Stroke refers to the brain injury that occurs following cerebral ischemia (Sharp et al., 1998). It has been suggested that acute ischemia increases extracellular concentrations of glutamate (Buchkremer-Ratzmann et al., 1998), reduces GABAergic function, and causes functional or structural impairment of GABAergic interneurons (Menon and Shorvon 2009). In addition, injury from stroke also causes selective cell death and apoptosis, changes in membrane potential properties, mitochondrial changes, receptor changes, deafferentation or collateral sprouting (Menon and Shorvon 2009). Together these impaired functional and structural relationships may lead to epileptogenesis- a process that causes a permanent neuronal plasticity change in previously normal brain tissue, leading to the onset of spontaneous recurrent epileptiform discharges (SREDs) (DeLorenzo et al., 2007). Certainly, this period of epileptogenesis represents an important clinical time-frame wherein intervention may be able to inhibit the plasticity changes that lead to the development of seizures.

Development of an in vivo model of stroke-induced AE has met limited success (Karhunen et al., 2005). Our lab has utilized the in vitro hippocampal neuronal culture model of stroke-induced AE (Sun et al., 2001; Sun et al., 2004). The in vitro model utilizes glutamate exposure to produce an injury similar to that seen secondary to ischemic stroke. Subsequent to injury, surviving neurons manifest SREDs analogous to epileptic seizures (Sun et al., 2001; Sun et al., 2004). While this model provides insight into some of the molecular mechanisms involved in epileptogenesis following stroke, it is limited by lack of normal anatomical morphology and circuitry that is important in excitatory feedback in the brain. In contrast, animal models provide proper morphology and neuronal feedback, but they are often time consuming and cost restrictive for use in rapid screening of novel therapeutic compounds. Some of these limitations could be overcome while still using an in vitro system by utilizing organotypic hippocampal slice cultures (OHSCs). OHSCs have been shown to manifest intact neuronal morphology, cellular and anatomical relations and network connections (Noraberg et al., 2005; Sundstrom et al., 2005; Zimmer and Gahwiler 1984). OHSCs have been used to study the acute physiological effects (Albus et al., 2008; Wahab et al., 2010) and some of the morphological changes that occur following excitotoxic injury (Routbort et al., 1999; Thomas et al., 2005). Glutamate toxicity is an important aspect of the ischemic cascade (Buchkremer-Ratzmann et al., 1998). OHSC models have also established similarities in cell death patterns in oxygen glucose deprivation and glutamate injury, suggesting NMDA mediated cell death in both injuries (Lipski et al., 2007, Noraberg et al., 2005). However, a thorough characterization of the physiological changes that occur after excitotoxicity particularly its effect on seizure genesis has not been explored.

The seizure events seen in extracellular recordings expressed many characteristics of overt electrographic epileptic seizures. Seizure events started and terminated spontaneously and were synchronized in nature, as they represented a population of synchronized neurons (figure 3). The seizure events produced by excitotoxic glutamate injury manifested larger spike amplitude than would be seen if the activity originated from a single neuron, suggesting that the activity occurred in a group of neurons. The typical seizure pattern observed in OHSCs after glutamate injury also included an abrupt onset of activity superimposed on a large field potential shift, consistent with the classic paroxysmal depolarizing shift associated with electrographic seizure discharges in both in vitro and in vivo models of epilepsy (Bausch et al., 2006; Dichter and Ayala 1987). Finally, the seizures produced by glutamate injury responded to the anticonvulsant drugs phenobarbital and phenytoin, but not to ethosuximide (figure 7). These results demonstrated that OHSCs subjected to injury by glutamate exposure could be transformed into neuronal networks manifesting seizures for the life of the culture, producing an in vitro model of epilepsy.

Various studies of OHSCs have confirmed that cells develop and mature similar to age-matched cells in vivo (Bahr et al., 1995, Stoppini et al. 1991). In this respect, OHSCs are often used at times comparable to the appropriate in vivo age. Our OHSCs are cultured at P8 and injured at DIV 21, thereby showing the equivalent cellular maturity of young adult rats at P29. OHSCs from P8 pups have been shown to develop mature synaptic properties within a few days in culture (Muller et al., 1993). In addition, OHSCs have been shown to respond to seizure inducing stimuli over a wide range of time points anywhere from 7-56 days in culture (Albus et al., 2008). We therefore believe our slice cultures are at an appropriate age to respond to a glutamate injury at DIV 21.

