PRPF31-retinitis pigmentosa: Challenges and opportunities for clinical translation

Abstract

Mutations in pre-mRNA processing factor 31 cause autosomal dominant retinitis pigmentosa (PRPF31-RP), for which there is currently no efficient treatment, making this disease a prime target for the development of novel therapeutic strategies. PRPF31-RP exhibits incomplete penetrance due to haploinsufficiency, in which reduced levels of gene expression from the mutated allele result in disease. A variety of model systems have been used in the investigation of disease etiology and therapy development. In this review, we discuss recent advances in both in vivo and in vitro model systems, evaluating their advantages and limitations in the context of therapy development for PRPF31-RP. Additionally, we describe the latest approaches for treatment, including AAV-mediated gene augmentation, genome editing, and late-stage therapies such as optogenetics, cell transplantation, and retinal prostheses.

Introduction

Photoreceptor degeneration is the ultimate cause of visual loss in many retinal disorders, including inherited retinal dystrophies (IRDs). IRDs comprise a heterogeneous group of diseases with wide genetic, allelic, and phenotypic variation. More than 250 IRD genes have been identified^1,2. Retinitis Pigmentosa (RP, OMIM 268000), first described by Donders in 1855³, is the most common cause of irreversible inherited visual dysfunction. RP affects approximately 1 in 3,000 to 5,000 people, with identified mutations in over 70 genes ^2,4,5. Night blindness and loss of peripheral vision, due to rod photoreceptor degeneration, are common early symptoms of RP. Over time, vision loss ultimately progresses to the central field. The advancement of rod dystrophy, together with cone photoreceptor loss several years or decades later, leads to the accumulation of intraretinal pigment deposits and retinal vasculature attenuation ^6-10.

Approximately 30-40% of RP patients are diagnosed with autosomal dominant RP (adRP), with mutations in the photopigment rhodopsin (RHO) being the most common cause. Mutation in the precursor mRNA (pre-mRNA) splicing factor 31 gene (PRPF31, RP11, OMIM 606419) has been described as the second most common adRP gene mutation in most populations ^8,11-15. PRPF31-RP therefore represents a major unmet clinical need and a prime target for therapeutic intervention¹⁶. In this manuscript, we review recent advances in the development of treatments for PRPF31-RP, including current understanding of disease mechanisms, the development of model systems in which to create treatments, as well as therapeutic approaches that show promise for treating PRPF31-RP.

PRPF31 function

Pre-mRNA undergoes crucial splicing steps which remove intronic regions and join together coding exons, before being transformed into mature mRNA. The splicing of pre-mRNA occurs almost exclusively in the nucleus, in an RNA/protein complex called the spliceosome. The spliceosome is composed of five small nuclear RNAs (snRNAs) called U1, U2, U4, U5, U6, and more than 200 protein factors^6,17-21. The assembly of snRNAs and protein factors results in RNA-protein complexes (small nuclear ribonucleoproteins, or snRPs). snRP complexes then combine with unmodified pre-mRNA and various other proteins to form a spliceosome. Spliceosome assembly occurs on the pre-mRNA at each exon-intron junction (Figure 1). Following assembly, the spliceosome then catalyzes the removal of introns and ligation of flanking exons.

Figure 1. Simplified diagram of PRPF31 in the assembly of the spliceosome.

A) U1snRNP recognizes and binds to the splice donor site of an intron, and U2snRNP associates with the branch site, forming the pre-spliceosome complex A. Meanwhile, the U4/U6.U5 tri-snRNP, which contains PRPF31, forms. B) The U4/U6.U5 tri-snRNP is then recruited to complex A, forming the pre-catalytic spliceosome complex B. C) Dissociation of U1snRNP and U4snRNP from the precatalytic B complex activates the spliceosome, resulting in the catalytic B complex. Introns are subsequently removed through two transesterification reactions.

PRP31, the Saccharomyces cerevisiae homologue of PRPF31, was first identified in 1996 and was subsequently shown to be essential for cell viability through the formation of the U4/U6.U5 tri-snRNP. Deletion of PRP31 inhibits the tri-snRNP assembly and blocks pre-mRNA splicing^12,22. During each round of splicing, the U4/U6.U5 tri-snRNP complex is assembled through the interaction of the U4/U6 di-snRNP with U5 snRNP, a process mediated by splicing factors PRPF31 and PRPF6 ^23-34. PRPF31 is also involved in additional cellular pathways beyond mRNA splicing, with a direct implication in photoreceptor ciliogenesis, cellular adhesion and severe RPE defects ³⁵. Numerous U4/U6.U5 tri-snRNP-associated PRPF genes have been linked to RP, including PRPF3 (RP18, MIM 607301), PRPF4 (RP70, MIM 615922), PRPF6 (RP60, MIM 613979), PRPF8 (RP13, MIM 607300), PRPF31 (RP11, MIM 606419), and SNRNP200 (RP33, MIM 601664). PAP-1 (RP9, MIM 607331) is a non-snRNP splicing factor whose loss of function also leads to onset of adRP³⁶.

Mutations in any of these ubiquitously expressed PRPFs may affect spliceosome assembly. Although PRPFs are ubiquitously expressed, surprisingly, mutations in PRPF3, PRPF8, PRPF31, PAP-1, and SNRNP200 only cause retina-specific cellular decay, for reasons that are not yet fully understood^25-28,30,37. Examination of snRNA expression levels in a panel of five human tissues (brain, retina, heart, testis and skeletal muscles) revealed that the retina expresses the highest amount of the spliceosomal snRNAs (7 times more), and the highest volume of processed pre-mRNA within the whole body ³⁸. Also, the retina is a highly active tissue with a high metabolic rate exceeding even that of the brain ^39-41. This may partially explain retinal sensitivity to PRPF deficiency.

