Abstracts

Rationale: Somatostatin (SS) neurons in the dentate hilus provide a major GABAergic innervation to the outer molecular layer of the dentate gyrus, and these hilar GABA neurons are among the most vulnerable neurons in temporal lobe epilepsy. Despite persistent loss of these neurons, a substantial SS/GABAergic innervation gradually reappears in the dentate molecular layer during the chronic stage. The goal of this study was to determine the source of this innervation and test the hypothesis that the axons could originate from SS neurons in stratum (s.) oriens of CA1 rather than the dentate gyrus. Methods: Sustained seizures were induced in SS-Cre mice by the administration of pilocarpine (Peng et al., J. Neurosci. 24:8629-8639, 2004). Following status epilepticus, the mice recovered and then developed spontaneous seizures within 1-2 weeks. At 5 weeks following pilocarpine-treatment, a Cre-dependent viral vector expressing channelrhodopsin 2 and yellow fluorescent protein (YFP) was injected stereotaxically in s. oriens of CA1 to selectively label SS neurons in this region that expressed Cre recombinase. Age-matched control SS-Cre mice received similar injections. At 3 weeks following these injections, the mice were prepared for histological study. Cre-dependent labeling of SS neurons at the injection site in s. oriens was determined by analysis of YFP localization in neuronal cell bodies and their axonal projections. Results: Cre-activated YFP labeling of SS neurons was confined to s. oriens of CA1 in both control and pilocarpine-treated mice. In control animals, a prominent band of YFP-labeled axon terminals was present in s. lacunosum-moleculare of CA1, consistent with the normal innervation of this region by SS neurons in s. oriens, but only a small number of GFP-labeled axons projected into the molecular layer of the dentate gyrus. In contrast, in the pilocarpine-treated mice, a dense plexus of YFP-labeled axons extended into the dentate molecular layer, and many of the labeled axons were continuous with those in s. lacunosum-moleculare. Immunohistochemistry confirmed that SS neurons in the hilus were severely depleted in the pilocarpine-treated mice, but that SS-labeled terminals were present in the molecular layer in a pattern resembling the YFP labeling. Conclusions: The findings demonstrate that remaining GABA neurons in the hippocampus are capable of extensive axonal sprouting in response to a decrease in GABAergic innervation in an adjacent region that they normally do not innervate. A unique GABAergic circuit appears to have been created between GABA neurons in s. oriens of the hippocampus and granule cells of the dentate gyrus. While the reorganized axons are expected to be functional, they are unlikely to control excitability of the granule cells in the epileptic animals in response to the normal inputs to the dentate gyrus. The findings are significant for they demonstrate that the changes in GABA neurons in temporal lobe epilepsy include not only neuronal loss and local sprouting but also the creation of novel, aberrant GABAergic circuitry.