A 3D human retina model reveals a non–cell-autonomous and non-neuronal mechanism of photoreceptor loss in a lysosomal storage disorder

Disruption of the photoreceptor–retinal pigment epithelium (RPE) interface, with loss of photoreceptor outer segments (POSs) in the retina, is a pathological hallmark of several neurodegenerative and retinal diseases, including lysosomal storage disorders like juvenile neuronal ceroid lipofuscinosis (CLN3) disease. However, in vitro stem cell models that enable investigation of the photoreceptor-RPE interface are lacking. Here, we developed a 3D human pluripotent stem cell–derived retina organoid–RPE (hPSC-RO-RPE) model to investigate the photoreceptor-RPE interface in healthy and diseased tissues. Using this 3D hPSC-RO-RPE retina model, we showed that the most common disease-causing CLN3 mutation ( CLN3 Δ ex7-8 ) leads to reduced levels of acid ceramidase (AC), sphingosine 1 phosphate, and POS loss. Furthermore, by analyzing control versus CLN3 mutant ( CLN3 Δ ex7-8 ) hPSC-derived RPE monoculture and hPSC-RO-RPE model, (i) we identified a pathogenic role of AC-mediated lysosomal sphingolipid metabolism in promoting CLN3 disease pathobiology, and (ii) we showed a cell autonomous role of the RPE dysfunction in promoting POS loss/retina degeneration in CLN3 disease. We validated the molecular and structural changes observed in the CLN3 Δ ex7-8 hPSC-RO-RPE model in the CLN3 miniswine model ( CLN3 Δ ex7-8 ) and donor eyes from two patients with CLN3 disease. High-resolution retinal imaging of the living eye in two patients with CLN3 disease suggested decreased RPE autofluorescence in early-stage CLN3 disease. Treatment with recombinant human acid ceramidase (rhAC) ameliorated photoreceptor degeneration in both the CLN3 disease RO-RPE model and CLN3 miniswine eyes. These findings suggest that rhAC could be a therapeutic approach for retinal degeneration in CLN3 disease.