Cellular and Subcellular Optogenetic Approaches Towards Neuroprotection and Vision Restoration

Abstract

Optogenetics is defined as the combination of genetic and optical methods to induce or inhibit well-defined events in isolated cells, tissues, or animals. While optogenetics within ophthalmology has been primarily applied towards treating inherited retinal disease, there are a myriad of other applications that hold great promise for a variety of eye diseases including cellular regeneration, modulation of mitochondria and metabolism, regulation of intraocular pressure, and pain control. Supported by primary data from the authors’ work with in vitro and in vivo applications, we introduce a novel approach to metabolic regulation, Opsins to Restore Cellular ATP (ORCA). We review the fundamental constructs for ophthalmic optogenetics, present current therapeutic approaches and clinical trials, and discuss the future of subcellular and signaling pathway applications for neuroprotection and vision restoration.

1. Introduction

Optogenetics combines genetic and optical methods to induce or inhibit well-defined events in cells, tissues, or animals(Deisseroth, 2015). It is predicated on three key elements: (1) identification of naturally occurring or engineering of novel light-sensitive proteins (primarily these have been based on microbial opsins utilizing carotenoid chromophores) that upon light stimulation increase or decrease a defined molecular activity, such as ion channel opening, binding affinity, or enzymatic activity; (2) genetic delivery and expression of these constructs in a target cell or tissue; and (3) stimulation with sufficiently strong and targeted light to activate or inhibit these proteins in specific cells or tissues(Deisseroth, 2015).

Optogenetics is considered one of the most groundbreaking innovations in neuroscience in recent years; some neuroscientists believe its impact will be on par with the discovery of magnetic resonance imaging(Adamantidis et al., 2015). While several groups had previously reported the expression of light-sensitive channels in neurons(Nagel et al., 2003, 2002; Zemelman et al., 2002), the method for functional manipulation was introduced by Ed Boyden and Karl Deisseroth in 2005,(Boyden et al., 2005) with the term “optogenetics” coined a year later(Boyden et al., 2005; Deisseroth et al., 2006). Optogenetics achieved remarkable progress from concept to tool in the ensuing ten years thanks in a large part to Boyden and Deisseroth’s open-source philosophy in sharing optogenetic reagents and methods.

Applications for optogenetics in eye and vision research have been envisioned since this field’s beginnings, as one of the first optogenetics papers(Zemelman et al., 2002) suggested that it might be better suited for retina-targeted studies(Adamantidis et al., 2015). In ophthalmology, the primary application of optogenetics has been in converting subpopulations of light-insensitive retinal neurons into light-sensitive artificial photoreceptors. In essence, this application of optogenetics allows one to jump-start the process of light transduction in the retina when native photoreceptors and/or other crucial retinal neural elements are damaged. The primary target of optogenetic clinical trials to date within ophthalmology has been inherited retinal disease (IRD). As calculated in 2020, autosomal recessive IRDs alone were estimated to affect approximately 5.5 million individuals (~1 in 1,380), with panel-based genetic testing able to identify a genetic cause in the known 300+ associated genes for approximately 67% of patients(Duncan et al., 2018). Treating inherited retinal disease is complicated by a complex and dynamic genetic landscape including various inheritance patterns of disease, genetic and phenotypic heterogeneity, incomplete penetrance and variable expressivity, uniparental disomy, genetic mosaicism, copy number variations, and epigenetic changes(Jin et al., 2007; Rivolta et al., 2002; Wood et al., 2018a). Due to these inherent challenges, optogenetics has several advantages(Zuzic et al., 2022) to treat disease when (a) the genetic mutation is unknown, (b) gene augmentation or replacement has not yet proven feasible, or (c) remaining retinal cells have undergone anatomical remodeling(Wood et al., 2018b). Optogenetics can also be useful for retinal degenerative disease irrespective of any underlying genetic mutations, including for age-related macular degeneration (AMD).

While ophthalmic optogenetics has primarily been envisioned for retinal dystrophic and degenerative disease, there are a myriad of other applications that hold great promise. Herein, we first focus on the fundamental tools required for effective optogenetic applications in the field of ophthalmology by outlining current approaches and highlighting current clinical trials. We then discuss important and novel applications of ophthalmic optogenetics including non-visual uses and the advances that have been made for subcellular applications and signal pathway modulation approaches that represent new directions for the field.

2. Essential Elements for Ophthalmic Optogenetics

2.1. Element #1: Light-sensitive proteins

The breadth and complexity of light-activated molecules available for optogenetic applications is a testament to nature’s innovative potential. Naturally occurring light sensitive molecules available for optogenetic application thus far include rhodopsins, xanthopsins (photoactive yellow proteins), light-oxygen-voltage (LOV) sensors, blue light-sensors that use flavin adenine dinucleotide (BLUF), cryptochromes, phytochromes, and UV-B receptors (UVR8)(Rost et al., 2017). The most commonly used light-sensitive molecules for optogenetics are rhodopsins. Opsins are proteins that bind to light-reactive chemicals that undergo photoisomerization and mediated responses including phototransduction, phototaxis, and circadian rhythms. Rhodopsins specifically are transmembrane proteins with an average size 30–65 kDa that covalently bind the chromophore retinal, a carotenoid derivative, and are ubiquitously found in three domains of life: archaea, eubacteria, and eukaryotes.

Rhodopsin molecules are classified by type, with microbial rhodopsins referred to as type 1 (microbial) rhodopsins, and metazoan (eukaryotic organisms with mitochondria) rhodopsins referred to as type 2 (animal) rhodopsins(Sharma et al., 2006). The main distinction between these two classes of opsins is the principle cellular event elicited. In both classes, the absorption of photons elevates the electron state in retinal (also known as retinaldehyde) which enacts cis-to-trans retinal photoisomerization. This induces a conformational change in the transmembrane opsin, culminating in either the transport of an ion across the cell membrane with type 1 rhodopsins or the activation of G protein-coupled receptors with type 2 rhodopsins. In both rhodopsin families, retinal is covalently bound to a seven-transmembrane-helix domain protein (opsin) through a conserved lysine residue in the seventh helix(Sharma et al., 2006). However, despite many similarities, there are notable differences.

Common type 1 (microbial) rhodopsins function as (a) light-driven proton transporters in haloarchaea (bacteriorhodopsin, the first opsin discovered by Oesterhelt and Stoeckenius in 1971(Oesterhelt and Stoeckenius, 1971; Russell et al., 2017)), marine Archaea (euryarchaeotic rhodopsin), and bacteria (proteorhodopsin, xanthorhodopsin); (b) light-driven chloride pumps in haloarchaea (halorhodopsin from Salinibacter ruber); and (c) receptors for phototaxis or photo-adaptation (sensory rhodopsins including channelrhodopsins)(Sharma et al., 2006). In type 1 rhodopsins, retinal photo-isomerizes from all-trans to 13-cis, and rather than undergoing ‘bleaching,’ 13-cis retinal spontaneously relaxes back to its all-trans isoform without the need for chromophore recycling(Ebrey, 2002).

In contrast, type 2 (animal) rhodopsins are coupled with G proteins and have a wide variety of functions including (a) photoreception (rod or cone opsin), (b) circadian rhythm modulation (melanopsin), (c) neural tissue modulation (neuropsin), (d) brain and visceral organ modulation (encephalopsin), (e) photoisomerization within the retinal pigment epithelium (RPE) and Mueller cells in the vertebrate retina (RGR opsin), and (f) modulation of the RPE cells of the vertebrate retina (peropsin)(Terakita, 2005). As opposed to type 1 rhodopsins, retinal in type 2 rhodopsins photo-isomerizes from 11-cis to all-trans as in the visual cycle(Ebrey, 2002) and undergoes photobleaching requiring recycling to regenerate 11-cis retinal (Fig. 1). In mammals, such recycling is principally conducted by the RPE, and to a smaller degree by Muller glial cells(Kaylor et al., 2013). Humans have 9 genes dedicated to type 2 opsins (Terakita, 2005) in the visual system. These genes encode three distinct opsins in cones, short (blue, peak absorption at 420 nm), medium (green, 530 nm), and long wavelength sensing opsins (red, 560 nm), and a more abundant rod-expressed opsin (rhodopsin, 500 nm)(Wang et al., 2014).

Figure 1: Opsin Classification.

Opsins are seven-pass-transmembrane proteins that may be classified as either type 1 (channels or pumps) or type 2 (G-Protein coupled receptors) and are sensitive to light through their interaction with the chromophore retinal.In addition to their structural differences, thephotoisomerization of retinal is unique between the two types of opsins.

