A Systematic Review of Optogenetic Vision Restoration: History, Challenges, and New Inventions from Bench to Bedside

Abstract

Blindness due to rod-cone dystrophies is a significant comorbidity and cause of reduced quality of life worldwide. Optogenetics uses adeno-associated viral (AAV) vectors to bypass lost photoreceptors and transfect remnant cell populations of the degenerated retina aiming to restore vision via the ectopic expression of opsins. The optogenetic targeting of retinal ganglion cells (RGCs) has been remarkably successful and several studies have advanced to clinical trials over the recent years. The inner retina and specifically ON bipolar cells represent even more appealing targets due to their intrinsically coded tasks in parallel processing and fine-tuning of visual signals before reaching the output: RGCs. However, present success with pursuing inner and outer retinal cells for optogenetic vision restoration is limited by multiple factors, including AAV tropism, promoter specificity, and retinal morphofunctional remodeling. Here we provide a review of the evolution of optogenetics, its greatest challenges, and solutions from bench to bedside.

In the broadest possible sense, optogenetic technology combines optics, genetics, and bioengineering to study intact, living neural circuits by introducing new genetic information to the target cells and artificially expressing light-sensitive proteins with the purpose of stimulating them by light (Deisseroth et al. 2006; Simunovic et al. 2019).

Evolution refined nature's apparatus for translating light stimuli into electrical signaling—with the aid of photoreceptive neurons expressing light-sensitive proteins—from simple phototaxis into image-forming vision. Mammalian vision is triggered by light entering the eye and activating opsins naturally occurring in the photoreceptors of the retina (i.e., rhodopsin and cone opsins), which bind a retinoid chromophore that changes conformation and thus initiates a chain of molecular events initially in the photoreceptor, later processed in three cellular and two plexiform layers of the neural retina to eventually be conveyed by the optic nerve into the brain. Therefore, a very obvious field that is continuously developing and benefiting from optogenetics today is vision science with a very straightforward aim: vision restoration.

The concept underpinning optogenetic vision restoration in inherited retinal degenerations (IRDs) is to use virus vectors (most commonly at present—adeno-associated virus [AAV]) to deliver opsins into surviving retinal neurons, which normally respond to synaptic transmitter molecules, not photons. Expression of the ectopic, light-sensitive transgene is driven and restricted by cell-specific promoters with the purpose of turning second- and/or third-order neurons into light sensors or rarely to resensitize degenerated, “dormant” photoreceptors.

Here we provide a systematic review of the history and evolution of optogenetics, its greatest challenges, and solutions from preclinical to clinical studies.

OPSINS

There are two broad types of opsins suitable for vision restoration: type 1 or bacterial opsins and type 2 or animal opsins. In addition, there are chemical photoswitches termed PORTLs (photoswitchable orthogonal remotely tethered ligands) that combine a photochromic ligand (typically azobenzene) with a tether, a glutamate molecule, and a covalent attachment “tag” such as SNAP-tag to impart light sensitivity to a G-protein-coupled receptor (GPCR). In this respect, azobenzene-based photoswitches are promising nanoscale tools for neuronal photo stimulation (Broichhagen et al. 2015).

Examples of type 1 opsins are channelrhodopsins (ChRs), halorhodopsins (NpHRs), and archeorhodopsins (Archs). NpHRs are chloride pumps most effectively activated by absorbing light in the yellow spectrum resulting in hyperpolarization of the cell caused by the increased influx of chloride ions. Archs, on the other hand, are light-activated, intra-, or extracellular proton pumps with highest sensitivity to yellow-to-green wavelengths, again resulting in cellular hyperpolarization (Duebel et al. 2015). Finally, ChRs are the only known light-gated ion channels (Lin 2011). The most appealing feature that makes ChRs widely used in optogenetic vision restoration is resistance to photobleaching. Bleaching desensitization refers to the temporary inhibition of phototransduction initiation in rod photoreceptors following exposure to bright light (photolysis) due to the photoisomerization of the chromophore 11-cis retinal into all-trans retinal (bleaching product) that remains bound to the apoprotein, opsin. Rod regeneration from bleaching depends on the removal of all-trans retinal and the delivery of 11-cis retinal to rhodopsin's chromophore binding site (Pepperberg 2003). In ChRs, the covalently bound all-trans retinal isomerizes to 13-cis retinal hence preventing bleaching, which is a well-known issue with type 2 opsins. ChRs employed in optogenetic vision restoration typically have peak light sensitivities in short wavelength illumination limiting their potential uses in natural environmental settings, where blue light is less abundant than longer wavelength light (Wyszecki and Stiles 1982). Additionally, ocular surfaces and especially ocular media exhibit significant absorbance properties in the blue, short wavelength spectrum (Norren and Vos 1974). Another challenge in using naturally occurring ChRs—and in fact all type 1 opsins—for optogenetic vision restoration is that their light sensitivities are not sufficient for activation in environments with typical ambient light levels. One might think that this challenge is easy to overcome by artificial means, incorporating amplifying or intensifying the light that enters the eye with goggles; however, this solution could pose another threat to the surviving photoreceptors in a patient: phototoxicity. Phototoxicity (type I and II) happens when the eye gets exposed to bright, short-to-medium wavelength light. The difference between the two types lies within exposure time and light intensity: type I phototoxicity is triggered by prolonged exposure, whereas type II phototoxicity is known to affect the retina after a shorter irradiation to more intense light (Ham et al. 1984). Therefore, using high-intensity light of the short wavelength spectrum to activate wild-type ChRs in retinal cells is a cause of significant concern. Today's ChR variants are modern designer proteins with modifications that are shifting their peak sensitivities to longer wavelength light, decreasing the risk of phototoxicity as well as better complementing the distribution of visible light spectra in naturally lit environments. The perfect optogenetic molecule would in fact be able to deliver operational limits that more faithfully mimic human cone vision in commonly encountered light conditions that can range from 10⁻⁴ (moonless night) to 10⁵ (direct sunlight). Currently though, type 1 opsins provide an operational range that is completely outpaced by the physiological human visual system. However, modern red-shifted designer molecules have allowed the development of devices including light-intensifying goggles, that with minimal risk of phototoxicity are able to modify and adjust the incoming light intensity to match the sensitivity of type 1 opsins (Yue et al. 2016). These devices are also able to modify other stimulus properties like wavelength, as well as spatial and temporal properties. Regarding temporal response characteristics, the ideal optogenetic molecule will function at response kinetics of at least the same order as retinal neurons.

