Gbx2 regulates the development of a non-GABAergic non-Glycinergic amacrine cell in the mammalian retina.

Abstract

Purpose : Inhibitory amacrine cells (ACs) make up the most diverse population of neuronal cell types in the retina. However, a lack of genetic tools to label and manipulate distinct AC subtypes has hindered our understanding of how they develop and wire into functional circuits. Molecular identification and characterization of AC subtypes will allow for a more comprehensive understanding of retinal neural circuits.

Methods : We used a Gbx2^creER; R26^LSL-TdTom mouse line to identify a novel AC subtype. For co-localization and morphological analysis, Gbx2+ ACs were analyzed by immunochemistry in adult retina cryosections and flat-mounts. To assess gap junction coupling, single Gbx2+ ACs were filled by electroporation of 2% Neurobiotin in a cell-attached configuration. For phenotypical and RNAseq analysis of the Gbx2 and Robo2 knockout retinas, we compared mutant mice, Gbx2^creER/flox or Gbx2^creER; Robo2^F/F, with heterozygous control mice, Gbx2^creER/+ or Gbx2^creER; Robo2^F/F.

Results : We identified a population of mouse ACs that selectively express the transcription factor Gbx2. Gbx2+ ACs were found in the inner nuclear layer and ganglion cell layer, and have dendritic fields that stratify in sublamina 3 and 5 within the inner plexiform layer. Interestingly, Gbx2+ ACs do not express the inhibitory neurotransmitters, GABA or Glycine, suggesting they are a novel non-GABAergic non-Glycinergic (nGnG) AC subtype. However, Gbx2+ ACs exhibit gap junction coupled connections to neighboring retinal cells, and may provide synaptic output through electrical synapses. In the Gbx2 mutant mice, the Gbx2+ ACs maintain their nGnG characteristics, but their dendrites mis-stratify and exhibit ectopic projections into the outer layers of the IPL (n=4 mice/genotype, p<0.05; t-test). In other brain regions, Gbx2 controls neural circuit formation by upregulating the expression of the axon guidance receptor, Robo2. Conditional deletion of Robo2 from the Gbx2+ ACs, showed similar ectopic dendritic projections as seen in the Gbx2 mutants, suggesting Gbx2 may mediate development of these ACs through Robo receptors (n=3 mice/genotype, p<0.05; t-test). Ongoing experiments are investigating additional Gbx2 effectors by performing RNAseq on Gbx2 mutant and control ACs.

Conclusions : These results genetically identify a novel Gbx2+ nGnG AC subtype, and provide a model for studying transcriptional control of retinal circuit development.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.