Single axon excitatory postsynaptic potentials in neocortical interneurons exhibit pronounced paired pulse facilitation

In slices of adult rat somatosensory/motor cortex, paired recordings were made from pyramidal and non-pyramidal neurons. Single axon excitatory postsynaptic potentials evoked in the non-pyramidal neuron by action potentials in the pyramidal neuron were large and fast and demonstrated large fluctuations in amplitude, with coefficients of variation between 0.1 and 1.25. Excitatory postsynaptic potential amplitude distributions included a large number of apparent failures of transmission as well as some extremely large events. This contrasted dramatically with the relatively narrow distribution of amplitudes for pyramid-pyramid connections in neocortex. Excitatory postsynaptic potentials increased in amplitude with postsynaptic membrane hyperpolarization. Very small changes in the coefficient of variation when mean amplitudes increased substantially were consistent with the increase being due to a change in quantal amplitude. These excitatory postsynaptic potentials displayed profound paired pulse facilitation. Moreover, third and fourth spikes in a presynaptic burst also evoked large responses. This facilitation was associated with a decrease in the proportion of apparent failures in transmission and a change in the shape of the excitatory postsynaptic potential amplitude distribution, both indicative of an increase in the probability of transmitter release. However a large change in the mean amplitude was not associated with a similar change in the inverse square of the coefficient of variation. The result of this third test, taken in isolation, might therefore suggest that quantal amplitude had increased with paired-pulse facilitation. However, of the three tests applied, this last is the most heavily model-dependent and produced a result inconsistent with the results of the other two tests. The possibility is therefore discussed that both the shape of the excitatory postsynaptic potential amplitude distribution and the failure of coefficient of variation analysis to detect an apparently presynaptic change might result from the release at these synapses being poorly fit by a simple model. Based on a more complex model of synaptic release proposed by Faber and Korn [ Faber and Korn (1991) Biophys. J. 60, 1288–1294] and a hypothesis proposed by Scharfman et al. [ Scharfman et al. (1990) Neuroscience 37, 693–707], two hypotheses arising from the present study are discussed: (i) that branch point failure contributes to the pattern of synaptic activation at the