More Opsin, or a Different Target? The Delivery Question Behind This Year's Optogenetics Nobel

Last week the 2026 Nobel Prize in Physiology or Medicine went to Karl Deisseroth, Peter Hegemann, and Georg Nagel for their discoveries concerning light-gated ion channels and optogenetics ^[1].

Figure 1 In vivo optogenetics for control of animal behavior using light ^[1].

Three days later, on October 8, the New England Journal of Medicine published what reads almost like that prize's clinical sequel: results from a ten-patient cohort receiving an optogenetic therapy for advanced retinitis pigmentosa, led by José-Alain Sahel at the University of Pittsburgh and Botond Roska at IOB in Basel ^[2]. In other words, the basic discovery and its first real test in blind patients arrived in the same news cycle. That is rare, and it is worth pausing on.

The trial design pairs a gene therapy that makes surviving retinal ganglion cells responsive to light with light-stimulating goggles that supply the stimulus. Across the ten participants, all blind from advanced RP, the primary safety endpoint was met, and six showed a clinically meaningful increase in light sensitivity. Several could detect or locate objects, find doorways, and follow lines while wearing the goggles, and brain recordings confirmed that visual signals were reaching the cortex. The team was careful about the ceiling: this did not restore normal sight, and the gains are partial.

Figure 2. The therapy combines gene therapy with light stimulating goggles to make surviving retinal ganglion cells responsive to light ^[2].

That combination of a supported mechanism, a real signal, and an honest limit is where the interesting work begins. For those of us building delivery tools and retinal models, a result like this is less a finish line than a sharper question. When a treatment improves light detection but visual function remains limited, what is actually holding it back? The answer is not academic, because it decides where a program should spend its next three years.

What Limits Optogenetic Vision Restoration: Light Sensitivity or Retinal Circuit Processing?

There are two very different stories that can produce the same partial result, and they suggest different development priorities.

Hypothesis 1 — The retrofitted cells are not sensitive enough to light.

If responsiveness is the limiting factor, the path forward is one the field knows well. You engineer better opsins, raise and stabilize expression, widen the transduction footprint so more of the target population carries the construct, and improve how efficiently the stimulation device drives those cells. This is fundamentally a problem of effective light responsiveness, which AAV-based delivery can help address through capsid choice, promoter design, and dose, within tolerability constraints ^[3].

Hypothesis 2 — Light sensitivity is being reintroduced at the wrong stage of the circuit.

This story is harder. Many optogenetic strategies, including one that confers light sensitivity on ganglion cells, place the new input at the retina's output stage rather than at its front end ^[4]. Ganglion cells are the layer that normally reports the results of retinal computation, including the spatial filtering, the contrast handling, and the center-surround organization that turns raw photons into structured signal. When you make that output layer directly light-responsive, you may be bypassing much of that native processing. If this mismatch between the input and the circuit is what caps visual quality, then adding more expression will not close the gap. You would be sending a louder message down a channel that was never designed to carry it. Crucially, the inner retinal circuitry that performs this computation remains functionally intact even after advanced photoreceptor degeneration, which is precisely why where you reintroduce light sensitivity matters [5].

Figure 3. Optogenetic interrogation of rd1 retinal circuits ^[5].

Can Higher Opsin Expression Overcome the Limits of Ganglion Cell Targeting?

Both mechanisms are almost certainly in play at once, and the balance probably shifts between individuals and disease stages. That is exactly why they are worth separating before a program commits its budget to one assumption. So here is the mechanistic question I would put to the scientists and R&D teams reading this. If you stimulate an output-layer cell directly and bypass part of the retina's native processing, how much of the remaining visual-quality gap can you actually recover by continuing to push opsin expression higher?

The honest answer is that nobody closes this question from a clinical readout alone. A cohort that gains light sensitivity but plateaus on object recognition is consistent with either story. Controlled preclinical comparisons can help distinguish their contributions by varying expression or cell targeting while accounting for other factors, including stimulation conditions and the state of the surviving retina.

Stress-Testing Hypotheses Before the Clinic

You cannot separate these two failure modes from a clinical readout alone. A cohort that gains light sensitivity but plateaus on object recognition is consistent with either story. This is why preclinical AAV delivery and controlled modeling platforms are non-negotiable before committing to a clinical path.