PRPF31 splicing and retinal disease

Located on human chromosome 19q13.4, pathogenic variants of PRPF31 are a major cause of adRP, and the second most common genetic cause of non-syndromic adRP in most populations. PRPF31 was first identified as the gene underlying RP11 in 2001²⁵. PRPF31-RP accounts for 5.5-8.9% of cases in the US ^42,43, 7.6-8.1% of adRP cases in Spanish populations ^44,45, 8% of adRP cases of French Canadian populations ⁴⁶, 14.1% of adRP cases in Japanese populations ⁴⁷, 10.5% of adRP cases in Belgian families ⁴⁸, 6.7% adRP cases in French populations ⁴⁹, and 9.4-11.1% of Chinese cases ^50-52. Currently, there are more than 130 presumed causative mutations in PRPF31-RP reported to date, leading to alteration of PRPF31 expression, including frameshift, splice site mutations, nonsense mutations, missense mutations, small insertions, small deletions, small indels, and/or large-scale insertions or deletions ¹¹. Reduction of PRPF31 protein expression leads to a reduction in the rate at which spliceosome activation occurs. The development of RP11 is thought to be a result of haploinsufficiency, where a single mutant copy of a gene causes disease by reducing the amount of gene product produced below necessary levels ^53-56.

PRPF31-RP may also result from dominant negative mutations, in which production of a mutant protein interferes with function of the normal protein, for example through the formation of complexes between normal and abnormal protein that disrupt RNA splicing. This loss of PRPF31 function can lead to the accumulation of abnormally spliced RNA and the death of cells in the retina, including photoreceptors and RPE cells, ultimately resulting in loss of vision³³.

Interestingly, some PRPF31 mutation carriers develop RP, while others remain healthy their entire life. This "all or none” pattern of incomplete penetrance has been observed in the majority of families carrying PRPF31 mutations, and is thought to be a result of varying levels of PRPF31 expressed by the unaffected allele. Expression of the PRPF31 allele varies by around 5-fold, following a continuous distribution⁵⁷. A heterozygous carrier of a PRPF31 mutation may be asymptomatic if the wt allele produces sufficient levels of PRPF31, or may express the disease if PRPF31 levels fall beneath a critical level. Cis and trans genetic factors may play a role in modulating PRPF31 mRNA expression levels. The MSR1 minisatellite repeat element is located upstream of the PRPF31 promoter with either three or four copies present. It has been suggested that alleles with 4 copies are related to non-penetrance^58,59. CNOT3 is another modifier gene that is located downstream of PRPF31, and it has been proposed that high CNOT3 gene expression suppresses PRPF31 expression, although studies on CNOT3 have been inconclusive⁶⁰. Despite the autosomal dominant inheritance of the disease, it appears to skip generations^53,61. As a result of differential expression levels of the wild-type allele inherited from the non-RP parent, asymptomatic carriers for PRPF31 mutations are common within affected families⁵³.

Typical symptoms of PRPF31 in patients include night blindness (nyctalopia), with on onset of symptoms in childhood or adolescence, which is followed by a gradual narrowing of the visual field, and ultimately a loss of visual acuity in adulthood^6,7,45,62. Retinal imaging reveals a characteristic pattern of peripheral hyperpigmentation known as "bone-spicule pigmentation," optic disk pallor, and thinning of retinal blood vessels ⁶³. Optical coherence tomography (OCT) shows disruptions in the retinal structure, specifically a loss of the junction between the inner and outer segments of the retina (IS/OS), followed by thinning of the outer nuclear layer (ONL) ^64,65. Electroretinogram (ERG) recordings often show a reduction in the amplitude of the a- and b-waves, indicating increasing loss of retinal function as the disease progresses ^49,66. Additionally, PRPF31-RP is characterized by primary degeneration of rods and secondary loss of cones, which is evident through histopathological examination. Changes in the retinal pigmented epithelium (RPE) and activation of retinal glia are also commonly observed ³³.

Models of PRPF31-RP

Appropriate disease models of PRPF31-RP are critical to the investigation of disease mechanisms, and for the development of therapies that will translate efficiently into patients. Researchers have successfully used nonvertebrate, vertebrate, mammalian, and in vitro models to investigate PRPF31-RP, including yeast, zebrafish, drosophila, rodents, human cells, induced pluripotent stem cells (iPSC), and retinal organoids. The following sections describe the characteristics and the advantages and disadvantages of each model, along with the outcomes from research carried out in each of these models (Figure 2).

Figure 2. Models used in translational PRPF31 research.

Researchers have successfully used nonvertebrate, vertebrate, mammalian, and in vitro models to investigate PRPF31-RP, including yeast, zebrafish, drosophila, rodents, human cells, induced pluripotent stem cells, and retinal organoids. The individual characteristics, advantages and disadvantages of each model, along with the outcomes from research carried out in each of these systems, are summarized in the text.

Yeast

Yeast is a simple model, which rapidly proliferates, and it is easy to culture and manipulate in the laboratory. It also has a relatively small genome size. Yeast has been intensely studied, making it a key model organism for studying eukaryotic cellular processes ⁶⁷. Over 30 different proteins are essential for mRNA splicing in Saccharomyces cerevisiae. In 1996, the PRPF31 ortholog PRP31 was identified by temperature-sensitive screening of splicing defective yeast strains^68,69. PRP31 mutant yeast strains are defective in splicing in vitro and in vivo indicating that Prp31 directly contributes to the splicing pathway ^12,22.

