Therapeutic Gene Editing in Inherited Retinal Disorders

Abstract

Since the development of CRISPR/Cas9 gene editing in 2012, therapeutic editing research has produced several phase 1-2a trials. Here we provide an overview of the mechanisms and applications of various gene-editing technologies including adeno-associated virus vectors, lentiviruses, CRISPR/Cas9 systems, base and prime editing, antisense oligonucleotides, short-hairpin RNAs, Cas13, and adenosine deaminase acting on RNA for the treatment of various inherited retinal diseases (IRDs). We outline the various stages of clinical trials using these technologies and the impacts they have made in advancing the practice of medicine.

The eye is an extraordinary organ that harbors several characteristics conducive for the implementation of gene therapy. The retinal–blood barrier, which excludes ocular immune cell penetration, and the presence of cytokines and signaling molecules, which promotes regulatory T-cell activation and immunosuppression, confer a unique immune-privileged state to the eye (Medawar 1948; Stein-Streilein and Taylor 2007; Caspi 2010). Anatomically, the eye consists of a limited volume of cells, and the organization of these cells into their functional layers is amenable to the delivery of gene therapy vectors. The bilateral nature of human eyes is of particular interest for clinical trials as it enables one eye to serve as the treatment recipient and the other to serve as the control. Here we provide an overview of the current applications of gene editing in the treatment of retinal diseases and include an update on promising developments in recent clinical trials and cutting-edge technologies.

STRATEGIES IN GENE THERAPY AND EDITING

Adeno-Associated Virus

Historically, the adeno-associated virus (AAV) vector has served as the delivery modality of choice for the treatment of ocular disease as it exhibits low immunogenicity and high transduction efficiency in retinal cells (Timmers et al. 2020). Although AAV-based therapies have found success in the treatment of some retinal diseases (i.e., RPE65-associated Leber congenital amaurosis [LCA] summarized in Table 1; Bainbridge et al. 2008, 2015; Maguire et al. 2008; Cideciyan et al. 2013; Russell et al. 2017; Duncan et al. 2018), the genetic payloads delivered by AAV-based therapies are restricted to ∼4.5 kb due to the small size of the AAV genome (Bainbridge et al. 2008, 2015). Thus, retinal disease-causing alterations in larger genes such as ABCA4, USH2A, and CEP290, have been difficult to correct using conventional AAV-based methods. In addition, AAV-based strategies rely on the supplementation of a wild-type (WT) copy of the mutant gene, which is not conducive to the treatment of autosomal-dominant retinal diseases. Finally, gene therapy applications of AAV vectors often require removal of the rep open reading frame (ORF) to prevent vector integration into the host genome. Thus, AAV vectors are conventionally expressed as episomes with waning therapeutic efficacy over time (Trapani and Auricchio 2018). Some groups have circumvented these obstacles by engineering dual AAV systems that carry segmented transgenes or by constructing a truncated, yet functional form of the gene of interest that is operable within the size constraints of the AAV vector (Lai et al. 2009).

Table 1.

Description of various gene-editing strategies that have been developed in recent years along with any associated ongoing clinical trials

Gene-editing strategy	Diseases	Clinical trial status	Notes
Adeno-associated virus (AAV) vector	RPE65-associated Leber congenital amaurosis (LCA) (FDA approved; NCT00999609) CEP290-associated LCA (phase 2; NCT03872479)	CMS and FDA approved	Low immunogenicity and high transduction efficiency in retinal pigmented epithelium Therapeutic gene is limited to ∼4.5 kb in size Difficult to apply to autosomal-dominant retinal diseases Waning therapeutic efficacy over time
Lentiviruses	Stargardt patients (NCT01367444)	Phase 1-2a; terminated as sponsor decided to stop development of product	Larger genetic payloads of up to ∼9 kb Unwanted insertional mutations in host
CRISPR/Cas system	RHO-adRP (preclinical) Stargardt macular degeneration LCA	Preclinical	Off-target effects Inability to cleave single-stranded DNA (ssDNA) (Cas9) Requires specific G-rich protospacer-adjacent motif (PAM) sequence localization (Cas9) Inefficient in mitotically inactive cells (Cas9)
Base editing	Muscular atrophy; inherited liver and skin diseases; RPE65-associated LCA; sickle cell disease; ALL; α1 antitrypsin deficiency; Stargardt disease	Preclinical	Independent of double-strand break (DSB) formation and homology-directed repair (HDR)-dependent DNA repair High fidelity and efficiency Minimal off-target mutations and indel events
Prime editing			Facilitates reverse transcriptase (RT) hybridization and mutation correction on the opposing strand Does not require HDR-mediated DNA repair or DSB formation Limited to transition mutations Dependent on PAM position near the desired incision site
Antisense oligonucleotides	RHO-adRP (preclinical, phase 1/2; NCT04855045) CEP290-associated LCA (preclinical) Usher syndrome type 2 (phase 1/2; NCT03780257) SCA7 (preclinical murine model)	Preclinical, phase 1/2	May be used in treatment of inherited retinal diseases (IRDs) Unscalable to all mutated genes causing disease
Short-hairpin RNAs	RHO-adRP (preclinical canine model)	Preclinical	Require delivery via a viral vector
Cas13 and ADAR	Usher syndrome type 2 (preclinical)	Preclinical	Improved specificity and off-target effects compared to previous RNAi approaches