Exogenous opsins commonly used in modern ophthalmic optogenetics are type 1 opsins, including channelrhodopsins (ChRs) and halorhodopsins (NpHR) (Fig. 2). Channelrhodopsins (ChRs) are non-selectively permeable ion channels allowing the passage of H⁺, Na⁺, K⁺ and Ca²⁺ and are most sensitive to blue light. These were first derived from green algae by Georg Nagel, Peter Hegemann, Ernst Bamberg, and others in Germany in 2002(Nagel et al., 2003, 2002). One well studied species of ChR is Channelrhodopsin-2 (ChR2), with a peak in its absorption spectra at 470 nm; this was the principle opsin used in cultured rat hippocampal neurons in the seminal optogenetic publication by Boyden et al.(Boyden et al., 2005) This ChR has some characteristics ideally suited for retinal signaling, with a rapid on-rate (~1.2ms) and a moderate channel closing rate (~13.5ms)(Lin, 2010). However, its sensitivity to blue light makes it less ideal for ophthalmic translation, as blue light has diminished penetration and potential for cellular toxicity. To address this limitation, newer iterations of ChRs have been designed, including ChR variants with red-shifted light absorption and activation(Douar et al., 2016; Klapoetke et al., 2014),(Erbguth et al., 2012; Sengupta et al., 2016),(Govorunova et al., 2011; Zhang et al., 2008), limited desensitization, greater light sensitivity, as well as increased channel opening and closing rates ranging from 0.86–1.92 ms opening and 7.9–90 ms closing rates, with the ideal situation of the fastest opening/closing states inversely balanced against light sensitivity (Lin, 2010). Another well-characterized opsin, Halorhodopsin (NpHR)(Bamberg et al., 1993; Han and Boyden, 2007), selectively pumps Cl− ions into cells, lending a hyperpolarized cellular state. NpHR, first described by Matsuno-Yagi and Mukohata in 1977(Matsuno-Yagi and Mukohata, 1977), is most sensitive to green-yellow light (525–650 nm) and has been used with some efficacy to activate remnant cone cell bodies in mice and human ex vivo retinas (Han and Boyden, 2007),(Busskamp et al., 2010). A so-called ‘multicharacteristic opsin’ (MC-010) has also been developed which has shown sensitivity to a broad range of ambient light(Wright et al., 2017).

Figure 2: The electromagnetic spectrum, visible light, and subject opsins.

Opsins commonly used in ophthalmology range between the blue and red spectrum. Both G-protein coupled receptor (GPCR) and channel opsins have been placed at their respective sites within the visible spectrum and coordinated with their respective peak absorption spectra.

Several type 2 opsins have also been used in ophthalmic optogenetics and may be directly targeted towards various cell types of interest (discussed further below in section 2.2). One natively expressed type 2 opsin is melanopsin (OPN4), with an absorption maximum at 467 nm(Matsuyama et al., 2012). This opsin mainly confers photoactivation to a subset of intrinsically photosensitive retinal ganglion cells (ipRGCs). The exogenous delivery of melanopsin to non-ipRGCs has shown promise in conferring light-sensitivity to other cell types in mice(Lin et al., 2008). As one therapeutic example, a faster, next-generation engineered melanopsin chimera, known as melanopsin-mGluR6, was used to confer light sensitivity to mouse retinal bipolar cells(van Wyk et al., 2015). Human rhodopsin has also been utilized to stimulate bipolar cells in mice, as demonstrated by two groups(Gaub et al., 2015; Kapetanovic et al., 2017). Overall, the benefit of using type 2 opsins is the large signal amplification transmitted via G-protein-coupled cascades, but a major limitation remains. Rhodopsin bleaching in cells with limited access to healthy RPE that normally refresh the supply of the retinal co-factor may render these necessary chromophores unavailable, thus limiting the potential therapeutic benefit of type 2 opsins.

2.2. Element #2: Delivery of Optogenetic Therapy

Optogenetics relies on exogenous expression of the genes encoding light-responsive proteins to specific cell populations of interest, similar to the approach taken by gene therapy. While there are numerous methods to introduce DNA into cells including nanoparticles and other viral-free transfection mechanisms(Russell et al., 2017), most modern clinical applications utilize viral vectors to deliver opsin-encoding genes into host cells, most commonly via lentiviruses and adeno-associated viruses (AAVs). Following cellular transduction, these proteins are produced via host cellular machinery and shuttled to their ultimate location (e.g., plasma membrane) by native trafficking processes (Fig. 3). While a comprehensive review of viral-based gene therapy is beyond the scope of this paper, we will briefly discuss approaches and considerations in using lentiviruses and AAVs in ophthalmology.

Figure 3: General Schematic of Gene Therapy.

Adeno-associated viruses (AAV) may be used to deliver a viral payload (genetically engineered plasmid containing the gene for your protein of interest - i.e: a transmembrane opsin) into a cell; thereby allowing for exogenous expression of transmembrane opsins.

Lentiviruses are integrating, complex, single-stranded, positive-sense, enveloped RNA retroviruses that may be engineered in larger vectors for delivery of genes up to 9–10 kb in size. As opposed to AAVs, lentiviral vectors integrate into the host genome, which presents a small concern for off-target effects of insertion on the local genome, and a theoretical potential for germline mutation(Chan et al., 2017). Still, lentiviral vectors show promise as they are currently being used in several clinical trials including Stargardt’s disease and RP due to Usher syndrome type 1B(Scholl et al., 2016).

AAVs are small (25-nm) non-enveloped viruses with relatively low immunogenicity, and contain a linear single-stranded DNA genome that relies on host cell machinery to synthesize the complementary strand, and typically remains episomal without host-integration(Daya and Berns, 2008). The AAV2 variant has to date been used the most within ophthalmology(Scholl et al., 2016), but AAV8 variants may have improved photoreceptor transduction by subretinal injection in monkeys(Vandenberghe et al., 2011). Currently, the maximum genomic packaging size for AAV vectors is approximately 4.5 kb, with transgene sizes topping out at approximately 3–3.5 kb, including promoter and expression-stabilizing elements(MacLaren et al., 2016). To deliver larger genes (such as the USH2A gene), one may employ two AAV vectors encoding for different components of the gene, or revert to larger viral vectors such as lentiviruses. Fortunately, most opsin genes fit into AAV cassettes. AAV-based gene delivery achieved the landmark designation as the first ocular gene therapy product gaining approval by the U.S. Food and Drug Administration in 2017 (voretigene neparvovec-rzyl, Spark Therapeutics, Philadelphia, PA, USA). For all of these reasons, adeno-associated viruses (AAVs) and AAV-variants are the most used—and most promising—viral vectors for ophthalmic optogenetics(Mendell et al., 2021).

Targeting specific cell types with AAV is complex due to species specificity of viral variants and promoters. Modification of these features (to be discussed further below in section 3) can result in variable outcomes. Even the route of delivery may have species dependent outcomes. expression. For example, intravitreal injection is the standard approach to target RGCs by AAV in many species including in non-human primates(Boye et al., 2016; Ye et al., 2015; Yin et al., 2011). However, photoreceptors may also be targeted by intravitreal injection in mice(Jomary et al., 1997; Kay et al., 2013; Pavlou et al., 2021) and dogs(Boyd et al., 2016; Pavlou et al., 2021), and fewer studies have been successful in targeting non-human primate photoreceptors by intravitreal injection(Dalkara et al., 2013; Pavlou et al., 2021). Transduction of non-neuronal cells of the ciliary body also occurs with intravitreal AAV injection, whereas subretinal and suprachoroidal injection may result in more widespread outer retinal expression(Mendell et al., 2021) (Fig 4).

Figure 4: Method of Optogenetic Gene Delivery.

Intravitreal injection (A) allows for clinic-based administration and exposes all intraocular surfaces (and predominantly the inner retina to the therapeutic vector). Subretinal injection (B) is performed as a surgical procedure and delivers the therapeutic vector under the neurosensory retina where they are exposed to the retinal pigment epithelium (RPE) and outer retina. Suprachoroidal injection by a microneedle (C) delivers the therapeutic vector into to the suprachoroidal space, a virtual space between choroid and sclera. Reprinted with open access and approval through creative commons (creativecommons.org)

Disadvantages of viral-based gene therapy include variable cellular specificity, variable efficacy of viral transduction, potential immunogenicity, uncertain long-term stability, off-target viral effects due to leaky promoters, and possible toxicity due to gene product buildup(Wood et al., 2018b). Immunogenicity has been a key hurdle in translating AAV gene therapy to the clinic and is dependent upon numerous factors including the transgene cargo, the promoter, the specific serotype viral envelope, the dose of AAV, the route of administration, and possible manufacturing details (Botto et al., 2022). There is general consensus that inflammation is common in clinical applications of AAV gene therapy and is most directly correlated with the vector dose(Chan et al., 2021).

Significant work is being undertaken to harmonize understanding of this hurdle in AAV gene therapy and standardize treatment protocols when inflammation arises(Chan et al., 2021). There have also been significant advances in high-throughput DNA synthesis, sequencing technologies, machine learning, and directed evolution to design viral capsids with improved cellular specificity, enhanced viral transduction, and decreased immune responses(Wec et al., 2021).