However, yeast is evolutionarily distant from humans and significantly different from humans in key aspects. Yeast are simple cells and thus are not able to model complex tissues like the retina. In addition, yeast PRP31 has only 20% sequence similarity to human PRPF31 with significant differences in protein structure, including variations in post-translational modifications and protein-protein interactions, which may affect the function and concentration of certain key proteins in the cell ^22,25,70. Thus, while yeast is a useful model for understanding the basic function of Prp31 in pre-mRNA splicing, it may not accurately reflect the pathogenesis of PRPF31-RP in humans.

Drosophila

Drosophila is widely used as a model organism for studying gene function and human disease due to its rapid breeding, genetic manipulability, and low maintenance cost. The Drosophila melanogaster homolog Prp31 encodes a protein that is 57% similar to human PRPF31, and it has been found to play a crucial role in mRNA splicing and mitosis ⁷¹. Reduction of endogenous Prp31 by RNA interference (RNAi)-mediated knockdown in fly eyes led to morphological defects and photoreceptor degeneration, with signs of neurodegeneration at the subcellular level⁷¹. In another study, heterozygous Prp31 mutant flies were more susceptible to light damage, resulting in light-induced retinal degeneration similar to the phenotype observed in RP11 patients⁷². Furthermore, studies have shown that Prp31 also plays a role in mitosis, independent of the mRNA splicing process⁷³.

While flies have been a critical model in understanding mRNA splicing and the role of PRPF31 in retinal health and disease, they are a non-mammalian system with eyes that are markedly different from human eyes in terms of their structure, anatomy, and physiology, and the Prpf31-RP fly model may not accurately reflect the clinical features and progression of the disease in humans^71,72,74,75. Flies are insects with compound eyes and ommatodia, while humans are mammals with rod and cone photoreceptors. Many aspects of vision and the phototransduction cascade differ between these species. Flies lack an immune system similar to humans, making it difficult to accurately model important factors such as immune response in the context of gene therapies. Moreover, widely used viruses such as AAVs do not infect flies. Therefore, accurately modeling important factors for the success of novel gene therapies cannot be achieved in flies.

Zebrafish

Zebrafish are highly valuable for retinal research due to their short reproductive cycles, the availability of transgenic lines, and their transparent bodies, which facilitate easy visualization. Additionally, zebrafish share a high degree of genetic similarity with humans. Thus, zebrafish have been used in a number of studies investigating the pathogenesis of PRPF31-RP.

Linder et al. silenced prpf31 in zebrafish using an antisense morpholino, causing severe embryonic malformations (curled body axis, gross head misdevelopment, cardiac edema) and death at the embryonic stage. While eye development was generally not affected, sublethal administration of morpholinos resulted in defective photoreceptor outer segments. Sublethal prpf31 morphants also had weakened optokinetic nystagmus and impaired visual function. Gene expression analysis showed that many down-regulated genes were related to RP, while up-regulated genes were involved in splicing. The study thus suggests that prpf31 knockdown affects retinal photoreceptor cells by disrupting tri-snRNP function ⁷⁶. Yin et al. injected mutant prpf31 cDNA into zebrafish. Two reported pathogenic mutations, SP117 and AD5, were studied. Injection of SP117 resulted in mislocalized protein, but not photoreceptor degeneration, while AD5 injection did result in rod degeneration and apoptosis, suggesting that SP117 may cause disease through haploinsufficiency while AD5 may play a dominant negative role in the pathogenesis of RP ⁷⁷. In another study, Li et al. used CRISPR/Cas9 to knock out prpf31 in zebrafish. They observed similar morphological abnormalities as the Linder et al. study, including microphthalmia, smaller heads and curved bodies in prpf31+/− embryos that could be rescued by wild-type human PRPF31 but not mutant forms. Splicing deficiency resulted in exon skipping and abnormal expression of genes involved in spindle formation, DNA repair and homologous recombination repair, resulting in abnormal mitosis and the accumulation of DNA damage in retinal progenitor cells at an early embryonic stage, as well as down-regulation of genes involved in retinal differentiation and up-regulation of genes involved in RNA splicing, mRNA processing and gene expression. In contrast to other studies, no mis-splicing of phototransduction genes was observed. The authors noted that genes involved in spindle formation also play a role in ciliogenesis, which suggests that prpf31 deficiency may affect cilia formation as well⁷⁸.

Key differences between zebrafish and humans prevent the use of zebrafish in the development of therapeutic approaches for PRPF31-RP. For example, zebrafish are capable of regenerating photoreceptors, while humans are not. Additionally, zebrafish lack critical eye structures, such as the macula, which is responsible for central high-acuity vision in humans. And AAV vectors, which are the leading vector for retinal gene transfer, do not infect zebrafish.

Mice and primary culture of mouse retinal cells

Mice have been widely used to study the pathogenic mechanisms of PRPF31-RP due to their cost-effectiveness and well-studied genome, as well as the availability of transgenic strains. And, a number of studies have used primary cultured mouse retinal cells to investigate PRPF31 in vitro.

In 2005, Yuan and colleagues used immunoprecipitation and microarray chip technologies to identify potential target genes of PRPF31. Among the 146 genes that co-precipitated with PRPF31, a considerable number were specific to the retina. Mouse primary retinal neuronal cultures were established and transfected with wild-type or mutant PRPF31. The transfection of PRPF31 mutants resulted in a significant decrease in RHO splicing, reduction of rhodopsin expression and rod apoptosis⁷⁹.

