The fully developed neural retina comprises high-specialized cells that transform light stimuli into electrochemical signals. Retinal progenitor cells (RPC) differentiate into neuronal subtypes in a specific and phylogenetical conserved order during development. Complementary processes guide the RPC into its cell fate, with noticeable involvement of the transcription factors [1], [2], [3], [4]. Alterations in this genetic program generate visual deficits, as observed in foveal hypoplasia, a retinal disorder in which the fovea lacks full development.

Retinitis pigmentosa (RP) is the main degenerative cause of irreversible blindness worldwide. The RP is a genetic group of disorders that leads to a progressive and massive loss of photoreceptors as a consequence of malfunction in the outer segment (OS) and retinal pigmented epithelium (RPE) [5]. Prognosis depends on the origin of gene mutation, which may cause alterations in phototransduction, cell trafficking, or rhodopsin turnover. Despite different gene mutations, rods are usually affected before the cones. Both rod and cone cell loss may depend on apoptosis, autophagy, or necroptosis due to oxidative stress, metabolic stress, and abnormal calcium regulation [6], [7].

Genetic characterization of RP is intricate but valuable for treatment. Whole-exome sequencing has been used for RP characterization, but the application is complex due to the number of involved genes [5]. In this context, in vitro models that predict the type of gene modification could be crucial, mainly when it is derived from the patient’s cells. Several mutations only reveal their harmful effects in differentiated cells. Therapeutic approaches are possible once in vitro differentiation is established [8], [9], [10], encompassing personalized gene correction and control of secondary consequences related to the DNA edition. Due to the shared developmental origin, the retina has several similarities with the brain, including cell stratification, interplay with the immune system, and similar response to insults. Those characteristics mirror the retina to other degenerative diseases, such as Alzheimer’s disease (AD) and Parkinson’s disease (PD). The clinical manifestation of AD in the retina includes ganglion cells atrophy and reduced blood flow and macula area. In turn, patients with PD present a decrease in visual acuity, swelling in photosensitive neurons, and loss of dopaminergic amacrine cells [11].

Models mimicking retinal development may subserve in-depth study of retinal dystrophies, especially with other technologies, such as bioprinting and microfluidics. Furthermore, considering the shared embryological origin with the brain, in vitro models of the retina could be helpful to the early diagnosis of age-related diseases, such as AD and PD. This review addresses recent advances and promising perspectives on these topics.

In vitro models have been developed to study the underlying mechanisms of diseases in the central nervous system (CNS). A notable evolution was achieved in the last decades from cultures obtained from neonatal and adult animal-derived primary cells, embryonic stem cells (ESCs), and recently, induced pluripotent stem cells (iPSCs) [12]. In 1998 Thomson et al. observed that human ESCs (hESCs) derived from human blastocysts remain proliferative and able to differentiate into cells of the three embryonic germ layers [13].

In 2006 Takahashi and Yamanaka published the first work on reprogramming adult fibroblasts by the expressing Sox2, Klf4, Oct3/4 and c-Myc genes, turning the somatic cells into iPSCs [14], a notable achievement not restricted to animal models. Indeed, human iPSC (hiPSC) provides a similar potential to hESC to obtain any cell type, including neurons, avoiding ethics issues. Afterward, several studies aimed to use hiPSCs for different proposals, including cell replacement therapies [15], [16], [17], [18]. iPSC cells can derive two-dimensional (2D) neuron differentiation and three-dimensional (3D) cell culture, including organoids. 2D cell culture can help understand neurological disease mechanisms and model neuronal development. However, to study the interaction between multiple types of neurons and cell organization, including extracellular matrix (ECM) interactions [19], 3D organoid has been considered a gold standard [20]. Lancaster (2013) was a pioneer in developing brain organoids, which presented an organized structure with different cell types that mimic the fetal brain [21]. The improvement of this technology opens an immensity of possibilities, such as establishing a Neanderthal cortical in vitro model [22], and obtaining other complex structures, such as the eye or retina.