Lentiviruses

Lentiviruses are retroviruses with the ability to accommodate delivery of larger genetic payloads of up to ∼9 kb and have been viewed as attractive alternatives for the treatment of retinal diseases mediated by mutations in large causative genes (Coffin et al. 1997). However, lentiviral transduction results in integration of foreign DNA into the host, which can lead to unwanted insertional mutations (Trapani et al. 2014). The development of integration-deficient lentiviral vectors (IDLVs) enables expression of the lentiviral vector as a stable episome and has been successfully applied to the treatment of retinal dystrophy in a murine model (Yáñez-Muñoz et al. 2006).

DNA EDITING

CRISPR/Cas9

The CRISPR/Cas9 system is the current state-of-the-art technique for genomic surgery. Conventional applications center on a Cas9 endonuclease that is directed across the genome by a single-guide RNA (sgRNA) (Jinek et al. 2012). The sgRNA hybridizes with a complementary target DNA strand, and the Cas9 endonuclease delivers a precise molecular incision. This Cas9-mediated DNA cleavage relies on positioning near a spacer sequence or protospacer-adjacent motif (PAM), which must be directly downstream of the incision site. These DNA double-strand breaks (DSBs) subsequently activate one of two DNA repair pathways, the homology-directed repair (HDR) or nonhomologous end joining (NHEJ), which are exploited to generate a desired genetic alteration.

Cas 12

Cas12a (previously known as Cpf1) and its orthologs possess unique properties that expand on the repertoire of CRISPR/Cas applications. Compared to Cas9, they are more compact in size, generate staggered double-stranded DNA (dsDNA) breaks instead of blunt-ended dsDNA breaks, and reportedly have lower overall off-target effects (Anzalone et al. 2020). In addition, Cas12 can also cleave single-stranded DNA (ssDNA). Recent studies demonstrate that following recognition of a complementary DNA sequence, Cas12 can indiscriminately cleave surrounding ssDNA molecules, a property that has been exploited for the development of novel diagnostic technologies (Broughton et al. 2020). One key limitation of Cas9 and its orthologs is their heavy reliance on G-rich PAM sequences for localization. In contrast, Cas12a and its variants primarily recognize T-rich PAM sequences. Thus, a combination of Cas9 and Cas12 applications may expand the areas of the genome that are amenable to gene editing in retinal diseases.

CasX, also known as Cas12e, harbors many of the same features as the other Cas12 proteins in its family (Liu et al. 2019). However, it has an even lighter molecular weight with a size of only ∼980 amino acids (aa), which may facilitate efficient delivery. Furthermore, the mechanism by which the CasX protein processes and cleaves DNA differs from its predecessors. Scribe Therapeutics is optimizing this novel editor toward the treatment of several ophthalmologic diseases including retinitis pigmentosa, Stargardt macular degeneration, and LCA.

Cas14

Cas14 is a remarkably compact endonuclease identified from uncultivated archaea bacteria with a size ranging from 400 to 700 aa. In contrast to prior Cas proteins, it is guided by an RNA molecule and selectively binds and cleaves ssDNA. Strikingly, its function and localization are not dependent on the proximity of a PAM sequence, and like Cas12a, harbors potential to be used in the future for exquisite diagnostic applications.

