Short-term plasticity (STP) is the process by which a synapse changes its efficacy in a history-dependent manner. It is hypothesized that STP’s information processing capabilities are connected to the way it implements temporal filtering in response to a sequence of presynaptic events [1,2]. Depression (STD) and facilitation (STF) refer to the decrease and increase (e.g., Figs. 1-A, B) of the efficacy in the synaptic response to a presynaptic spike as compared to the previous one. Sequences of presynaptic spikes create synaptic _temporal filters_ (TFs) (e.g., Fig. 1). STD and STF lead to low-pass and high-pass (Figs. 1-A, B) TFs. The presence of both STD and STF may lead to band-pass TFs (Fig. 1-C). These synaptic TFs are communicated to the postsynaptic membrane potential. However, it is unknown whether the properties of the postsynaptic TFs (PSTFs) are inherited from the synaptic TFs or there is additional processing involved.
We report on the results of a computational study aimed at identifying the types of TFs and PSTFs in response to periodic presynaptic inputs and their dependence on the biophysical and dynamic properties of the participating components. We implement biophysically plausible (conductance-based) computational models of a synapse with STP [3], which drives a post-synaptic cell. We characterize the TFs that arise in the synapse. We determine the conditions under which synaptic low-, high- and band-pass filters arise in terms of the STP and synaptic time constants and the presynaptic input frequency. We also determine how the long-term time constants of the synaptic envelope responses (Fig. 1, black curves) depend on the STP time constants, which operate at the single event level. While the envelopes for the low- and high-pass TFs have a single time constant that depends on depression or facilitation time constants, accordingly, the envelope for the band-pass TF has three time constants that depends on the time constants for (i) depression only, (ii) facilitation only, and (iii) a combination of both. This is in contrast to the naïve expectation that there would be only two time constants involved (depression and facilitation only). We then extend our study to include the postsynaptic cell. We identify and characterize the different types of PSTFs in terms of the properties of the input TF and the properties of the receiving cell. We show that while under certain conditions, the PSTFs are qualitatively a copy of the synaptic TFs and share many of the TF’s dynamic properties, in other biophysical conditions, the PSTF exhibit a higher degree of complexity, which involve a multiplicity of time scales (e.g., depression/facilitation, synaptic, membrane, ionic currents, summation).
Overall, our results highlight the complexity of TFs and PSTFs, which reflects the complexity of the underlying interactions. This has implications for the understanding of network filters and the development of effective synaptic decoding algorithms.
Acknowledgment:
Funded in part by a grant (DMS-1608077 to HGR) and a Graduate Research Fellowship (to YM) both from the NSF .
References:
1\. Zucker R, Regehr W. Short-term synaptic plasticity. Annu Rev Physiol. 2002, 64, 355-405.
2\. Fortune E, Rose G. Short-term synaptic plasticity as a temporal filter. Trends Neurosci. 2001, 24.7, 381-385. 3\. Tsodyks M, Pawelzik K, Markram H. Neural networks with dynamic synapses. Neural Comput. 1998, 10.4, 821-835.
