Dynamics of electrosensory feedback: short-term plasticity and inhibition in a parallel fiber pathway.

The dynamics of neuronal feedback pathways are generally not well understood. This is due to the complexity arising from the combined dynamics of closed-loop feedback systems and the synaptic plasticity of feedback connections. Here, we investigate the short-term synaptic dynamics underlying the parallel fiber feedback pathway to a primary electrosensory nucleus in the weakly electric fish, Apteronotus leptorhynchus. In open-loop conditions, the dynamics of this pathway arise from a monosynaptic excitatory connection and a disynaptic (feed-forward) inhibitory connection to pyramidal neurons in the electrosensory lateral line lobe (ELL). In a brain slice preparation of the ELL, we characterized the synaptic responses of pyramidal neurons to short trains of electrical stimuli delivered to the parallel fibers of the dorsal molecular layer. Stimulus trains consisted of 20 pulses, at either random intervals or constant intervals, with varying mean frequencies. With random trains, pyramidal neuron responses were well described by a single exponential function of the inter-stimulus interval-suggesting a single facilitation-like process underlies these synaptic dynamics. However, responses to periodic (constant interval) trains deviated from this simple description. Random and periodic stimulus trains delivered when the feed-forward inhibitory component of this pathway was pharmacologically blocked revealed that inhibition and depression also contribute to the observed dynamics. We formulated a simple model of the parallel fiber synaptic dynamics that provided an accurate description of our data. The model dynamics resulted from a combination of three distinct processes. Two of the processes are the classically-described synaptic facilitation and depression, and the third is a novel description of feed-forward inhibition. An analysis of this model suggests that synaptic pathways combining plasticity with feed-forward inhibition can be easily tuned to signal different types of transient stimuli and thus lead to diverse and nonintuitive filtering properties.