Microelectrode arrays (MEAs) are a standard tool in neuroscience for recording activity from multiple neurons in-vitro and in-vivo. In contrast to intracellularly recorded action potentials (APs), the shape of extracellular APs (EAP) has large variability (Nam and Wheeler 2011). The EAP shape is affected by the relative location of signal source and recording electrode, as well as the properties of extracellular medium. The maximum signal peak can be either positive or negative, while the signal shape can be monophasic, biphasic or triphasic. The width of the EAP is also believed to depend on the neuronal type. However, a direct correlation of the EAP signal shape with respect to neuronal location and type has not been established. We have developed a protocol for identifying the exact origin of MEA-recorded spikes.
We plated dissociated neuronal cultures on a CMOS-based high-density microelectrode array (HDMEA) containing 11,011 electrodes, which enables subcellular-resolution recordings (Frey et al. 2010). The neuronal activity from all electrodes was recorded, followed by immunostaining for beta-III-Tubulin (TUJ1) and GABA. The full array (2 mm x 1.8 mm) was then imaged with Metamorph software. An image registration protocol was used to align the immunofluorescence pictures to the HDMEA's electrode coordinates. We analyzed and compared the overlaid images (electrodes and TUJ1 immunostaining) with the HDMEA signals of selected areas. We used spike sorting algorithms to assign the detected APs to putative neurons and extracted the spike templates for each putative neuron. TUJ1 stained dendrites and axons. Neuritic fascicles were very strongly labeled, whereas putative single axons had weaker staining. Many spikes were recorded in, seemingly, areas containing only axons. These spikes were triphasic, although the peak of the negative phase was variable. Finally, we show the spike shape properties of different GABAergic subtypes.