At many glutamatergic synapses, the decay of the EPSC accelerates as the release probability is lowered (Trussel et al. 1993; Mennerick & Zorumski, 1995; Silver et al. 1996). This non-linearity is thought to arise from spillover of glutamate between neighbouring active zones, but the mechanism by which this is achieved is unclear. We have investigated this issue by combining electrophysiological recordings from the cerebellar mossy fibre to granule cell synapse, where spillover currents can be observed in isolation (DiGregorio et al. 2002), with a model of glutamate diffusion.

Thin parasagittal slices were cut from the vermal cerebellum from P25 rats killed by decapitation. Whole-cell patch-clamp recordings were made from granule cells, and mossy fibres were stimulated extracellularly (Silver et al. 1996). We recorded pharmacologically isolated non-NMDA currents at 36-37 °C. Diffusion in the synaptic cleft was simulated using an analytical solution to Fick’s second law for the region bound by two parallel planes (Otis et al. 1996) with stochastic release from 25 sites.

The acceleration of the EPSC in lowered [Ca²⁺]_o was quantified by the weighted decay, determined from the integral of the peak-normalised EPSC. Relative to [Ca²⁺]_o = 2 mM, the weighted decay was reduced to 96 ± 5.8, 89 ± 7.4 and 82 ± 9.0 % in [Ca²⁺]_o = 1.5, 1.25 and 1 mM, respectively (means ± S.E.M., n = 13). Simulations of glutamate diffusion following the stochastic release of quanta of neurotransmitter indicated that the release probability determines the amplitude but not the mean waveform of the glutamate transient. Simulations indicated that the glutamate concentration due to spillover attained on a particular trial is likely to be proportional to the release probability (Pearson correlation coefficient = 0.94), and thus to the peak EPSC amplitude. The isolated spillover current, when plotted against the peak EPSC amplitude for each [Ca²⁺]_o could be fitted with a Hill equation, giving a mean Hill coefficient of 1.46 ± 0.14 (n = 9) for synaptic receptors. This value is close to previous estimates from patches, consistent with the acceleration being due to non-linear activation of postsynaptic receptors.

This work was supported by The Wellcome Trust and the European Commision. R.A.S. is the recipient of a Wellcome Trust Senior Research Fellowship.