Building epileptiform activity from hippocampus cells models

PosterEpilepsy is one of the most common neurological disorders, characterized by recurrent seizures. The pathophysiological interpretation of specific electrophysiologicial patterns recorded during interictal (outside seizures) or ictal (seizures) periods is of crucial importance as electrophysiologicial recording (scalp and intracerebral EEG) remains one of the gold standard technique in the investigation of drug-resistant epilepsy. The general objective of this work is to bridge between single unit activity (cellular level) and field activity (network level) using computational modeling. We currently focus on the role of the hippocampus (CA1 and CA3 sub-fields) often involved in Temporal Lobe Epilepsy. Our approach is based on the use of “minimal” but still biologically-relevant neuron models. For pyramidal cells of CA1, we elaborated a two-compartment model by reducing a more complex model [Poirazi and al, 2003]. This simplification consists in two steps. First, we retained the main active voltage gated currents (related to sodium, potassium and calcium channels) known to be involved in transitions from single spike (normal) to burst activity (epileptic).This selection was performed on the basis of a literature review about experimental models of epilepsy. Second, we reduced the number of compartments from 183 to 2 (somatic and dendritic). For CA3 pyramidal cells, we implemented the reduced Traub model [Traub et al, 1991] proposed by Pinsky and Rinzel [Pinsky and Rinzel, 1994]. This two-compartment representation uses the sodium and potassium delayed rectifier current in the soma and the calcium and potassium calcium-activated current into a dendrite-like compartment. Finally, as far as interneurons are concerned, only basket and OLM cells are considered [Wang and al, 1996, 2002]. These cells are represented using single-compartment models. Network simulations use AMPA, NMDA and GABA synapses. A comparison with experimental data (animal models) allowed us to evaluate our computational models. Results showed that the reduction still preserve electrophysiologically-relevant features. In particular, blocking A type potassium current leads the CA1 pyramidal model to switch from normal to burst activity, as in the 4AP (4- aminopyridine) experimental model. Synaptic transmission was also validated by connecting cells into networks and by analyzing emerging network activity (like synaptic spatio-temporal summation processes). Regarding computational resources, simple networks that included up to 4,000 cells were built. Satisfactory computation time was obtained on a standard PC (about a few minutes for 10 seconds of activity)