Aim: 

The cerebellum is classically depicted as a well defined circuit, whose basic function can be understood by considering the anatomical organization and the sign of neuronal interactions. However, local connectivity and neuronal and synaptic properties suggest the emergence of complex spatio-temporal dynamics. Here, experimental measurements are combined with computational models to investigate network interactions.

Methods: 

Patch-clamp whole-cell recordings from granule and Golgi cells have been used to develop detailed models of the neurons and synapses. MEA and voltage-sensitive dye recordings in acute cerebellar slices were used to investigate the spatio-temporal structure of responses. Finally, we have developed a large-scale computational model of the cerebellar granular layer network (NEURON) and evaluated its ability to reproduce the spatio-temporal patterns of activity recorded in vitro and in vivo.

Results: 

The granular layer network showed a series of remarkable properties in response to mossy fiber stimulation, that could be faithfully reproduced by the model:

Granule cells emitted spikes for a limited time period (< 5 ms) before inhibition through the Golgi cell loop interrupted the output (time-window).

Granule cells optimally transmitted spikes when the input frequency was higher than 100 Hz (high-pass filtering).

Sustained and diffused mossy fiber activity generated oscillations through the reverberant Golgi cell loops (oscillation and resonance).

Activity was organized with maximum excitation in the core and inhibition in the periphery of activated areas (center-surround).

Conclusions: 

These results suggest that the granular layer can perform complex transformations on incoming signals behaving as an adaptable spatio-temporal filter. LTP and LTD at the mossy fiber – granule cells relay, by regulating the release machinery and postsynaptic responsiveness, could contribute to organize time-window, oscillation, filtering, and center-surround properties.