Inhibitory Circuits of the CA3-Dentate Gyrus System

Abstract

Neuronal inhibition is an essential feature of all brain areas. Besides controlling the average rate of neuronal firing, it also controls the precise timing of action potentials and mediates several types of network oscillations that are related to cognition. Inhibition is provided primarily by interneurons that release gamma-Aminobutyric acid (GABA). Many interneuron subtypes have been identified based on their morphology, electrophysiology, and molecular markers. Here we characterize a novel interneuron subtype of the hippocampus that is primarily found in the CA3 area. It colocalizes the molecular interneuron markers somatostatin (Sst) and glutamate decarboxylase (Gad) but also solute carrier family 17 member 8 (Slc17a8). Slc17a8 is a gene encoding the vesicular glutamate transporter 3 and is therefore a marker of glutamatergic neurons. We used patch seq to transcriptomically and electrophysiologically characterize this Sst+/Slc17a8+ interneuron subtype, finding that it is electrophysiologically not clearly distinguishable from other interneuron subtypes. To investigate its functional role, future studies should establish methods to specifically target them with optogenetic constructs. We tested a transgenic mouse line that was developed to guide expression to Sst interneurons. <br /> Transgenic mouse lines are widely used to express constructs that allow targeted manipulation of neuronal activity. We used the SST-Cre mouse line to express channelrhodopsin attached to yellow fluorescent protein by intracranial viral injection in somatostatin positive interneurons of CA3. Viral transduction resulted in widespread axon signal in contralateral hippocampus and optogenetic activation caused strong excitatory postsynaptic currents in contralateral CA1 pyramidal cells. At the injection site, somatostatin negative cells in the pyramidal cell layer were expressing the viral construct. In other CA3 layers almost all expressing cells were also somatostatin positive. These data show that the mouse line is unsuitable to optogenetically study somatostatin interneurons in CA3 because it also targets pyramidal cells. Pyramidal cells however, do not express Slc17a8 and an intersectional strategy with a mouse line expressing another recombinase in Slc17a8⁺ interneurons could in the future specifically target Sst⁺/Slc17a8⁺ interneurons. <br /> While circuit level experiments are key to understanding behavior, only small parts of the network are accessible for manipulation and measurement at the same time. Computational modeling can be a powerful tool and provide a more complete picture of the entire simulated network. Experimentally well constrained models can provide testable hypotheses. Many of the inputs into CA3 come from the dentate gyrus. The dentate gyrus has the special property that its outputs are pattern separated. We implemented a circuit model of the dentate gyrus to study the role feedback inhibition plays during pattern separation. <br /> Feedback inhibition is a specific type of inhibition that is activated by the same principal cells that receive the inhibition. For the model implementation we used data from optogenetic experiments to constrain properties of feedback inhibition. Pattern separation is a neuronal computation that decreases the similarity of the network output as compared to the networks input pattern. It is known to occur in the dentate gyrus and inhibition has been shown to support pattern separation. Accordingly, we found that removing feedback inhibition from the model impaired pattern separation. The size of this effect, however, depended on the frequency of oscillatory activity we imposed on the input pattern. At higher frequencies, feedback inhibition had a stronger effect. This was not the case for feedforward inhibition, which had similar effects regardless of frequency. These findings highlight the role of input frequency for pattern separation and suggest that different circuit motifs engage differently depending on the oscillatory state of the upstream area. Behaviorally our model predicts that interfering with oscillatory activity could affect hippocampal pattern separation.