Statistical model relating CA3 burst probability to recovery from burst-induced depression at recurrent collateral synapses

When neuronal excitability is increased in area CA3 of the hippocampus in v itro, the pyramidal cells generate periodic bursts of action potentials tha t are synchronized across the network. We have previously provided evidence that synaptic depression at the excitatory recurrent collateral synapses i n the CA3 network terminates each population burst so that the next burst c annot begin until these synapses have recovered. These findings raise the p ossibility that burst timing can be described in terms of the probability o f recovery of this population of synapses. Here we demonstrate that when ne uronal excitability is changed in the CA3 network, the mean and variance of the interburst interval change in a manner that is consistent with a timin g mechanism comprised of a pool of exponentially relaxing pacemakers. The r elaxation time constant of these pacemakers is the same as the time constan t describing the recovery from activity-dependent depression of recurrent c ollateral synapses. Recovery was estimated from the rate of spontaneous tra nsmitter release versus time elapsed since the last CA3 burst. Pharmacologi cal and long-term alterations of synaptic strength and network excitability affected CA3 burst timing as predicted by the cumulative binomial distribu tion if the burst pace-maker consists of a pool of recovering recurrent syn apses. These findings indicate that the recovery of a pool of synapses from burst-induced depression is a sufficient explanation for burst timing in t he in vitro CA3 neuronal network. These findings also demonstrate how infor mation regarding the nature of a pacemaker can be derived from the temporal pattern of synchronous network activity. This information could also be ex tracted from less accessible networks such as those generating interictal e pileptiform discharges in vivo.