Base Editing

CRISPR/Cas9 approaches for the correction of point mutations initially centered on the delivery of a WT Cas9 and an sgRNA directed toward the site of the point mutation followed by HDR using a functional donor sequence (Chapman et al. 2012). Unfortunately, HDR-mediated repair is highly inefficient, particularly in mitotically inactive cells. In addition, Cas9-mediated DSB formation is susceptible to widespread, inadvertent indels that may produce unwanted deleterious effects (Lin et al. 2014; Zhang et al. 2015).

Recent advancements have revealed a new class of gene-editing tools that are capable of editing ssDNA independent of DSB formation and HDR-dependent DNA repair. Termed base editors, they include cytosine base editors (CBEs), which consist of a catalytically inactivated Cas9 protein (dCas9) linked to a cytosine deaminase (rAPOBEC1), and adenine base editors (ABEs), which consist of a dCas9 linked to an adenine deaminase (Komor et al. 2016; Gaudelli et al. 2017). CBEs and ABEs perform point changes of C•G to T•A and vice versa, respectively, with high fidelity and efficiency and have been successfully applied to a variety of genetic diseases including muscular atrophy and inherited liver and skin diseases in preclinical studies (Rossidis et al. 2018; Villiger et al. 2018; Osborn et al. 2020).

Suh et al. demonstrated the utility of base editors in the treatment of RPE65-associated LCA caused by an underlying point mutation. Specifically, the group engineered a lentiviral vector containing a codon-optimized ABE variant (ABEmax) and an sgRNA directed toward a point mutation in exon 3 (c.130C > T; p.R44X) of the Rpe65 gene. In rd12 mice, lentiviral transduction via subretinal injection corrected the de novo nonsense mutation with a maximum 29% efficiency, recovered visual chromophore production, and restored retinal function as assessed by scotopic electroretinography (ERG), optomotor responses (OMRs), and visually evoked potentials (VEPs) (Suh et al. 2021). Strikingly, these improvements were achieved with minimal off-target mutations and indel events. Recently, base editing was demonstrated to be efficacious in long-lasting restoration of cone functionality and survival in mice, building on the success of AAV-mediated treatment of retinal pigment epithelium (RPE)-LCA. The patients with LCA who were treated with the AAV-mediated therapy often developed continued retinal degeneration after treatment (Cideciyan et al. 2013; Bainbridge et al. 2015). Base editing offers promising development of treatment for a variety of retinal diseases on a long-term basis (Caruso et al. 2022; Choi et al. 2022).

BEAM Therapeutics is leveraging base-editing techniques in the treatment of a variety of genetic diseases including sickle cell disease, acute lymphoblastic leukemia (ALL), α1 antitrypsin deficiency, and Stargardt disease, an inherited retinal disease caused by a wide array of mutations in the ABC4A gene (Tanna et al. 2017).

Prime Editing

While base editing is able to overcome a critical obstacle in its avoidance of DSB formation and HDR-dependent repair, its applications are limited to transition mutations and are dependent on PAM position near the desired incision site. Prime editing (PE) relies on reverse transcriptase (RT)-mediated repair to expand on the repertoire of desired mutational corrections including deletions, insertions, and all types of transitions and transversions, and, like base editing, PE does not require HDR-mediated DNA repair or DSB formation. PE consists of an RT, a pegRNA, and an SpCas9 nickase modified with an H840A mutation that causes the nickase to preferentially generate a single-strand break (SSB) along the PAM-containing strand (da Costa et al. 2021). The pegRNA contains an sgRNA directed toward the mutation site and an extension segment containing a spacer sequence, an sgRNA scaffold, a primer-binding sequence (PBS), and a reverse transcription template (RTT). The spacer sequence extends the 3′ end of the sgRNA, facilitating RT hybridization and mutation correction on the opposing strand (Fig. 1).

Figure 1.

Prime editing is a novel strategy that uses Cas 9 nickase, pegRNA, and reverse transcriptase to edit target DNA. (1) The target DNA binds to the prime editing complex via protospacer-adjacent motif (PAM), and (2) the Cas 9 nickase creates a single-strand break (SSB) in the target DNA. (3) The reverse transcriptase reads the RNA and synthesizes new DNA to the edited strand. (4) The edited strand and unedited strand now have a base pair mismatch, and then a guide RNA (gRNA) directs the prime editor complex to create another SSB in the unedited strand. (5) The cell remakes the nicked strand using the edited strand as a template, creating a double-stranded section of edited DNA (in yellow). (RTT) Reverse transcription template, (PBS) primer-binding sequence.

