Restoring vision with optogenetics

When Robin McCoy was 50, she was diagnosed with the disease that had taken her mother’s sight. McCoy had inherited a genetic mutation that causes retinitis pigmentosa (RP), a condition in which the retina’s light-sensing cells — its rods and cones — die or stop responding to light.

First, as her rods fell silent, McCoy’s night vision deteriorated. Then, as her cones followed suit, her daytime vision faltered. She developed tunnel vision, which worsened over time.

The day McCoy’s optometrist told her she was legally blind — in 2019, when McCoy was 58 — she had driven to his office. That was the last time she drove. The following year, she began using a cane after she fell and broke her hand, having walked into a parked car that she simply hadn’t seen.

Researchers are exploring several strategies for either halting RP’s progression or restoring affected people’s sight, and McCoy started applying to join clinical trials. She wasn’t picky about what approaches they were testing. “I really didn’t care,” she says. “I just wanted to not go completely blind.”

After one hopeful e-mail enquiry, McCoy received a call back the next day. It was Sheila Nirenberg, a neuroscientist at Weill Cornell Medicine in New York City. Through Bionic Sight — a New York-based start-up that emerged in 2019 from her academic work — Nirenberg is running an early-stage clinical trial of a technique called optogenetic therapy.

The aim is to overcome the loss or dysfunction of the eye’s light-reactive cells by introducing a light-sensing protein into surviving retinal cells. Depending on which cells are targeted, this will either restore their natural responsivity to light, or create a new visual pathway by making normally light-insensitive retinal neurons respond to illumination.

In February 2024, McCoy attended a clinic on Long Island, New York, and had a harmless genetically modified virus injected into her right eye. By then her visual field spanned just 5 degrees, which she likens to viewing the world through two drinking straws.

At her nine-month follow-up with Nirenberg, McCoy was jubilant: “My vision got so bad a couple of years ago, I could not read a normal book. Now — it’s crazy — I can sit down on the beach in San Diego and read a book again.” She learnt that week that her visual field had increased to 30 degrees. On an optometrist’s chart of rows of shrinking letters, she’d gained two lines of vision. And she’d made huge progress in walking through a maze of tiles laid across the floor of Nirenberg’s laboratory. At first, she says, “it took me over a minute to get through in the lowest light, and I made four mistakes. Now I’m going through in 30 seconds, no mistakes.”

Wider results of the trial are not yet published, but in January the US Food and Drug Administration (FDA) designated the treatment as a ‘regenerative medicine advanced therapy’, which should accelerate its development. It marks a significant step in the 20-year pursuit of optogenetic therapies for RP. Clinical trials have previously had mixed success, with several companies exiting the space.

Now, many academic and industry researchers are investigating variations on the idea, in parallel with Nirenberg’s work. Differences include which light-sensitive protein is introduced; which cells this protein is placed in; and whether to use auxiliary technologies such as goggles that deliver precise flashes of light to better activate the protein. “It’s a convergence of many fields of science,” says José-Alain Sahel, an ophthalmologist at the University of Pittsburgh in Pennsylvania.

In this context, Nirenberg’s success is surprising. McCoy’s visual gains, and those of others in the Bionic Sight trial, go beyond what might be expected from what is one of the simpler trials under way. Nirenberg herself was initially confused. “It was working in a way that I didn’t understand,” she says. “It kept me up at night, thinking, ‘Why is this working?’ ”

The right way to do it

RP is a collection of genetic conditions caused by hundreds of mutations in more than 80 genes, and preventing carriers from going blind could require tailored approaches for each form. Optogenetic therapy, however, is designed to restore lost vision, and therefore could help anyone with RP.

Optogenetic therapy got its start at Wayne State University in Detroit, Michigan, in the lab of visual neuroscientist Zhuo-Han Pan. When he moved there in 1999, he says, artificial electrical implants were the darlings of vision restoration efforts. But having observed that some retinal neurons are spared in RP, Pan thought that it might be possible to generate higher-resolution visual signals by making the surviving cells sensitive to light. “The idea is very simple,” he says, but no one knew how to do it then.

That changed in 2003, when scientists in Germany described channelrhodopsin-2, a protein that forms a channel through the membrane of cells in green algae, and that is activated by blue light¹. Channelrhodopsin revolutionized experimental neuroscience, enabling researchers to get neurons to fire by exposing them to pulses of light. For Pan, however, it was just the right molecule for testing his vision-restoration theory.