Bujakowska et al. established both a Prpf31 knock-in mouse model carrying the p.A216P mutation and a Prpf31^−/− knock-out mouse using Cre-Lox recombination. Homozygous Prpf31^−/− and Prpf31^A216P/A216P genotypes were found to be embryonically lethal, while heterozygous Prpf31^+/− and Prpf31^+/A216P mice were monitored until 12 and 18 months of age respectively and did not exhibit any histological or functional evidence of any retinal phenotype. No evidence of ERG changes or retinal abnormalities was observed in either of the models. Based on these findings, the researchers concluded that the PRPF31^A216P mutation did not have a dominant negative effect, and that one wild-type allele of PRPF31 was sufficient to support the health of the retina in this strain of mice ⁸⁰.

A study conducted by Graziotto et al. in 2011 revealed that Prpf31^+/− mice (knockout mice) displayed RPE cell degeneration at 12 months of age, characterized by the loss of the basal infoldings, accumulation of amorphous deposits, and cell vacuolization between the RPE and Bruch’s membrane. This finding implies that the photoreceptor degeneration typically observed in PRP31-RP may be a secondary result of pathological changes in pigment epithelial cells⁸¹.

The maintenance of normal retinal function relies on the regular elimination of shed photoreceptor outer segments by RPE⁸². In 2014, Farkas et al. reported defective phagocytosis in primary RPE cultures from Prpf31^+/− mice, causing disrupted diurnal rhythmicity and mis-localization of phagocytic receptors⁸³. Knockdown of PRPF31 in human RPE cells showed a similar decrease in phagocytosis, but the phagocytic function of peritoneal macrophages was not affected. Similarly to the Graziotte et al. study, this suggests that the Prpf31 mutation specifically affects the phagocytic function of pigment epithelial cells, which may be the first type of retinal cell affected by the mutation.

Valdes-Sanchez et al. found that Prpf31^{+/ A216P} mice had impaired pigment epithelial function, drusen-like deposits in their retina, and phenotypes resembling age-related macular degeneration. The mutant A216P Prpf31 protein attracted the wild-type Prpf31 protein to cytoplasmic aggregates, depleting functional Prpf31 from the nucleus. The researchers identified 1333 differentially expressed genes in the RPE cells of Prpf31^{+/ A216P} mice, including up-regulated Hsp70 co-localized with the mutant Prpf31 polymer in the RPE cytoplasm. This indicated activation of the unfolded protein response. Additionally, 11% of genes in the RPEs of Prpf31^{+/ A216P} mice were differentially spliced, including Prpf31 itself and genes related to inflammation, oxidative stress, retinol metabolism, cilia formation, and apoptosis ⁸⁴.

In all of the mouse models studied, photoreceptor dysfunction was not obvious. Currently, the underlying cause for the degeneration of RPE cells, but not photoreceptors, in Prpf31^+/− mice remains unclear. It has been observed that the expression level of Prpf31 in mouse RPE cells was higher than in neural retina cells ^35,81,83,84, suggesting that RPE cells may have a greater need for Prpf31 and consequently may be more sensitive to a decrease in Prpf31 levels. Moreover, it is possible that the more RPE-specific degeneration seen here may be due to different requirements for PRPF31 between mice and humans, and that the mouse retina may not require as much Prpf31 as human retina^85,86. Additionally, the lifespan of mice may be too short for the development of profound photoreceptor degeneration caused by PRPF31^+/− retinitis pigmentosa⁸⁷.

In 2022, our research group developed mouse models of PRPF31-RP with early-onset morphological and functional impairments similar to those in patients using AAV-mediated CRISPR/Cas9 knockout⁸⁸. We used CRISPR/Cas9 to target the early coding exons of Prpf31 in the retina of postnatal and adult mice delivered using AAV administered via various delivery routes, including intravitreal and subretinal injection. When Prpf31-knockout vectors were delivered intravitreally, the inner retina was primarily affected, leading to reduced thickness and loss of cells of the inner plexiform layer, as well as a decrease in the amplitude of the ERG b-wave. Subretinal injection resulted in retinal pallor and black-brown pigmentation, and in contrast to previously reported transgenic mouse lines, caused the dramatic degeneration of photoreceptors and RPE. ERGs showed significant reduction in the amplitudes of a-, b-, and c-waves, along with prolonged c-wave implicit times. Using the subretinal injection model, we also investigated the efficacy of AAV-Prpf31 gene augmentation therapy using this model. Co-injection of Prpf31 gene augmentation vectors with KO vectors resulted in reduced retinal pallor and pigmentation, preserved retinal structure, and improved a-, b-, and c-wave responses indicating functional improvement of photoreceptor (evaluated by a-, b- waves) and RPE (evaluated by c-waves). These models offer new opportunities for exploring the pathogenesis of PRPF31-RP and developing therapeutic strategies.

Overall, mice models have been crucial to the study of PRPF31-RP, although there are considerable disadvantages to their use. Mice lack a fovea, making the study of high acuity vision unattainable. Additionally, the immune response in mice is differs significantly from that of humans. Because transgenic Prpf31−/+ mice do not exhibit photoreceptor degeneration, their usefulness in investigating photoreceptor-based retinal disease is limited. And, while the murine CRISPR model shows severe photoreceptor degeneration, and it has been useful in showing proof-of-concept for gene augmentation as a treatment for PRPF31-RP, it is not a model of haploinsufficiency.