Type 2 or mammalian opsins also have the potential to be used in optogenetic vision restoration. They do in general have higher light sensitivities than type 1 opsins, although high-reaction kinetics is a feature of bacterial opsins that is hard to beat. Quite intriguingly, all type 1 opsins exhibit faster response kinetics than photoreceptors (Busskamp et al. 2010), ∼35 msec for dark adapted cones (Hood and Birch 1993a) and ∼190 msec for dark adapted rods (Hood and Birch 1993b). Human opsins—melanopsin, rhodopsin, and cone opsins—may also be ectopically expressed in cells surviving photoreceptor loss in IRDs. Human opsins are G-protein-coupled receptors that are presumed to naturally enhance sensitivity by signal amplification. Besides having excellent light sensitivity, these opsins have another major advantage from a therapeutic viewpoint. These proteins are naturally expressed by human tissues and are less likely to be recognized as foreign antigens by the patients’ immune system; therefore, these are not anticipated to elicit a significant inflammatory (uveitis) response. Reaction kinetics—especially in melanopsin, which is the slowest known animal opsin (De Silva et al. 2017)—is in fact a painful tradeoff when it comes to choosing from type 1 and 2 opsins in optogenetics. A further issue with type 2 opsins—as mentioned above—is bleaching and recovery happening at a particularly slow rate in rhodopsin, which is a process fundamentally dependent on the retinal pigment epithelium (RPE). It remains unclear whether the RPE will continue to deliver the 11-cis retinal chromophore to the retina after the photoreceptors have degenerated. It remains possible that the RPE will down-regulate this function as the photoreceptors are lost. However, the complement of intact photoreceptor outer segments in a healthy retina represents a huge “sink” for 11-cis retinal, and in the absence of these cells, the amount of 11-cis required for “charging” the optogenetic opsin when expressed in inner retinal neurons should only be a small portion of that required for supplying an intact outer nuclear layer. In addition, a Müller glia–driven cone photopigment recycling pathway (Wang and Kefalov 2011) may be sufficient when it comes to employing cone opsins in optogenetic vision restoration.

VECTORS

AAV vectors are the best studied and most often used gene delivery vectors in optogenetics because of their lack of pathogenicity, extremely low incidence of host genome integration, and long-term episomal persistence in the target tissue/cell.

Current AAV production methods generate titers (>10e13) high enough to be used in human subjects with the small volumes required for intraocular administration (Ferreira et al. 2014). AAV was accidentally discovered in 1965 as a contaminant in an adenovirus preparation and by the 1980s it was recognized and repurposed “from defective virus to effective vector” (Gonçalves 2005). AAV is a dependovirus and without the presence of a helper virus like adenovirus (hence the name) it is unable to replicate, which is another reassurance of its safe application as a vector in human gene therapy. In addition to its natural role as a “helper” virus that requires coinfection with adenovirus for replication, the deletion of the “rep” and “cap” genes from the therapeutic construct further cripple its replication capabilities.

Of the nine naturally occurring AAV serotypes, 12 human and >100 animal AAV capsid variants have been identified to date. While all AAV serotypes are able to infect the RPE, some show tropism toward different retinal cell types (e.g., AAV2, 5, 7–9 are capable of transfecting photoreceptors) (Day et al. 2014), while AAV2 shows a propensity to transfect inner retinal cells (Yin et al. 2011). The RPE may be particularly permissive to AAV transduction from its apical (subretinal) face as these are highly phagocytic epithelia with elaborate apical microvilli and lamellipodia.

Fortunately, opsins are encoded by genes short enough to be packaged into AAV vectors, including the required promoter elements and inverted terminal repeats (ITRs). AAV vectors are known to have a maximum gene packaging capacity of 4.7–5 kb without compromised transduction efficiency and truncated or fragmented genomes (Wu et al. 2010). It has been shown that when a cargo of larger than 5.3 kb was packaged into AAVs, their capacity to transduce targets was significantly reduced, due to a preferential degradation process of oversized vector particles (Grieger and Samulski 2005). Although the same study by Grieger and Samulski found evidence that successful protein expression from overpackaged AAVs with fragmented genomes was possible after the reconstitution of overlapping fragments, this is not an acceptable option for clinical application with strict FDA requirements for complete sequence validation in the clinical grade good manufacturing practice (GMP) vector product. Due to this inconvenient limit in encapsidated genome size, however, it is nearly impossible to apply AAV-based gene therapy to treat a variety of diseases like congenital stationery night blindness type 2A caused by mutations in the CACNA1F gene (Carr 1974), which is longer than 35 kb and thus impossible to fit into an AAV capsid. Understandably, there is a great deal of interest in increasing the size of therapeutic transgenes while at the same time benefiting from the numerous advantages of AAV transduction. As a result, a so-called dual AAV vector system has been and is to date under development to address the aforementioned issue (Chamberlain et al. 2016).