2.3. Element #3: Light

At very high temperatures in the sun, hydrogen (H₂) undergoes nuclear fusion to form helium (He), consequently releasing energy constituting the entire electromagnetic spectrum. Visible light, encompassing the 400–760 nanometer wavelengths of the electromagnetic spectrum, and in some cases infrared or ultraviolet wavelengths, serve two primary purposes in biologic systems: (1) the transduction of light to chemical energy, as used in photosynthesis, and (2) the transduction of light to initiate a signaling pathway, as used in phototaxis, photoperiodism, photomorphogenesis, and vision in its various forms(Ebrey, 2002). Life forms have designed, prototyped, and implemented a multitude of proteins to interface with visible (and to a lesser extent non-visible) light in their unique environments, and the application of these light sensitive proteins is fundamental to the field of optogenetics (Fig. 2).

Light stimulation parameters that are both safe and effective for cellular stimulation are critical for human application of optogenetics. Light can be toxic to retinal cells depending on the wavelength, power, exposure duration, pupil size, and focus. Typically, non-modified microbial opsins require intense light activation (between 10¹⁴ and 10¹⁷ photons/cm²/second) which for some opsins, especially those activated by shorter wavelengths, may exceed the safety threshold of artificial radiation for the human retina(International Commission on Non-Ionizing Radiation Protection, 2013). Therefore, there is an emphasis to engineer opsins that respond to the less damaging, longer (red-shifted) wavelengths of light with enhanced kinetics and efficacy for optimum performance and safety.

Light delivery for optogenetic restoration of vision would ideally be driven simply by the normal environmental illumination of everyday exposure, but is also being explored using specific delivery systems encompassing video acquisition and projector technologies that translate an environmental scene into optogenetic-compatible stimuli, using other specific wavelengths, enhancing contrast, or other approaches. For example, light stimulating “neural deciphering” goggles have been utilized to activate RGCs in visually meaningful patterns(Nirenberg and Pandarinath, 2012). These approaches are discussed further below in the context of current clinical trials.

2.4. Alternative Element: Chemical Photoswitches

An alternative to genetically encoded opsins is the leveraging of endogenous channels rendered light sensitive by chemical modification, i.e., chemical “photoswitches.” Chemical photoswitches are small, synthetic, light-sensitive molecules containing azobenzene derivatives that undergo cis-trans photoisomerization. These molecules bind native channels in the plasma membrane and cause them to membrane depolarization upon the absorption of photons (Tochitsky and Kramer, 2015) (Fig. 5). While a complete discussion of chemical photoswitches is beyond the scope of this paper, a few of the leading approaches will be discussed herein.

Figure 5: Photoswitch Technology.

The proposed mechanism of action of a photoswitch involves using a photoisomerizable molecule that enters the cell and interfaces with an endogenous channel to bestow light sensitivity. On exposure to light, isomerization of the molecule either blocks the channel or unblocks the channel allowing for direct regulation of the channel activity, and therefore electrical regulation of the cell through light stimulation.

The first-generation photoswitches(Banghart et al., 2004) required the concomitant delivery of engineered ion channels, which limited their therapeutic potential. One such genetically and chemically engineered channel is the light-gated ionotropic glutamate receptor (LiGluR), which has an introduced cysteine residue allowing for binding of an azobenzene-based “photoswitch” called maleimideazobenzene-glutamate (MAG). MAG has the capacity to reversibly activate the LiGluR(Caporale et al., 2011),(Carroll et al., 2015; Kienzler et al., 2013). MAG-induced LiGluR activation has been shown to restore light-sensitivity in the outer retinal degeneration Rd1 mouse when expressed in retinal ganglion cells (RGCs)(Caporale et al., 2011; Kienzler et al., 2013). Newer photoswitches were subsequently designed including acrylamide-azobenzene-quaternary-ammonium (AAQ) to target endogenous ion channels in mouse retinal neurons(Polosukhina et al., 2010). AAQ is a photoisomerizable molecule responsive to UV light that confers light sensitivity by blocking endogenous voltage-gated K+ channels, largely targeting mouse inhibitory amacrine cells; however, AAQ has poor photosensitivity, fast reversal, and quite unstable (half-life less than 6 hours)(Polosukhina et al., 2012). Next generation photoswitches were then developed to improve clinical applicability, including diethylamine-azobenzene-quaternary ammonium (DENAQ). DENAQ has red-shifted activation spectra and improved photosensitivity, and interfaces with RGC hyperpolarization-activated cation (HCN) channels. These channels are preferentially upregulated in RGCs of animals with retinal degeneration, including the rd1 mouse(He et al., 2014). DENAQ showed some efficacy in treating mouse models of retinitis pigmentosa and elicited light-responses under light-levels exemplary of daytime conditions. However, high concentrations of the molecule were required (300 μM-1 mM), and the half-life remained short at about 2 days, necessitating high-dose repeated administration with potential for toxicity(Tochitsky et al., 2014). The most recent photoswitch in this lineage is benzylethylamine-azobenzene-quaternary ammonium (BENAQ), an improved photoswitch that is 20x more potent than DENAQ, longer-lasting (a single intravitreal injection confers a month of light-response to mice and rabbit retina), and considerably safer, with no detected toxicity at 10 times the concentration required for light sensitivity(Tochitsky et al., 2017). Like DENAQ, BENAQ preferentially binds RGC HCN channels, and thus is selective for retinas undergoing degeneration with no apparent effect on retinas from healthy (wild-type) animals(Tochitsky et al., 2017). Thus, photoswitchable molecules also hold promise as a new group of light-activated therapeutics for the treatment of retinal degeneration.

3. Current Therapeutic approaches for optogenetics in the eye

The current optogenetic approaches for ophthalmic diseases may be theoretically tailored to treat broad disease phenotypes. When treating retinal degenerations or dystrophies, there are 3 main sub-populations of retinal neurons that one may choose to preferentially target depending on the desired cellular activation, the opsin of choice, and the mode of administration: (1) remnant cones (2) retinal bipolar cells, and (3) RGCs (Fig. 6). Other approaches to target RGCs are envisioned to promote visual system repair, e.g. optic nerve regeneration. In addition to targeting specific subpopulations of retinal neurons, it is also possible to direct opsins to a myriad of cell types in the mouse eye including Mueller glial cells and corneal afferent neurons(Mecum et al., 2021).

Figure 6: Opsin and Neuronal Sub-Population of Choice.

The three most commonly targeted retinal cell types in optogenetics include retinal ganglion cells, bipolar cells, and photoreceptors (rods and cones). While in theory any opsin may be expressed in any location, several of the most historically utilized opsin and retinal cell combination are shown for illustrative purposes. Depending on the translocated ions, the cells ectopically expressing these optogenes either depolarize or hyperpolarize.

3.1. Therapeutic approach: Cone targeting for light transduction

Targeting remnant cones is an appealing approach where feasible. Many retinal dystrophies are of a rod-cone nature, exhibiting early rod degeneration with a relative preservation of cones(Cotter and Noell, 1984). In rodent models, cone degeneration follows, characterized by the initial formation of remnant cone cell bodies with outer segments that no longer respond to light, followed by the destruction of the entire cone cell(Cotter and Noell, 1984; Lin et al., 2009). Thus, one potential optogenetic approach in early to intermediate rod-cone dystrophies would be to target the light-insensitive but viable remnant cone cell bodies with hyperpolarizing opsins, allowing for photoreceptor initiated light-transduction and subsequent preservation of native retinal signal processing. This approach was shown to be feasible using mouse models and ex vivo human retinas by Botond Roska’s group at the Friedrich Miescher Institute for Biomedical Research in Basel, Switzerland in collaboration with José Sahel and Serge Picaud’s group at the Insitut de la Vision in Paris, France(Busskamp et al., 2010).

By delivering to remnant cones a red-shifted modified halorhodopsin (driving an inward chloride current) called enhanced Natronomonas 9haraonic Halorhodopsin (eNpHR)(Busskamp et al., 2010; Cotter and Noell, 1984; Lin et al., 2009), red light-induced activity was observed at the ganglion cell level (with preservation of ‘ON’ and ‘OFF’ pathways), in the visual cortex, and at the behavioral level.

As opposed to targeting bipolar cells and RGCs, cone-targeting has the potential for the most realistic visual experience given the preservation of visual mapping as well as the potential for high acuity. However, the limitation of targeting cones is that viable remnant cone cell bodies are fundamental to this approach, and in many degenerative diseases such cones may not be present in adequate numbers, may soon die if the disease process is ongoing, or in degenerating may have led to significant retinal remodeling(Jones et al., 2003),(Marc and Jones, 2003).

When outer retinal cells (including cones) are targeted, subretinal injection is considered the most direct approach(Chan et al., 2017; Rowe-Rendleman and Durazo, 2014). Subretinal surgery employing submacular bleb formation via a 41 gauge cannula is a validated technique in the armamentarium of vitreoretinal surgery for displacement of subretinal hemorrhage,(Yiu and Mahmoud, 2014) macular translocation,(Aisenbrey et al., 2002) stem-cell implantation,(Schwartz et al., 2015) and gene therapy(Fischer et al., 2016). While subretinal delivery of gene products (including optogenetic tools) is a viable technique in the hands of skilled surgeons(Scruggs et al., 2021), there are still procedure-specific issues to consider when operating with fragile retinal architecture. These risks include the potential for outer retinal thinning, reactive gliosis, sub-anatomic functional loss, progressive chorioretinal atrophy(Gange et al., 2022), and intraoperative macular hole formation(Maguire et al., 2008; Nork et al., 2012; Ochakovski et al., 2017). An alternative approach is targeting outer retinal cells (including cones) via suprachoroidal injection with a dedicated suprachoroidal needle(Ding et al., 2019; Kansara et al., 2020).

