Disease mechanism and therapies for retinal degeneration

National Eye Institute

Investigators

Linked publications & trials

Abstract

We have continued our studies in mouse models where appropriate but have made a transition to retinal organoid models with the ultimate goal of displacing animal models, with major efforts directed at continuing improvement in the organoid models and expand their utility in retinal research. Role of different RPGR variants -- RPGR is expressed primarily as a default and an ORF15 variant, with further complexity among each variant group. During development, all cell types express the default variant but as photoreceptors begin to mature, a switch in splicing occurs that replaces the default with the ORF15 variant. The latter is believed to be photoreceptor specific10 and is a hot spot for disease causing mutations. We further investigated whether the default variant played any role at all in photoreceptor development or function. We generated a transgenic model that expressed recombinant RPGR default. Not only did the default variant fail to rescue the RPGR null phenotype but it seemed to be highly toxic even at moderately elevated expression levels in WT mouse background. These data suggest that the default variant play no functional role in adult photoreceptors and can in fact compete against the ORF15 variant thus interfering with its normal function. To determine if RPGR default played any role in retinal development, given the fact that it is expressed in the developing retina, we carried out germline ablation of the default variant only (in collaboration with the Dr. Lijin Dong, NEI Genetic Engineering Core). The knockout mice were phenotypically normal both systemically and in the retina. Both ERG and histology followed up to over 1 years showed normal retina function and cell survival. The only change we could identify was an increased levels of -catenin protein in multiple tissues, suggesting an upregulation of canonical Wnt signaling. Given the role of primary cilia as a Wnt transducing organelle, this finding isn’t surprising. Analyses of mouse vs. human RPGR at the protein level -- One of the little-known facts is that, despite thousands of papers having been written, neither mouse nor human RPGR ORF15 cDNA has never been ‘cloned” in the traditional sense of the term. No one has ever obtained a full-length cDNA from tissues, but rather it was pieced together based on a best guess. Twenty years later this problem has remained unresolved despite numerous efforts (including our own). One theory was that the RNA forms G-quadruplex, which reverse transcriptase cannot read through. The question then arises whether the lack of clear-cut efficacy in multiple clinical trials might be due to the imprecision of input coding DNA? We chose to seek answers by examining the RPGR proteins. We generated multiple domain specific antibodies for human and mouse RPGR, respectively and examined mouse, non-human primate and human donor retinal tissues by IHC and immunoblots. We made the following discoveries. In both mice and primates, RPGR-ORF15 is the only variant in photoreceptor connecting cilia of adult. In mice the isoform switch from default to ORF15 occurs around postnatal 12 or so. Proteins on immunoblots showed further heterogeneity, the most surprising of which is that a large fraction of RPGR does not carry the C-terminal domain. In human genetics, the C-terminal domain seems essential and any frameshift mutations that lose the C-terminus are deemed causal for disease. Furthermore, we have shown previously that the C-terminal interaction with TTLL5 is critical for RPGR polyglutamylation and function. Our working hypothesis is that the C-terminus is cleaved following RPGR glutamylation and will conduct labeling experiments to test this. The putative human RPGR-ORF15 sequence, when delivered into null mouse retinas, recapitulate the precise ciliary localization (not shown). We conclude that the assembled human RPGR-ORF15 sequence is approximately correct, though minor differences will need to be excluded at finer resolution. Going forward, isoform switch in human retinas will be determined using retinal organoids, as a proxy, at a series of ages. While several additional experiments are still pending, we anticipate that we will submit this study for publication before the end of the year. CCP5 maintains homeostasis of tubulin glutamylation in photoreceptors -- We were motivated to undertake this study based on our hypothesis that CCP5 might act as the opposing enzyme to TTLL513 to achieve dynamic equilibrium of RPGR glutamylation in photoreceptors. Complex patterns of protein glutamylation contribute to the “tubulin code” that confers versatility to stable microtubules in ciliary structures and neuronal axons. Addition of a single glutamate produces monoglutamylation, and elongation of the branched chains generates polyglutamylation structures. Tubulin tyrosine ligase-like (TTLL) proteins either initiate or elongate the glutamate side chains. Cytosolic carboxypeptidases (CCP) catalyze the reverse reactions and similarly have preferential activities either to shorten the chain or remove the branching point glutamate. We show that the broadly expressed CCP5, the only CCP known to remove the branch point glutamate, is uniquely required for maintaining homeostasis of microtubule glutamylation in photoreceptor cells. Upon loss of CCP5 in mice, cytoplasmic microtubules, normally devoid of significant glutamylation, become excessively glutamylated. Ciliary microtubules, normally modified by both glutamylation and glycylation in a state of physiological balance, are skewed towards predominant glutamylation at the expense of glycylation. Perturbation of microtubule glutamylation leads to a defect in trafficking of cone and rod opsins, and abnormal outer segment disc formation. Photoreceptors subsequently degenerate with cones affected more severely than rods. Despite its widespread expression, loss of CCP5 increases tubulin glutamylation only in photoreceptors and in testes suggesting redundant mechanism may exists in other tissues. These results show that, in photoreceptors, CCP5 is a critical deglutamulase responsible for removing the branch point glutamate in vivo, and support CCP5 mutations as a cause of human cone rod dystrophy or retinitis pigmentosa. This project was partly reviewed in the previous BSC where we presented biochemistry data. Following completion of the study, another group published similar results. Interestingly, RPGR glutamylation was completely unaltered upon loss of CCP5. Thus our original hypothesis was proven incorrect. Nevertheless, it gave us insights into CCP5 function and disease mechanism in the retina. We delayed our publication to seek greater insights in terms of photoreceptor pathophysiology, and fine tuning of protein levels in preparation AAV-mediated gene augmentation study (towards clinical translation). Bypassing an embryoid