ADMINISTRATION

Administration routes of optogenetic therapy carry challenges as well. The neural retina, like any nervous tissue, is extremely compact with small extracellular spaces and outer (OLMs) and inner limiting “membranes” (ILMs) on each side. It should be noted that neither of these structures are true cell membranes, as the ILM is formed by the “end feet” of Müller cells. The ILM is an interface between the retina and the vitreous composed of collagen fibers, glycosaminoglycans, laminin, and fibronectin. The OLM is similarly not a true cell membrane, but a row of junctional complexes formed of Crb1 and Crb2—linking the photoreceptor inner segments and Müller glia and demarking the inner border of the subretinal space. Despite not being “true” cell lipid-bilayer membranes, these structures both constitute barriers restricting viral penetration between retinal cells (Dalkara et al. 2009). Although photoreceptor degeneration significantly reduces retinal thickness, Müller glial hypertrophy creating the so-called “glial seal” only makes permeability more challenging (De Silva et al. 2017).

Two delivery routes have been applied in the eye: intravitreal and subretinal. Intravitreal injection of AAV vectors is the easiest and least invasive way of administration that most ophthalmologists are very competent at performing (Ochakovski et al. 2017). Although intravitreal injection represents the most efficient method of transducing the widest possible retinal area in small laboratory animals, a major limitation in larger animals and human subjects is that achieving a similar transduction efficiency may require AAV injections volumes and titers high enough to induce intraocular inflammation (Simunovic et al. 2019), whereas safe AAV dosing has never been able to produce panretinal transgene expression and was, until recently, still struggling to reach deeper layers of the tissue to access inner retinal neurons, photoreceptors, or the RPE from the vitreous cavity. The latter issue may be addressed by pars plana vitrectomy and surgical peeling of the ILM before AAV injection; however, these procedures require the expertise of vitreoretinal surgeons and may jeopardize surgical outcomes by serious side effects like cataracts (Do et al. 2008), or retinal detachment with a case study reporting a full-thickness macular hole in one patient (Maguire et al. 2008). Advances in vector design and directed evolution largely contribute to the development of AAV capsid variants that are able to penetrate full retinal depth and reach all layers as well as the RPE via intravitreal administration (Dalkara et al. 2013; Byrne et al. 2020).

Subretinal injection is a feasible alternative to intravitreal administration to access photoreceptors and the RPE. Indeed, it has been the gene delivery route of choice in clinical trials (Ochakovski et al. 2017). The major setback of the procedure is the necessity of a pars plana vitrectomy as well as piercing through the retina with a needle and dispensing the AAV into an artificially detached subretinal space, often described as a bleb. This approach results in retinal damage at the piercing and detachment sites and transgene expression is often restricted to the bleb area. Although iatrogenic retinal detachment is agreed to have a devastating effect on retinal function, it has been shown that spatial vision is swiftly recovered following the procedure. Other visual functions, for example, color matching, were found to be temporarily disturbed as well, which in combination with outer retinal thinning confirmed via optical coherence tomographic (OCT) findings, suggest photoreceptor loss or outer segment shortening in some patients after surgery (Simunovic et al. 2017). Therefore, engineering capsid modifications enabling enhanced retinal penetration of the AAVs from the vitreous cavity would be more favorable. The most effective recombinant AAV serotypes to date deployed in nonhuman primate (NHP) preclinical and human clinical trials are AAV2/2(4YF) (Petrs-Silva et al. 2011) and AAV2.7m8 (Dalkara et al. 2013). However, in vivo–directed evolution of AAVs in NHP retina recently revealed promising, more efficient alternatives (Byrne et al. 2020).

PROMOTERS

The choice of vector serotype, viral capsid modifications exhibiting different tissue and cell tropisms, and the vector delivery routes—with intravitreal injections preferentially targeting ganglion cells, whereas subretinal injections target outer retinal cells and the RPE—are meant to specify therapeutic targets to a certain extent. Promoters driving ectopic transgene expression can potentially enhance cell-type specificity, although a major advantage of ubiquitous promoters is that they might be able to withstand retinal morphofunctional remodeling caused by retinal degeneration, which is known to largely alter endogenous promoter activity and gene expression (Hackam et al. 2004). Hence, the most straightforward way to transfect the retina is via ubiquitous promoters like the cytomegalovirus (CMV) enhancer/chicken β actin/rabbit β globulin (CAG) promoter, with or without a downstream woodchuck hepatitis virus post-translational regulatory element (WPRE), which is known to improve and increase expression in retinal cells (Patrício et al. 2017).