Recently, Zhi et al. (2022) engineered a dual AAV8 split-PE system, which consists of Split-ABE- and Split-CBE-containing vectors. Split-ABE was then generated by two distinct mechanisms of splicing either by mRNA trans-splicing or intern-mediated protein slicing, the latter of which produces a Split-ABE that is more efficient in genome editing in vivo. Using this approach, the group delivered the dual AAV8 split-PE system into adult mouse retina via subretinal injection and successfully generated a G•A transversion in the Dnmt1 gene at the p.P55Q, c.G164T position. The transversion efficiency of the dual system was only 1.71%.

RNA Editing

CRISPR/Cas9 and other relevant genome surgery approaches result in permanent alterations to the patient's genomic DNA. Many patients often opt for strategies that ensure the integrity of the genome and limit permanent modifications to their underlying DNA. Thus, approaches that modulate disease processes at the RNA level are highly desirable and can address these patient-centered concerns. RNA-editing strategies include degradation of pathological mRNA molecules, modulation of pre-mRNA splicing, and, more recently, individual base editing of mRNA transcripts.

Antisense Oligonucleotides

Antisense oligonucleotides (ASOs) are synthetic small RNA molecules or nucleotide analogs that bind to complementary pre-mRNA sequences. Depending on the chemical modification made to these molecules, ASOs can initiate a variety of biochemical activities following pre-mRNA hybridization including mRNA degradation, translational obstruction, and exon exclusion/inclusion (Hammond and Wood 2011).

ASOs with the ability to degrade mRNA are of particular interest in the treatment of autosomal-dominant inherited retinal diseases (IRDs), which have been unamenable to gene replacement strategies in mouse models of retinal dystrophies (Wu et al. 2022). These ASO moieties hybridize to complementary pre-mRNA molecules and recruit ribonuclease H (RNase H1) to cleave the bound pre-mRNA molecule. Several ASO therapies centered on an mRNA degradation strategy have been developed for various diseases, and one such therapy is currently in phase 2 of clinical trials testing for the treatment of rhodopsin-mediated autosomal-dominant retinitis pigmentosa (RHO-adRP) (NCT04123626).

Other ASO moieties can facilitate skipping of pathological exon or pseudoexon containing stop codon mutations. By avoiding these harmful segments, an appropriate reading frame can be reestablished, and the production of functional protein with the required domains can be restored. ProQR Therapeutics is employing this strategy in the treatment of LCA caused by an underlying c.2991 + 1655 > G intronic mutation in the CEP290 gene and in the treatment of Usher syndrome type 2 with underlying exon 13 mutation.

Short-Hairpin RNAs

Short-hairpin RNAs (shRNAs) are small RNA molecules that initiate the intracellular RNA interference (RNAi) pathway. Specifically, the RNase III Dicer complex is recruited to process the shRNAs into small-interfering RNAs (siRNAs). These RNA molecules can then hybridize to complementary mRNA molecules and recruit the RNA-induced silencing complex (RISC) for mRNA degradation. Importantly, while siRNAs can be delivered as discrete RNA molecules, shRNAs require delivery via a lentiviral DNA vector (Lambeth and Smith 2013).

Cas13 and ADARs

Recent advances in CRISPR/Cas therapies have yielded unique Cas proteins with the capacity to precisely target and degrade RNA transcripts. Abudayyeh et al. isolated an RNA-guided, RNA endonuclease from the Leptotrichia shahii bacterium (Abudayyeh et al. 2017; Cabral et al. 2017; Sengillo et al. 2017). Previously termed C2c2, Cas13a relies on an sgRNA to localize to and target mRNA transcripts for degradation, achieving the same efficacy as previous RNAi approaches but with improved specificity and off-target effects.

While the aforementioned RNA-based therapies center on transcript degradation, recent advances in Cas13 systems have yielded the possibility of base editing at the mRNA level. These approaches employ a catalytically inactive dCas13 and an adenosine deaminase acting on RNA (ADAR). dCas13 guides the ADAR to the desired editing site to facilitate A·I trans conversions. As of now, however, ADARs are only able to facilitate A·I conversions and therefore can only correct A > G mutations.

