Cortical pyramidal and parvalbumin cells exhibit distinct spatiotemporal extracellular electric potentials.

Brain circuits are composed of diverse cell types with distinct morphologies, connections, and distribution of ion channels. Modeling suggests that the spatial distribution of the extracellular voltage during a spike depends on cellular morphology, connectivity, and identity. However, experimental evidence from the intact brain is lacking. Here, we combined high-density recordings from hippocampal region CA1 and neocortex of freely-moving mice with optogenetic tagging of parvalbumin-immunoreactive (PV) cells. We used ground truth tagging of the recorded pyramidal cells (PYR) and PV cells to construct binary classification models. Features derived from single-channel waveforms or from spike-timing alone allowed near-perfect classification of PYR and PV cells. To determine whether there is unique information in the spatial distribution of the extracellular potentials, we removed all single-channel waveform information from the multi-channel waveforms using an event-based delta transformation. We found that spatiotemporal features derived from the transformed waveforms yield accurate classification. The extracellular analogue of the spatial distribution of the initial depolarization phase provided the highest contribution to the spatial-based prediction. Compared to PV cell spikes, PYR spikes exhibited higher spatial synchrony at the beginning of the extracellular spike and lower synchrony at the trough. The successful classification of PYR and PV cells based on purely spatial features provides direct experimental evidence that spikes of distinct cell types are associated with distinct spatial distributions of extracellular potentials. Significance statement It is not clear if and how neuronal morphology, cell type, and synaptic inputs are mapped to the spatial distribution of the extracellular voltage during spikes. Here we show that spatial information alone allows accurate differentiation between pyramidal cells and parvalbumin-immunoreactive cells in neocortex and hippocampus of freely-moving mice. The ability to distinguish cell types based on spatiotemporal properties of extracellular potentials suggests that neurons with distinct morphology, connectivity, and ion channel distributions create unique and learnable extracellular patterns. Further research may reveal whether unique spatial information is characteristic of other cell types.