Photoreceptor degeneration in most IRDs typically progresses in a rod-to-cone pattern, where cone demise is a bystander effect of rod death. However, degenerated cone cells not only remain for almost a year following the complete loss of rods in the rd1 mouse model of early onset, recessive retinitis pigmentosa, but they were also shown to maintain active gene expression, although at altered levels (Narayan et al. 2019). Also, imaging of human patients with advanced IRD revealed degenerated but remnant cone photoreceptors often referenced as dormant cone cells (Wong and Kwok 2016). In theory, cone dormancy enables therapeutic intervention with the optogenetic reactivation of cone photosensitivity using specific promoters like cone arrestin (mCAR and hCAR) or a more efficient chimeric promoter consisting of an enhancer element of inter-photoreceptor retinoid-binding protein promoter and a minimal sequence of the human transducin α-subunit promoter (IRBPe/GNAT2) (Dyka et al. 2014). However, additional treatment might be necessary to improve their viability as cones represent a gradually degenerating cell population.

Retinal ganglion cells (RGCs) are the easiest targets for optogenetics partially because their proximity to the retinal surface makes them intravitreally accessible for most vectors and also because their long-term persistence leaves them unaffected despite retinal degeneration. However, RGCs are third-order neurons, and the majority of visual signal processing is completed prior to them; thus, the extent of vision restoration targeting these cells over second-order neurons is currently being debated and remains unclear. The only deployed and effective RGC-specific promoter to date is represented by the nonselective, panneuronal human synapsin 1 (hSyn1) promoter (Simunovic et al. 2019), stimulating discussions over the necessity of ON-selective promoters.

Visual signal processing first occurs at the level of second-order neurons, including the bipolar cells, referred to as the retina's switchboard by Thomas Euler (Euler et al. 2014). Each of the more than 10 types of bipolar cells plays a distinct role in transforming the photoreceptor input in a unique way, thus creating individual channels that encode stimulus properties including polarity, contrast, temporal profile, and chromatic composition. Finally, the axonal endings of each bipolar cell type systematically stratify at different inner plexiform layer (IPL) strata and by this means provide synaptic input to specific groups of RGCs and amacrine cells. Parallel processing and signal refinement abilities make bipolar cells the most desired targets in optogenetic vision restoration strategies to date. However, present success in bipolar cell targeting is limited by multiple factors including promoter specificity. At present, the most widely used ON bipolar cell-specific promoter is represented by the mGluR6 promoter and its modified, enhanced variants.

Numerous optogenetic preclinical trials have been and are being conducted in murine, canine, and NHP models to date. These may be divided into categories based on the targeted cell population (ubiquitous, photoreceptors, bipolar cells, ganglion cells), the expressed optogenetic proteins (type 1 or 2 opsins), the delivery route, and the capsid variant used. Major studies until 2017 are summarized in Table 1 (Simunovic et al. 2019). Studies and clinical trials published between 2017 and 2022 are listed and summarized in Table 1.

Table 1.

A summary of major studies of optogenetic approaches to vision restoration in animal models

