UC Irvine

Communication between different sub regions of the hippocampus is fundamental to learning and memory. However accurate knowledge about information transfer between sub regions in individual axons is lacking. MEMS devices with microtunnels connecting two sub networks have begun to approach this problem but common 10 µm wide tunnels frequently produce ambiguous interfering spikes from multiple axons. To reduce this complexity, we compared polydimethylsiloxane (PDMS) microtunnel devices each with a separate tunnel width of 2.5, 5 or 10 μm bridging two wells aligned over a multi electrode array (MEA). After 2-3 weeks of culturing primary rat neurons, with dentate gyrus on one side and hippocampal CA3 on the other of the chamber, spontaneous activity in the axons inside the tunnels was recorded. We report electrophysiological, exploratory data analysis for feature clustering and visual evidence to support our expectation that compared to 10 µm wide tunnels, 2.5 µm wide tunnels have fewer axons per tunnel. Clustering measures comparing the variations of spike height and width for different tunnel widths revealed tighter clusters of spikes with less height and width variation for narrow tunnels. Wider tunnels tended toward more diffuse clusters from a continuum of spike heights and widths. Standard deviations for multiple cluster measures, such as Average Dissimilarity, Silhouette Value Derivative (S) and Separation Factor, support a conclusion that 2.5 μm wide tunnels with fewer axons enable more precise determinations of individual action potential peaks, their propagation direction and hence timing of information transfer between sub networks.