DISEASE-SPECIFIC ADVANCEMENTS

RPE65-Associated LCA

The RPE65 gene encodes a crucial retinoid isomerase that regenerates 11-cis-retinal for rod and cone photoreceptors in the classical RPE visual cycle. Mutations in the RPE65 gene among several other genes have been extensively linked to LCA, a recessively inherited retinal disease characterized by severe vision impairment, nystagmus, and diminished ERG responses by the first year of life (Gu et al. 1997; Perrault et al. 1999).

In 2017, Voretigene neparvovec-rzyl (Luxturna) was approved by the U.S. Food and Drug Administration (FDA) as the first gene-replacement therapy for the treatment of RPE-associated LCA. Luxturna is an AAV serotype 2 (AAV2) vector that contains a WT copy of the RPE65 gene and is delivered via subretinal injection to restore RPE65 production by RPE cells (Bainbridge et al. 2008). The AAV2 vector is expressed intracellularly as a stable episome in nonmitotic RPE cells. Thus, the therapy could conceivably serve as a one-time treatment solution for patients with RPE-associated LCA. In clinical trials, Luxturna-treated patients demonstrated dramatic improvements in bilateral multi-luminance mobility testing (MLMT) compared to control patients following 1 year of treatment. Moreover, patients also reported improvements in visual field testing, best corrected visual acuity (BCVA), and overall navigational ability in both light and dark conditions (Maguire et al. 2019). These exciting clinical developments in the landmark case of Luxturna provide compelling support for future investigation and the use of gene therapy strategies in the treatment of retinal diseases.

CEP290-Associated LCA

The CEP290 gene encodes a critical regulator protein that directs ciliary trafficking and synthesis. These processes are vital to the function of the photoreceptor outer segment, which is dependent on effective transport of proteins and lipids synthesized in the photoreceptor inner segment. Mutations in CEP290 lead to a form of LCA, LCA10, which is highlighted by nystagmus, compromised ERG response, and severe vision loss by the first year of life (Perrault et al. 2007). Mutations in the CEP290 primarily exhibit an autosomal-recessive pattern of inheritance and have been implicated in ∼30% of all LCA patients. The c.2991 + 1655A > G mutation in the intervening sequence in intron 26 (IVS26) represents the most frequent underlying mutation for LCA10 and produces an mRNA with a premature stop codon (p.Cys998*). Subsequently, both a nonfunctional truncated CEP290 protein and a full length-WT CEP290 protein are produced, resulting in haploinsufficiency (Ruan et al. 2017).

Gene augmentation approaches in the treatment of CEP290-mediated LCA have been hindered by the size of the CEP290 gene (∼8 kb), which is not able to be accommodated by the size restrictions of the AAV vector. However, one group circumvented this size limitation by developing a truncated, yet fully functional version of the gene, termed miniCEP290. The truncated gene was successfully packaged and delivered by an AAV vector with demonstrable improvement in photoreceptor viability in a murine model (Zhang et al. 2018).

In the realm of gene editing, Maeder et al. (2019) developed an AAV5 vector containing a Staphylococcus aureus Cas9 (saCas9) and two sgRNAs flanking the IVS26 mutation in the CEP290 gene. This construct, termed EDIT-101, removes the IVS26-containing segment in the CEP290 gene to restore mRNA splicing and functional CEP290 protein expression. The group demonstrated that EDIT-101 was able to recover CEP290 mRNA splicing and protein function in an ex vivo, postmortem human retinal model with a productive edit rate of ∼17%. These results were recapitulated in HuCEP290 mice, a human CEP290 IVS26 knock-in mouse model. Because mouse retina harbors lower proportions of cones (∼3%), the group further evaluated the efficacy of EDIT-101 in a cynomolgus monkey nonhuman primate (NHP) model, which better recapitulates the human eye (Carter-Dawson and LaVail 1979; Curcio et al. 1987). In this NHP model, administration of EDIT-101 via subretinal injection in the perifoveal region achieved a maximum 28% productive edit rate with a vector concentration of 1 × 10¹² vgml⁻¹.

In 2019, Editas Medicine spearheaded the phase 1/2 Brilliance clinical trial (NCT03872479) to assess the safety, tolerability, and efficacy of EDIT-101 (Fig. 2) in LCA patients aged 3 and older with at least one mutation in the CEP290 gene (including c.2991 + 1655A > G) and a visual acuity no better than 0.4 LogMAR (NCT03872479). Primary outcome measures include drug toxicity and adverse effects assessed every 3 months for up to 1 year. Secondary measures including BCVA, color vision, full-field stimulus testing (FST), microperimetry, and quality of life scores, among other metrics. The efficacy component of the study consists of five subjects, three of which received a single dose of EDIT-101 at the moderate-level dose and two at the low-level dose (1.1 × 10¹²vg/mL and 6 × 10¹¹vg/mL, respectively). Unfortunately, initial results have been inconclusive with only two patients demonstrating improvement in outcome measures at 6 months following treatment. Furthermore, only one patient achieved clinically meaningful improvement, an increase of 0.7logMAR in BCVA. Reported adverse effects include retinal tears and hemorrhages.