Study	Model	Vector	Mode of delivery	Promoter/enhancer	Opsin	Functional estimates	Comments
Ubiquitous promotors
Bi et al. 2006	Murine rd1	AAV2	Intravitreal (P1 or P60–360)	CAG/WPRE	ChR2	Patch clamp MEA VEP	Light sensitivity estimated from patch clamp recordings 9/13 eyes demonstrated VEPs Sensitivity maintained at up to 6 mo
Zhang et al. 2009	Murine rd1	AAV2	Intravitreal (P80)	CMV/WPRE	ChR2 and NpHR	Patch clamp MEA	Light sensitivity estimated from patch clamp recordings NpHR-expressing cells 20 times less sensitive than ChR2-expressing cells Sensitivity maintained at up to 4 mo
Tomita et al. 2007	Murine RCS	AAV2	Intravitreal (P300)	CAG	ChR2	Electroretinogram Visually evoked cortical potentials	Chr2 expressed in 28% of RGCs Expression also seen in the inner plexiform and inner nuclear layers
Tomita et al. 2010	Murine RCS	AAV2	Intravitreal (P180)	CAG	ChR2	Visually evoked cortical potentials Behavioral response (optomotor)	Recovery of visually evoked cortical potentials commencing 2 wk following injection and peaking at 6 wk Evidence of functional rescue behaviorally at high photopic levels (2.25 × 1015 photon cm−2)
Tomita et al. 2014	Murine RCS	AAV2	Intravitreal (age not specified)	CAG	mVChR1	Visually evoked cortical potentials at 4 and 12 mo Behavioral response (optomotor)	Recovery of VEP observed up to 12 mo postinjection
Sato et al. 2017	Murine ChR2-expressing/MNU-induced degeneration	AAV2	Intravitreal (age not specified)	CAG	mVChR1 (+ChR2)	Patch clamp recordings ERG VEP	Coexpression of mVChR1 and Chr2 in 12% of RGCs Lack of response to longer wavelengths proposed to be due to lack of chromophore in mVChR1/ChR2 coexpressing cells
Lin et al. 2008	Murine rd1	AAV2	Intravitreal (P30)	CMV	OPN4	Patch clamping PLR Light avoidance Two AFC visual discrimination	Light sensitivity estimated from PLR Sensitivity retained at 11 mo
Liu et al. 2016	Murine rd1	AAV2/8	Subretinal (P30)	CMV	OPN4	ERG VEP Light avoidance	Functional rescue at day 30; none at day 45
De Silva et al. 2017	Murine rd1	AAV2/8(Y773F)	Subretinal (P42–45)	CAG/WPRE	OPN4	MEA LASER cortical imaging PLR Light avoidance Object recognition test	Threshold for light sensitivity estimated by MEA Function retained at up to 13 mo following injection
Photoreceptor-specific promotors
Busskamp et al. 2010	Murine rd1 CNGA3−/−/Rho−/− double-knockout	AAV2.1	Subretinal (P53–264)	hRHO hRO mCAR	eNpHR	Patch clamp MEA VEP Behavior (“dark light box”) Optomotor response	ON, OFF, and ON–OFF RGC responses were demonstrated Transient and sustained RGC responses preserved Lateral inhibition and directional selectivity demonstrated Cortical activity and optomotor responses evident in rd1, but not CNGA3−/−/Rho−/− double-knockout mice
Bipolar cell-specific promoters
Lagali et al. 2008	Murine rd1	Electroporation	Subretinal (P0–1)	mGRM6/SV40	ChR2	Patch clamp MEA VEP Behavior (“light-induced locomotion”) Optomotor response	About 7% of ON bipolar cells expressed ChR2 Threshold estimated by MEA
Doroudchi et al. 2011	Murine rd rd10 rd16	AAV8 (capsid mutation Y733F)	Subretinal (P ≅ 56)	mGRM6/SV40	ChR2	MEA (P290) Behavior (“water maze task”)	Sustained benefit evident at up to 10 mo No evidence of a significant local or system immune response Threshold for light sensitivity estimated by MEA
Cronin et al. 2014	Murine rd1	AAV2/8 or AAV2/8BP2	Subretinal (P = 70)	“4X” mGRM6 (4 enhancer elements)	ChR2	MEA (P84–98)	Transduction efficiency 59% Electrophysiological responses suggest responses from ON, OFF, and ON–OFF ganglion cells
Macé et al. 2015	Murine rd1	AAV2 (7m8 mutant)	Intravitreal (P = 28–56)	mGRM6/SV40	ChR2	MEA (P = 132–324 d) Extracellular cortical recordings (P = 245–346 d) Light-induced locomotory behavior (P > 105 d)	Threshold estimated by MEA ON and OFF responses demonstrated Cortical activity and improvements in visually guided behavior demonstrated
van Wyk et al. 2015	Murine rd1	AAV2 (tyrosine mutant)	Intravitreal and subretinal (P > 168)	mGRM6/SV40/WPRE	mGluR6/OPN4 chimera	Patch clamped human embryonic kidney cells Cell-attached recordings from RGCs Whole-cell patch clamp recordings from bipolar cells Intrinsic signal imaging from visual cortex Behavioral assessment (optomotor reflex and water maze)	Similar efficacy demonstrated for intravitreal and subretinal routes Expression demonstrated at up to 8 mo 12% transduction of target ON bipolar cells Reported sensitivity of transgenic rd1 mice expressing mGluR6/OPN4 similar to transfected mice Response kinetics similar to native bipolar cells Responses are “sign inverting”
Cehajic-Kapetanovic et al. 2015	Murine rd1	AAV2/2	Intravitreal (P56–70)	mGRM6 or CAG (i.e., selective and nonselective)	Rho	MEA (P112–154) Multielectrode probe recording from LGN Behavioral (light-induced locomotion, simulated predator-induced locomotion)	Threshold estimated from MEA Behavioral responses elicited under low photopic illumination levels Selective expression in bipolar cells results in increased responses at behavioral tests compared to nonselective expression
Gaub et al. 2015	Murine rd1	AAV2/2, 4 YF	Intravitreal (P21–42)	mGRM6	Rho and ChR2 (H134R)	MEA (P63–112) VEPs Behavior (light avoidance, modified water maze, and temporal light pattern discrimination)	Only direct comparison of type 1 and type 2 opsin Rhodopsin at least 1000 times more sensitive than ChR2 Firing rate in treated retinas similar to wild-type ON and OFF RGC responses demonstrated
Ganglion cell-specific promoters
Thyagarajan et al. 2010	Murine rd1	Chr2 × rd1 cross	NA (cross-bred)	NA (cross-bred)	ChR2	MEA Cortical imaging (optical) Behavior (optomotor response, water maze)	ChR2 expressed in 30%–40% of RGCs No evidence of functional rescue at behavioral tasks
Sengupta et al. 2016	Murine rd1	AAV2	Intravitreal (P28–35)	hSyn1	ReaChR	Patch clamp recordings from RGCs MEA (P ≥ 60) Cortical responses (extracellular) Behavior (light avoidance)	Thresholds 1000× below threshold for photochemical damage Functional responses also recorded from macaque and human RGCs—their temporal response characteristics were preserved (i.e., transient vs. sustained)
Ameline et al. 2017	Canine RPE65	AAV2	Intravitreal following vitrectomy	hSyn1	Chr2 or OPN4	MEA (21 mo postinjection)	No evidence of injection-induced thinning on optical coherence tomography

Table adapted from Simunovic et al. 2019, with permission, from Elsevier © 2019.

