 1968). In the rat, most of the pyramidal cells destined for layer V are generated at E16 and E17 (Ignacio, 1995) and are migrating to the cortical plate at the time of radiation (Ferrer, 1993b), whereas layer II/III neurons are generated later and are likely to be trapped near their periventricular origin due to radial glia damage. Therefore, it is not surprising that disorganization of both hippocampus and cortex (including subcortical heterotopias) increases with increasing radiation doses. It is interesting to note the dissociation between not only the severity of pathology and the expression of spontaneous epileptogenicity but also between in vitro and in vivo electrophysiological results: extracellular recordings from neocortical slices of rats treated with 225cGy on E17 showed evidence of hyperexcitability (Roper, 1997). Single-cell recordings from the dysplastic neocortex (Roper, 1997) showed reduced inhibition in that model that was correlated with a reduction of parvalbumin- and calbindin-reactive (i.e., probably inhibitory) neurons in the dysplastic normotopic neocortex (Roper, 1999). Therefore, changes in the inhibitory mechanisms may significantly contribute to the epileptogenicity. However, rats exposed to 175cGy on E17 with histological findings resemble most closely the model used for the in vitro studies (225cGy on E17) showed only rare epileptiform discharges and no spontaneous seizures.

At this point there has been no clear explanation of the dissociation between pathology and epileptogenicity in some animal models of CD. Therefore, the question remains: would these animals become epileptic under the right circumstance? If so, what are the conditions that may transform a potentially epileptogenic pathology to an epileptic phenotype? Data from human studies suggest that some patients with congenital/perinatal dysplastic lesions do not express epilepsy till later in life and if they do, epilepsy appears after some type of a trigger. In order to address these questions we recently studied the in vivo epileptogenicity in rats exposed to in utero radiation (Oghlakian, 2006): The majority of the radiated rats did not display spontaneous epileptic activities, but a second hit (a single treatment with a low dose of the proconvulsant agent pentylenetetrazole, PTZ) in rats that did not develop spontaneous ictal patterns rendered the majority of dysplastic rats (but not age matched control rats) into epileptic ones as these rats started to exhibit spontaneous epileptic seizures after a single injection of PTZ (Oghlakian, 2006). These results are interesting as they mirror the natural history in a significant number of patients with CD (thought to be due to prenatal/congenital or perinatal insults) in whom the epileptic phenotype does not develop till an otherwise nonepileptic stressor such as trauma, acute infection, stress, sleep deprivation … leads to the transformation of a nonepileptic pathology into an epileptic phenotype. As the mechanisms of this transformation from a dormant pathology into an epileptic phenotype are unknown further studies are needed.

As the pathological abnormalities in the in utero injury models are diffuse it is difficult to use these animals for studies that aim to directly correlate intrinsic epileptogenicity with histopathological/cellular, and/or molecular changes but these models are of interest for the studying of the dissociation between pathology and epileptogenicity and for the studying of the mechanisms underlying the transformation of nonepileptic dysplastic lesions into epileptic ones.

Neonatal lesion models

The hypothesis of an association between neonatal injuries and development of CD has been supported by extensive experimental data. Cowen et al. (1970) first reported that a closed head injury in rat pups could result in focal microdysgenesis. Thereafter, several studies reported on the occurrence of various types of malformations in the cortex of neonatal rats (up to 3–4 days postnatal) as a result of different types of injuries that include freeze lesions, laser and ionizing radiations, electrocoagulation, electroshock, focal injection of ibotenate, focal aspiration, and punctures (Dvorak and Feit, 1977; Ferrer and Catala, 1991; Humphreys et al., 1991; Suzuki and Choi, 1991; Ferrer, 1993a; Ferrer et al., 1993; Marret et al., 1995). The description of the early models focused mainly on the pathological characterization of the lesions with no attempt for electrophysiological studies till the second half of the last decade. Therefore, the functional significance (from an epilepsy perspective) of most of these elegant pathological models remains unknown.

One of the models of focal CD in the rat that was extensively studied is the neonatal freeze lesion model (Prince et al., 1997): The application of a deep-freeze probe to the skull of newborn rats results in a dysplastic lesion similar to polymicrogyria with spontaneous in vitro epileptogenicity in the immediate vicinity of the lesion (Jacobs et al., 1996; Prince et al., 1997; Jacobs et al., 1999). Although neocortical freeze lesions show histological characteristics of some forms of CD, and brain slices show in vitro hyperexcitability (Jacobs et al., 1996, 1999; DeFazio and Hablitz, 2000; Hablitz and DeFazio, 2000), recent studies showed a lack of epileptogenicity in vivo: Holmes et al. (1999) reported no significant differences in seizure susceptibility, as measured by after discharge threshold and kindling rate, between animals subjected to neocortical freeze lesions (1 or 3 lesions) at birth and age matched control animals. Long-term EEG monitoring has failed to find in vivo epileptiform discharges or behavioral seizures in this model (Kellinghaus, in press).

This illustrates some of the main problems with some animal models of epileptic pathologies where there is dissociation between pathology and electrophysiology.

Despite extensive studies, it is safe to say that most of the animal models using various species (that ranged from drosophila to primates) to this day have failed to meet the validation criteria as defined above. Further research is needed to determine the effect of a “second hit” on the potential transformation of various models of “dormant” pathology into an active (spontaneous) epileptic pathologies. These studies would reproduce some of the features that are characteristic of the natural history of some forms of human pathology and would ultimately lead to the identification of some of the cellular mechanisms of epileptogenesis and the future design of “preventative” interventions.

CONCLUSION

Despite more than 50 years of extensive research using human tissue and various animal models of epileptic pathologies and seizures there remain many gaps that require to be filled in order to make headways towards better evaluation and treatment of epilepsy:

• 1
  A significant percentage of patients with pathologically proven neocortical lesions (at least 25% of the patients who undergo surgical resection for CD) have normal preoperative high-resolution MRI studies. An improvement in conventional MRI technology may lead to a better visualization of some of these potentially epileptic lesions. In addition, there is a need for imaging technology with cellular and molecular resolutions.
• 2
  Recent studies using direct electrocorticographic recording techniques report on a variable correlation between MRI-identified pathology and in situ epileptogenicity: The center of some dysplastic lesions with significant imaging abnormalities is at times less epileptic than the histological borders the same lesions where the MRI does not reveal anatomical and/or signal abnormalities. This illustrates the possible dissociation between MRI pathology and epileptogenicity, and highlights the need for molecular imaging technology for the visualization of the epileptic neuron and not necessarily the morphologically abnormal neurons or regions. Thus, the importance of an identification of the molecular substrates of epileptogenicity through careful collaborative work between human tissue epilepsy researchers, and investigators using animal models of seizures and epilepsy.
• 3
  As many animal models have been described but were not fully characterized, a careful characterization of these animals using state of the art clinical evaluation techniques (imaging and electrophysiological), and novel research methods is warranted. This would result in the availability of “better” animal models to specific epileptic pathologies that will be used for the validation of in vitro human tissue results and later on the testing of novel therapeutic and diagnostic approaches prior to their introduction in humans.

Acknowledgments

Acknowledgment:  This work was supported by National Institutes of Health Grants K08 NS-02046 and R21 NS42354 to Imad M. Najm.