Abstracts

We used a computational model of simultaneous discharges of a population of pyramidal cells in the hippocampal area CA3 to estimate the relationship between the network connectivity and the probability of spontaneous bursts. Prior investigations (Traub and Miles, [underline]Neuronal Networks of the Hippocampus[/underline], Cambridge University Press, 1991; Lytton et al. [underline]Computer models of hippocampal circuit changes of the kindling model of epilepsy[/underline] [italic]Artif Intell Med[/italic]. 13:81-97 1998) have supported a nonlinear relationship between connectivity and the probability of synchronous population discharges. In our experiments we wished to simulate the growth of recurrent synaptic connections after an epileptogenic insult (sprouting) in order to model the effects on sprouting on network activity., Our artificial network model employed an array of 10000 pyramidal cells and 225 interneurons. Interactions between the neurons take place via a randomly generated mesh of synapses that includes activity-dependent depression. The strength of glutamatergic synaptic connections is modifiable according to rules governing long-term synaptic plasticity. The network was initially generated with sparsely connected neurons, simulating the network prior to sprouting. Then after each experiment we increased network connectivity to simulate sprouting of synaptic connections. This was accomplished by increasing the maximum feasible radius of synaptic connections and randomly generating additional glutamatergic synapses onto principal cells and interneurons., We performed experiments for networks with various connectivities, focusing on spontaneous bursting behavior. An average number of incoming connections per neuron in separate experiments ranged between 35 and 208. The relationship between the connectivity and the average incoming connections count was close to linear. Interestingly the relationship between network connectivity and spontaneous bursting behavior was very nonlinear. This agrees with our findings regarding seizure frequency vs. time following an epileptogenic injury, (see abstract by Williams PA et al.), and with prior modeling studies employing more detailed neuronal models (Traub and Miles 1991)., The rapid increase in seizure probability that eventually follows an epileptogenic insult may be the result of a more linear increase in local connectivity arising from axonal sprouting. The modeling data suggest that a similar and cotemporaneous change in the pattern of interictal spikes should also be evident. Further, the evolution of interictal spikes over time may reflect changes in the underlying epileptogenic network that may help in clinical decision making regarding long-term seizure probability. These hypotheses are being tested using continuous radiotelemetric EEG monitoring., (Supported by NIH (NINDS).)