HUMAN CELLULAR MODELS FOR RETINAL DISEASE

Abstract

Purpose:

To provide a concise review of induced pluripotent stem cells (iPSCs) and retinal organoids as models for human retinal diseases and their role in gene discovery and treatment of inherited retinal diseases (IRDs).

Methods:

A PubMed literature review was performed for models of human retinal disease, including animal models and human pluripotent stem cell–derived models.

Results:

There is a growing body of research on retinal disease using human pluripotent stem cells. This is a significant change from just a decade ago when most research was performed on animal models. The advent of induced pluripotent stem cells has permitted not only the generation of two-dimensional human cell cultures such as RPE but also more recently the generation of three-dimensional retinal organoids that better reflect the multicellular laminar architecture of the human retina.

Conclusion:

Modern stem cell techniques are improving our ability to model human retinal disease in vitro, especially with the use of patient-derived induced pluripotent stem cells. In the future, a personalized approach may be used in which the individual’s unique genotype can be modeled in two-dimensional culture or three-dimensional organoids and then rescued with an optimized therapy before treating the patient.

Retinal degenerative conditions are a major cause of visual impairment. These conditions range from diseases with multifactorial etiologies such as age-related macular degeneration to monogenic genetic causes implicated in most inherited retinal diseases (IRDs). Approximately 300 coding genes¹ are implicated in IRDs at this time but that number continues to grow. With the advent of gene therapy for RPE65-mediated retinal dystrophy,² knowledge of the specific disease-causing genetic variants in patients plays an increasingly important role in the diagnosis and management of disease. With advances in human stem cell technology, it has now become possible to use human stem cells for molecular diagnosis, disease modeling, and treatment optimization. This review will detail the implications of human-induced pluripotent stem cells (iPSCs) and retinal organoids (ROs) as models for human retinal diseases and their role in gene discovery and treatment of IRDs (Figure 1).

Fig. 1.

Schematic describing the use of patient-derived iPSCs and ROs to study retinal disease. Patient-derived cells can be differentiated from normal, unaffected subjects and compared with those isolated from diseased patients afflicted with IRDs. The use of stem cell technology allows one to examine the structural and functional consequences of disease. Immunohistochemistry (IHC) can be used to examine expression of cellular markers, whereas 2D electron microscopy (2D EM) and 3D scanning electron microscopy (3D SEM) can be instrumental to study the retinal ultrastructure. Whole cell patch clamp experiments can elucidate the functional electrophysiologic features of retinal cells.

Importance of Genetic Diagnosis

The genetic diagnosis in patients afflicted with retinal degenerative phenotypes can be established with next-generation sequencing. Exome-based approaches, which can capture all protein-coding regions, have been shown to be an effective preliminary diagnostic technique, but this approach has a detection rate of 60% to 70%.³ Patients with pathogenic nonexomic, or noncoding, variants are missed by such approaches. These variants produce disease by affecting transcription and splicing of genes and affect epigenetic regulation in the retina.⁴ Genome sequencing allows identification of variations across the entire genome and allows identification of both coding and noncoding variants. Although genome sequencing analysis provides a superior rate of diagnosis,⁵ a large proportion of patients still remain undiagnosed. To solve the missing heritability in IRDs, a patient-centered approach that focuses on noncoding and regulatory regions in disease genes of interest will be critical. This is especially important in the current era of genome editing technology where identification of disease-causing variants is central to enrollment in possible treatment trials.

Limitations of Nonhuman Models Systems

Access to tissue affected by disease is essential to understand the disease process, but this is not possible in retinal disease. To overcome this limitation, model systems have been developed to elucidate the genetic causes and cellular mechanisms underlying human retinal disease. The models to date have taught us much about retinal diseases, but they are not without their own shortcomings. Retinal cell culture lines display poor expression levels of essential disease genes, whereas retinal culture systems grow in a disorganized manner and have immense variability between samples.⁶ Animal models more closely depict the clinical manifestations of the corresponding disease in humans.⁷ Large animal models, particularly primates that have a macula, are best suited for understanding human disease, but they are difficult to manipulate genetically and only a limited number of spontaneously arising models of IRDs exist. Thus, rodent models, particularly mice, have become the most widely used models of human retinal disease. The main limitation is that the architecture of the rodent eye and retina differ from those of the human, and in many retinal diseases, mouse models do not faithfully replicate the human condition.⁸ The advent of stem cell–based technologies has revolutionized our ability to decipher the genetic causes of IRDs and overcome limitations presented by previous model systems.