3.2. Therapeutic approach: Retinal bipolar cell targeting for light transduction

Retinal bipolar cell targeting is an attractive optogenetic approach when the outer retina is severely damaged, with significant middle-retinal damage and/or retinal remodeling not yet occurring(Henriksen et al., 2014). However, there are limitations to this approach. A human-like fovea is only present in primates where foveal bipolar cells are displaced(McGregor et al., 2018) and parafoveal bipolar cells do not follow the typical retinal mosaic seen elsewhere in the retina(McGregor, 2019), leading to distortions in retinotopic mapping.(Chang and Goldberg, 2012; “Chapter 8 - restoring vision to the blind: Evaluating visual function, endpoints,” 2014). Thus far, channelrhodopsins (ChRs)(Doroudchi et al., 2011; Macé et al., 2015) and multicharacteristic opsin(Wright et al., 2017), human rhodopsin(Kapetanovic et al., 2017) have been effectively utilized in this approach in mouse models. While intravitreal injection with modified AAV variants(Dalkara et al., 2013),(Petrs-Silva et al., 2011) have been developed to decrease inner limiting membrane (ILM) binding and increase widespread outer retinal transduction efficiency in mouse models and nonhuman primate retinas, retinal bipolar cells are likely most efficiently targeted through the subretinal approach(Mendell et al., 2021).

Theoretically, targeting bipolar cells represents a compromise between (a) preserving as much native signal processing as possible, such as ‘ON’ and ‘OFF’ signaling(Cronin et al., 2014; Macé et al., 2015), and (b) application in late-stage disease. However, there is some debate over the efficacy of bipolar cell targeting in small animal studies. On the one hand, arecent study using mice directly compared targeting melanopsin or red-shifted channelrhodopsin to either bipolar cells or RGCs using cell type specific promotors and found that targeting bipolar cells resulted in faster kinetics and flatter intensity-response relationships compared with RGCs(Gilhooley et al., 2022). On the other hand, Chr expression in mouse RGCs using a CAG promoter has been shown to be more effective at visual restoration than Chr expression in mouse ON-bipolar cells using mGluR6 promoter(Lu et al., 2020).

3.3. Therapeutic approach: Retinal ganglion cell targeting for light transduction

RGC targeting with optogenetics is a viable approach, particularly (a) when the outer and middle retinal layers have suffered irreversible damage and (b) in mid- to late-stage disease where significant remodeling of the retinal architecture has occurred(Chang and Goldberg, 2012; Henriksen et al., 2014; Levin et al., 2011). Optogenetic delivery and activation of channelrhodopsins (Chr2) in mouse RGCs using the thy-1 promoter has proven feasibility in promoting vision restoration(Arcuri et al., 2020; G et al., 2013). However, there are important caveats to consider. There are more than 20 types of RGCs in humans, with each representing a different feature of the visual experience(Levin et al., 2011). Furthermore, RGCs, like bipolar cells(McGregor, 2019; McGregor et al., 2018), are not arranged in a straightforward mosaic in the fovea as they are elsewhere in the retina(McGregor et al., 2020), which may also create distorted image formation(Chang and Goldberg, 2012; “Chapter 2 - Restoring Vision to the Blind: Optogenetics,” 2014). Additionally, it is preferable to target the RGC cell body as opposed to the relatively long and spatially imprecise RGC axon, but this may require cell compartment-specific targeting motifs(Chang and Goldberg, 2012; “Chapter 2 - Restoring Vision to the Blind: Optogenetics,” 2014). When RGCs are transduced with depolarizing opsins, all RGCs elicit solely ‘ON’ responses forcing transduced OFF RGCs to respond to the presence of light instead of activation from the absence of light. Thus, the ability to optogenetically activate RGC OFF or ON-OFF sub-populations in a spatiotemporal pattern that is visually meaningful may be a difficult task. This may be overcome by using highly controlled light patterns with electronic goggles(Nirenberg and Pandarinath, 2012), or by differentially transducing separate RGC sub-types(Chang and Goldberg, 2012; “Chapter 2 - Restoring Vision to the Blind: Optogenetics,” 2014). Although its unknown how visual situations arising solely from RGC activation would be interpreted by the brain, there is also evidence that visual learning can offset some of the inaccuracies of artificial vision, as demonstrated by retinal prosthesis studies(Golden et al., 2019; Palanker et al., 2020). Still, optogenetic RGC delivery of channelrhodopsins(Bi et al., 2006; Ivanova et al., 2010; Tomita et al., 2010, 2007; Zhang et al., 2009), melanopsin(Lin et al., 2008), chemically engineered light-gated endogenous receptors(Caporale et al., 2011), a and “medium wavelength opsin” (MW-opsin)(Berry et al., 2019) have proved effective in marmoset, mouse, and rats models, and for patients with severely compromised vision(Sahel et al., 2021), this imprecision may be a successful trade-off. This efficacy was shown by reinstating variable degrees of visual cortex light responsiveness, pupillary reflexes, and light-avoidance behavior. Optogenetically targeting RGCs has the potential benefit of being a more straightforward task via intravitreal injection, now through machine learning and mouse models, there are improved viral vectors with increased ILM penetrance and expression efficiency and decreased immunogenecity(Cronin et al., 2014; Wec et al., 2021). Furthermore, it may be the only feasible optogenetic approach in diseases where the remaining retina is damaged and/or significant retinal remodeling has occurred(Henriksen et al., 2014; Marc and Jones, 2003),(Jones et al., 2003).

3.4. Therapeutic approach: Retinal ganglion cell targeting for optic nerve regeneration

In optic neuropathies, RGCs fail to recover after insult, resulting in the degeneration of first their axons followed by their somas. Many studies have been done to try to attenuate this process by reactivating signaling pathways involved in axonal growth and/or inhibiting pathways preventing axonal growth. For example, during development and throughout adulthood, neurotrophic factors, which may come from either the retina itself or from the optic nerve and RGCs’ targets in the brain, have been shown to stimulate axon growth and support RGC survival. Because of this, studies have analyzed the benefits of delivering neurotrophic factors to the retina to promote RGC regeneration; however, it appears that there is a narrow window of opportunity for these neurotrophic factors alone to have an effect(Benowitz et al., 2017). Previous data from rodent models suggested that physiologic levels of electrical activity could strongly potentiate neurotrophic factor responsiveness for survival and axon growth(Corredor et al., 2012; Corredor and Goldberg, 2009; Goldberg et al., 2002). Providing extra electrical activity to the retina has similarly been demonstrated in clinical trials to improve performance on visual acuity and visual field testing in human patients with optic neuropathies(Fujikado et al., 2006; Gall et al., 2016). Could delivery of depolarizing stimuli be done with optogenetic approaches instead of direct electrical stimulation? There is some preliminary data suggesting optogenetic targeting of RGCs with externally delivered LED-based blue light to activate cation flow from Chr2, a depolarization stimulus, can facilitate optic nerve regeneration following crush injury in mice(Arcuri et al., 2020). The mechanism for this effect may be through stimulation of activity-dependent signaling pathways in RGCs(Corredor and Goldberg, 2009; Moore and Goldberg, 2010), of particular interest in retinopathy scenarios in which photoreceptor loss precedes RGC axon degeneration and retinal activity loss, an approach that clearly deserves further study. Of particular interest, would be the delivery of red shifted opsins to RGCs, which could avoid the potential cell/retinal toxicity effects from continuous blue light stimulation.

In recent decades, investigations have also highlighted a range of potentially meaningful manipulations for RGC axon regeneration, showing effective and long range RGC axon regeneration in neurodegenerative animal models, extensively reviewed elsewhere(Williams et al., 2020). These investigations have largely focused on delivering genetic manipulations to RGCs in vivo via posterior segment AAV injections, for example targeting regeneration signaling cascades via PTEN suppression/deletion(Park et al., 2008), pro-survival pathways such as BCL2 over-expression(Chen et al., 1997), and the expression/downregulation of transcription factors such as KLFs(Moore et al., 2009). However, these therapies have yet to be tested in a clinical trial setting. This is in part because many of these genetic manipulations are pro-oncogenic in nature, which may provide a significant hurdle to adopting these therapies for human clinical trials. In line with this hurdle many efforts have been made towards AAV and promoter engineering for selective targeting of just RGCs, which are postmitotic cells and incapable of proliferating, with some promising success(Byrne et al., 2020). Another hurdle for these axon regeneration approaches is having an on and off switch for the therapeutic intervention, as maintaining a perpetual pro-regenerative retinal environment could lend to off target brain innervation, non-functional and/or deleterious consequences(Bray et al., 2017). This might be overcome with recent innovations in the optogenetic field, in which light activated switches have been modified to control signaling pathways, secondary messenger production, and transcriptional activity(Rost et al., 2017). Thus, combining innovative AAV approaches and optogenetic controls for axon growth promoting genetic manipulations could pave the way for overcoming safety concerns in future human trials focused on RGC axon and optic nerve regeneration. Interestingly, in rodents, light stimulation after optic nerve crush in combination with regeneration therapy via PTEN deletion, has provided evidence that light stimulated activation of visual pathways can guide and promote greater regeneration efficacy than regenerative therapy alone(Lim et al., 2016). Thus, it’s also possible that in addition to improving safety in human trials, traditional inotropic optogenetic tools could be applied in combination with pro-regenerative genetic manipulations to provide the greatest efficacy for optic nerve regeneration. Overall, while many of these approaches are still in the laboratory development stage and with many challenges still to overcome for human trials, efforts are underway to test these therapies for safety and efficacy in large animals, using pig(Holt et al., 2018) and primate regeneration models.