He loaded the channelrhodopsin gene into a virus targeted to neurons, and delivered it into the eyes of blind mice. Pan saw not only that the protein made retinal neurons respond to flashes of blue light, but also that a signal made it to the rodents’ visual cortex².

This was a crucial proof of concept. But turning optogenetics into an effective therapy demanded refinements. Key among them was to work out which cells should be made light-sensitive.

“People think of the eye as being like a camera, and that the retina is just the film in the back,” says Nirenberg. “But it’s not — it’s a little image-processing device.” This processing is achieved by a network of cells (see ‘A question of targeting’).

The key light-sensing cells of the retina are rods and cones. When light strikes them, these photoreceptors are inhibited, which changes the signals they relay to a layer of neurons called bipolar cells. ON bipolar cells are activated by light, and OFF bipolar cells are inhibited by it. The cells’ opposing functions help the retina to detect edges and contrast to better distinguish objects.

Bipolar cells then drive activity in retinal ganglion cells, the long axons of which form the optic nerve, and it is their firing patterns that are transmitted to the brain. Taking in signals from both kinds of bipolar cell enables retinal ganglion cells to respond to complex features in the visual world.

A central question that optogenetics researchers face is which of these various cells should receive the foreign light-sensitive protein. The closer to the photoreceptors one intervenes, the more natural retinal signal processing is retained, explains Botond Roska, a neuroscientist at the University of Basel in Switzerland. In theory, then, the best option for reinstating high-acuity daytime vision is to target an inhibitory light-sensing protein to surviving cones that have lost their photosensitivity — but a person with RP might have few of these left to work with.

The next best thing would be to place an excitatory light-sensing protein, such as channelrhodopsin, into ON bipolar cells, Roska says, but this is complicated by the need to leave OFF bipolar cells alone. “We don’t want to activate the ON and OFF cells simultaneously, because this never happens in nature.”

The final choice is to place channelrhodopsin into retinal ganglion cells, so that light activates them directly. However, this skips all the complex signal processing that the retina ordinarily does before this point. Combined with the physical arrangement of ganglion cells in the retina, Roska says this significantly limits the richness of vision that could be achieved.

His work in mice backs this up. In 2008, Roska showed that introducing channelrhodopsin into ON bipolar cells in a mouse model of RP restored visual ability³. And in 2010, he showed that inserting an inhibitory light sensor into cones restored higher-acuity vision⁴.

But researchers racing to try this in the clinic faced a major barrier. Although viruses targeting human retinal ganglion cells existed, viruses that target human ON bipolar cells or cones did not.

The light-sensitive proteins being delivered also had issues. Cones generate colour vision because they each contain one of three distinct sensors that respond to red, green or blue light. Channelrhodopsin, by contrast, is activated only by blue light, limiting the best possible outcome to monochromatic vision.

More concerningly, channelrhodopsin is picky about the intensities of light it responds to. “When you put channelrhodopsin in human eyes, you have to use a very, very bright light,” Pan says. Dynamic range is limited too. There is a mere 100-to-1000-fold difference between the intensity of light needed to activate the channel and the intensity that saturates it. Cones, by comparison, respond to ranges of intensities that are orders of magnitude larger, which is crucial to allowing people to see in both dim and bright conditions.

These issues prompted efforts to bioengineer channelrhodopsin derivatives without these limitations. The resulting proteins are sensitive to different wavelengths and to lower intensities of light, but so far none has improved the protein’s dynamic range.

To compensate, some researchers have turned to wearable technology. For instance, some optogenetic therapies involve goggles that contain cameras pointed at the world and light emitters that point into the wearer’s eyes. The device converts what it sees into flashes of light of the correct wavelength and intensity to maximally activate the inserted protein. Other efforts use goggles that simply regulate the amount of light entering the eye and hold it in the protein’s functional range. To discover which combination of cell type, light-sensing protein and auxiliary technology works best requires human volunteers.

Clinical trials matter

Despite a flurry of commercial interest and some initial trials in the mid-2010s, the only published clinical result in optogenetics so far is a 2021 report about a participant in a trial by Paris-based Gensight Biologics⁵. Working with Roska and Sahel, Gensight targeted retinal ganglion cells and introduced a channelrhodopsin derivative called ChrimsonR, which is sensitive to amber light. Such light is less damaging to the eye than is blue light, and the company used goggles that converted recorded images into flashes of amber light.

