Abstracts

Rationale: To understand the factors that lead to interictal vs. ictal activity, we are analyzing the spatial propagation of synchronous network activity. We hypothesize that the pattern of propagation may determine whether network activation is brief or sustained. Methods: We use computer models of the CA3 neural network to simulate network activity. The models include the following assumptions: the majority of neurons are pyramidal cells with small number of interneurons, the majority of connections are excitatory recurrent collateral synapses, the model is two-dimensional, connectivity between cells is random with uniform connection probability that decreases with distance, and there are real edges. Propagation of activity in real networks is recorded from organotypic slice cultures of rat hippocampus as electrophysiological data synchronized with the high speed video frames of calcium sensitive signal acquired at 125-250 Hz. The video data is digitally processed to remove slow glial transients and intensify and extract neuronal calcium signals. Results: In computational simulations we observed an array of propagation patterns that were influenced by network parameters including the distribution of synaptic strengths and connectivity patterns. We distinguished the following types of bursts: 1) individual bursts that propagated rapidly and uniformly across the network 2) seizure like activity, initiated by a burst that was followed by a long series of smaller and longer bursts that propagated more slowly and inhomogeneously through the network. In the in vitro recordings we observed similar behaviors. The majority of recorded bursts propagated rapidly and uniformly through the entire CA3 network. Ictal events started with a preictal burst closely followed by a long array of weak bursts. Calcium imaging revealed that preictal bursts were localized to specific areas of network, and that activity slowly spread to the rest of the network returning to the origin over a time span of several seconds. Conclusions: Computational models and calcium imaging both indicate that seizure propagation is slower and less homogenous than the propagation of interictal bursts. Because both discharges can occur in the same preparation, the parameters that lead to these different propagation patterns must be dynamic. Understanding the determinants of these different propagation patterns should elucidate some of the mechanisms of seizure initiation.