Abstract
Purpose:
Components of neural circuits are often repurposed so that the same hardware can be used for distinct computations. The demands of the visual system depend on the mean illumination - we struggle to collect photons on a moonless night, and we see fine spatial details in daylight - yet the same retinal ganglion cells (RGCs) encode visual information across all these conditions. Our goals were to determine whether RGCs perform different computations at different levels of illumination and to uncover the synaptic mechanisms that reconfigure retinal circuits.
Methods:
We recorded spike responses and excitatory input currents from an identified mouse RGC across a range of levels of background illumination while presenting visual stimuli to assess how the cell integrated inputs across space. To track the mechanistic basis of the computational changes, we made whole-cell recordings from several locations in the upstream circuit using both light and pharmacology.
Results:
The RGC integrated linearly over its receptive field (“X”-cell behavior) in dim conditions, but the cell integrated spatial inputs nonlinearly (“Y”-cell behavior) in brighter conditions. This computational change was the result of a change in excitatory inputs received from On cone bipolar cells, which transitioned from linear to rectified with increasing luminance. Current-clamp recordings from presynaptic bipolar cells did not show the same change in rectification, so the site of the change was localized to the cone bipolar to RGC synapse. Recordings of resting membrane potential throughout the circuit revealed that while rod bipolar cells depolarize with increasing luminance, AII amacrine cells and On cone bipolar cell axon terminals hyperpolarize. This luminance-dependent transition was the result of changes of tonic excitatory output from rod bipolar cells which reduced drive on the electrically coupled network and resulted in increased rectification of the On cone bipolar cell output.
Conclusions:
Light-dependent interactions between parallel circuits upstream of RGCs can alter the features of a visual signal encoded by at least one class of RGC through a novel mechanism involving gap junctions in the inner retina.