Human cells, induced pluripotent stem cells, and retinal organoids

Human cells may offer a more accurate model of human biology compared to animals, moreover, utilizing human cells in research circumvents ethical concerns associated with animal experimentation. The use of human cells in research may facilitate the development of personalized medicines, in which therapies can be customized based on the individual patient's genetic makeup and unique disease characteristics.

Working in HeLA cells, researchers discovered a correlation between PRPF31 and the splicing complex⁸⁹. They also demonstrated that PRPF31 plays a role in the segregation of chromosomes during mitosis⁷³. In HEK cells, it was found that a significant number of genes that co-immunoprecipitated with PRPF31 were specific to the retina⁷⁹. Knockdown of PRPF31 caused phagocytosis disruption in the ARPE-19 human RPE cell line⁸³. In lymphocyte lines obtained from peripheral blood of PRPF31+/− RP patients, the efficiency of tri-snRNP assembly was reduced to approximately 40% compared to normal controls³⁸. Nonetheless, cultured cell lines grown in-vitro lack interactions with other cell types, and therefore cannot fully replicate the in-vivo condition. As a result, it is challenging to accurately represent the complex biology of RP patients.

The emergence of induced pluripotent stem cells (iPSCs) and organoid cultures has introduced new methods for modeling human diseases in-vitro. Buskin et al. isolated skin fibroblasts from PRPF31^+/− RP patients, which were differentiated into RPE cells and retinal organoids ³⁵. PRPF31 expression levels were found to be reduced in the PRPF31^+/− RP organoids, and RPE cells were the most severely affected retinal cell type. Impaired splicing was observed in the RPE and photoreceptors of organoids, but not in PRPF31^+/− iPSC or fibroblasts, which aligns with the retina-specific phenotype observed in PRPF31-RP patients. PRPF31^{+/ −} RPEs exhibited various defects, including loss of apico-basal polarity, impairment of the tight epithelial barrier, reduction and shortening of microvilli, defective phagocytosis, and the appearance of large uncharacterized basal deposits under the RPE. Additionally, PRPF31^+/− photoreceptor cells had more apoptotic nuclei and stress vacuoles than normal controls. The incidence and length of cilia were significantly reduced in PRPF31+/− RPEs and photoreceptors, and the splicing of several cilia-related genes (e.g., SF1, SART1/Snu66 and DDX5) was affected. Ciliary deficiency may be a direct consequence of retina-specific spliceosome dysregulation, which in turn, may exacerbate the phenotype resulting from splicing dysfunction. While mutated PRPF31 was found in the RPE, it was absent in other retinal cells. This may indicate that mutated PRPF31 has a dominant negative effect only on RPE, leading to more severe pathological changes compared to haploinsufficiency ³⁵.

In 2019, a study from Brydon et al. demonstrated that AAV-mediated PRPF31 gene therapy had a positive impact on the function and morphology of RPE cells derived from iPSCs with PRPF31 mutations⁹⁰. AAV-mediated PRPF31 gene therapy improved the morphology, phagocytosis, ciliogenesis, barrier function and other key functions of PRPF31+/− iPSC derived retinal pigment epithelial cells, which conceptually verified the protective effect of AAV-PRPF31 gene augmentation therapy.

In 2022, Rodrigues et al. created retinal organoids and RPE cells derived from iPSCs, and showed that reduced levels of PRPF31 were associated with disorganized RPE cells and reduction of RPE phagocytosis. In contrast to transgenic mouse models, these organoids also modeled the human phenotype with the death of rod photoreceptors followed by cone cell death. RPE and photoreceptor cells from asymptomatic patients did not show RP phenotypes, and correction of the pathogenic mutation by CRISPR/Cas9 restored normal morphology, supporting haploinsuffiency as the mechanism underlying retinal disease. Transcriptome profiles revealed differentially expressed and mis-spliced genes involved in cell adhesion, phototransduction and cilia structure. This study also provided proof-of-concept for gene therapies, as a rescue of the defective phenotypes by PRPF31 gene augmentation was successful in both rod and cone cells ⁹¹.

Ex vivo human retinal culture

Ex vivo retinal culture is widely used in the field of retinal research, bridging the gap between in-vitro cell/organoid culture and in-vivo experiments.

Leila et al. effectively reduced expression levels of PRPF31 by transfecting siRNA into human retinal explants. This knockdown led to the reduction of the outer nuclear layer thickness and a decrease in the expression of retina-specific genes related to light conduction, photoreceptor structure, and transcription, without affecting cell activities⁹². Pormehr et al. also used RNA interference to recreate PRPF31 haploinsufficiency in human organotypic retinal cultures, resulting in photoreceptor degeneration. They found that reduction of PRPF31 caused mis-splicing of genes involved in RNA processing (PRPF3, PRPF8, PRPF4, and PRPF19) and phototransduction (RHO, ROM1, FSCN2, GNAT2, and GNAT1)⁹³.

A significant drawback of human ex vivo retinal culture is the limited survival window, as all retinal cell types degenerate in culture over time^94-97. Ex vivo culture also lacks critical anatomical structures such as the vitreous, and cannot model immune response.

Large animal models

The creation of large animal models remain an important goal and unmet need in the study of PRPF31-RP. Non-human primates (NHPs) most closely resemble human retinal structure and function, but they involve substantial financial burden and ethical issues^85,86. NHPs are also far less readily available than mice or other animal models. No primate models of PRPF31-RP currently exist. Pigs and dogs may represent opportunities for the creation of large animal models, although pig or dog models of PRPF31-RP do not currently exist either.

