A deconvolution method to improve automated 3D-analysis of dendritic spines: application to a mouse model of Huntington's disease

Dendritic spines are postsynaptic structures the morphology of which correlates with the strength of synaptic efficacy. Measurements of spine density and spine morphology are achievable using recent imaging and bioinformatics tools. The three-dimensional automated analysis requires optimization of image acquisition and treatment. Here, we studied the critical steps for optimal confocal microscopy imaging of dendritic spines. We characterize the deconvolution process and show that it improves spine morphology analysis. With this method, images of dendritic spines from medium spiny neurons are automatically detected by the software Neuronstudio, which retrieves spine density as well as spine diameter and volume. This approach is illustrated with three-dimensional analysis of dendritic spines in a mouse model of Huntington's disease: the transgenic R6/2 mice. In symptomatic mutant mice, we confirm the decrease in spine density, and the method brings further information and show a decrease in spine volume and dendrite diameter. Moreover, we show a significant decrease in spine density at presymptomatic age which so far has gone unnoticed.