Cell replacement has long been viewed as one of the most promising routes to restoring vision in late-stage retinal degeneration. Yet despite encouraging preclinical work, photoreceptor transplantation has repeatedly run into the same obstacle: poor integration. Many transplanted cells survive in the subretinal space, but fewer than one percent typically integrate into the host retina, limiting functional recovery and raising questions about whether the field has been using the right donor cells.
A new study by researchers at the Scheie Eye Institute, University of Pennsylvania, takes aim at that question by examining the developmental diversity of photoreceptor precursor cells in the neonatal mouse retina. Rather than treating early postnatal photoreceptor precursors as a single population, the team used single-cell RNA sequencing and genetic lineage tracing to ask whether this critical developmental window contains distinct cell states, and whether any of them may be especially relevant for future regenerative therapies.
The investigators focused on postnatal day 2 to day 6 mouse retinas, a period previously associated with improved donor-cell survival and integration in transplantation studies. By analysing Crx-positive cells – a population enriched for developing photoreceptors – they identified three transcriptionally distinct precursor subgroups. These were interpreted as early, intermediate, and late photoreceptor precursor states, marked respectively by enrichment of Dll1, Neurod4, and Prom1.
The distinction was not merely computational. RNA-FISH (fluorescence in situ hybridization) confirmed that these markers had different spatial expression patterns within the developing neuroblast layer. Dll1 was expressed broadly, Neurod4 appeared more medially, and Prom1 was strongest in the apical neuroblast layer, consistent with a maturation trajectory toward the future outer nuclear layer.
Lineage tracing then showed that all three labelled populations could give rise to mature photoreceptors. At postnatal day 30, Dll1-, Neurod4-, and Prom1-lineage cells were found in the outer nuclear layer and co-expressed photoreceptor markers, with most adopting rod photoreceptor morphology. Prom1-labelled cells showed the strongest bias toward the photoreceptor fate, whereas Dll1 and Neurod4 lineages also contributed to some inner nuclear layer cell types, including amacrine, bipolar, and Müller glial cells.
The study also provided evidence that these three subpopulations may represent sequential developmental states rather than unrelated parallel routes. Flow cytometry and qPCR of lineage-traced cells across postnatal days 4, 6, and 8 showed progressive changes in marker expression. Dll1-labelled cells, for example, increasingly acquired Prom1 and CD73 expression over time, suggesting movement toward later photoreceptor maturation.
Importantly, the research did not stop at mouse models. By reanalyzing published single-cell datasets from human retinal organoids, the researchers found CRX-positive precursor populations bearing similarities to the murine early, intermediate, and late groups. This raises the possibility that comparable precursor states may exist in human stem-cell-derived systems – a crucial consideration if organoids are to serve as a scalable source of donor cells.
The findings do not yet identify the optimal transplantation cell type, but they do help to add clarity to the question. Current isolation strategies often rely on broad photoreceptor markers such as CD73, CD133, Crx, or Nrl, which may pool together cells at different developmental stages. A more refined approach could allow researchers to isolate the precursor state best suited for survival, migration, synaptic connectivity, and functional integration.