While germline editing may be used to produce a PRPF31^−/+ primate that more precisely models the haploinsufficiency of PRPF31 patients, creating and maintaining such primates would be costly and time-consuming. Alternatively, generating PRPF31 mutant NHP models by intraocular injection of AAV-mediated CRISPR/SaCas9 may be relatively achievable and rapidly completed, allowing for larger-scale analysis. Xi et al. (our group) designed and expressed AAV-CRISPR/Cas9 constructs in rhesus and human ex vivo retinas, and subsequently demonstrated editing of PRPF31 with decreased protein expression, strongly supporting the feasibility of using the PRPF31-KO vectors to establish a rapidly available in-vivo NHP model of PRPF31-RP⁸⁸.

Potential treatment avenues for PRPF31-RP

No cure for PRPF31-RP currently exist. Treatment is limited to low-vision aids, rehabilitation programs, and monitoring of disease progression. However, recent proof-of-concept studies in cells, organoids and mouse models have demonstrated that gene therapy may treat PRPF31-RP by addressing the underlying genetic cause of the disease. Gene therapy approaches include gene augmentation, genome editing, and optogenetics. Later in disease progression, once cells have died, gene therapy is no longer an option, however, cell therapies may hold promise to replace missing retinal cells. And retinal implants may represent a method to restore visual signaling to the brain, bypassing missing photoreceptors (Figure 3).

Figure 3. Potential treatments for PRPF31-RP.

A) AAV-gene augmentation, via subretinal, intravitreal and suprachoroidal injections can deliver a healthy copy of PRPF31 in order to boost protein expression levels and prevent photoreceptor degeneration. B) Genome editing may be used to rewrite the mutation underlying disease or to augment PRPF31 expression levels. C) Late-stage therapies may restore vision following the loss of photoreceptors. These approaches include optogenetics, which aims to create artificial photoreceptors through ectopic expression of light-sensitive proteins in remaining retinal cells, cellular therapies, which may replace missing cells, and retinal implants, which could transmit electrical signals to the brain in the absence of photoreceptors.

Gene augmentation therapy

PRPF31-RP is thought to represent a naturally occurring proof-of-concept for gene augmentation therapy, as the expression of disease is related to levels of PRPF31 expression from the unaffected allele. It follows that by increasing levels of PRPF31, the disease may be effectively treated. Gene augmentation is accomplished through the delivery of a wildtype copy of the PRPF31 coding sequence. The majority of gene therapy trials for ocular diseases have used gene augmentation, including the FDA-approved retinal gene therapy drug Luxturna ⁹⁸.

However, gene augmentation for PRPF31-RP has some potential drawbacks. It is not effective in late-stage cases where photoreceptors have already been lost. Gene augmentation therapy may also only provide a temporary fix for genetic diseases, as the therapeutic gene may not be expressed long-term or may become silenced over time⁹⁹. Targeting photoreceptor cells and RPE, the primary cell types involved in PRPF31-RP, can be challenging. Finally, gene augmentation carries a risk of potential side effects such as off-target expression, inflammation, and toxicity.

Currently, most clinical trials for retinal gene therapy employ adeno-associated viruses as the gene delivery vehicle, as they have been shown to be safe, with relatively low immunogenicity compared to other viruses and tropism for retinal neurons and RPE^100,101. However, AAV vector administration has been associated with inflammation, and many people have neutralizing antibodies against AAVs that may reduce efficiency of gene expression. Nanoparticles and virus-like particles, while currently less commonly used due to lower efficiency, have lower immunogenicity and may allow for vector readministration.

The AAV capsid serotype and promoter play a crucial role in determining expression levels in a given cell type, and efforts are being made to engineer new AAV capsids and promoters to enhance gene therapy efficacy¹⁰². Rational design, directed evolution, and machine learning methods are all being used to create AAV vectors with improved efficiency and specificity. The route of vector administration also plays a critical role in determining the efficiency of gene delivery, as well the cell types infected. Subretinal injections, intravitreal injections, and suprachoroidal injections have all been used to deliver gene therapies to the eye, and each of these approaches has benefits and drawbacks.

Subretinal injections involve administering the viral vector in a subretinal bleb underneath the neurosensory retina. Compared to other delivery routes, subretinal injections are more efficient at targeting cells in the outer retinal layers and can specifically target areas such as the macula^103,104. However, concerns have been raised about the effect of subretinal injection on the thinning of the outer nuclear layer (ONL) following injection in some cases^{105 106}. The formation of a bleb leads to the separation of the photoreceptor and RPE layers, compromising the function and survival of the photoreceptors. This is of particular concern in a degenerating retina, such as in IRDs¹⁰⁷. Additionally, subretinal injection only transduces a limited number of retinal cells in close proximity to the subretinal bleb, which restricts its efficacy in treating pan-retinal diseases.

Intravitreal injection, in contrast, involves injecting a therapeutic gene directly into the vitreous humor of the eye. Intravitreal injections may treat widespread areas of the retina without a retinal bleb or detachment. However, intravitreal injection also presents some challenges. Pre-existing immunity to AAV resulting from past exposure to natural AAV serotypes can produce neutralizing antibodies that decrease transduction efficacy¹⁰⁸. The introduction of a vector to the vitreous body can trigger an innate and adaptive immune response¹⁰⁹. Additionally, the inner limiting membrane and the dense retinal extracellular matrix can prevent the vector from reaching cells in the outer retina⁸⁵.

The suprachoroidal space is potential space located between the scleral wall of the eye and the choroidal vasculature. Suprachoroidal injections have been tested as an alternative delivery approach to target the outer retina via a less invasive injection^110,111. However, following suprachoroidal AAV8 delivery in non-human primates, transgene expression was reported as transient, and delivery to the macula may be difficult in large animal eyes, as the needle is placed at the periphery of the eye.

