Abstract
Purpose :
The neural circuits of the retina extract salient visual features from the environment. Amacrine cells are a diverse class of retinal interneurons (~63 types), critical for visual feature computations. Most amacrine cells lack axons, and receive and send signals through their dendrites. The dendritic arbors of amacrine cells compartmentalize input-output transformations in cell-type-specific patterns (i.e., local processing). The mechanisms limiting dendritic signal spread and shaping local processing are poorly understood. Here, we test how synaptic inhibition and arbor morphology co-operate to extract behaviorally relevant visual features in VGluT3-expressing (VG3) amacrine cells.
Methods :
We analyze signal processing in VG3 dendrites by two-photon calcium imaging in VG3-Cre Ai148 mice. We use sparse CRISPR genome editing in VG3 cells to disrupt synaptic inhibition without the network effects confounding pharmacological manipulations. We combine experimental results and computational modeling based on electron microscopy reconstructions of VG3 dendrites arbors and synaptic input patterns to probe how inhibition and arbor morphology shape dendritic processing.
Results :
Removal of the gamma2 subunit of GABAA receptors, which mediates synaptic localization increases signal spread in VG3 dendrite arbors. This results in abnormal mixing of contrast information (i.e., ON and OFF) along the vertical arbor dimension and reduces the precision with which stimulus positions are encoded along the horizontal arbor dimension. GABAergic inhibition also contributes to the stimulus size selectivity of VG3 dendrites. This contribution is specific to OFF stimuli. We use computational modeling to probe how the specific distribution of inhibitory synapses co-operates with the arbor morphology to pattern dendritic integration.
Conclusions :
Our findings reveal how synaptic inhibition and arbor morphology shape dendritic integration to generate and compartmentalize behaviorally relevant feature preferences (e.g., looming).
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.