Abstracts

RATIONALE: While the neuronal substrate of epileptic seizures involves paroxysmal bursting of neurons, the clinical manifestations result from spread of activity from the local circuits to involve regional and remote brain regions. Recently there has been a growing interest in neural stimulation to reduce the frequency of seizures. Neural networks models are attractive systems to address the influences of these interventions.
METHODS: Neurons are modeled as single compartment units using Av-Ron-Rinzel[ssquote]s reduced model equations. The simulated multi-segmental neuronal network comprises a group of serially connected local neural sub-networks which form a chain loop. In a sub-network each neuron receives input from randomly selected cells from the same network. Some neurons have additional connections to neurons in adjacent sub-networks. This hierarchical structure allows for the efficient implementation of the simulations on a computer cluster. The parameters of the neurons and synapses were selected to allow for propagation of bursts of action potentials along the sub-network. To initiate bursting activity, selected neurons in the sub-network receive random (Poisson) excitatory inputs.
RESULTS: At the beginning of the simulations all neurons are in the resting state (stable node point). When a stimulus (external current or background activity) is applied for the first time it spreads relatively rapidly through the loop of sub-networks but does not initiate recurrent activity in a loop. Subsequent stimuli, however, usually do initiate such recurrent loop activity in a loop. The observed frequency of the triggered repetitive activity is dependent on the length of the loop and does not exceed 3Hz. The minimal length of the loop in which self-sustained oscillation is observed is 13 (for the selected values of model parameters). In short loops consisting of 16 or less sub-networks, the repetitive activity is in the range of 2.5 - 3 Hz. In longer loops, lower frequencies of repetitive bursting are observed; different modes of oscillation also exist. When recurrent activity propagates through the chain loop, external current delivered locally alters the propagation process and may result in cessation of that activity.
CONCLUSIONS: The simulation of bursting activity in network models allows for efficient investigation of propagation of bursting activity. Simulations show that application of the external stimulus may result in termination of propagation of epileptiform activity. Factors that affect spike frequency in neurons play an important role in the termination of this activity.
Support: NIH grant NS 38958