Genome Editing

Genome editing allows precise modifications to an organism's genome at a specific location. A multitude of genome editing methods now exist, including engineered nucleases such as zinc finger nucleases, transcription activator-like effector nucleases, and the widely used CRISPR-Cas system. CRISPR-Cas systems use guide RNAs to direct a nuclease to a specific location in the DNA, allowing for the creation of a double strand break and the elimination of the mutation ^112-114. In order to replace the mutated sequence with the wildtype sequence, a segment of donor DNA must also be provided, which serves as a template for homologous recombination. Prime editing has recently emerged as a new technology for direct replacement of a specific genetic sequence, without the need for creating double strand breaks. Prime editing uses a catalytically impaired Cas9 endonuclease that nicks genomic DNA fused to a reverse transcriptase, along with a prime editing guide (pegRNA) which both recognizes the genetic location and provides the template DNA to replace the targeted DNA. In prime editing, nicked DNA can be used to initiate reverse transcription of the pegRNA, resulting in incorporation of an edit into one strand of the genome. The cellular mismatch repair mechanism is then used to copy the edit into the complementary strand.

Base editors may also be used to rewrite mutations underlying PRPF31-RP without double strand breaks. These tools can be used for correcting single base pair mutations, which is accomplished through directly converting one base pair into another. Base editors are fusions between catalytically impaired Cas nuclease and a base-modification enzyme, such as a deaminase enzyme. Following gRNA binding, displaced base pairs can be modified by the enzyme, to convert one base pair into another. For example, cytosine base editors convert C-G base pairs into T-A base pairs, and adenine base pairs convert A-T base pairs to G-C base pairs. Cas13 has also been developed as a tool to edit RNA.

CRISPR activation (CRISPRa) is a tool for upregulating gene expression by recruiting transcriptional activators to the target gene. The CRISPRa system consists of a gRNA that targets a specific DNA sequence, as well as a catalytically inactive version of the Cas9 protein (dCas9) fused to transcriptional activators such as VP64 or p65. When the CRISPRa complex binds to the target gene, the transcriptional activators recruit other factors that initiate transcription, leading to increased gene expression ^115,116. CRISPRa has been considered in PRPF31 to increase levels of wildtype PRPF31 from the unaffected allele, although in the absence of any mechanism by which to avoid activation of the mutated allele, mutated PRPF31 is also likely to be overexpressed, which represents a significant safety concern.

While genome editing approaches hold promise for the treatment of PRPF31-RP, challenges lay ahead, including the risk of off-target editing, lack of efficiency, and risk of immunogenicity and toxicity. Genome editing tools are often at or above the carrying capacity of AAV vectors, which presents a challenge for their delivery to the retina. New, smaller editing tools have been created, however, and strategies employing dual vectors may allow for a split gene to be delivered in 2 complementary vectors.

Therapeutic options for advanced PRPF31-RP

Late-stage PRPF31-RP is characterized by the permanent loss of photoreceptors and significant vision loss. In this stage of disease, gene augmentation or genome editing are unable to treat photoreceptors that have already died. However, alternative approaches have emerged as promising treatments for restoring vision, even after the loss of photosensitive cells. Optogenetics involves using light-sensitive proteins to stimulate remaining retinal cells and produce visual signals, while cell therapies aim replace missing retinal cells. Prosthetic devices bypass photoreceptors and stimulate remaining retinal ganglion cells to send signals to the brain and restore vision.

Optogenetics

Optogenetics offers a promising approach to treat PRPF31-RP by introducing light-sensitive proteins called opsins into surviving retinal cells. These opsins can convert light into electrical signals. Viral vectors may be used to deliver opsin genes to functional retinal cells, such as ganglion or bipolar cells. Once expressed, these opsins can be activated by light to stimulate the neural circuitry of the retina and send visual signals to the brain ¹¹⁷.

Recently, Sahel et al. reported on the use of optogenetic therapy to partially restore visual function in an RP patient by injecting an AAV vector encoding ChrimsonR, which infected remaining retinal ganglion cells, rendering them sensitive to light stimulation via a pair of goggles that interpret the visual scene and project light onto the retina. After several months of training, the patient was able to detect and recognize objects, locate light sources, and distinguish between different shades of gray, demonstrating the promise of this approach for treating PRPF31-RP¹¹⁸.

However, current optogenetics approaches have significant limitations. Many optogenetic effectors, such as ChrimsonR, have limited light sensitivity, requiring signal amplification such as goggles. Other G-protein coupled receptors have been explored as alternative, more sensitive opsins, however, these may have lessened ability to source chromophore outside of photoreceptors or slower response time. The search thus continues for optimal optogenetic tools with sufficient sensitivity, fast kinetics supporting meaningful vision, and efficient expression outside of photoreceptors. Finally, it remains to be seen what degree of visual acuity may be recovered following delivery of opsins to second or third order neurons, as this bypasses some of the signal processing that normally occurs in the retina.

Cell therapy

Cell-based therapy may treat IRDs, such as PRPF31-RP, by replacing damaged or missing retinal cells with healthy cells. The transplantation of healthy stem or progenitor cells into the retina, which can differentiate into a variety of retinal cell types and potentially integrate into the existing circuitry, could potentially restore vision in retinas that have lost photoreceptors^119,120.