3.5. Therapeutic approach: Mueller glial cell targeting for neuronal protection and regeneration

Mueller glial (MG) cells are the predominant glial cell type in the retina functioning to maintain retinal homeostasis and provide support to retinal neurons(Salman et al., 2021). The dynamic role of MG including their capacity to coordinate specialized cellular microenvironments plays an important role in retinal health, disease, and regeneration(Karl and Reh, 2010). MG respond to retinal injury through a process called ‘reactive gliosis’ which is driven by changes in gene expression that promote both neuroprotective and neurotoxic functions(Ashouri et al., 2021). Given that MG cells extend across photoreceptor and RGC retinal layers, they function to compartmentalize and direct these signaling responses(Cameron and Goldberg, 2016) (Fig. 7).

Figure 7: Activated glia and immune cells in the retina acquire protective and degenerative phenotypes in neuronal disease and injury.

The extent to which cells polarize toward one phenotype could lead to zones of protective or deleterious signaling. Alternatively, reactive cells may express both signal types, compartmentalize such signaling across different processes extending into different tissue zones, and promote repair in one area and degeneration in another. Reprinted with permission from the American Association For The Advancement Of Science.

The role of MG in retinal degenerative disease has been explored in mice by optogenetically ablating MG cells, using an MG-selective AAV (ShH10) to deliver light-activated generation of oxygen radicals, (Byrne et al., 2013), which showed that MG are critical to maintaining retinal architecture and retinal cell survival (Byrne et al., 2013). MG have also been preferentially targeted through novel AAV vectors to increase neurotrophic factor (including glial cell-derived neurotrophic factor GDNF) expression resulting in retinal neural support(Dalkara et al., 2011), again supporting the role of MG in promoting photoreceptor survival.

Interestingly, activated MG have also been shown to be a source of stem cells in injured zebra fish retinas, supporting the production of neural progenitors and the regeneration of RGCs(Yurco and Cameron, 2005). This activity is in contrast with mammalian retinas, where upon activation via injury, MG gliosis leads to scar formation and no evident regeneration(Bringmann et al., 2009). Interestingly, when purified from rodent retinas, MGs are capable of producing neural progenitors and retinal neurons in vitro(Das et al., 2006), suggesting that the key features of regeneration are present but dormant in mammalian retinas in vivo. Unlocking the key mechanistic differences between zebrafish and mammalian MG activation for neural regenerative will provide new approaches for neural regeneration beyond axon regeneration. Resident MG has been proposed as an innate pool of RGC-producing cells for scenarios where advanced RGCs loss has already occurred, i.e., in advanced stages of glaucoma. Currently, the best way to activate MGs in mammalian retinas is via induced-injury, activating apoptosis and scar forming events in the retina. Thus, it’s possible that upon identifying the key neural regeneration stimulating factors of MGs in human retinas, optogenetics will be looked at as one of the principal tools to dissect the mechanisms that underlie MG activation. There is evidence that activation of diverse signaling pathways, including BMP-SMAD, Pax6, and Ascl1 transcription factor signaling pathways, growth factor signaling such as EGFR, FGF, IGF and TGF, and other secreted factors such as BMP, SHH, and glutamate(Salman et al., 2021). Overall, while MG targeting for optogenetic therapies focused on regeneration is hypothetical and needs to be assessed in preclinical and laboratory experiments, the high number, high-contact with other retinal cells, and the proven capacity to enhance protection of MGs, present a strong argument to continue the development of MG targeted optogenetic approaches for retinal dystrophy and degenerative disease.

3.6. Therapeutic approach: Corneal afferent neuron targeting for pain control

In addition to retinal neural and glial cell targeting, it is possible to optogenetically control specific neuronal populations in anterior segment of the eye. Neuropathic pain in mouse models have implicated r-hydroxytryptamin (5-HT) projections from the dorsal raphe nucleus to somatostatin expressing interneurons in the amygdala (Zhou, Nat. Neurosci 2019). Inhibiting neuropathic pain by optogenetics has been proposed by targeting afferent neurons associated with pain(Liu and Wang, 2019). Chronic corneal neuropathic pain is debilitating and is associated significant morbidity. Hyperactivity of corneal afferent neurons drives most ocular surface related pain in dry eye disease and chronic corneal exposure(Meng and Kurose, 2013). Mecum et. al. generated a mouse model of corneal neuropathy by crossing the voltage gated sodium channel Nav1.8, expressed by primary afferent corneal neurons with light sensitive, proton pump archaerhodopsin-3/eGFP mice. In a lacrimal gland excised animal, which develops chronic dry eye and induced neuropathic pain, they showed that light-activated corneal neurons alleviated the pain in the mice, concluding that targeting Nav1.8⁺ afferent neurons maybe clinically useful for treating corneal pain (Mecum et al. 2021). Optogenetic stimulation resulted in corneal afferent neuron inhibition and apparent reduction of behaviors correlated with dry eye pain(Mecum et al., 2021), showing optogenetic use for dissecting mechanism of cornea-related neuropathic pain and possible neuronal targets for treatment.

4. Active and Enrolling Clinical Trials

There are currently four companies with optogenetic product candidates undergoing active clinical trials (Table 1): (1) Nanoscope therapeutics, (2) Gensight biologics, (3) Applied genetic technologies corporation (AGTC) in collaboration with Bionic Sight, and (4) Allergan through Retrosense therapeutics. There are several other companies with preclinical data that may support future clinical trials including Acucela Inc, LambdaVision Inc, and Novartis through Vedere Bio and Arctos Medical.

Table 1:

Active Clinical Trials in Retinal Optogenetics.

Company	Opsin	Peak Absorption Spectrum	Disease	Vector	Mode	Cell Type	Trial NCT
Nanoscope Therapeutics	MCO-010	Broad, Ambient	RP	AAV2	Intravitreal injection	Bipolar	04945772
Gensight Biologics	ChR-tdT (Chr)	590 nm	RP	AAV2	Intravitreal injection	RGC	03326336
Applied Genetic Technologies Corporation	ChronosFP (Chr)	Short Wavelength	RP	AAV2	Intravitreal injection	RGC	04278131
Allergan (Retrosense Therapeutics)	ChR2 (Chr)	Short Wavelength	RP	AAV2	Intravitreal injection	RGC	02556736

Nanoscope Therapeutics has developed an engineered microbial opsin called multicharacteristic opsin (MCO-010) which is sensitive to a broad range of ambient and polychromatic light(Wright et al., 2017), potentially allowing for visual restoration without accessory light stimulation goggles. The opsin is packaged into an AAV2 vector and delivered via intravitreal injection to preferentially target retinal bipolar cells. MCO-010 has received orphan drug designations for RP and Stargardt disease. Following preliminary evidence from Nanoscope’s Phase 1/2a study in 11 patients, the safety and efficacy of a single intravitreal injection of MCO-010 is being evaluated in 27 patients with RP within the phase 2b RESTORE trial (NCT04945772).

GenSight biologics has developed a therapeutic system comprising an optogenetic vector (GS030-Drug Product (GS030-DP)) in combination with light stimulation goggles (GS030-Medical Device (GS030-MD)). The AAV2 vector (GSO3O-DP) is injected intravitreally and delivers ChrimsonR-tdTomato (ChR-tdT), a modified channelrhodopsin with enhanced absorption spectra and kinetics) preferentially to RGCs. Light activation at peak sensitivity of 590 nm (amber light) is aided by goggles (GS030-MD) that capture images from the real world and transforms them into monochromatic images in the form of 595nm light pulses projected onto the retina. ChR-tdT is red-shifted more than any previous channelrhodopsin which theoretically allows for greater safety compared to shorter (blue) wavelengths and decreased remnant photoreceptor crosstalk. The PIONEER trial (NCT03326336) is evaluating the safety and efficacy of intravitreal injection of GS030-DP in combination with GS030-MD in patients with RP. Results from one 58-year-old male patient with prior light perception vision due to RP were recently published showing that the patient was able to locate objects and count the number of cups placed on a table in front of him(Sahel et al., 2021). Visual function was corroborated by electroencephalogram (EEG) recordings showing visual cortex excitation. The study is ongoing, and one-year results are expected in 2023.

