The study focused on Müller glia, support cells that can revert to a progenitor-like state and replace damaged retinal neurons in animals such as zebrafish. Mammalian Müller glia normally stop short of that regenerative program.
How PROX1 blocks the process
The researchers found PROX1 in Müller glia from degenerating mouse and human retinas, but not in the same cells in regenerating zebrafish retinas. In mice, the glia did not appear to make the protein themselves. Instead, PROX1 moved into them from neighboring retinal neurons.
Blocking that transfer let Müller glia in injured mouse retinas re-enter the cell cycle and produce retinal neurons. The team also found the response was much smaller than natural regeneration in zebrafish, suggesting that PROX1 is one barrier rather than the whole switch.
What the eye injection did
For the treatment experiments, the team injected an adeno-associated virus called AAV2 into the eye. The vector carried a gene for an antibody designed to trap PROX1 outside cells before it could enter Müller glia.
The results depended on the disease model and timing. In mice with a fast-progressing form of retinitis pigmentosa, the treatment generated photoreceptors and briefly improved visual function when given early. The effect was gone by postnatal day 60, and treatment did not restore vision when it began after retinal degeneration was already complete.
In a slower-progressing model tied to the Rp1 gene, treated mice maintained normal visual acuity or recovered it within one month after treatment. The researchers also detected newly generated rod photoreceptors.
The improvement had a time limit
The paper says the visual improvement in the slower model did not persist beyond six months after treatment. That wording is more cautious than KAIST's March 2025 news release, which described vision restoration lasting more than six months.
The authors linked the loss of benefit to silencing of the promoter used in the viral construct and said a redesigned promoter would be needed for longer-lasting expression. They also reported that blocking PROX1 alone produced far fewer proliferating Müller glia or progenitor cells than were seen in regenerating zebrafish retinas.
No people received the treatment. The paper reports observations in donated human retinal tissue, but its regeneration and vision experiments were conducted in mice. Five authors disclosed that they were co-founders or employees of Celliaz Inc., a company developing anti-PROX1 therapies for retinal disease.
