Non-Gaussianity of neurotransmitters co-released from mammalian adrenal chromaffin cells.

Abstract

While synaptic currents in computational neuroscience are conventionally modeled as Gaussian processes, there tends to be theoretical assumption that non-Gaussian Lévy processes can better describe the stochastic nature of neurotransmitter release in real neurophysiological scenarios. To support this view, we conduct statistical inference with the recordings of the co-release currents of two neurotransmitters from mammalian adrenal chromaffin cells by two steps. First, both the deterministic part and the random part of the current time series are separated by local weighted regression based on the individual vesicle releases and the entire co-release process, respectively. By fitting the resultant deterministic parts in individual release by the double exponential function and the counterparts in the entire co-release process by the truncated Fourier series, the procedure of separation we adopt is validated. And then, the statistical analysis based on the quantile-quantile plot and the empirical characteristic function reveals that the distribution of the random parts dramatically deviates from Gaussian distribution but matches well with certain non-Gaussian alpha stable distribution. Thus, the present study provides significant evidence for the non-Gaussian nature about neurotransmitter release from biophysical experiment.