Applied Genetic Technologies Corporation (AGTC) in collaboration with Bionic Sight have created a different therapeutic system comprising an opsin and light stimulation goggles. The drug product (BS01) is an AAV2 vector containing another version of an enhanced channelrhodopsin called ChronosFP which may have improved absorption spectra and kinetics. BS01 is delivered via intravitreal injection and intended to preferentially target RGCs. Following transfection, proprietary light stimulation “neural deciphering” goggles are intended to activate cells in a manner meaningful for RGCs to create a natural visual experience(Nirenberg and Pandarinath, 2012). The therapeutic system is being evaluated in an open-label dose escalation Phase 1/2 clinical trial (NCT04278131) in patients with advanced RP.

Allergan acquired Retrosense Therapeutics in 2016 for its optogenetic technology (RST-001) for retinal dystrophic and degenerative disease. RST-001 is an AAV2 vector encoding the Type 1 opsin channelrhodopsin-2 (ChR2) which is sensitive to short wavelengths of light. RST-001 is administered via intravitreal injection and intended to preferentially transfect RGCs. Retrosense is undergoing a Phase I/IIa, open-label, Dose-Escalation Study of Safety and Tolerability of Uniocular Intravitreal RST-001 in patients with RP (NCT02556736). In this nonrandomized dose-escalation study, 1 of 3 doses of an intravitreal injection of RST-001 will be performed in 21 patients. The primary outcome is safety, and results are pending.

Several other companies are investigating optogenetic therapies for retinal disease in pre-clinical studies. Acucela Inc, a Kubota Pharmaceutical company, has developed an AAV2 vector encoding human rhodopsin (hRho). Administered via intravitreal injection, the vector will preferentially transduce retinal ON bipolar cells. Similar to the approach by Nanoscope, Acucela believes that targeting bipolar cells will provide superior signal quality and amplification, perhaps with lower light levels, and possibly without then need for external light-emitting devices. Given that rhodopsin is an endogenous protein native to human retina, there is a theoretically lower risk of immunologic reaction. Potential hurdles include the fact that human rhodopsin typically requires contact with intact RPE cells to recycle the carotenoid pigment and the opsin kinetics are slow to support dynamic visual experiences. Preclinical trials are currently underway. LambdaVision, Inc has developed an optogenetically endowed sheet which serves as a light-sensitive retinal implant(Alamusi et al., 2017; Maya-Vetencourt et al., 2017). The protein-based subretinal implant is coated with bacteriorhodopsin (a light-activated proton pump) which creates an ion gradient stimulating retinal bipolar cells and RGCs. The subretinal implant consists of multiple layers of oriented bacteriorhodopsin that are between two ion-permeable membranes. Novartis has also recently acquired Vedere Bio and Arctos Medical to explore novel optogenetic approaches for retinal disease.

5. New Directions for Ophthalmic Optogenetics: Subcellular Approaches

Subcellular targeting of optogenetic tools(Rost et al., 2017) is an exciting application and potential strength of optogenetics within ophthalmology. By employing a host of unique targeting motifs, e.g. short peptides encoding intracellular retention signals, specific targeting or anchoring signals, and fusions with full-length organellar proteins(Rost et al., 2017), one may traffic optogenetic actuators to specific subcellular domains other than surface plasma membranes.

5.1. Subcellular Optogenetic Approach: Targeting cellular compartments

Simply expressing optogenetic constructs in cells allows for the proteins to reside in many different compartments within the cells, making the light modulation less specific for any single signaling pathway and potentially affecting many cellular functions. In contrast, these constructs could be targeted to distinct cellular compartments to make the manipulation more specific. Neurons are both morphologically and physiologically polarized. Therefore, targeting optogenetic constructs to compartments including various neuronal plasma membrane compartments such as the cell body, axons, dendrites, synaptic vesicles, and post/presynaptic terminals may influence the cellular response to the manipulation. This principle is ensconced in the discussion of Mueller glia above, in which localizing signaling to the cell’s inner versus out compartment could preferentially affect the biology of RGCs versus photoreceptors, respectively (Fig. 7).

An important example of subcellular compartmentalized signaling that influences RGC survival comes from studies by us and others. For example, a series of studies exploring the localization of cAMP and PKA signaling in rodent models, which were previously shown to mediate the positive effects of electrical activity in RGCs(Corredor et al., 2012; Goldberg et al., 2002; Stiles et al., 2014), demonstrated that a scaffolding protein, A-Kinase anchoring protein 6 (AKAP6, also called mAKAP), localizes to the perinucleus and is critical for compartmentalized cAMP signaling, which influences RGC outgrowth and survival in rats and mice(Boczek et al., 2019; Wang et al., 2015) (Fig. 8). Subcellular optical readouts using FRET sensors were critical to proving that this local compartment’s importance in neuroprotection in rat and mouse e models in vivo(Boczek et al., 2019); reversing this optical approach to provide local, optogenetics-driven signaling would be a major step forward.

Figure 8: Displacement of PDE4D3 from mAKAPα elevates cAMP signaling in the perinuclear region of the cell and promotes RGC neurite extension.

A. Grayscale images of mCherry fluorescence for hippocampal neurons transfected with mCherry or 4D3(E)-mCherry expression plasmids and cultured for 2 d in defined media. Scale bar, 100 μm.

B. Mean lengths of the longest neurite are shown for four independent experiments (different colors). *p ≤ 0.05, **p ≤ 0.01. Reprinted with open access and approval through creative commons (creativecommons.org)

This local, optogenetics-driven compartmented signaling was successfully leveraged in another series of experiments by targeting constructs to a specific compartment in trabecular meshwork cells. The most common way to target an optogenetics protein to a specific compartment is by using localization sequences or short peptides from proteins that typically reside in the target compartments and are responsible for subcellular trafficking and/or retention there. For example, the CAAX domain from the c-terminus of the K-Ras protein will target a construct to the plasma membrane.(Idevall-Hagren et al., 2012; Nagai et al., 2004; Toettcher et al., 2011; Zhang et al., 2014) In trabecular meshwork cells, membrane targeting of a protein that regulates actin cytoskeleton structure using the CAAX sequence was shown to be successful in reducing IOP in mouse model systems(Kowal et al., 2021; Prosseda et al., 2020) (Fig. 9). Other targeting sequences allow for even more specificity of the membrane compartments of a neuron. To target the axon of rodent cortical neurons, the c-terminus of myosin VI was used(Lewis et al., 2011). Even more specifically, the axon initial segment can be targeted by using the N-terminus of ankyrin G in rabbit retinas (Greenberg et al., 2011) or the C-terminal domain of Nav proteins which bind to ankyrin G in rodents(Grubb and Burrone, 2010; Wu et al., 2011). Postsynaptic and dendritic spines can be targeted using motifs from the Post Synaptic Density protein(Gradinaru et al., 2007; Greenberg et al., 2011; Hayashi-Takagi et al., 2015; Leitz and Kavalali, 2014; Mao et al., 2008; Sinnen et al., 2017)as observed in mice, rabbits, and rats. Using the targeting sequence from synaptophysin or synaptobrevin, an optogenetic protein can be targeted to synaptic vesicles. This has been shown to successfully acidify synaptic vesicles in order to increase transmitter loading and transmitter release in rodents(Rost et al., 2015) and to use light activation to inhibit transmitter release essentially inhibiting synapse transmission in c. elegans, rats, and mice.(Hermann et al., 2015; Lin et al., 2013; Shu et al., 2011) Interestingly, targeting excitatory cation channel channelrhodopsin-2 (hChR2) to the soma and inhibitory chloride pump enhanced halorhodopsin (eNpHR) to the dendrites, recreated center-surround RGC responses.(Greenberg et al., 2011) Additional motifs were subsequently identified that enhanced the optogenetic-mediated center-surround response of RGCs(Wu et al., 2013).

Figure 9: Optogenetic recruitment of OCRL to the plasma membrane.

Stimulation with blue light recruits mCh-Cry2-OCRL to CIBN-CAAX-GFP localized at the plasma membrane

- CIBN-GFP was modified with the targeting sequence CAAX, causing the construct to localize to the plasma membrane. Stimulation with blue light caused Cry2-OCRL-mCherry to interact with CIBN-CAAX-GFP and therefore localize to the plasma membrane. CIBN-GFP was modified with the cilia targeting sequence from the Somatostatin receptor 3 protein (SSTR3), causing the construct to localize to the plasma membrane. Stimulation with blue light caused Cry2-OCRL-mCherry to therefore localize to the primary cilia. Reprinted with author access through Translational Vision Science & Technology.

