Abstract
Purpose: :
To connect cell morphology to cell type specific rules in the dynamic interactions of excitation, inhibition and spiking in genetically-identified ganglion cell subtypes in the mouse retina.
Methods: :
A conditional reporter system was used to generate mice that express GFP in subsets of retinal ganglion cells. Two-photon imaging was used to detect the GFP-positive ganglion cells in light-adapted wholemount retinas. Whole-cell voltage-clamp recordings were used to characterize the excitatory and inhibitory inputs during light stimulation. Flashing or moving discs were used to explore how the ON and OFF inputs shaped the spiking output while being filled with Neurobiotin. Retinas were fixed, and morphological analysis was carried out using markers for cholinergic strata (ChAT) and GFP. Qualitative features were defined, then recognized in measurements in a high throughput fashion. Associative data mining techniques were used to identify relationships between morphology, excitation, inhibition and spiking data. These relationship data were further refined and summarized.
Results: :
An expert system was built up using measurements from over 250 genetically-identified cells. The system exploited connections (1) within different electrophysiological recordings (inhibition, excitation, spiking) and (2) between electrophysiology and morphology. For example, it found that if a cell receives excitation after a small white disc flash stimulus, it will always receive inhibition after a large black flash. It was also able to predict morphology from a few electrophysiological recordings, which contain about five qualitative features, such as a significant excitatory OFF input for a full-field flash .
Conclusions: :
Two-photon imaging of genetically-labeled retinas made it possible to repeatedly target the genetically and morphologically identified ganglion cell types for physiological recordings. This enabled us to find relationships between excitation, inhibition and spiking in genetically identified ganglion cells. Our results suggest that each ganglion cell type can be equivalently defined by morphology and well-defined electrophysiological measurements.
Keywords: retinal connections, networks, circuitry • computational modeling • gene/expression