Figure 2.

Mechanisms of two therapies for CEP290. (A) Mutations in CEP290 cause incorrect splicing, which results in a truncated CEP290 protein. The CEP290 protein is a crucial component of protein trafficking in cilia and results in vision loss due to the malfunction of the outer segment of the photoreceptors. (B) QR-110 is an RNA therapeutic designed by ProQR, which binds to the mutated intronic sequence in CEP290 to prevent improper splicing and allow for correct protein formation. (C) EDIT-101 is a gene-editing therapy developed by Editas Medicine, which removes the disease-causing mutation, allowing for correct splicing and full-length CEP290 formation.

Toward an RNA-editing approach, Dulla et al. (2018) developed an optimized ASO capable of restoring normal RNA splicing in LCA10 with an underlying CEP290 c.2991 + 1655A > G mutation. Also known as QR-110 and sepofarsen, this construct does not require AAV packaging but can be delivered directly via intravitreal injection (IVT). In preclinical studies, QR-110 treatment successfully corrected CEP290 c.2991 + 1655A > G mutations in human retinal organoids and demonstrated robust bioavailability in WT mice and rabbit retinas and excellent tolerability in an NHP model.

ProQR Therapeutics is currently conducting the phase 2/3 Illuminate clinical trial (NCT03913143) to assess the safety, tolerability, and efficacy of sepofarsen in the treatment of LCA10 caused by underlying CEP290 p.Cys998X mutation (Fig. 2). Thirty-six patients aged 8 and older received one treatment dose of sepofarsen at 0 mo and 3 mo and every 6 mo thereafter, up to 24 mo. The primary outcome measure is the mean change in BCVA following 12 mo of treatment, and secondary outcome measures include navigational ability, FST, ERG, ocular adverse events, and changes in photoreceptor outer/inner layer, among others. While an initial case study reported sustained visual improvement in one enrolled patient 15 mo following treatment, the Illuminate trial ultimately failed to meet primary outcome measure (Cideciyan et al. 2019). In addition, differences in secondary outcome measures between the treatment and control groups were not statistically significant. Although sepofarsen was overall well-tolerated, adverse events included retinal thinning and cystoid macular edema (CME), which were consistent with findings from phase 1/2 of the trial.

RHO-adRP

Retinitis pigmentosa (RP) is a group of rare inherited retinal disorders initiated by rod photoreceptor degeneration with progression to a wide range of disease phenotypes from mild night blindness to severe reductions in visual acuity (Verbakel et al. 2018). In the early stages of RP, patients report diminishing night vision that then progresses to deterioration of peripheral vision, termed “tunnel vision.” In later stages, significant rod photoreceptor loss subsequently leads to secondary cone photoreceptor death, which may compromise daytime visual acuity and result in complete blindness (Narayan et al. 2016).

RP can be precipitated by a compendium of defects in the photoreceptors and RPE. In particular, the rhodopsin (RHO) gene encodes a G-protein coupled receptor that serves as the primary visual pigment for rod photoreceptors (Nathans et al. 1986). Mutations in the RHO gene exhibit an autosomal-dominant hereditary pattern of retinitis pigmentosa and were the first to be associated with the development of RP. Rhodopsin-mediated autosomal-dominant retinitis pigmentosa (RHO-adRP) accounts for almost 25% of all autosomal-dominant retinitis pigmentosa (adRP) cases (Jacobson et al. 1991).

Conventional gene therapy approaches for the treatment of RHO-adRP centered on attenuating expression of the mutant RHO gene, which is directly implicated in the retinal degeneration process. QR-1123 is an ASO therapy designed to treat RHO-adRP with underlying P23H mutation in the RHO gene by targeting P23H mutant RHO transcripts for degradation (Murray et al. 2015). In preclinical studies, intravitreal injection of QR-1123 effectively attenuated P23H mutant RHO expression while preserving WT RHO expression, restored ERG response, and augmented outer nuclear layer (ONL) thickness in a P23H mutant murine model. In 2018, ProQR Therapeutics acquired QR-1123 from Ionis Pharmaceuticals and initiated a phase 1/2 Aurora clinical trial (NCT04123626) to assess the safety, tolerability, and efficacy of QR-1123 in the treatment of patients with P23H mutated RHO-adRP. Phase 1 results indicate that QR-1123 is well tolerated with an excellent overall safety profile.

