Functional dissection of a cortical microcircuit for spatial computation

Abstract

In mammals, spatial learning and memory depend on neural processing carried out in the hippocampal formation. Interestingly, extracellular recordings from behaving animals have shown that cells in this region exhibit spatially modulated activity patterns, thus providing insights into the neural activity underlying spatial behaviour. One area within the hippocampal formation, layer II of the medial entorhinal cortex, houses cells that encode a grid-like map of space using a firing rate code. At the same time, oscillatory signals at distinct theta (4–12 Hz) and gamma (30–120 Hz) frequencies are also present in layer II, providing a substrate for a timing code. To understand how layer II of the medial entorhinal cortex produces these outputs I sought to characterise the electrical properties and functional computational architecture of its microcircuitry. The functionality of any neural circuit depends on the electrical properties of its constituent cells. Because the grid cells in layer II are likely to be stellate cells, I used the perforated patch-clamp technique to accurately assess the intrinsic excitable properties of this cell type. Compared to whole-cell recordings, these recordings indicate that some intrinsic properties of stellate cells, such as spike clustering, which is revealed to be robust, are more likely to play a functional role in circuit computation. Conversely, other intrinsic properties, such as spontaneous membrane potential fluctuations, which are confirmed to be insufficiently stable to support reliable interference patterns, are revealed to be less likely than other, more robust electrical properties to play a direct role in circuit function. The characteristic connectivity profiles of different cell types are also critical for circuit function. To investigate cell type-specific connectivity in layer II I used optogenetic stimulation in combination with in vitro electrophysiology to record synaptic activity in different cell types while selectively activating distinct subpopulations of cells with light. Using this method I found that connections between stellate cells are absent or very rare and that communication between stellate cells is instead mediated by strong feedback inhibition from fast-spiking interneurons. Dissecting oscillatory activity in neural circuits may be important for establishing functionally relevant circuit architecture and dynamics but is difficult in vivo. I accomplished this in vitro by recapitulating the interacting theta and gamma rhythms that are observed in vivo with an optogenetic method. I found that locally driving a subset of neurons in the layer II microcircuit at theta frequency with a light stimiulus produced a nested field rhythm at gamma frequency that was also evident as rhythmic inhibition onto stellate cells. Critically, these interacting rhythms closely resembled those recorded from behaving animals. In addition, I found that this thetanested gamma is sufficiently regular to act as a clock-like reference signal, indicating its potential role in implementing a timing code. To functionally dissect the circuit I performed multiple simultaneous whole-cell patch-clamp recordings during circuit activation. These recordings revealed how feedback interactions between stellate cells and fast-spiking interneurons underpin the theta-nested gamma rhythm. Together, these results suggest that feedback inhibition in layer II acts as a common substrate for theta-nested gamma oscillations and possibly also grid firing fields, thereby providing a framework for understanding how computations are carried out in layer II of the medial entorhinal cortex.