Simulation of Regenerative Ca2+ Signals for Coincidence Detection for Cerebellar Learning

Tomokazu Doi (0151069)

Synaptic plasticity is believed to be a molecular basis of learning and memory. Cerebellar Purkinje cells receive two excitatory inputs, numerous parallel fibers (PFs) and only one climbing fiber (CF). Cerebellar learning theories require that long-term depression (LTD) at PF synapses should be induced when PF input followed by CF input. Some studies, however, reported that strong PF stimulus without CF activation induced large Ca2+ signals and LTD.

To address the issue, I developed a Ca2+ signaling model by using GENESIS simulator with kinetikit interface. The model contains 49 variables and 95 parameters, most of which were determined from the literature. The 9 remaining were assumed so as to reproduce the PF-CF temporal window of Ca2+ response in Ca2+ imaging studies.

My model predicted that metabotropic pathway has a 100-ms delay in PF input to inositol 1, 4, 5-triphosphate (IP3) production and that IP3 receptors detect the coincidence of IP3 accumulation and CF-evoked Ca2+ entry. IP3 concentration determines Ca2+ threshold for regenerative Ca2+ release from internal stores. The model reproduces three quantitatively different Ca2+ dynamics depending on low, medium and high IP3 concentrations, which coherently explains various LTD induction studies. The intermediate IP3 is only physiological state where PF input followed by CF input is required for LTD. The results strongly supports that LTD is the molecular basis of cerebellar learning.

Although the time-dependency of Ca2+ signaling is sensitive to amplitude of inputs, the model showed that LTD stabilizes the time window by tuning of PF-mediated Ca2+ entry.