Abstract

The retina transforms visual sensation into perception. Extracted visual features are encoded by retinal ganglion cells (RGCs), the output neurons of the eye, and sent to the brain in parallel processing channels. Morphologically, RGCs fall into more than fifty diverse types, which innervate distinct brain areas. Such visual pathways differentially regulate various behaviors. However, the genetic determinants of RGC type diversity are unknown and thus we lack genetic access to study visual pathways. A generation of a more comprehensive RGC type atlas integrating molecular, morphological and functional properties is essential to dissect the functional architecture of the visual system. In a collaborative effort, I used single cell transcriptomics to molecularly classify RGCs during larval and adult stages. RGC types segregate into many discrete transcriptional clusters each with a unique molecular composition. Relatedness of clusters revealed a molecular taxonomy, in which RGC types are arranged into major RGC groups that comprise subclasses and diversify into individual types. This organization of RGC type diversification underlies a code of gene expression patterns, composed primarily of transcription factors. Differential gene expression analysis identified dozens of novel cluster-specific genetic markers for RGC types. Comparison of transcriptional signatures revealed that larval RGCs exhibit higher molecular diversity, which facilitates segregation of similar types, while adult RGCs maintain a core molecular identity suggesting a tight correspondence between larval and adult RGC types. Next, I mapped transcriptional clusters to RGC morphotypes. Select candidate markers were exploited as genetic entry points in a CRISPR-Cas9 transgenesis approach. To restrict labeling specifically to cluster-specific RGC types, I established a genetic intersection with a broad RGC marker. This intersectional transgenic approach allowed to correspond various clusters to distinct morphologically classified RGC types. I generated two transgenic lines using RGC subclass markers, one of which is based on the transcription factor eomesa expressed by RGC types routing to visual areas in hypothalamus, pretectum and tectum. Based on homologies to RGC types characterized in other species, I hypothesized that eomesa+ RGCs constitute intrinsically photosensitive RGCs and have non-image forming functions. I tested this hypothesis by characterizing their response profiles to a battery of visual stimuli and found that they are not tuned to canonical pattern stimuli. Rather eomesa+ RGCs encode ambient luminance levels corroborating my hypothesis. I further tested their necessity in non-image forming behavior, specifically visual background adaptation, which by initial investigation appears to not be affected by chemogenetic ablation of eomesa+ RGCs. In conclusion, this thesis presents a strong foundation for a RGC type atlas and reconciles molecular, morphological and functional features of discrete cell types. This comprehensive molecular classification of RGC types, together with the identified markers and newly established transgenic tools, provides a rich resource towards a better understanding of visual pathway function.