One critical limitation for therapies such as QR-1123 is that they are not scalable to the over 150 reported mutations in the RHO gene that have been shown to cause RHO-adRP. Thus, a more conventional paradigm involves eliminating expression of both the patient WT and mutant RHO genes followed by gene replacement with an exogenous WT RHO gene (Meng et al. 2020). Such a strategy can be expanded to treat all forms of RHO-adRP without designing individually tailored therapies for each underlying mutation.

Cideciyan et al. (2018) developed a single AAV system containing both a WT copy of the human RHO gene and a component encoding an shRNA that targets the human RHO gene. Importantly, the group engineered the codon-modified RHO gene to be resistant to degradation by the vector-carrying shRNA (Gorbatyuk et al. 2007; O'Reilly et al. 2007; Palfi et al. 2012).

A single subretinal injection of the shRNA₈₂₀ candidate in a canine model for light-induced RHO-adRP degeneration model dramatically attenuated RHO mRNA and protein expression in mutant RHO retinas. Next, the group delivered the AAV system containing the shRNA₈₂₀ and the codon-modified human RHO gene and showed that the rescue RHO gene could be expressed at maximum 132% and 33%, at the mRNA and protein level, respectively, in comparison to control retinas. The group also demonstrated that vector administration could preserve retinal ONL thickness, photoreceptor cells, and the outer segment layers in treated retinas (Cideciyan et al. 2018). In 2018, IVERIC Bio acquired the rights to the therapy, now named IC-100, with plans to initiate a phase 1/2 clinical trial. However, subsequent preclinical studies in NHPs were unable to recapitulate the safety profiles previously reported in canine models and have stalled clinical trial proposals. IVERIC Bio is seeking potential collaborations as it turns its attention to other products in its portfolio. Natural history studies and IND-enabling activities for IC-100 are ongoing (Meng et al. 2020).

Using a CRISPR/Cas9-based approach in the treatment of RHO-adRP, Editas Medicine developed a dual AAV5 system to knock out and replace the mutant RHO gene. The dual system consists of a vector encoding an S. aureus Cas9 (SaCas) cRNA and a second encoding several gRNAs and a codon-optimized copy of the RHO cRNA. Widespread frameshift deletions generated by this dual system dramatically diminished RHO expression in human retinal explants, and a rescue RHO vector was sufficient to restore RHO expression similar to physiological levels (Diner et al. 2020).

USH2A

Biallelic mutations in the USH2A gene result in Usher syndrome type 2, an autosomal-recessive syndrome characterized by congenital hearing deficits and late-onset retinitis pigmentosa with late-onset, or nonsyndromic retinitis pigmentosa (nsRP) (van Wijk et al. 2004). USH2A encodes usherin, a transmembrane protein that is critical for photoreceptor outer layer and periciliary region structure (Géléoc and El-Amraoui 2020). ∼35% of USH2A mutations occur in exon 13, which can cause premature termination and inappropriate splicing resulting in dysfunctional usherin protein. Like CEP290-mediated LCA, Usher syndrome type 2 has been difficult to approach using AAV-based gene augmentation strategies due to the size of the USH2A gene.

Dulla et al. developed the novel ASO QR-421a, which promotes exon 13 skipping in the mutant USH2A gene, thereby restoring normal pre-mRNA splicing (Dulla et al. 2021). QR-421a treatment selectively induced exon 13 skipping in an induced pluripotent stem cell model derived from the photoreceptors of a patient with homozygous c.2299delG mutation in the USH2A gene. In addition, the group showed that QR-421a had high bioavailability in the photoreceptor outer layer of WT mice. Based on these promising results, ProQR Therapeutics initiated phase 1/2 Stellar clinical trial (NCT03780257) to assess the safety, tolerability, and efficacy of QR-421a (also termed ultevursen) in patients with exon 13 mutations in the USH2A gene. Results from phase 1/2 have been moderately positive. Ultevursen is well tolerated and is without significant adverse effects, and the 16 subjects that received a single dose of the therapy reported a mean 6.0 letter improvement in BCVA. ProQR has since initiated two phase 2/3 clinical trials, Sirius (NCT05158296) and Celeste (NCT05176717), with results expected in 2024.