The individual, who had been completely blind for two years, regained some rudimentary, dot-like vision. “Initially, what they see is like a starry sky,” Sahel says. “Then they have to align the dots and understand what it means.” With time and training, the volunteer learnt to identify, count and pick up objects laid out before him. Navigating an unfamiliar room using such limited visual information, however, was too difficult.

The trial involved 15 participants and their responses varied, says Roska, who is currently preparing a full report. But he sees it as an important demonstration that the brain can make some sense of signals as unnatural as those produced by light-sensitive retinal ganglion cells.

Pan, Roska and Sahel have all been refining their technologies. Pan has developed a more light-sensitive channel, CoChR3M, that allows blind mice to see under ambient light. Clinical testing by Ray Therapeutics in Berkeley, California, is assessing whether it has a similar effect in people’s retinal ganglion cells.

Pan is also pursuing ways of targeting ON bipolar cells, as is Roska, who has taken advantage of a technique for keeping human retinas alive for weeks post-mortem to screen for viruses and promoters that specifically target ON bipolar cells or cones. He is excited about results showing that introducing an inhibitory light-sensing protein into cones reconstituted the full array of retinal responses to various light stimuli.

Roska says his group has shown that around 60% of people with RP retain significant numbers of cones, suggesting that most people with the disease could benefit from cone-targeted optogenetics. This is the goal of the newly formed Basel-based company Rhygaze, through which Roska expects to begin clinical trials within three years.

One company claims to be ahead of the pack. Last October, Nanoscope Therapeutics in Dallas, Texas, announced that its randomized trial of optogenetic therapy in 18 people with RP was a success. Using a channelrhodopsin derivative called multi-characteristic opsin, the company says that it successfully targeted ON bipolar cells and that ten participants achieved significant visual improvements.

Nanoscope has presented its findings at conferences, but Roska and Sahel are frustrated that the company has not yet published detailed data in a peer-reviewed journal. And Nirenberg is underwhelmed by the extent of reported visual gains. The company did not respond to requests for comment from Nature.

A quest for answers

Having submitted data to the FDA, Nirenberg says she is now preparing a paper on the surprising improvement of McCoy and others in Bionic Sight’s ongoing trial.

When Nirenberg first conceived her optogenetic strategy, custom-designed goggles were an elemental part of it. Her target was retinal ganglion cells, and she thought that if those cells were made light-sensitive, then a device could transform camera images into coded flashes that activate those cells and send natural-seeming signals to the brain.

But right now, those goggles are on the back burner. “I tried them on,” says McCoy. “They were OK, but I didn’t need them.”

More from Nature Outlooks

Nirenberg thinks that in the trial participants who have gained the most, an introduced channelrhodopsin derivative called chronos is boosting residual retinal activity. The participants were not completely blind to begin with, indicating that they retain some functional photoreceptors. Nirenberg thinks that their retinal circuitry was still working, but too weakly to activate most retinal ganglion cells, leading to severely diminished sight. Placing the chronos protein in the retinal ganglion cells, she says, might excite them just enough that more respond to the remaining weak input from photoreceptors and bipolar cells, and fire off signals to the brain. The theory makes biological sense, but Nirenberg doesn’t know whether the chronos channel is responding to light, or whether its presence alone has a sufficient excitatory effect on the retinal ganglion cells.

If this is what is happening, it’s unclear why the effect has not been seen in other trials. Perhaps participants have had less residual photoreceptor activity to boost, or perhaps the biophysical properties of the channel used are crucial. It is also uncertain whether RP’s degenerative processes will continue and eventually wipe out these visual gains; researchers will have to closely monitor study participants over years to see how the disease progresses.

With multiple approaches under investigation and trials likely to examine each of them soon, the next three to five years should provide clinical data that address many key questions. Nirenberg is grateful to the volunteers who are helping them to get there. “The patients are wonderful collaborators,” she says. “They’re partners in this process.”

McCoy, too, is thinking of others. Tests have shown that she has passed her RP gene to her daughter, who is 43 and has three daughters of her own. “I’m doing this, not just for me,” McCoy says, “but for my daughter, my family, and all of my RP friends.”