If you lose retinal cells to disease, they stay lost. That rule is why conditions like retinitis pigmentosa and glaucoma are so hard to treat. Zebrafish donβt follow it. They regrow damaged retinal neurons using support cells called MΓΌller glia, and mammals have the same cells. They just donβt use them. A team at KAIST in South Korea may have found out why: a protein called Prox1, which neighboring neurons pass to MΓΌller glia after injury, keeps them from repairing the retina. When researchers blocked it in mice, new retinal cells formed and vision improved for a time. The results are early and limited to animals, but they shift the question from whether a mammalian retina can regenerate to what is stopping it.
Why the Human Retina Canβt Repair Itself
Once retinal cells die, they donβt come back. Thatβs the problem in retinitis pigmentosa, where rod photoreceptors degenerate, and in glaucoma, where retinal ganglion cells are lost. Current treatments can slow damage in some cases, but none replace the lost cells. Most patients never get that vision back.
It doesnβt have to work this way. Zebrafish, like other fish and amphibians, regrow damaged retinal neurons on their own. The cells responsible are MΓΌller glia, support cells that span the retina. After an injury, zebrafish MΓΌller glia reprogram into progenitor cells, multiply, and become new neurons, including rod photoreceptors and ganglion cells.
Mammals have MΓΌller glia too. They just donβt finish the job. After injury, mouse MΓΌller glia partly change and then return to a resting state, and no new neurons form.
Scientists had already found that mammalian MΓΌller glia lack certain activating factors, such as a gene called Ascl1. What they hadnβt identified was an active blocker. A KAIST team, publishing in Nature Communications in March 2025, found one: a protein called Prox1.
Prox1 is a transcription factor, a protein that switches genes on or off. During normal development, it tells progenitor cells to stop dividing and settle into their final role as neurons. Thatβs useful in an embryo and counterproductive in an injured adult retina.
The researchers found Prox1 building up in MΓΌller glia in injured mouse retinas and in a retina from a donor with retinitis pigmentosa. It did not build up in regenerating zebrafish. Where Prox1 accumulated, regeneration stalled, which made it the leading suspect.
How Researchers Removed the Block
The first surprise was where the Prox1 came from. Injured MΓΌller glia had more Prox1 protein, but their own Prox1 gene wasnβt making more of it. Messenger RNA levels stayed flat. The glia were taking Prox1 up from somewhere else.
That source turned out to be neighboring retinal neurons. Bipolar cells, horizontal cells, and amacrine cells all make Prox1, and bipolar cells are the largest group. Proteins like Prox1 can move from one cell to another, and MΓΌller glia absorb it after injury.
The team then tested this two ways.
Method 1: Delete the source.Β They removed theΒ Prox1Β gene from bipolar cells in mice. Prox1 levels in MΓΌller glia dropped, and after injury far more MΓΌller glia began dividing. The miceβs bipolar cell counts and visual acuity stayed normal, so the deletion didnβt harm healthy retinas.
Method 2: Intercept the protein. They built an antibody fragment that binds Prox1 outside cells and delivered it with AAV, a virus commonly used to carry genes into the retina. The antibody lowered Prox1 in MΓΌller glia without changing it in neurons, and MΓΌller glia proliferation increased again.
The reverse experiments pointed the same way. Injecting Prox1 protein back into mouse eyes cut MΓΌller glia division. Injecting it into zebrafish eyes reduced progenitor cells and slowed their normal regeneration.
Single-cell sequencing confirmed what the dividing cells were. In injured retinas lacking Prox1 transfer, MΓΌller glia shifted into retinal progenitor cells, the same cell type zebrafish use to rebuild. A cluster of these cells appeared only in the treated retinas, and it linked to the lineages of bipolar and amacrine neurons.
What Happened in the Mouse Models
The team tested the antibody on two mouse models of retinitis pigmentosa, and the results depended on how fast the disease moves. Both models showed the same Prox1 buildup in MΓΌller glia seen in injured retinas.
Fast-progressing disease (rd10 mice). Treated mice had more rod photoreceptors than untreated ones. Cell death didnβt decrease, so the rods werenβt simply being saved. Labeling showed newborn rods, which appeared only in treated retinas. MΓΌller glia had produced new cells, including rod bipolar cells.
Function followed. Treated mice had stronger electrical responses from the retina and better visual acuity than controls. But the gain disappeared by two months of age. The new rods carried the same genetic mutation and degenerated within about a week. Treating mice at two months, after most degeneration had already happened, produced no recovery at all.