(rd) Retinal degeneration, (RCS) Royal College of Surgeons, (MNU) N-methyl-N-nitrosourea, (CNGA3) α-subunit of the cone cGMP-gated cation channel, (Rho) rhodopsin, (AAV) adeno-associated virus, (P) days postnatal, (CAG) CMV enhancer/chicken β actin/rabbit β globulin, (CMV) cytomegalovirus, (WPRE) Woodchuck hepatitis virus post-translational regulatory element, (hRHO) human rod opsin, (hRO) human red opsin, (mCAR) mouse cone arrestin, (mGRM6) metabotropic glutamate receptor 6, (SV40) simian virus 40, (hSyn1) human synactin 1, (ChR2) channel rhodopsin 2, (OPN4) melanopsin, (eNpHR) halorhodopsin, (ChR2/H134R) humanized channelrhodopsin 2, (OPN4/mGluR6) melanopsin/glutamate receptor chimera, (Rho) rhodopsin, (MEA) multielectrode array, (VEP) visual evoked potential.

NEW PROMOTERS

Toward the end of 2017, a report was published from the Dalkara laboratory on a new promoter with great potential for optogenetic vision restoration. The new promoter sequence was identified based on the regulatory region of the human γ-synuclein gene (SNCG) granting strong transgene expression in RGCs across species. Intravitreal injections of SNCG-driven channelrhodopsin-Ca²⁺-permeable channelrhodopsin (CatCh) with green fluorescent protein (GFP) packaged into AAV2 vectors showed panretinal expression and obvious CatCh responses in the retina and cortex of rd1 mice. When injected into NHPs, the same construct resulted in strong expression in perifoveolar ganglion cells that responded strongly to light levels safe for the human eye, even 6 months after injection. With the same vector and plasmid construct, the new SNCG promoter was compared to the ubiquitous CMV promoter with the latter requiring higher activation thresholds and leaving much more unresponsive cells based on multielectrode array (MEA) recordings, suggesting that cell-type-specific promoters are key for the clinical translation of optogenetics (Chaffiol et al. 2017).

Ferrari and colleagues administered intravitreal injections of AAV2-hSyn-ReaChR-mCitrine to target RGCs in rd1 mice as well as AAV2-SNCG-CatCh-GFP in NHPs. Since transgene expression is also present in axons, it is unclear whether RGCs will be responding only to a stimulation area covering their somas and dendrites, or to any stimulation reaching their axons, severely impairing spatial resolution. Using MEAs, they recorded responses of mouse and macaque retinas to random checkerboard patterns after optogenetic treatment and found that transduced RGCs were only sensitive to a small region of the visual space. They then created a simplified model based on this small receptive field, which predicted RGC responses to complex stimuli accurately. From this model, in a simulation, they estimated how the entire light-sensing RGC population would respond to letters of different sizes, similar to a human visual acuity test. Then the maximal acuity expected by a patient could be estimated and the obtained acuity was found to be above the limit of legal blindness (Ferrari et al. 2020).

Developing methods for noninvasive cone targeting led to the identification of a strongly cone-specific promoter—PR1.7—in both healthy and degenerating retinas across species: wild-type and rd10 mice as well as NHPs in vivo and human retinas in vitro (Khabou et al. 2018). Khabou and colleagues used Jaws, a hyperpolarizing microbial opsin (Chuong et al. 2014), to evaluate its potential for optogenetic vision restoration. The intravitreal injection of AAV2-7m8-PR1.7-Jaws-GFP in macaques produced robust and highly cone-specific GFP expression in the foveola and resulted in measurable optogenetic reactivation of cones (Khabou et al. 2018).

Jüttner et al. created a library of 230 AAV-promoter constructs, with synthetic promoters created using four innovative and independent design strategies. They also demonstrated that some of the identified promoters specifically target neuronal and glial cell types in the murine and NHP retina in vivo and in the human retina in vitro, applications for recording and stimulation, as well as the intersectional and combinatorial labeling of cell types (Jüttner et al. 2019).

Hulliger and colleagues are “Empowering retinal gene therapy with a specific promoter for human rod and cone ON bipolar cells.” They describe the design and functional assessment of 770En_454P(hGRM6), a human GRM6 gene-derived, short promoter that drives strong and specific expression in the ON bipolar cells of the human retina. 770En_454P(hGRM6)-driven medium wavelength cone opsin (MW-opsin) expression in ON bipolar cells resulted in lasting restoration of the optomotor reflex in rd1 mice (Hulliger et al. 2020).

VECTOR DEVELOPMENT

Maddalena and colleagues addressed another issue in AAV-based gene therapy: packaging capacity. The known 5 kb limit in transfer capacity can be expanded to about 9 kb using dual AAV vectors, although this strategy would still not enable effective treatment of certain types of Usher or Alström syndromes caused by mutations in very large genes. To overcome this limitation, they have generated a triple AAV vector system with a maximal transfer capacity of about 14 kb (Maddalena et al. 2018).

Dias and colleagues compared intravitreally administered AAV transfection and penetration efficacy in mice versus rats when adjuvant pharmacological agents are applied to enhance the preservation and penetration of viral particles. They used wild-type AAV2 and recombinant AAV2 with a quadruple Y-F mutation carrying the ubiquitous CBA promoter driving GFP expression and applied tyrosine kinase inhibitors like imatinib and genistein, which were highly effective and promoted even outer retinal transgene expression in mice, but not in rats. They conclude that eye size and ILM differences in mice versus rats are too large to achieve similar results in the two species and thus further vector and administration strategy development is needed (Dias et al. 2019).

