High-pass Filtering Through Short-Term Synaptic Facilitation Amplifies Low-Frequency Modulation of Bursting Input

Regular rhythmic activity typically produces stereotypical synaptic responses, masking dynamics due to short-term synaptic plasticity (STP). Multiple-frequency (e.g., Poisson-like) inputs unveil canonical STP effects where facilitation or depression, respectively, favor high- or low-frequency inputs and a mix of both favors intermediate frequencies. Notably, regular activity with multiple oscillatory components can produce synaptic responses that are not readily surmisable from canonical STP responses. In the responses of rhythmically activated muscles of the lobster ( Homarus americanus ), of either sex, slow modulation of bursting inputs, consisting of periodic changes in burst frequency and spike number, is amplified by dynamic neuromuscular synapses. Using a simple STP model, we demonstrate that facilitation enhances the difference (contrast) in responses to strong and weak bursts, while depression diminishes it. Non-intuitively, such changes in contrast imply that high-pass filtering enhances low-frequency components of the modulated bursting, whereas low-pass filtering attenuates them. For mixtures of facilitation and depression, our modeling results suggest a complex dependence of the readout of slow modulation on overall release probability and recovery times for vesicle depletion and calcium accumulation. Notably, these effects are reduced when the recovery time of STP exceeds the burst period and thereby allows a memory of prior activity across consecutive bursts. Additionally, with memory across bursts, response contrast does not change proportionally with input contrast and depends on the number of bursts per slow modulation cycle. Finally, a biophysical model of a postsynaptic cell demonstrates that simple subthreshold voltage-gated conductances can substantially contribute to the readout of low-frequency modulation. Significance Statement Neuronal communication is shaped by synaptic dynamics that filter signals, passing some frequencies while suppressing others. While frequency filtering is well understood for simple, repetitive activity, its effect on patterned signals containing multiple frequency bands is less clear. Using a simple model of short-term synaptic plasticity, we show that high-pass filtering of bursts can unexpectedly amplify slow modulation of bursting oscillations, whereas low-pass filtering reduces it. These effects depend on the specific frequency components and the time constants of synaptic recovery. Inspired by experimental results from a motor system, our findings apply broadly to contexts where multiple frequency bands interact, such as in sensory processing or during the coupling of brain oscillations at different frequencies.