Pathway specificity of dendritic spine morphology in identified synapses onto rat hippocampal CA1 neurons in organotypic slices.

The output of the hippocampus is largely determined by interaction of the three excitatory pathways that impinge on CA1 pyramidal neurons. These synapses, formed by axons of: (1) CA3 pyramidal neurons; (2) neurons of the entorhinal cortex (EC); and (3) neighboring CA1 neurons, are all potentially plastic. Here, we take advantage of the accessibility of the organotypic slice preparation to identify the type of spines with which each of these pathways forms synapses, at different developmental stages. Recent reports have shown that morphology of dendritic spines is activity-dependent with large mushroom spines being thought to represent stronger synaptic connections than thin or stubby spines. Although in a wide range of preparations, mushroom spines represent only 15% of spines across the whole dendritic tree, we find that this proportion is highly pathway specific. Thus in organotypic slices, the axons of CA3 neurons form synapses with mushroom spines on CA1 neurons in approximately 50% of cases, whereas this spine type is rare (<10%) in either of the other two pathways. This high proportion of mushroom spines only occurs after spontaneous excitatory activity in the CA1 cells increases over the second week in vitro. Previous studies suggest that pathway specificity also occurs in vivo. In tissue fixed in vivo, it is the synapses of distal apical dendrites thought to be formed by axons originating in the EC that are richer in mushroom spines. Hence, contrary to previous suggestions, the proportion of mushroom spines is clearly not an intrinsic property of the pathway but rather a characteristic dependent on the environment. We suggest that this is most likely a result of the previous activity of the synapses. The fact that, despite the large differences in pathway specificity between preparations, the overall proportion of different spine types remains unchanged, suggests a strong influence of homeostasis across the network. (c) 2006 Wiley-Liss, Inc.