Iannella Nicolangelo, Tuckwell Henry C, Tanaka Shigeru
Laboratory for Visual Neurocomputing, Brain Science Institute, RIKEN, 2-1 Hirosawa Wako-shi, Saitama 351-0198, Japan.
Math Biosci. 2004 Mar-Apr;188:117-32. doi: 10.1016/j.mbs.2003.10.002.
We have developed a non-linear stochastic PDE (partial differential equation) model of a rat layer 2/3 somatosensory pyramidal neuron which approximates several of the dynamical properties of these cells. The model distinguishes telodendrites, a myelinated axon, initial segment, hillock, soma and a simplified dendritic tree. Distributions and properties of excitatory and inhibitory synapses were included, in accordance with recent anatomical and physiological findings. Using simulation methods, we aim to show that the spatial separation between regions of spatially distributed randomly activated excitatory and inhibitory synaptic inputs may be an important parameter which can influence neuronal firing properties. Due to the complexity of the problem, with respect to configurations of spatially and temporally activated excitatory and inhibitory synaptic inputs, we consider two simple configurations in which the spatial region of activated excitatory and inhibitory synaptic inputs overlap and when they are far from each. In the first, denoted configuration A, activated excitatory and inhibitory synapses were located close to the soma. In the second, denoted configuration B, active inhibitory synapses were close the soma, while active excitatory synapses were located on distal regions of the dendrite. For the first configuration, we find that increases in the mean rate of inhibition results in an increase in the width of the firing rate tuning curves, and that for particular mean input frequencies of excitation, increasing the mean input rate of inhibition does not always imply that the neuron fires at a slower rate. Furthermore, we observed for mean input frequencies of excitation between 15 and 60 (Hz), that increasing the mean rate of inhibition resulted in the linearization of the firing rate over this interval. For configuration B, no increase in width nor a linearization effect via inhibition was observed. These differences indicate that the distance between regions of active excitatory and inhibitory synapses may be an important factor to consider in determining how the interaction between excitation and inhibition contributes to neuronal firing.
我们开发了一种大鼠第2/3层体感锥体神经元的非线性随机偏微分方程模型，该模型近似于这些细胞的几种动力学特性。该模型区分了终末树突、有髓轴突、轴突起始段、轴丘、胞体和简化的树突树。根据最近的解剖学和生理学研究结果，纳入了兴奋性和抑制性突触的分布及特性。通过模拟方法，我们旨在表明空间分布的随机激活的兴奋性和抑制性突触输入区域之间的空间分离可能是一个重要参数，它会影响神经元的放电特性。由于该问题的复杂性，考虑到空间和时间上激活的兴奋性和抑制性突触输入的配置，我们研究了两种简单配置：激活的兴奋性和抑制性突触输入的空间区域重叠以及它们彼此远离的情况。第一种配置称为A配置，激活的兴奋性和抑制性突触靠近胞体。第二种配置称为B配置，活跃的抑制性突触靠近胞体，而活跃的兴奋性突触位于树突的远端区域。对于第一种配置，我们发现抑制平均速率的增加会导致放电速率调谐曲线宽度的增加，并且对于特定的平均兴奋输入频率，增加抑制平均输入速率并不总是意味着神经元以较慢的速率放电。此外, 我们观察到对于15至60（Hz）之间的平均兴奋输入频率，抑制平均速率的增加导致该区间内放电速率的线性化。对于B配置，未观察到宽度增加或通过抑制产生的线性化效应。这些差异表明，活跃的兴奋性和抑制性突触区域之间的距离可能是在确定兴奋与抑制之间的相互作用如何影响神经元放电时需要考虑的一个重要因素。