This is the work our group at Cyagen is set up to support. On the delivery side, the gene therapy platform uses deep-learning capsid models that predict pan-ocular expression and ocular penetrance, along with proprietary ocular-tropic capsids such as PN168. In one representative study, PN168 carrying an hGRK1-driven construct was delivered intravitreally in cynomolgus monkeys at 5E11 vg/eye and transduced rod photoreceptors, with tissue analyzed at four weeks.

Figure 4. A representative Cyagen study evaluating intravitreal delivery of PN168-hGRK1-GUCY2D-FLAG.

To test expression limits, ddPCR quantifies vector genomes and biodistribution, while protein measurements and light-evoked responses establish functional expression.

You can hold the cell target constant, vary vector dose, and ask whether visual-task performance keeps climbing or flattens. A plateau despite increased functional expression would prompt investigation of additional constraints, including circuit processing.

On the modeling and readout side, the retinal disease platform supplies the other half. An MNU-induced RP model reproduces progressive photoreceptor degeneration, outer nuclear layer thickness gives a quantitative structural endpoint, and both intravitreal and subretinal routes allow you to deliver to different layers on purpose.

Ocular phenotyping through ERG, fundus imaging, and retinal flat mounts turns those experiments into functional and anatomical data rather than anecdote.

With delivery and modeling in the same hands, you can run the comparison the two-bottleneck framing demands: move the transduced cell type or the spatial pattern of stimulation while keeping expression roughly constant, and see whether performance jumps. If it does, the limit was never about magnitude in the first place.

👉 [Explore Our Gene Therapy CRO Services]

👉 [Boost Your Study Through Our Comprehensive Ophthalmic CRO Solutions]

Beyond the Retina: Choosing Cell Targets for CNS Gene Delivery

The retina is accessible, but this tension exists everywhere in CNS-focused gene delivery. The recurring fork is whether the signal is too weak or whether it is being injected at the wrong point in a circuit that expects something more specific.

Cyagen supports these investigations with a broader portfolio of neural-circuit research tools, including ChR2-based tools that build on this year’s Nobel-recognized discoveries. For example, the RCL-ChR2_H134R/EYFP mouse (I001027) enables Cre-dependent optical activation of defined cell populations, while the AAV9-hSyn-DIO-hChR2(H134R)-mCherry vector offers flexibility for local delivery and comparisons across brain regions with an appropriate Cre source. These complementary approaches help researchers examine how targeted neuronal activation shapes circuit function.

The Nobel recognition and new clinical findings bring the field back to a practical question: does the next advance require stronger activation, a different cellular target, or both? Answering it requires measuring expression alongside the function it produces.

👉 [Discuss Your Next Preclinical Study with Our Experts]

Reference

[1] Nobel Assembly at Karolinska Institutet. Press release: Nobel Prize in Physiology or Medicine 2026 [Internet]. Stockholm: Nobel Prize Outreach; 2026 Oct 5 [cited 2026 Oct 8].

[2] Institute of Molecular and Clinical Ophthalmology Basel. Optogenetic therapy shows safe step forward in vision restoration [Internet]. Basel: IOB; 2026 Oct 7 [cited 2026 Oct 8].

[3] Sahel JA, Boulanger-Scemama E, Pagot C, Arleo A, Galluppi F, Martel JN, Esposti SD, Delaux A, de Saint Aubert JB, de Montleau C, Gutman E, Audo I, Duebel J, Picaud S, Dalkara D, Blouin L, Taiel M, Roska B. Partial recovery of visual function in a blind patient after optogenetic therapy. Nat Med. 2021 Jul;27(7):1223-1229. doi: 10.1038/s41591-021-01351-4. Epub 2021 May 24. PMID: 34031601.

[4] Berry MH, Holt A, Broichhagen J, Donthamsetti P, Flannery JG, Isacoff EY. Photopharmacology for vision restoration. Curr Opin Pharmacol. 2022 Aug;65:102259. doi: 10.1016/j.coph.2022.102259. Epub 2022 Jun 21. PMID: 35749908.

[5] Rodgers J, Hughes S, Lindner M, Allen AE, Ebrahimi AS, Storchi R, Peirson SN, Lucas RJ, Hankins MW. Functional integrity of visual coding following advanced photoreceptor degeneration. Curr Biol. 2023 Feb 6;33(3):474-486.e5. doi: 10.1016/j.cub.2022.12.026. Epub 2023 Jan 10. Erratum in: Curr Biol. 2024 Jan 8;34(1):224-233. doi: 10.1016/j.cub.2023.12.015. PMID: 36630957; PMCID: PMC7619324.