5.2. Subcellular Optogenetic Approach: Targeting cilia for intraocular pressure control

Lowering intraocular pressure (IOP) is the only current treatment for glaucomatous optic neuropathy. The trabecular meshwork (TM) serves as the major level of resistance for aqueous outflow which maintains appropriate intraocular pressures, and targeting the TM maybe be beneficial for lowering intraocular pressure to reduce glaucomatous RGC loss. In two complimentary studies, Prosseda et al.(Prosseda et al., 2020) and Kowal et al.(Kowal et al., 2021) showed that IOP can be regulated using two disease model systems by modulating the function of trabecular meshwork using optogenetics (Fig. 10). In both cells in culture and in mice, these experiments used a blue light stimulated cryptochrome optogenetics system, which may be modified for clinical studies but provides proof of concept that an optogenetics system can be stimulated in the trabecular meshwork. Importantly, in mice there is evidence to suggest that channelrhodopsin can be used to activate transient receptor potential villanoid receptors(Jeong et al., 2018), which play a role in regulating IOP(Patel et al., 2021). Although it may seem that this approach will only be beneficial during an awake state, there have been studies in patients that investigated the use of night lights or light masks that emit the appropriate wavelength of light to attenuate macular degeneration (McKeague et al. 2014, Robinson et al. 2018) and diabetic retinopathy (Arden et al. 2010, Arden et al. 2011, Sahni et al. 2017). We propose that a similar mechanism could be used to maintain the optogenetics treatment during sleep. Therefore, using optogenetics to localize or regulate a protein within trabecular meshwork cells may be a novel approach to preventing further damage to the retina when typical glaucoma therapeutic approaches are not feasible.

Figure 10: Trabecular meshwork targeting and functional rescue using CRY2/CIBN optogenetics.

A. AAV2-s injected into the anterior chamber was used to transduce trabecular meshwork with the optogenetics CRY2/CIBN system. Plasma membrane targeting AAV2-s-CIBN-EGFP-CAAX was observed in the trabecular meshwork 4 weeks post-injection. Control trabecular meshwork did not have any GFP signal.

B. Outflow facility was measured by perfusion. Blue light stimulation of the CRY2/CIBN system, which caused OCRL to localize to the plasma membrane, rescued the decreased outflow facility observed in the glaucomatous eyes of the Lowe syndrome mouse model.

C. Additionally, tonometer readings indicate that the intraocular pressure was reduced with blue light stimulation. Reprinted with author access through Translational Vision Science & Technology.

D. AAV2-s injected into the anterior chamber was used to transduce trabecular meshwork with the optogenetics CRY2/CIBN system. Plasma membrane targeting AAV2-s-CIBN-EGFP-CAAX was observed in the trabecular meshwork 4 weeks post-injection. Control trabecular meshwork did not have any GFP signal.

E. Outflow facility was measured by perfusion. Blue light stimulation of the CRY2/CIBN system, which caused OCRL to localize to the plasma membrane, rescued the decreased outflow facility observed in the glaucomatous eyes of the Lowe syndrome mouse model.

F. Additionally, tonometer readings indicate that the intraocular pressure was reduced with blue light stimulation. Reprinted with author access through Translational Vision Science & Technology.

5.3. Subcellular Optogenetic Approach: Targeting organelles and opsins to restore cellular ATP (ORCA)

Subcellular targeting of optogenetic tools may also be used to modulate the structure and/or function of specific organelles, including endoplasmic reticulum, golgi apparatus, autophagosomes, lysosomes, and even mitochondria(Rost et al., 2017). Our group has explored applications for subcellular optogenetics in retinal diseases, with a novel focus of optogenetic modulation of mitochondria and metabolism.

Mitochondria are double membrane bound organelles found in all eukaryotic cells that can dynamically change in number per cell, size, and position, to distribute and provide a diversity of cellular functions, including adenosine triphosphate (ATP) production via the electron transport chain (ETC), intracellular calcium buffering and signaling, redox reactions and reactive oxygen species (ROS) signaling, apoptosis coordination, iron-sulfur cluster synthesis, and protein signaling events(Wallace and Chalkia, 2013). Given that these functions serve to maintain generalized cellular homeostasis and with the neural retina and RPE being amongst the most metabolically active tissues in the body, it’s not surprising that mitochondrial dysfunction has been shown to strongly contribute to inherited retinal disease(Zhu et al., 2017), diabetic retinopathy (DR)(Alam et al., 2015; Zhu et al., 2017), retinopathy of prematurity (ROP)(Beharry et al., 2016), glaucoma(Kamel et al., 2017), and age-related macular degeneration (AMD)(Fisher and Ferrington, 2018). In many of these retinal diseases the sequela of mitochondrial dysfunction has been linked to progressive neurodegeneration and cell loss(Eells, 2019; Lefevere et al., 2017), with similar hallmarks found in traumatic brain and spinal cord injury(Merlini et al., 2022; Wang et al., 2021). Thus, to understand the role of mitochondrial function in retinal and nerve degenerative events, we have been investigating the role of mitochondria in models of glaucomatous injury as well as manipulating mitochondrial biology to identify therapeutic interventions for axon regeneration.

As mitochondrial size is thought to modulate many mitochondrial functions, and with apoptosis signaling linked to mitochondrial fragmentation in neurons(Knott et al., 2008), we aimed to identify interventions that suppress mitochondrial fragmentation in RGCs. We initially highlighted the importance of mitochondria for RGC growth and patterning during development in rats, showing increased mitochondrial size and metabolic activity in axon growth and development(Steketee et al., 2012). We also found that manipulating mitochondria in rat RGCs by increasing mitochondrial size, with pharmacological and genetic interventions, could decrease responses to axon growth repulsion cues and promote greater axonal growth in vitro(Kreymerman et al., 2019; Steketee et al., 2012)(Fig. 11a–b). However, manipulation of mitochondrial size alone could not improve RGC survival or promote axon growth in vivo in optic nerve injury models (Fig. 11c–d)(Kreymerman et al., 2019), despite others showing modest improvements in retinal cell survival with pharmacological targeting of mitochondrial fragmentation in glaucoma mouse models(Park et al., 2011). Additionally, although mitochondria size has been linked to alterations in mitochondrial based metabolism which plays a central role in maintaining cell health(Youle and van der Bliek, 2012), it’s not clear whether our manipulations of mitochondrial size played a therapeutically meaningful role in shifting metabolic activities.

Figure 11: MTP18 a gene known that promotes mitochondrial fragmentation was knocked down to promote neurite outgrowth on inhibitory substrate but is not sufficient for promoting axon regeneration after optic nerve crush injury.

A. Representative images of cultured RGCs with neurite growth shown after electroporation with a non-targeting scramble siRNA or MTP18 targeting siRNA and seeding on PDL + Laminin or PDL + Laminin + CSPGs (3 μg/ml) for 72 hrs (scale bar 100 μm).

B. Average neurite length of cultured RGCs electroporated with scramble siRNA or MTP18 siRNA and seeded onto different concentrations of CSPGs. All points are N = 4 repeat electroporation experiments per point, normalized to average neurite length of scramble siRNA seeded onto PDL + Laminin only treated wells (significance between scramble and MTP18 siRNA treatments at each CSPG concentration was determined by Student’s t-test, *p ≤ 0.05).

C. Scrambled and MTP18 siRNA-electroporated RGCs were plated at the same density as in neurite outgrowth experiments and were assayed for living and dead cells, using calcein AM (green labeled live cells) and sytox (orange labeled dead cells). Tiled images were captured per well and green and red cells were counted and graphed as percent ratio, live over dead cells, normalized to scrambled siRNA controls. (N = 4 repeat electroporation experiments, significance tested for by Student’s t-test, p ≥ 0.05.)

D. Representative fluorescent images of P40 rats intravitreally injected with AAV2 viruses expressing shRNA against MTP18 (Anti-MTP18) or control shRNA against a non-present luciferase gene (Anti-luciferase) prior to optic nerve crush. CTB Alexa-555 labeled axons after optic nerves crush show no significant regeneration. Mitochondrial (mito.) mTurquoise labeling from prior virus transduction and merged images with CTB, show degenerate labeling past the crush site and bright linear labeling of some preserved axons, just prior to the crush site (left side of images, scale bar 500 μm). Reprinted with open access and approval through creative commons (creativecommons.org)

Our recent efforts have focused on addressing this gap by identifying direct manipulations of mitochondria metabolism and energetics, including a novel optogenetic approach aimed at increasing mitochondrial ATP production. To see if we could also apply optogenetic tools to modulate mitochondrial function and metabolism within ocular cells, we inserted type 1 proton (H+) pumping opsins into the inner mitochondrial membrane of retinal cells to promote a light-generated mitochondrial H+ gradient for ATP production, independent of the ETC (Fig. 12A). We attempted this approach with three unique proton pumping opsins (Fig. 12B), each of which were designed to contain a fused N-terminal inner mitochondrial membrane localization sequence (IMMLS) to drive mitochondrial integration, and a C-terminal Flag tag for antibody detection of modified opsins. Human RPE (hRPE) cells were then transfected with modified opsins (Fig. 12C), and cells were imaged to confirm that these modified opsins insert into mitochondria by immunostaining for flag tagged opsins (red) in cells with green fluorescent protein (GFP) labeled mitochondria (Fig. 12D). The mitochondrial-modified opsins were tested for their capability to increasing ATP levels by light, exogenous to normal ETC function, by shutting down ETC activity initiation and thus ATP with a complex 1 inhibitor, rotenone (Fig. 12E). Under these conditions our preliminary results suggest that these newly engineered opsins are capable of producing ATP in the presence of light stimulation despite the presence of ETC inhibition (LS opsin) (Fig. 12F–H) when compared to all other conditions i.e., cells alone with no light stimulation (NLS Control), light stimulation with no opsins (LS Control), and opsins with no light stimulation (NLS Opsin). This was the case with all tested opsins bacteriorhodopsin (bR) (Fig. 12F), deltarhodopsin (dR) (Fig. 12G), and xanthorhodopsin (xR) (Fig. 12H), however the standard deviation was high in some cases warranting further experiments. Nonetheless, this proof of concept termed ‘Opsins to Restore Cellular ATP’ (ORCA) shows that optogenetics can also be used to augment cellular metabolism and could be a very promising application for retinal degenerative and dystrophic disease. Further studies are underway in the lab to explore the observed effects in different cell types, models of retinal dysfunction/disease, and for RGC axon regeneration.