Advances in Stem Cell Techniques

The first derivation of human embryonic stem cells⁹ spurred the development of pluripotent stem cells as a model system to study diseases. Then, the ability to reprogram mouse somatic cells¹⁰ and human fibroblasts¹¹ into iPSCs along with evidence that cells from diseased patients could be reprogrammed¹² opened up a new frontier to better understand human disease. Shortly after this discovery, methods to differentiate iPSCs into retinal cells^13,14 demonstrated their promise as a source of disease-relevant retinal tissue. Subsequent patient-specific studies revealed that a disease phenotype could be observed in patient iPSC-derived retinal cells.^15,16 Although iPSC were initially produced from adult fibroblasts, they can now be produced from less invasive samples such as blood and urine, although there is no consensus on the preferred somatic cell source for iPSC derivation. The next breakthrough ushered in a new generation of retinal models, based on organoids that more closely replicate in vivo development. This three-dimensional (3D) optic vesicle model is termed ROs¹⁷ and exhibits functional photoreceptors.¹⁸ Significant advances have been made in the morphological and molecular characterization of human ROs and the demonstration of their utility in understanding the normal development of the human retina and changes manifested in disease. The 3D architecture of ROs recapitulates a spatiotem-poral organization of the retina that is well structured and has functionality for in vitro studies. In the next section, the advances in iPSC technology to address IRDs will be examined.

Modeling Retinal Degeneration Using Patient-Derived Induced Pluripotent Stem Cells

Stem cell technology allows the discovery of disease-causing variants and the assessment of pathogenicity of identified variants without the need to directly assay retinal tissue in diseased individuals. Moreover, derivation of iPSCs from patients with known pathogenic variants provides an option for a personalized medicinal approach in which the individual’s unique variation can be understood more precisely and rescued with gene therapy in vitro before transitioning to therapy in the patient. There are various methods to derive iPSCs through directed differentiation methods, which involve the addition of factors such as growth factors and small molecules that mimic the development process in vivo. For the sake of this review these methods will not be discussed in detail.

The benefit of iPSC-based methods is that it can narrow candidate variants to a focused few that can be rapidly tested in vitro to determine the genetic etiology in patients lacking a diagnosis. More importantly, the necessity and sufficiency of causative variants can be definitely assessed from patient-derived stem cells. Patient-derived cells from unaffected family members can be genetically edited to harbor the disease-causing variant. Sufficiency of the proposed variant in producing disease can be established if the diseased phenotype of the patient-derived cells is replicated in the genetically edited unaffected family member cell line. However, if the putative disease-causing variant can be corrected in patient-derived cells using gene editing technology and the disease phenotype is reversed, the necessity of the proposed variant can be established. Synthetic introduction of variants into wild-type cells is a powerful means of providing convincing evidence of a variant’s pathogenicity, especially for noncoding variants.

For the most common IRD, retinitis pigmentosa, the genetic diagnosis and underlying molecular mechanisms that drive disease pathogenesis remain unclear despite the phenotypic commonality in patients. In retinitis pigmentosa, the use of iPSC modeling was instrumental in identifying novel gene variants that contribute to disease in photoreceptors¹⁹ and RPE.²⁰ Usher syndrome is the most common ciliopathy that leads to retinal degeneration and blindness with most of the disease caused by the gene USH2A. Despite its disease relevance, there is a paucity in understanding of the normal function of the protein encoded by USH2A, usherin. Gene correction of the two most common USH2A variants from patient-derived iPSCs revealed novel mechanisms of aberrant mRNA levels associated with the disease when comparing iPSCs derived from patient with the disease and the corrected isogenic counterpart.²¹ These illustrative examples indicate the power of iPSC technology to understand monogenic IRDs. In the future, this technology can be expanded to study multiple variants in multigenic diseases, such as age-related macular degeneration.

The utility of identifying the pathogenic variant extends past therapeutic intervention. Correcting a known variant at the iPSC stage allows one to create isogenic control lines with the same genetic background as the patient. These model lines can be used to prototype cell-specific gene editing strategies. Various gene editing technologies (CRISPR-Cas9, prime editing, or transposases)²² can be used to correct variants in iPSC lines, and their pathogenicity can be confirmed. A seminal work was the use of CRISPR-Cas9 to correct pathogenic variants for three different IRD-causative genes.²³ In a case of a patient with RPGR-associated IRD, the third most common IRD gene implicated in X-linked retinitis pigmentosa; the pathogenic variant was corrected using CRISPR-Cas9 technology, and photoreceptors were restored in both structure and electrophysiology as compared to healthy controls. This finding showed that gene correction led to a rescued photoreceptor structure, reversed ciliopathy, and restored the expression level of retinal-related genes, thus providing proof-of-concept for future therapeutic interventions.²⁴ Gene editing technologies combined with autologous iPSCs provide a promising research basis for a personalized transplantation strategy for IRDs. Furthermore, iPSC technology can also be used for phenotypic drug discovery, wherein one can screen for compounds to reverse a disease-associated phenotype without previous knowledge of the mechanisms of degeneration.

A major goal of modeling the human retina using iPSC technology is representation of the heterogeneous population of cells that compose the retina. A major advantage of gene-corrected patient iPSC-derived cells for transplantation is reduced immunogenicity. However, derivation to a single cell type do not emulate the complex interplay between the many retinal cell types. This shortcoming of iPSC technology can be addressed with 3D RO models that now introduce vascularization and interaction with the RPE.²⁵ Moreover, because ROs recapitulate retinal development, they can reveal gene expression dynamics that change over the course of differentiation. Research conducted in 2D iPSC models can be repeated in 3D ROs to better understand the complex interplay of various cell types in degenerative phenotypes.

