Abstracts

Rationale: Ictogenesis is difficult to study experimentally because seizures are abrupt, unpredictable, and infrequent. Computational models of epileptic neural networks could serve as important tools for understanding ictogenesis, because all network parameters at the time of ictogenesis can be stored and then studied in detail. However, no distributed computational models exhibit spontaneous transitions between interictal and ictal activity. We describe such a model, which elucidated a new candidate mechanism for interictal to ictal transitions. Methods: The model is comprised of 100 x 100 MacGregor (integrate and fire) pyramidal cells, interspersed with a 20 x 20 array of interneurons. Excitatory synaptic connections undergo activity-dependent short-term depression and recovery, so that the amount of glutamate released at a synapse depends on both the probability of release of a glutamate vesicle and the number of releasable glutamate vesicles currently available at the synapse. Neurons are linked to other neurons using the following 3 strategies for synaptic connectivity.The first strategy is uniform connectivity, in which every neuron is synaptically connected to neurons in its local neighborhood, with the connection probability falling exponentially with distance. The second strategy is the small world network model, where the majority of connections are local, and a small percentage of connections are long range. The third strategy is a scale free network, in which some neurons have a great many connnections, but most neurons have few. Results: We performed 25 simulations of 60000 iterations for each network connectivity strategy (75 simulations total).  The network generated spontaneous interictal spikes over a wide range of connectivity and excitability parameters.  Spontaneous seizures occurred under a more restricted range of tested parameters.  Spontaneous interictal to ictal transitions occurred most rarely, and only in networks with small world connectivity.  Interictal-ictal transition mechanism: synchronous activity was terminated by activity-dependent synaptic depression. Full synaptic recovery engendered interictal population spikes that spread via long-distance synapses. When synaptic recovery was incomplete, postsynaptic neurons only fired in response to coincident activation of multiple presynaptic terminals. Only local connections were sufficiently dense to spread activity under these conditions, so that network activity coalesced into traveling waves whose velocity varied with synaptic recovery.  Sustained ictal traveling waves occurred when synaptic recovery, wave velocity, and network dimensions caused waves to re-enter a previously depressed area at ictogenic levels of synaptic recovery.   Conclusions: Distributed neural networks can generate both spikes and spontaneous seizures without external inputs or parameter adjustments. In simulations to date, small world network connectivity is most conducive to seizure generation.  A new mechanism of transition from interictal to ictal activity is described. Funding: NIH Grant R01NS086364, 2R37NS077908-05A1