AAV variant development via in vivo–directed evolution created a replication-incompetent AAV library, guided by deep sequencing and controlled by GFP barcoded capsids that went through six rounds of in vivo secondary selection in primates (Byrne et al. 2020). The selection process involved intravitreal administration of the library and recovery of genomes from the outer retina followed by next-generation sequencing of each round. Directed evolution resulted in vectors with redirected tropism to the outer retina and increased gene delivery to retinal cells expanding the pool of effective AAV vectors for primate retina and potentially enabling less invasive administration of gene therapy in human patients (Byrne et al. 2020).

NEW OPTOGENETIC PROTEINS

Retinitis pigmentosa causes degeneration of photoreceptors, whereas other retinal cells survive. It has been shown that the expression of light-activated signaling proteins in those surviving cells could restore vision. Berry et al. used a retinal G-protein-coupled receptor, mGluR2, which was chemically engineered to respond to light. In RGCs of rd1 mice, photoswitch-charged mGluR2 (“SNAG-mGluR2”) evoked robust OFF responses to light, but not in wild-type retinas, suggesting selectivity for RGCs that have lost photoreceptor input. SNAG-mGluR2 enabled animals to discriminate between high-contrast patterns. Simultaneous viral delivery of the inhibitory SNAG-mGluR2 and excitatory light-activated ionotropic glutamate receptor LiGluR restored ON, OFF, and ON–OFF light responses and improved visual acuity (Berry et al. 2017).

Wright and colleagues have developed the so-called multicharacteristic opsin (MCO) to restore visually guided behavior as shown via the radial water maze test in rd10 mice using intravitreal AAV delivery and the mGluR6 promoter (Wright et al. 2017). Batabyal et al. used the same AAV2-mGluR6-MCO-mCherry construct to intravitreally inject rd10 mice and reported that MCO was highly sensitive and functional in ambient light, reporter expression was detectable 6 mo after injection, and water maze and optomotor assays showed significant changes 8 wk postinjection (Batabyal et al. 2021). Intravitreal injections of the construct in wild-type dogs produced transgene expression in the inner nuclear layer with no detectable inflammatory response in the dog retina. The virus vector was secreted and cleared from the body via urine and feces in about 3–13 wk postinjection. Moreover, intravitreal injection of AAV-MCO resulted in a few off-target expressions in the mesenteric lymph node, liver, spleen, and testis. The group concluded that AAV2-mGluR6-MCO-mCherry gene therapy is safe (Tchedre et al. 2021). MCO is currently being tested in two first-in-human clinical trials.

Ganjawala et al. reported that the optimization of the kinetics of a recently described ChR variant resulted in the identification of two highly light-sensitive Chloromonas oogama ChR mutants: CoChR-L112C and CoChR-H94E/L112C/K264T. C57Bl6 and TKO (Opn4^−/−Gnat1^−/−Cnga3^−/− triple-knockout) mice received intraocular injections of AAV2 carrying the CoChR mutants under control of the CAG promoter. The improved light sensitivity was confirmed by ex vivo electrophysiological recordings of the AAV transfected retina. Also, the CoChR mutants were found to restore the optomotor reflex in ambient light conditions, eliminating a major setback in ChR-based optogenetic vision restoration (Ganjawala et al. 2019).

Lu and coworkers compared the CAG promoter to an improved mGluR6 promoter (In4s-In3-200En-mGluR500P [Lu et al. 2016]) by driving CoChR-GFP expression using the AAV2.7m8-Y444F vector in a mouse model of retinal degeneration. They reported that, based on MEA recordings, pupillary light response, and optomotor tests, CAG promoter-driven CoChR expression restored higher light and contrast sensitivity to the degenerated mouse retina (Lu et al. 2020).

Optogenetic strategies are restricted by either low light sensitivity or slow kinetics and lack adaptation to changes in ambient light. Berry et al. found that the mammalian MW-opsin under control of the RGC-specific hSyn promoter delivered intravitreally via AAV2/2(4YF) conquers these limitations and is able to grant visually guided behavior in dim light conditions. MW-opsin allows rd1 mice to discriminate temporal and spatial light patterns displayed on an LCD screen, exhibits adaptation to increments and decrements in ambient light, and restores novel object exploration in an open field arena providing the kinetics, sensitivity, and adaptation needed to restore patterned vision (Berry et al. 2019).

McClements et al. reported successful expression of rod and cone opsin in ON bipolar cells of the rd1 retina via both subretinal and intravitreal delivery routes when using the well-known 4xGrm6 promoter and various AAV capsid variants: AAV2.4YF, AAV8.BP2, and AAV2.7m8 (McClements et al. 2021).

ChrimsonR

Gauvain and colleagues describe the selection of an optogenetic construct in NHPs as groundwork for an ongoing clinical trial (reference number was not mentioned) testing AAV2.7m8-ChR-tdTomato in retinitis pigmentosa patients. They showed that the AAV2.7m8 vector had a higher transfection efficacy than AAV2 in RGCs and that ChrimsonR fused to tdTomato was expressed more powerfully than ChrimsonR alone. The tested vector doses transfected robust numbers of RGCs in the perifovea, with no significant inflammation. After recording RGC responses, they estimated a visual acuity of 20/249, above the level of legal blindness (20/400) (Gauvain et al. 2021).

Chaffiol et al. demonstrated that a single intravitreal injection of AAV2.7m8-CAG-ChrimsonR-tdTomato can efficiently target perifoveal RGCs and, 20 mo after injection, optogene expression and cortical visually evoked potentials were still detectable and sufficient in macaques (Chaffiol et al. 2022). ChrimsonR is currently being tested in a clinical trial.

