DGCs are highly inhibited, and their intrinsic excitability is much lower than that of hippocampal CA1 pyramidal neurons. DGCs are proposed to restrict spread of excitatory activity to neurons of the hippocampus proper, but impaired inhibition of DGCs could allow seizure spread into the hippocampus. The GABAergic inhibition of DGCs is compromised in animals in SE [50–52]; the frequency of synaptic currents is reduced, and their amplitude is also smaller. In addition, GABAergic inhibition of CA1 pyramidal neurons of animals in SE is also attenuated [53–55]. The entorhinal cortex provides substantial input into the hippocampus. Thus, these alterations likely allow activity from the entorhinal cortex to spread into the hippocampus and follow a reentrant loop that sustains seizure activity.

Biochemical studies have found reduced surface expression of γ2 subunit-containing GABA receptors, which are clustered at the synapses in the hippocampi of animals in SE [51, 55]. Development of resistance to benzodiazepines, which are the commonly used first-line therapy for SE, is a significant clinical problem that is also replicated in animal models of SE [41, 51, 56, 57]. Benzodiazepines exert their action via γ2 subunit-containing GABA receptors. The reduced expression of functional γ2 subunit-containing GABA receptors thus helps to explain a rapid reduction in the efficacy of diazepam to terminate SE [50, 53, 54, 58].

GABA receptors are constantly trafficked between the surface membrane and the intracellular pool. In a cell culture model of SE, repetitive action potential firing can be induced by incubation of cultures in a medium lacking Mg²⁺ ions or containing high extracellular potassium and NMDA. In these systems, the reduced surface expression of γ2 subunit-containing GABA receptors is associated with accelerated internalization of these receptors [51, 56, 59]. Similar to whole animal experiments, blocking the activity of protein phosphatases is sufficient to prevent the functional down-regulation of synaptic GABA receptors in cell culture models of SE [56]. Thus, preventing mechanisms that destabilize receptors at the surface membrane may provide targets to restore GABAergic inhibition.

Concomitant with reduced inhibitory neurotransmission, excitatory neurotransmission of hippocampal principal neurons mediated by NMDA and AMPA receptors is strengthened during GCSE [60]. There is internalization of GluA2 subunit-containing AMPA receptors during SE, which likely contributes to the calcium influx and activation of cellular signaling associated with cell death and epileptogenesis. The excitatory neurotransmission mediated by NMDA receptors is also augmented during GCSE [60]. Synaptic NMDA receptor-mediated currents recorded from DGCs of animals in GCSE were larger and had altered kinetics of decay. The cell surface expression of NR1 and NR2B subunits of NMDA receptors is also increased in the hippocampi of animals in GCSE.

Taken together, the dynamic movement of GABA receptors from the surface membrane toward intracellular compartments coincides with the insertion of NMDA and AMPA receptors into the surface membrane. Under normal conditions, mechanisms that maintain homeostatic synaptic plasticity ensure that when the network activity is enhanced, excitatory neurotransmission is downscaled and inhibitory neurotransmission is increased, such that the target firing rate is restored. Studies in animal models of SE, however, indicate that a failure in maintaining homeostatic synaptic plasticity contributes to prolonged seizures. If excitatory neurotransmission through AMPA or NMDA receptors is blocked, the seizures terminate [60–62]. NMDA antagonists also enhance the efficacy of diazepam, and a combination of NMDA antagonists and diazepam can be effective in terminating even prolonged SE [63, 64].

The hippocampus is the region most susceptible to cell death following SE [41, 69, 70]. Extensive degeneration of CA3 and CA1 pyramidal neurons has been observed in different animal models of GCSE, whereas CA2 pyramidal neurons and DGCs are mostly preserved [65, 71–73]. In addition, middle layers of entorhinal cortex also suffer cell loss [74]. In these animal models, fluoro-jade, a marker of cell death, could be detected in hippocampal principal neurons as early as 3–4 days following SE. Additional cell loss may occur due to recurrent spontaneous seizures which appear following a seizure-free latent period of variable duration after SE.

GABAergic interneurons are also susceptible to neurodegeneration. Loss of inhibitory interneurons present in the hilus, particularly those expressing somatostatin, is observed in animal models of SE [75–78]. In contrast, parvalbumin- or cholecystokinin-positive interneurons are spared. This yields a complex modulation of inhibition, resulting in the net loss of the inhibitory inputs to the DGCs.

It is noteworthy that young animals are resistant to neurodegeneration following SE [79–81]. There is also a strong correlation between neurodegeneration following SE and the development of epilepsy, and young animals incurring minimal cell death following SE do not develop epilepsy [82].

It is reasonable to consider that systemic factors, such as elevated body temperature, and oxygen and glucose supply to the brain that are insufficient to cope with the increased demand of SE, may contribute to neurodegeneration. Cell death was still observed, however, in animals in which these parameters were maintained in the normal range during SE [83]. Instead, there is compelling evidence that glutamate-induced excitotoxicity plays a central role in cell death [84]. Canadian researchers reported that an acute neurologic syndrome, including seizures and SE, was due to ingestion of mussels from Prince Edward Island contaminated by domoic acid [85, 86]. Domoic acid is structurally similar to glutamate and probably activates glutamate-mediated excitation. Furthermore, blockade of NMDA receptors during experimental SE prevents cell death [87, 88]. Glutamate receptor-mediated calcium overload is likely to cause stress to mitochondria and endoplasmic reticulum, which also causes cell death [89, 90]. The breakdown of the blood–brain barrier during SE causes activation of glia and inflammation [91], which also appears to contribute to cell death.