Changes in synaptic structure and efficacy including dendritic spine number and morphology have been shown to underlie neuronal activity and size. Moreover, the shapes of individual dendritic spines were proposed to correlate with their capacity for structural change. Spine numbers and morphology were reported to parallel memory formation in the rat using a water maze but, so far, there is no information on spine counts or shape in the radial arm maze (RAM), a frequently used paradigm for the evaluation of complex memory formation in the rodent.

24 male Sprague-Dawley rats were divided into three groups, 8 were trained, 8 remained untrained in the RAM and 8 rats served as cage controls. Dendritic spine numbers and individual spine forms were counted in CA1, CA3 areas and dentate gyrus of hippocampus using a DIL dye method with subsequent quantification by the Neuronstudio software and the image J program.

Working memory errors (WME) and latency in the RAM were decreased along the training period indicating that animals performed the task. Total spine density was significantly increased following training in the RAM as compared to untrained rats and cage controls. The number of mushroom spines was significantly increased in the trained as compared to untrained and cage controls. Negative significant correlations between spine density and WME were observed in CA1 basal dendrites and in CA3 apical and basal dendrites. In addition, there was a significant negative correlation between spine density and latency in CA3 basal dendrites.

The study shows that spine numbers are significantly increased in the trained group, an observation that may suggest the use of this method representing a morphological parameter for memory formation studies in the RAM. Herein, correlations between WME and latency in the RAM and spine density revealed a link between spine numbers and performance in the RAM.