Dendrites of classes of hippocampal neurons differ in structural complexity and branching patterns

Abstract

Dendrites of reconstructed hippocampal neurons were analyzed for morphometric, topologic, and fractal parameters (n = 32 quantities) to investigate neuronal groupings and growth characteristics with a common set of assumptions. The structures studied included CA1 and CA3 pyramidal cells, interneurons, and granule cells from young animals (71 cells in total). Most of the cells showed no characteristic fractal dimension; rather, the scaling relation could be well represented by a two-parameter fit, of which one parameter showed a significant difference between cell classes. Other significant quantities that differentiated cell classes were related to the complexity of the dendritic tree (number of branch points and maximal terminal branch order) and the cell's electrical properties such as the mean attenuation between the soma and terminals. Principal components analysis produced combined measures of only slightly greater discriminative power than the best individual measures, indicating that the elementary quantities capture most of the structural variation between hippocampal cell groups. Another finding was that for all cells the mean segment length increased with dendritic branch order, which is consistent with decreasing branching probability as a function of the path distance from the soma. Analysis of another set of CA1 pyramidal neurons from aged animals (n = 15; 22-24 months) showed only a few significant differences than those from young animals (n = 11; a subset of n = 71) of which the most important was a straightening of the paths between terminals and the soma. The quantities analyzed in these reconstructed hippocampal neurons may reflect both intrinsic neuronal characteristics and extrinsic influences. Hippocampal cell groupings (i.e., pyramidal cells as opposed to dentate granule cells and interneurons) were significantly differentiated by most parameters. These differences and parameter values may be critical for understanding and generating synthetic neuronal populations for modelling studies.