Figure 12: ORCA Schematic and rationale.

A. Model for the ORCA approach: inserting type 1 proton (H+) pumping opsins into the inner mitochondrial membrane will allow ocular cells exposed to light to generate a mitochondrial H+ gradient for subsequent ATP production via ATP synthase but independent of the rest of the electron transport chain (ETC) machinery.

B. To test this hypothesis, we selected three different H+ pumping opsins not previously tested in the ophthalmology field. bR, dR, and xR generalized structures shown, along with activation wavelengths.

C. Human retinal pigment epithelial (hRPE) cells were transfected with each plasmid using lipofectamine. The plasmid sat extrachromosomally in the nucleus, read by host hRPE machinery, and shuttled to the inner mitochondrial membrane

D. Transfected cells show that opsins colocalize with mitochondria. Immunofluorescent staining showed opsins (red), detected by anti-flag antibody (F1804, Sigma-Aldrich) and Alexa Fluor^™ Plus 555 (A32727, ThermoFisher Sci.), colocalize with mitochondria (green), in cells transfected with a plasmid encoding modified opsins and a mitochondrial targeted GFP.

E. We selected toxins that inhibit the mitochondrial electron transport chain (ETC) to simulate clinical mitochondrial dysfunction. Rotenone inhibits Complex I, decreasing the buildup of a transmembrane proton gradient.

F. hRPE plated on 96 well plates were transfected with modified opsins or empty plasmids (control plasmid), exposed to 1 uM Rotenone for 24hrs, and then exposed to 1 hr of light stimulation. ATP levels where then detected on a plate reader using a luminescence assay (CellTiter-Glo^® 2.0, G9241, Promega), according to manufacturer specifications. Comparing LS_bR Plasmid to LS_Control Plasmid suggests that the H+ opsin bacteriorhodopsin (bR) may have an effect in increasing hRPE ATP in the presence of rotenone.

G. A similar trend was observed for deltarhodopsin (dR).

H. A similar trend was observed for xanthorhodopsin (xR).

6. Conclusions

Francis Crick’s statement in 1999 anticipated the field of optogenetics: “one of the next requirements [for neuroscientific tools] is to be able to turn the firing of one or more types of neurons on or off in the alert animal in a rapid manner. The ideal signal would be light, probably at an infrared wavelength to allow the light to penetrate far enough.”(Crick, 1999) However, the field has now shown promise and indeed delivered on that promise well beyond applications as a neuroscientific tool. Optogenetics improves our understanding of the role of individual functional components within complex neural circuits that can be targeted for repair(Ivanova et al., 2016). Current therapeutic approaches and clinical trials hold great promise for the treatment of ophthalmic disease using optogenetics, independent of any patient’s underlying a mutation-agnostic manner. Extension to exogenous control of cellular and compartmented signaling, tissue homeostasis, and regulation of metabolism, as we introduce here with ORCA, are some of the additional, exciting directions that basic and translational research stand to leverage with optogenetics approaches.

However, the field will benefit from further directing optogenetic tools to specific subcellular compartments and specific sub-populations of cells for more nuanced light-based control. Other elements of optogenetic approaches are worth improving. The primary limitation is the high light levels required to activate type 1 opsins which may require an external light source for optimal excitation. While many red-shifted opsins have been created(Douar et al., 2016; Klapoetke et al., 2014),(Erbguth et al., 2012; Sengupta et al., 2016),(Govorunova et al., 2011; Zhang et al., 2008), the continued design and implementation of red-shifted and infrared (IR) opsins would be beneficial for both cellular safety and in minimizing cross-talk between remnant light-responsive photoreceptors (that have no far-red or IR absorption)(Chang and Goldberg, 2012; “Chapter 2 - Restoring Vision to the Blind: Optogenetics,” 2014). By engineering and/or discovering novel opsins with longer channel open time, increased permeability, and improved light sensitivity parameters, one could also lower the threshold of light required for stimulation. Several novel opsins have been recently utilized that facilitate improved kinetics under ambient lighting conditions in mice including the MW-opsin(Berry et al., 2019) and the ChR variant Chloromonas oogama (CoChR)(Ganjawala et al., 2019).

In the absence of sophisticated heterotopic mapping of opsins and maintaining of center-surround, on/off signaling, and other complex features that make light transduction visually meaningful, the quality of vision produced by optogenetic tools may also be relatively limited. While viral-based gene therapy made a significant advancement with Luxturna (voretigene neparvovec-rzyl) to treat patients with RPE65 mutations, there are still many questions surrounding the immunogenicity, safety, efficacy, and longevity surrounding gene therapy. Improved viral vectors with decreased immunogenicity that are capable of specific cell-targeting, along with increased and stable expression will be of great benefit(Cronin et al., 2014; Ivanova et al., 2010; Wec et al., 2021). The further development of surgical tools, including intraoperative OCT and novel surgical techniques for subretinal surgery may improve the safety and efficacy of therapeutic agents used for subretinal delivery(Wood et al., 2019). In addition to naturally occurring type 1 and 2 opsins, there are many synthetic opsins and opsin-like molecules under development to overcome barriers such as photon capture efficiency, photocycle refractory time, and wavelength sensitivity(Henriksen et al., 2014; Rost et al., 2017). Additionally, the advent of long-lasting photoswitches without the need for frequent re-administration has the potential to validate optogenetics as a clinical tool that is reliable and also reversible. Recent innovations in the design of viral capsids(Botto et al., 2022) and cell-specific promoters(Hulliger et al., 2020) promise improved clinical application of optogenetic therapies. Finally, retinal remodeling(Marc et al., 2014; Pfeiffer et al., 2020) results in anatomically and physiologically disrupted cellular connections which in late stages resembles central nervous system (CNS) proteinopathies, and remains a major limitation to many restorative and regenerative therapeutic approaches.

Despite all potential drawbacks, the future of optogenetics is promising. Both laboratory science and patients will benefit from a larger library of clinically useful optogenetic approaches. In particular, light-based sub-cellular and second messenger signaling regulation is an optogenetic application poised for ophthalmic scientific and clinical utility. Overall, vision research and clinical ophthalmology together have great potential to elevate the utility of optogenetics.

Article Highlights.

• While optogenetics within ophthalmology has been primarily applied towards treating inherited retinal disease, there are a myriad of other applications that hold great promise for a variety of eye diseases including cellular regeneration, modulation of mitochondria and metabolism, regulation of intraocular pressure, and pain control.

• Several optogenetics techniques can be applied to treat retinal diseases. The current tools and approaches are reviewed in detail including the potential target cells and sub-cellular targets.

• Exciting advancements have been made in the clinical utility of ophthalmic optogenetics. Here we outline the details of several clinical trials are underway to bring this technology to patients.

Abbreviations

IRD

Inherited Retinal Disease

AMD

Age-Related Macular Degeneration

RPE

Retinal Pigment Epithelium

ChR

Channelrhodopsin

NpHR

Halorhodopsins

OPN4

Melanopsin

RGC

Retinal Ganglion Cell

LiGluR

Light-Gated Ionotropic Glutamate Receptor

AAQ

Acrylamide-Azobenzene-Quaternary-Ammonium

DENAQ

Diethylamine-Azobenzene-Quaternary-Ammonium

DENAQ

Benzylethylamine-Azobenzene-Quaternary-Ammonium

AAV

Adeno-associated viruse

ILM

Internal Limiting Membrane

MG

Mueller Glia

MCO-010

Multicharacteristic Opsin

ChR-tdT

ChrimsonR-tdTomato

ETC

Electron Transport Chain

ROS

Reactive Oxygen Species

ATP

Adenosine Triphosphate

IOP

Intraocular Pressure

TM

Trabecular Meshwork

bR

Bacteriorhodopsin

dR

Deltarhodopsin

xR

Xanthorhodopsin