Organotypic hippocampal slice cultures have been utilized in many models of excitotoxicity, including kainic acid (Routbort et al., 1999), NMDA (Ring et al., 2010), and oxygen glucose deprivation injuries (Lipski et al., 2007). However, few of these studies have examined the long-term physiological changes that occur in OHSCs after excitotoxic injury, thereby exploring their potential as a model of acquired epilepsy. Recently, Bausch and McNamara (Bausch and McNamara 2004) used a kainic acid injury paradigm in OHSCs and found no significant difference in seizure rates from controls when recording from granule cells in physiological buffer 30 days after injury. They were however, able to show that the granule cells were more hyperexcitable and there was considerable mossy fiber sprouting in the injured slices. Various other studies have used excitotoxic injuries to study neuroprotective agents and morphological changes (Boscia et al., 2006; Cho et al., 2007; Lipski et al., 2007; Ring et al., 2010; Routbort et al., 1999; Thomas et al., 2005). Our model takes this a step further by characterizing the physiological changes after an excitotoxic injury with glutamate. A study by Lahtinen et al. (Lahtinen et al., 2001) utilized a higher concentration of glutamate (10 mM) to examine acute electrophysiological consequences of glutamate injury in OHSCs. Within 2 hours of the injury, they found increased hyperexcitability in the CA3 cell region. In addition, they reported attenuation of cell death with TTX following glutamate injury, indicating that some cell death during the initial 24 hour period may be caused by neuronal excitability following the initial injury. Indeed, clinical and in vivo studies have shown that seizures often occur acutely following an ischemic injury (Menon and Shorvon, 2009, Karhunen et al., 2006), but this does not necessarily lead to epileptogenesis and acquired epilepsy. This would suggest that the post-insult activity observed by Lahtinen et al. is not necessarily indicative of chronic epileptogenesis. In contrast, our study employs a more moderate glutamate injury (3.5mM glutamate for 35 minutes) and we examine cell death at both 24 hours and 72 hours. We establish that cell death subsides following the 24 hour period, as the PI uptake is significantly decreased at 72 hours. In addition, we examine epileptogenesis in OHSCs 9-12 days following glutamate injury indicating long lasting changes in neuronal excitability following glutamate injury. It is important to stress this issue, since our study is the first demonstration of the development of spontaneous recurrent epileptiform discharges (epilepsy) in the OHSCs and represents the first model of post injury acquired “epilepsy” in this model.

Our model utilizes glutamate as a mode of excitotoxicity, as excessive glutamate concentration is an important aspect of the ischemic prenumbra in stroke (Davalos et al., 1997). The involvement of glutamate in epileptogenesis has been implicated in whole animal, (Croucher et al., 1988; Croucher and Bradford 1990; Rice and DeLorenzo 1998) slice, (Anderson et al., 1990; Stasheff et al., 1989) and cell culture (Sombati and Delorenzo 1995; Sun et al., 2001) models of epilepsy. To induce epileptogenesis, these models all used continuous neuronal spiking produced by seizures, (Rice and DeLorenzo 1998) repeated high-frequency excitation (Croucher et al., 1988; Croucher and Bradford 1990; Stasheff et al., 1989), or low extracellular magnesium environments (Anderson et al., 1986; Sombati and Delorenzo 1995) while only one used a glutamate-induced prolonged, reversible depolarization as used in this study (Sun et al., 2001; Sun et al., 2004). Many of these models have implicated activation of the N-methyl-D-aspartate receptor (NMDAR) for epileptogenesis (Croucher et al., 1988; Croucher and Bradford 1990; DeLorenzo et al., 1998; Rice and DeLorenzo 1998; Stasheff et al., 1989). Interestingly, epileptiform discharges have also been produced by growing OHSCs in culture in the presence of tetrodotoxin (TTX) to block activity or D(–)-2-amino-5-phosphonopentanoic acid (D-APV) to block NMDAR activation (Bausch et al., 2006). Removal of these agents resulted in the expression of seizure-like activity (Bausch et al., 2006). This distinctly different culture model uses the inhibition of glutamate receptors to induce hyperexcitability. Control OHSCs displayed a low occurrence of seizures, possibly as a result of collateral synaptic connections that are known to form after long term culture of hippocampal tissue (Bausch and McNamara 2000; McBain et al., 1989). Although the mechanism producing hyperexcitability in this model has not been fully delineated, it has been shown that inhibition of glutamate receptors in neurons in culture produces alterations in NMDAR subunit expression that are regulated by synaptic activity during development (Hoffmann et al., 2000; Yashiro and Philpot 2008). It is possible that alterations in NMDAR subunit expression may underlie the development of hyperexcitability in this model. Though glutamate exposure may induce changes in receptor subunit expression in the glutamate injury–induced epileptogenesis model, these potential changes probably occur through a separate mechanism. It will be interesting to investigate the mechanism of glutamate injury-induced epileptogenesis in future studies.

The potential role of selective neuronal death in glutamate injury–induced epileptogenesis requires further investigation, especially in light of the fact that inhibitory neurons are typically less vulnerable to excitotoxicity than excitatory neurons (Tecoma and Choi 1989). Although differential cell death may affect the balance between the number of inhibitory and excitatory neurons, resulting in a larger number of surviving inhibitory neurons, (Tecoma and Choi 1989) the glutamate-induced injury produced “epilepsy” in the OHSCs despite the potential alterations in neuronal subpopulations. Our experiments suggest that the severity of injury is not a large factor in epileptogenesis, as there was no difference in the degree of cell death in those OHSCs that developed seizures and those that did not (figure 6). Further studies are needed to determine the role of selective cell death in this model. In addition, the possible roles of gap junctions (Dudek et al., 1998), ischemia-induced alterations in second-messenger systems, and gene changes (Morris et al., 2000) in mediating epileptogenesis represent important future directions for research that can be conveniently studied in this system.