Retinal Organoids with Stratified Architecture

Human ROs arguably represent one the most important advances in our ability to study human retinal development and model human retinal disease. It is well known that the retina’s laminar architecture and cellular connectivity (through synapses, tight junctions, and other interactions) remains an essential aspect of its functionality. Retinal organoids have permitted the study of retinal cell biology in the context of this 3D architecture, and because they are PSC-derived tissues, one can model specific disease variants, perform gene editing, or engineer them to express fluorescent proteins or other tools as a biologic readout.

In 2011, the Sasai Laboratory reported the first generation of ROs from mouse embryonic stem cells,¹⁷ which ushered in the era of RO technology at the same time that other groups were beginning to report the generation of organoids for other bodily tissues.²⁶ After this report, numerous groups began to report on the generation of human ROs derived from embryonic stem cells or iPSCs^18,27-29 (Figure 2). These studies demonstrated the successful recapitulation of various aspects of retinal development including the creation of all five neuronal cell types including the four photoreceptor subtypes, photoreceptor outer segments, synaptic layers, and even electrophysiologic evidence of photoreceptor phototransduction. Subsequent reports have further expanded on the utility of the organoid system by identifying stages of organoid maturation,³⁰ characterization of outer plexiform layer synaptic maturation,³¹ and single-cell RNA sequencing analysis that confirms the verisimilitude of RO cell types to those of fetal and adult retina^32-35 (Figure 3).

Fig. 2.

Human ROs recapitulate retinal development in vitro. A. Representative light microscopic images of human PSCs and ROs generated from PSCs at various ages in days. Maturing inner segments and outer segments of photoreceptors appear as hair-like structures on the outer surface of the ROs (D130 onward). B and C. Representative immunofluorescence images of human RO sections revealing 3D organization of major retinal cell types mimicking their native laminar architecture. Images in (B) show low-magnification cross-sections, and images in (C) are at higher magnification. DAPI-stained nuclei in the ONL and the INL, with OPL appearing as a gap between the two nuclear layers. Cell-type specific markers were used to detect each cell type (cones, ARR3; rods, NRL-GFP line⁴⁰; bipolar cells, GNG13; horizontal cells, CALB; and nuclei, DAPI). INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer. Scale bars in panels (A–C) 100, 50, and 10 μm, respectively.

Fig. 3.

Single-cell RNA sequencing of adult human retina and mature human ROs identifies similar cell type expression. Two-dimensional representation of single-cell distribution in adult human retina (left, data from³²) and mature human ROs (right, data from³⁴). Single dots represent cells that are colored by their cell-type classification.

In addition to modeling human retinal development, human ROs have provided a window into disease pathogenesis and our ability to correct phenotypic defects. For example, Leber congenital amaurosis and X-linked retinitis pigmentosa human RO models have demonstrated photoreceptor abnormalities in CEP290,³⁶ dominant CRX,³⁷ RPGR,²⁴ and RP2³⁸ mutant organoids. These studies were also able to implement different methods to correct the abnormality, either with antisense oligonucleotide-mediated suppression of aberrant splicing for CEP290, CRISPR-based correction of the mutant iPSC line for RPGR, or AAV-based gene augmentation for CRX and RP2 organoids. The potential applications for RO technology will continue to expand as tools to probe retinal development and to test therapeutics in a highly personalized fashion. These efforts are greatly enhanced by the inherent scalability and efficiency of organoid production.

Current Limitations of Retinal Organoids

Retinal organoids remain attractive models to study human retinal development and single-gene photoreceptor degenerative diseases. However, the heterogeneity of cellular maturation and morphology³¹ can lead to difficulty in interpreting the course of “normal” development and the effects of genetic mutations. In addition, although some organoid protocols generate the RPE, they do not arise as a monolayer in contact with the photoreceptors as seen in vivo. Thus, diseases that arise from photoreceptor–RPE pathology, such as those affecting outer segment phagocytosis and the visual cycle, can be difficult to model with current organoid technologies. Other limitations relate to the proper development of some cell types in organoids, namely, retinal ganglion cells, immune-related cells, and blood vasculature. Although retinal ganglion cells develop in organoids, there is progressive loss of retinal ganglion cells over the course of organoid development³⁹ as the viability of retinal ganglion cells is dependent on their connectivity to visual cortex neurons. The addition of these cell types will require assembloid technology to bring together cells from different developmental lineages. Efforts are currently underway to address at least some of these limitations through the use of organoid–RPE cocultures and retina-on-a-chip technology.²⁵

Conclusions

Up until 15 years ago, the ability to study human retinal disease in vitro was relatively limited to primary cell cultures or fetal retina explants. With the advent of stem cell techniques including iPSCs, CRISPR engineering, and the ability to generate 3D organoids, human disease models have arguably risen to the forefront of retinal research. Furthermore, because iPSCs can be generated from patients, this permits highly specific and scalable testing of therapies in vitro on patient-derived 2D or 3D cultures. This will be especially advantageous for emerging therapies such as base editing that are specific to individual variants. In conjunction with the ability to use organoid-derived, corrected retinal cells for human cell therapy, these approaches have the ability to continue to pave the future of retinal disease research.