Developmental control of rod photoreceptor number via a light-dependent retrograde pathway from intrinsically photosensitive retinal ganglion cells

Photoreception, a form of sensory experience, is essential for normal development of the mammalian visual system. Detecting photons during development is a prerequisite for visual system function - from shaping vision’s first synapse and maturation of retinal vascular networks, to transcriptional establishment and maturation of cell types within the visual cortex. Consistent with this theme, we find that the lighting environment regulates developmental rod photoreceptor apoptosis via OPN4-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs). Using a combination of genetics, sensory environment manipulations, and computational approaches, we establish a molecular pathway in which light-dependent glutamate release from ipRGCs is detected via a transiently expressed kainate receptor (GRIK3) in rod precursors localized to the inner retina. Communication between ipRGCs and nascent inner retinal rods appears to be mediated by transient hybrid neurites projecting from ipRGCs that sense light before eye-opening. These structures, previously referred to as outer retinal dendrites (ORDs), span the ipRGC-rod precursor distance over the first postnatal week and contain the machinery for sensory detection (melanopsin, OPN4) and anterograde neurotransmitter release (Synaptophysin and VGLUT2). Computational and histological assessment of human mid-gestation development reveal conservation of several hallmarks of an ipRGC-to-rod precursor pathway, including displaced rod precursors, transient GRIK3 expression in the rod lineage, and the presence of ipRGCs with putative neurites projecting deep into the developing human retina. Thus, this analysis defines a retinal retrograde signaling pathway that links the sensory environment to rod precursors via ipRGC photoreceptors, allowing the visual system to adapt to distinct lighting environments prior to eye-opening. Opn4+/- x Opn4+/- harem crosses were used to produce wildtype and knockout littermate pups raised in similar lighting environments (~300 lux). On the day of birth (P0), tail and finger clippings were harvested and mice were genotyped prior to determine appropriate genotypes. At P4, mice were sacrificed between ZT6-7 and retina were dissected in ice-cold oxygenated Ames’ media (95% air, 5% oxygen). Multiple retinas were pooled together from animals of the same genotype (4 retina per sample, 2 animals), and tissue was snap frozen in Trizol (100uL). Post-hoc genotyping was then performed on each sample to confirm genotypes. RNA extraction, library preparation, and sequencing (Illumina HiSeq 2 x 150bp) were all performed at GeneWiz. Samples with >500ng of RNA and a RIN > 8.5 were selected for further processing. The average read depth per sample was 42.7 ± 3.9 million reads. Subsequently, adapters and low-quality reads were trimmed/filtered using Timmomatic v0.39. Reads were pseudoaligned to the mouse reference genome (GRCm38) and counts were quantified using Kallisto v0.46.1. Differential gene expression analysis was performed using Sleuth v0.30.8.