Currently, Locanabio is developing a Cas13d-centered RNA-editing approach to promote exon 13 skipping in the USH2A gene. In preliminary studies, a single AAV vector containing a catalytically inactive Cas13d (dCas13d) and a gRNA targeting a portion of exon 13 augmented the proportion of truncated usherin protein with exon 13 exclusion in HEK293T cells (Lardelli et al. 2021).

Spinocerebellar Ataxia

Polyglutamine spinocerebellar ataxias (SCAs) represent a heterogeneous group of six autosomal-dominant, neurodegenerative diseases characterized by a constellation of symptoms including ataxia, speech difficulties, impaired gait, and motor dysfunction (Duenas et al. 2006). The six disease groups, SCA1, SCA2, SCA3, SCA6, SCA7, and SCA17, are defined by CAG (glutamate) repeat expansions in the coding region of their associated genes, ATXN1, ATXN2, ATXN3, CACNA1A, ATXN7, and TBP, respectively (Paulson et al. 2017; Buijsen et al. 2019).

In particular, SCA7 distinctively presents with pigmentary regional dystrophy and progressive visual loss beginning in the macular region and spreading to the retinal periphery (Lebre and Brice 2003). These pathological changes can be clinically monitored via optical coherence tomography (OCT), ERG, and full-field ERG (ffERG), which demonstrate retinal thinning of the retinal nerve fiber layer, diminished amplitudes, and diminished 30 Hz cone flicker amplitudes, respectively (Campos-Romo et al. 2018; Park et al. 2020).

Niu et al. (2018) successfully developed an ASO against Ataxin-7, and delivery of the ASO in a SCA7 266Q knockin mouse model via intravitreal injection dramatically attenuated Ataxin-7 expression. In addition, mice receiving the ASO therapy recovered rod and cone function. Ramachandran developed short double-stranded RNA (dsRNA) molecules to target repeat expansion of SCA7 (Ramachandran et al. 2014). In preclinical studies, this RNAi strategy significantly reduced WT and mutant Ataxin-7 expression, improved disease phenotype in a murine model of SCA7, and demonstrated minimal toxic effects.

CONCLUSION

The past decade has seen an explosion in gene therapy and editing techniques with major improvements to existing technologies. While Luxturna remains the only FDA-approved AAV2 therapy for the treatment of RPE65-mediated LCA, editing tools including CRISPR/Cas9 and its emerging novel variants have expanded the ways that we may approach the treatment of IRDs. These technologies have opened the possibility for researchers to treat IRDs with varying inheritance patterns and disease pathologies, IRDs with underlying mutations in previously inoperable genes due to size constraints, and IRDs with aberrations at the DNA and/or RNA levels. Meanwhile, existing RNA-editing techniques such as RNAi continue to be refined and optimized, and several trials assessing the efficacy of ASOs and RNA-based therapies have since been initiated with promising results. While the majority of clinical trials have failed to translate preclinical findings in patients and the recent results from ProQR's Brilliance trial for the treatment of CEP290-mediated LCA represents one such trial, gene therapy and editing techniques are still in the early stages of development. Undoubtedly, gene therapy and editing will continue to encounter challenges and several rounds of refinement in clinical trials before it is shown to be safe and effective in the treatment of many IRDs. Simultaneously, gene therapy has garnered overwhelming public support and financial backing. Cas14a-Scribe Therapeutics recently closed a 100-million-dollar series B round, and other top gene therapy companies carry robust market caps. There are more gene therapy–based clinical trials than ever before with more than 20 recruiting, enrolling by invitation, and active (not recruiting) trials designated on Clinicaltrials.gov as of July 2022. Finally, the success story of Luxturna and the profound impact that it has had on patients with RPE65-mediated LCA indicate that gene therapy and editing are viable strategies in the treatment of IRDs and will inevitably become mainstays in the way we manage patient IRDs in the future.

COMPETING INTEREST STATEMENT

S.H.T. receives financial support from Abeona Therapeutics and Emendo. He is also the founder of Rejuvitas and is on the scientific and clinical advisory board for Nanoscope Therapeutics and Medical Excellence Capital.