Various types of stem cells, including embryonic stem cells (ESCs) ^121,122, induced pluripotent stem cells (iPSCs) ^123,124, and bone marrow-derived mesenchymal stem cells (BMSCs) ¹²⁵, have been extensively investigated by researchers for their potential to replace or repair damaged cells in diverse diseases, and clinical trials for macular degeneration have been launched to evaluate the safety and tolerability of ESC- or iPSC-derived RPE cell transplantation in patients ^126,127. However, the successful integration into the host retina and the reconstruction of neural circuitry and functionality are major challenges for photoreceptor cell transplantation. While iPSCs are an attractive option as a source for autologous stem cells, the successful translation of iPSCs into the broader population may prove challenging, due to variability between cell populations and the lack of optimized, standard protocols for scale up. And, restoring vision using a single type of retinal cell may be insufficient for PRPF31-RP, as this disease appears to affect multiple cell types.

Retinal implants

Retinal implants work by bypassing damaged photoreceptors and directly stimulating remaining ganglion cells, which transmit visual information to the brain. The implant consists of a small microchip that is surgically implanted into the retina, either epiretinally ^128-130, or subretinally ^131-134. In some cases, an external camera captures images of the external environment and sends these images to the chip. The chip then converts images into electrical signals, which are transmitted to retinal ganglion cells via an array of electrodes. The ganglion cells then in turn transmit electrical signals to the brain, where they are interpreted as visual information.

Several devices have received approval for commercial use in the European Economic Area and one in the United States, including Retina Implant Alpha IMS, from Retina Implant AG in Germany, Argus II Retinal Prosthesis System from Second Sight Medical Products Inc in Sylmar, CA, and IRIS II developed by Pixium Vision in Paris, France^135-142. However, none of these is currently available as they have been discontinued. The PRIMA implant, developed by Pixium Vision, is implanted subretinally in end-stage geographic atrophy. It is activated wireless by goggles capturing the visual scene. Positive results have been shown in Age-Related Macular Degeneration, but the implant has not been tested in RP¹⁴³.

Retinal implants are still a relatively new technology and are not suitable for all types of retinal disease. They also require extensive training to use effectively. The quality of vision provided by retinal implants is not yet as good as natural vision, but they may improve the ability of people with certain types of retinal disease to navigate their environment.

Delivery of trophic factors

Mutation-independent strategies aiming to promote photoreceptor survival, irrespective of the underlying genetic mutation, have emerged as promising approaches to maintain vision. One such strategy involves the delivery of rod-derived cone viability factor (RdCVF), a protein secreted by rods that promotes cone survival¹⁴⁴. As cones are responsible for central and color vision, their preservation is crucial for maintaining visual acuity and overall quality of life.

RdCVF has been shown to exert its neuroprotective effects by enhancing glucose uptake in cones¹⁴⁵. Therapeutic approaches involving RdCVF supplementation or gene therapy to stimulate its expression hold promise in decelerating the progression of retinal degeneration and preserving cone function across various genetic backgrounds¹⁴⁶. Other mutation-independent strategies include neuroprotective agents, anti-inflammatory approaches, and modulating metabolic pathways to support the survival of cones. Each of these strategies aim to target common downstream pathways of retinal degeneration, regardless of the specific gene defects involved.

Conclusions

The development of efficient therapies for PRPF31-RP requires deeper understanding of underlying disease mechanisms. Several key questions remain unanswered and are the focus of current studies across in vitro and in vivo models of PRPF31-RP. First, while PRPF31 is ubiquitously expressed, PRPF31-RP is a retina-specific disease, for reasons that are not understood. Several models have been proposed to explain the retinal specificity of PRPF31. Studies in human tissues have shown increased levels of snRNA and higher volumes of processed mRNA in retina compared to other tissues³⁸, which may explain retinal sensitivity to reduced PRPF31. Alternatively, studies in mice and human organoids as well as some zebrafish studies have indicated the importance of PRPF31 for retinal-specific functions, including ciliogenesis, phototransduction, and RPE cell adhesion^{35,78,84,90,91}, which suggests a mechanism by which the loss of PRPF31 may particularly affect the retina. However, loss of PRPF31 has also been strongly linked to missplicing of genes involved in mRNA splicing (leading to a feedback loop exacerbating splicing deficiencies)^78,84. And a wide spectrum of genes have been identified in other models following reduction of PRPF31 expression, including differentially expressed genes involved in spindle formation, DNA repair and homolgous recombination repair in zebrafish⁷⁸ and differentially spliced genes in Prpf31^+/A216P mice involved in splicing, inflammation, oxidative stress, and apoptosis⁸³. In short, the exact mechanisms by which reduced PRPF31 expression causes retinal degeneration are yet to be determined.

Additionally, the retinal cell types primarily affected by the disease are still a matter of debate. Rods, cones, and RPE cells have all been implicated, but the primary cell type(s) driving disease have not been conclusively identified. Studies in transgenic mice suggest a primary defect is in RPE cells^81,83,84, which would indicate that photoreceptor degeneration may be secondary to RPE loss. However, studies in human iPSC-derived organoids⁹¹ and CRISPR/Cas9 knockout mice⁸⁸ show dramatic loss of rods and cones as well as RPE dysfunction, supporting the direct involvement of photoreceptors in the development of disease, as well as RPE. Determining the cell types driving degeneration will be critical to the design of effective therapies.

Investigation into PRPF31-RP using in vitro and in vivo model systems has resulted in a better understanding of disease etiology and paved the way for treatments. While further study is essential for the development of effective therapies, a variety of therapeutic approaches, including gene augmentation, genome editing, and optogenetics, as well as cell therapies and retinal implants, hold promise. In the coming years, significant progress towards effective therapies is likely to be made.