NEW METHODS

McGregor et al. reported that AAV2-CAG-tdTomato-ChrimsonR and AAV2-CAG-GCaMP6s coinfection enables the combination of adaptive optics ophthalmoscopy with calcium imaging to optically record optogenetically restored RGC activity in the fovea of the living primate. This in vivo imaging method could be combined with any gene therapy to reduce the number of primates needed to evaluate restored retinal activity, while maximizing translational benefit by using an appropriate preclinical model of the human visual system (McGregor et al. 2020).

Gilhooley and colleagues compared two optogenetic tools: mammalian melanopsin (hOPN4) and microbial red-shifted channelrhodopsin (ReaChR) expressed within RGCs versus (mainly) bipolar cells in the degenerating retina. Constructs packaged into AAV2/2(quadY-F) were administered intravitreally into Cre/lox transgenic mice where expression was restricted only to cells expressing the Cre recombinase enzyme—target cell populations dominated by ON bipolar cells via floxed hOPN4 gene in L7-Cre mice or RGCs via floxed hOPN4 in Grik4-Cre mice—and was compared with nontargeted delivery using the CBA promoter. In summary, they found that bipolar-targeted optogenetic tools produced faster kinetics and flatter intensity-response curves compared with nontargeted or RGC-targeted tools. They concluded that both mammalian and microbial opsins granted better results when targeted to bipolar cells and thus bipolar cell targeting for vision restoration in IRDs has obvious advantages over RGC targeting (Gilhooley et al. 2022).

DOG MODEL

Nikonov and coworkers demonstrated restoration of retinal responses and partial restoration of vision in the rod-cone dystrophy 1 (Rcd1) dog model of retinitis pigmentosa after unilateral subretinal injection of the AAV9 vector, ensuring cone-specific expression by including the chimeric antigen receptor (CAR) promoter to drive the enhanced version of the optogenetic gene eNpHR, a light-gated hyperpolarizing chloride pump belonging to the halorhodopsin family (Nikonov et al. 2022).

ETHICS

Harris and Gilbert (2022) emphasize the importance of fair treatment and informed consent of patient volunteers and raise ethical issues to be resolved in high-profile, first-in-human clinical trials.

CLINICAL TRIALS

Clinical trials of optogenetics are underway.

Nanoscope Therapeutics’ product-vMCO-010, also known as AAV-delivered multicharacteristic opsin, is being tested in a phase 1 and 2 dose escalation study and a phase 2 efficacy and safety study: NCT04945772, NCT04919473.

In September 2021, GenSight announced positive results regarding safety of GS030 in human patients with retinitis pigmentosa. GS030 was delivered intravitreally using AAV2 carrying ChrimsonR targeting RGCs (NCT03326336).

A phase 1/2 trial for advanced retinitis pigmentosa is testing ChR as product RST-001 by RetroSense Therapeutics with an estimated completion date of April 2035.

MEASURES OF VISUAL ACUITY IN PRECLINICAL OPTOGENETIC STUDIES

The optomotor reflex test (OptoMotry) is commonly used for behavioral visual acuity assessment in the field of optogenetic vision restoration. Visual acuity by definition is a measure of ability to differentiate and recognize shapes and details of environmental objects at a given distance. It has multiple components, including contrast sensitivity (stationary and dynamic), color vision, image vision, depth perception, and visual memory. The optomotor reflex is the involuntary movement of the eyes and head in the direction of horizontally moving environmental object(s). In testing mice, the phenomenon relies on the proper functioning of horizontal cells (contrast sensitivity) and direction selective ganglion cells (horizontal motion), satisfying only one component of visual acuity as a collective term. Therefore, OptoMotry should not be the only means of visual acuity testing. Instead, we suggest the use of the following series of behavioral assays:

1. Light-dark paradigm or photophobia test to confirm light perception (Bourin and Hascoët 2003),

2. OptoMotry to confirm dynamic contrast sensitivity (using colored background lighting can provide information on color contrast sensitivity) (Abdeljalil et al. 2005),

3. Open field and object exploration test for image vision (Berry et al. 2019),

4. Visual cliff test for depth perception (binocular vision) (Fox 1965), and

5. Water maze test or the two-chamber active avoidance test for contrast sensitivity and visual memory (Frick et al. 2000; Berry et al. 2019).

The summarized data from these five tests should be used to interpret the extent of restored (or missing) visual acuity and be used as a standard for all laboratories to adhere to in the future allowing the comparison of results originating from laboratories around the world.

CONCLUDING REMARKS

IRDs are the leading cause of blindness and visual impairment, affecting between 1 in 3000–5000 individuals worldwide. Gene therapy might be a solution for some patients; however, it is not applicable to all known genotypes or to patients with advanced disease and unknown genotypes. Optogenetics proposes genotype-independent vision restoration, and unlike electronic retinal implants, it offers advantages in sensitivity, retinal area, and precision. Experiments in animal models confirm that the choice between type 1 versus type 2 opsins comes with a tradeoff of sensitivity versus reaction kinetics. Targeting interneurons might restore higher acuity vision than output neurons and using cell-specific promoters may be more advantageous than generic ones. Furthermore, many outstanding questions remain, including the optimal vector and surgical approach of delivery. Nevertheless, preclinical trials of optogenetics for vision restoration are promising and the results of the first phase 1/2 trials in humans are awaited.