Abstracts

Neuronal hyperactivity and intense depolarization of neurons during seizures is accompanied by enhanced calcium influx into neurons. Increased concentration of cytosolic calcium is thought to precede changes in synaptic strength. Little is known how changes in synaptic strength influence patterns of neuronal network activity during seizures. Direct measurements of synaptic facilitation in situ are in large part limited to single synapses. We use neural network models to investigate how these synaptic changes influence network behavior during seizures. Our network model is composed of excitatory and inhibitory neurons (Kudela et al., 2003). We adopt a phenomenological approach (Shouval et al., 2002) in which the synaptic strength change is fully determined by intracellular calcium concentration producing either synaptic potentiation or depression. Intracellular calcium is modeled in detail including calcium influx and efflux, binding, and uptake. The network is activated by random excitatory input. In order to induce the repetitive rhythmic activity, the strengths of inhibitory synaptic weights are gradually decreased. This results in recurrent synchronous bursts in all neurons in the network. The local field potential (LFP) modeled as an average membrane potential of all neurons in a network is calculated. The patterns of modeled LFPs can be compared with selected intracranial ictal EEG from humans. The addition of synaptic strength regulation in these network models affects the characteristics of simulated epileptiform activities. This is reflected in the increase of the amplitude of simulated LFPs. These increased levels of intracellular calcium in neurons produce synaptic potentiation, which amplifies LFP amplitude. After long periods ([gt]1 min) of simulated epileptiform activities we observed changes in temporal synchronization of bursts in neurons. Strong synaptic potentiation leads to irregular bursting, a quantitative change in patterns of bursting activities and a common pattern observed late in partial seizure evolution in man. Synaptic efficacy changes occurring rapidly after seizure onset and resulting from seizure activity can in turn affect dynamics of these seizures. This can result from either potentiation of existing synapses or creation of new active synaptic connections (i.e. conversion of silent synapses into active synapses). In network models, synaptic potentiation produced by raised intracellular Ca²⁺ levels in neurons is responsible for the late irregular bursting in simulated LFPs. A similar pattern is observed in human EEG prior to seizure termination. The amplitude and the time frequency characteristics of simulated LFPs are in agreement with those obtained for ictal EEG from humans. These neural network models can provide insights into potential mechanisms for seizure evolution and termination. (Supported by Epilepsy Foundation and NIH grant NS 38958.)