Abstract
Epilepsy affects around 1 % of the population, and up to a third of patients are resistant to antiepileptic drugs (AEDs). Epilepsy is defined as an enduring predisposition of the brain to generate seizures, and the disorder encompasses more than forty different syndromes, and many more electrographic patterns. Current pharmacological treatments are symptomatic – they prevent the ultimate expression (the seizure) of an often unknown brain dysfunction. Only two of the ~20 AEDs currently on the market are results of rational target-based drug design, despite many efforts to develop drugs that reduce neuronal excitation in the brain. This lack of success is a consequence of our incomplete understanding of the mechanisms of ictogenesis (seizure generation) in the brain. The detailed study of ictogenesis, and the involvement of potential drug target proteins, requires reduced model systems that can generate seizure-like events (SLEs). Brain slices of hippocampus and cortex can maintain such activity in vitro, but the experimental reproducibility has proven to be unsatisfactory. Most in vitro research has therefore focused on short (<200 ms) epileptiform events resembling interictal activity, which can be readily induced in slices but have dubious relevance in ictogenesis.
In this thesis, I will evaluate the historical role of brain slice seizure models, and argue that poor reproducibility of existing models have prevented their effective incorporation in research on AEDs and ictogenesis. Then, I will present experimental work describing a novel slice model using a simple archetypical cortical network – the olfactory bulb (OB). This system produced recurrent minute-long SLEs in virtually 100 % of slices, when exposed to low magnesium conditions. These events were analogous to the tonic phase of focal seizures in the cortex and hippocampus, which consists of a long-lasting negative shift in the field potential accompanied by sustained depolarisation of output neurons. SLEs in the OB could be recorded from single output neurons using whole-cell current clamp recording, and their stereotypical patterns of action potential firing could be recorded extracellularly. SLEs were dependent on fast glutamate neurotransmission and persistent Na+ currents, became more severe under blockade of inhibitory neurotransmission or glutamate uptake, and were sensitive to the clinical AED phenytoin. They recurred regularly at ~0.01 Hz for hours, until the ictogenic medium was washed out, and could be induced repeatedly without any apparent rundown. The SLEs propagated non-synaptically through the external plexiform layer, in a manner consistent with lateral diffusion of extracellular potassium, as occurs in the hippocampal low-Ca2+ model of non-synaptic seizures.
The second part of the thesis addresses the possible treatment of seizure disorders with antidepressant serotonin reuptake inhibitors (SSRIs). The experimental work was done using whole-cell and extracellular recordings in the OB low-Mg2+ seizure model, and field potential recording in the hippocampal low-Ca2+ model. SSRIs are widely used as antidepressants, and there is a wide-spread erroneous belief that SSRIs are proconvulsant. There is existing evidence from animal models that SSRIs are in fact anticonvulsant, and this action has been assumed to depend on serotonergic action. I found that the SSRIs fluoxetine and citalopram abolished SLEs in the OB. This anticonvulsant action was not dependent on serotonin neurotransmission. Instead, I found that the SSRIs potently inhibited sodium channels, at concentrations and time courses corresponding to their anticonvulsant action. SSRIs were also anticonvulsant in the hippocampus, suggesting that the finding may have wide clinical implications.
I suggest that the olfactory bulb may provide an unusually robust model of acute seizures, particularly suited for the study of non-synaptic propagation mechanisms and pharmacological AED studies. Furthermore, the results suggest that SSRIs exert an anticonvulsant action in the OB and hippocampus, and that this most likely occurs through sodium channel blockade rather than via serotonin reuptake inhibition.