body stage in organoid production improves efficiency -- The generation of retinal organoids from human pluripotent stem cells (hPSCs) has been established within the past decade, partially recapitulating in vivo retinal development. However, the efficiency of hPSC differentiation into retinal organoids using current protocols varies depending on the cell line used and the technical proficiency of the researchers involved. This variability presents a considerable challenge, particularly for those new to the field. To address these limitations, we developed an alternative differentiation protocol for the generation of retinal organoids. By bypassing embryoid body formation, incorporating nicotinamide treatment, and initiating differentiation from properly inter-spaced colonies, our method streamlines the process and accelerates differentiation. Subsequent long-term cultures of the generated organoids yield large-scale production with stratified retinal organoids containing the seven major retinal cell types, as well as extended inner and outer segment-like structures. This streamlined and reproducible protocol successfully generated retinal organoids from 11 hPSC lines in our laboratory without requiring extensive technical expertise. Two key variables that enable the success of this approach appear to be careful control of starting cell density and use of nicotinamide early in differentiation, which prevents massive cell death among some sensitive lines. Our approach offers a robust platform for large-scale retinal organoid production, facilitating research across various cell lines and experimental conditions. A 3D-printed stirred bioreactor at the initial phase of differentiation enhances retinal organoid production via improved oxygenation -- During early embryonic development, non-vascularized tissues are exposed to a state of physiological hypoxia, particularly in the developing central nervus system. Various studies have defined physiological hypoxia in a range of 1 – 6% oxygen levels, or 8 – 45 mmHg in terms of oxygen partial pressure17,18 whereas oxygen level below 1% is considered severe, non-physiological hypoxia detrimental to neural development. Current differentiation protocols for retinal organoids often yield inconsistent results with large cell line and batch variability. These protocols commonly use static culture methods relying on passive oxygen diffusion to reach the vessel bottom where adherent iPSCs initially differentiate. Static culture is standard for adherent monolayer cells and is presumed suitable19,20. We questioned this assumption given that, during differentiation, the monolayer iPSCs become highly structured and multi-layered, first as neural rosettes and then as optic vesicles. We hypothesized that the cellular oxygen consumption rate would exceed the rate of delivery via passive diffusion, particularly to inner regions of emerging optic vesicles. To test this hypothesis, we measured dissolved oxygen concentrations at the vessel bottom and found that oxygen dropped to <1 % within hours of media change, a level considered non-physiological hypoxia that imperils cell viability. This severe hypoxia caused optic vesicle degeneration, hypoxic marker expression, and necrosis. To address this problem, we developed a novel 3D-printed stirred bioreactor (SBR) that enhances oxygen diffusion, maintaining oxygen levels between ~4-6%. This approach significantly improved organoid yield, quality, and reproducibility. We conclude that non-physiological hypoxia, a previously unappreciated condition, is a limiting factor underlying inconsistent yield and quality in retinal organoid production. Physiological oxygenation levels can be restored by the SBR platform, resulting in greater reproducibility and improved yields. A minor shear stress from media flow did not cause any unwanted changes in the pattern of cell fate specification as shown by transcriptomic profiling. A hydrogel platform improves organoid culture during maturation and long-term maintenance -- Following our success for the initial adherent phase of culture, we turned our attention to the following floating phase of culture. Current methods show a high decay rate during maturation, with additional challenges posed by biofouling of dishes, organoid aggregation, restricted access to nutrients and oxygen, and complex maintenance requirements. These issues are pronounced both in en masse culture in 10-cm dish or isolated culture in multi-well plate. Some reported application of bioreactors was overly complex. Retinal organoids require non-adherent surfaces for proper growth, typically achieved by coating plastic surfaces with hydrophilic polymers. However, these coatings degrade over time, leading to biofouling as cells and proteins adhere to the vessel surface. Commercially supplied ultra-low attachment dishes, while more effective, are extremely expensive. Additionally, the stiff, non-permeable nature of plastic restricts oxygen and nutrient diffusion from the surface in contact with the organoids, creating an environment that does not mimic physiological conditions. We have developed a Multi-Well Agarose platform (MWA), a novel organoid culture device that overcomes the limitations of existing systems. The MWA is fabricated from biocompatible and transparent agarose, which is derived from natural polymers. The agarose hydrogel forms a porous mesh network with flexible fibers, creating pore sizes ranging from 100 to 500 nanometers, depending on the precursor concentration used during fabrication. Agarose hydrogels offer strong, biocompatible, long-lasting and anti-adhesive properties that are resistant to biofouling, ensuring an adhesion-free environment for long-term maintenance of organoids. The highly porous hydrogel scaffold allows enhanced oxygen and nutrient diffusion, as well as efficient waste removal. An optional feature permits enhanced media circulation through stir bar agitation on a magnetic stir plate, overcoming limitations of static culture by improving mass transfer and ensuring adequate nutrient and oxygen delivery to organoids. With this device, we have achieved much lower rate of decay, longer survival periods, enhanced productivity and more consistent outcomes. The MWA houses 80 organoids, close to the capacity of a 96-well plate. Media changes can be done quickly by suctioning off spent media from the central reservoir and adding fresh media to the same location. In terms of cost, agarose is relatively inexpensive and widely available. We have maintained organoids in this platform for over 8 months and saw no evidence of biofouling or structural degradation. This platform and the previous SBR device are the subjects of an issued patent (WO/2025/072826). A manuscript on this study will be submitted in two months. We further hypothesized that a stirred MWA might thus enhance RGC survival localized deep inside the organoid. This idea is being tested and is the subject of an ongoing follow-up study.