Slow-progressing disease (Rp1 mice). These mice start losing rods around two months of age and decline in vision over a year. Treated at two months, they kept normal visual acuity while control mice gradually lost it. Treated at six months, when vision was already declining, they regained acuity comparable to healthy littermates within a month. Newborn rods again appeared only in treated retinas.
Rod numbers and retinal responses did not return to normal. Cone photoreceptors, which handle daylight and color vision, did increase, along with the electrical responses they generate. Rods release a factor that keeps cones alive, so the authors suggest even a partial rod recovery may have been enough to protect cones. The vision gains faded after about six months.
Takeaway for patients: slower disease gave regeneration more time to work, and early treatment mattered.
Limits and What Comes Next
This is early work, and the headline overstates where it stands. Everything described so far happened in mice, and the treatment is not available to patients.
The human evidence is thin: The only human data in the study is a retina from one 79-year-old donor with retinitis pigmentosa, compared with one 83-year-old healthy donor. Prox1 was higher in the patientβs MΓΌller glia, which fits the mouse findings, but two donors canβt tell us how reliably this holds across people. The paper doesnβt report any human treatment.
The effect was temporary: The vision gains faded within about six months. The authors trace this to the promoter in their viral construct, which switches off over time and stops the antibody from being made. They say it needs to be redesigned for longer-lasting treatment.
Regeneration was modest: Blocking Prox1 produced roughly 3 to 5 dividing MΓΌller glia per retinal area in mice, compared with about 28 in zebrafish. The authors conclude that Prox1 is one barrier, not the only one. They point to Notch inhibition after reprogramming and sustained Yap/Taz activity as possible additions. Rod photoreceptors also didnβt return to normal numbers.
Several questions remain open: Researchers donβt yet know how much Prox1 is enough to block regeneration, how MΓΌller glia capture it from neighboring cells, or which proneural genes drive the conversion, since Ascl1 was not restored. Real-time tracking of MΓΌller glia is still needed to confirm how the new neurons form.
For patients, the practical point comes from the mouse results: treatment worked when started early. If this approach reaches people, it will likely depend on catching disease before most photoreceptors are gone. That makes regular eye exams and prompt evaluation of vision changes worth taking seriously now.
My Personal RX on Protecting Your Vision Before Damage Sets In
This research is in mice, but it points to something you can act on now. Human retinal cells donβt regrow on their own, and in the experiments, treatment worked best when started early. Protecting the vision you have is the real prescription.
Hereβs what Iβd tell you in my office.
- Book a Dilated Eye Exam: Glaucoma often has no early symptoms, and many retinal conditions are found only on exam. Go sooner and more often if you have diabetes or a family history of eye disease.
- Report Vision Changes Quickly: Trouble seeing in dim light or shrinking side vision needs prompt evaluation. Be wary of clinics selling unproven βretina regenerationβ injections.
- Know Your Family History: Some retinal diseases are inherited. Ask relatives about vision loss and tell your eye doctor.
- Eat for Your Eyes: Leafy greens, eggs, colorful vegetables, and fatty fish provide lutein, zeaxanthin, and omega-3s.
- Fill Gaps With Eye Max: My Eye Max supports macular health and circulation to the eye, alongside a good diet.
- Wear UV-Blocking Sunglasses: Choose lenses that block UVA and UVB, and wear them year-round.
- Protect Blood Vessels and Quit Smoking: The retina depends on healthy blood flow. Keep blood pressure, blood sugar, and cholesterol in range, and donβt smoke.
- Rest Your Eyes From Screens: Every 20 minutes, look at something 20 feet away for 20 seconds. It wonβt prevent retinal disease, but it eases strain.
- Treat the Retina as Part of the Brain: It is an extension of the central nervous system. My Gut Brain Reset Bundle pairs MindBiotic and Digestive Enzymes with Eye Max for gut, brain, and vision support.
- Learn the Gut-Brain Connection: My book Heal Your Gut, Save Your Brain shows how daily habits protect your nerve cells.
Sources:
- KAIST. (n.d.).Β KAIST develops retinal therapy to restore lost vision. Kaist.Ac.Kr. Retrieved October 27, 2026, from https://www.kaist.ac.kr/newsen/html/news/?mode=V&mng_no=45350
- Lee, E. J., Kim, M., Park, S., Shim, J. H., Cho, H.-J., Park, J. A., Park, K., Lee, D., Kim, J. H., Jeong, H., Matsuzaki, F., Kim, S.-Y., Kim, J., Yang, H., Lee, J.-S., & Kim, J. W. (2025). Restoration of retinal regenerative potential of MΓΌller glia by disrupting intercellular Prox1 transfer.Β Nature Communications,Β 16(1). https://doi.org/10.1038/s41467-025-58290-8
