Evelyne Sernagor

In the developing retina, spontaneous waves sweep across the layer of retinal ganglion cells (RGCs), the output cells of the retina. Experimental evidence indicates that retinal waves play a crucial role in guiding the refinement of visual connectivity. Using a large-scale, high-density array of 4,096 electrodes covering the RGC layer in the neonatal mouse, we have characterized wave spatiotemporal properties at unprecedented resolution from postnatal day (P) 2 to P13 (eye opening), when they disappear, replaced by visual experience. Wave dynamics undergo profound developmental changes. From initial gap junction communication during late gestation (Stage 1), they become controlled by directly interconnected cholinergic starburst amacrine cells (SACs), the only retinal cholinergic cells (Stage 2 waves). Stage 2 waves are initially wide spreading with random propagation patterns and low cellular recruitment. Around P6-7, they begin to shrink because of emerging GABAergic inhibitory connections and become denser, with many more immediate neighbouring RGCs recruited within waves. Direct connections between SACS withdraw at P10. Waves become then driven by newly formed glutamatergic connections originating from bipolar cells (Stage 3 waves, P10-eye opening). Recent observations from our lab are challenging the hypothesis that Stage 2 waves are driven by SACs. Indeed, we found a “novel” transient population of cholinergic cells present from P2-9. These cells co-exist with SACs, but they are larger, forming tight clusters in an annulus pattern around the optic disc at P2-3. That annulus expands towards the periphery with development, until the cell clusters disappear at P10, coinciding with the disappearance of Stage 2 waves, suggesting that they may represent a transient hyper-excitable cellular hub responsible for the generation of Stage 2 waves. In support, we found that wave origins follow a centrifugal pattern between P2-P9 as well. Stage 3 glutamatergic waves completely change spatiotemporal patterns, gradually becoming activity hotspots that tile the entire retinal surface. We propose that Stage 2 waves are important for guiding the establishment of eye-specific segregation and refinement of topographic maps in retinal central targets. Stage 3 waves, on the other hand, may be important for carving local retinal networks underlying the establishment of retinal receptive fields that become functional immediately after eye opening.