Retina organoids (ROs) mimic tissue development, maturation and environment [20], and increase the generation of photoreceptors [23]. Transcriptomic analysis revealed that the human fetal retina and hiPSCs-derived ROs have a similar cellular composition at specific developmental stages [24]. Different methods can be employed for ROs differentiation. In Eiraku M. et al. (2011), ESC was seeded in low-cell-adhesion plates for fast aggregates of cells, which resemble a primordial retina. After that, supplements were added for the optic cup self-formation and maturation (Fig. 1A), and the Rax gene - crucial in the development of the eyes [25] - was expressed after five days of culture [26]. Nakano’s protocol was based on modifications and improvements of Eiraku’s protocol [27]. Capowski compared the generation of ROs from multiple lines of hiPSC and hESC, using a hybrid method based on [23] to improve organoid formation. In this method, BMP4 treatment was used in the embryoid bodies until D24 - 30 to enhance optic vesicle formation, and the ROs were transferred into a suspension culture with retinal differentiation medium (RDM) supplemented lipids and retinoic acid (RA), required in several pathways for eye and retina development [28]. The authors showed three stages in ROs development and maturation (Fig. 1B). There are still some challenges to improving ROs protocols. For example, the use of RA remains under debate, whether it has a better response when supplemented throughout the entire incubation period or only in specific time windows [28], [29], 30].

RO protocols usually involve the presence of RPE cells (Fig. 1). In vivo, the RPE layer is located after the photoreceptor segments. RPE is essential in several functions, including photoreceptor differentiation and metabolic support, light absorption, and protection against photooxidation [31]. The RPE support to neuroretina tissue is so expressive that mutations in its proteins could lead to retinal dystrophies, such as RP and AMD [32], [33]. Although mature RPE cells are present in RO approximately in 63 DIV [34] - depending on the protocol - this layer is not adjacent to photoreceptors as observed in vivo. Some authors try to circumvent this characteristic by co-culturing RPE cells with RO, which already showed positive results by improving the differentiation of functional photoreceptors [35]. However, as far as we know, no published protocol showed a RO that relies on an optic cup-like structure, as discussed in [36] (Fig. 2).

Besides RPE-RO's lack of interaction, other points are still not well elucidated, such as the lack of microglial cells and vascularization in the CNS-derived organoids. Therefore, improving methods to circumvent these disadvantages may be crucial for obtaining a more realistic model.

In the ROs, the lack of microglial cells and vascularization is due to the different cell lineage origin [37], which makes the generation of these cells in brain organoids initially disregarded [38]. The microglia cells originate from the yolk sac, considered from the myeloid lineage [39], and colonize the CNS through the vasculature [40] before their differentiation into microglia, which are later separated from other bone marrow-derived progenitors by the blood-brain barrier (BBB) [38].

Microglia plays an essential role in CNS health and disease, supporting several processes in neuronal [41], [42] and vascular development. Microglia release anti-inflammatory cytokines and neuroprotective factors in physiological conditions [40]. In neurodegenerative disorders, the microglia become activated, presenting an amoeboid morphology [43], and releasing inflammatory mediators [44]. An imbalance of microglia functional polarization contributes to the progression of neurodegenerative disorders such as RP, AD and PD [38], [45], [46]. In RP, the activated microglia migrate to the degeneration site (outer nuclear layer, ONL), interacting with photoreceptors under stress [47], [48]. The literature brings up methods to improve the vascularization and microglia presence in ROs. One proposed method does not require manipulation of molecular pathways, such as the inhibition of dual-SMAD [49], and it seems efficient to generate mesoderm-derived progenitors into the organoid, which later differentiated into functional microglia cells. A similar result was obtained using the “unguided organoid” protocol (Fig. 2D) [50]. Other authors suggest co-culturing microglial cells or mesodermal progenitors after RO formation to obtain functional microglia, which respond to harmful external stimuli [51]. The second approach differentiates microglia cells and generates an endothelial cell system that could mimic functional vascularization in brain organoids using co-culture approaches (Fig. 2B) [37].

The vascularization process is essential for the retina since it is one of the most energy-demanding tissues due to phototransduction [52]. In ROs, the nutrition supply hardly penetrates the inner layers by diffusion, which may cause a gradual loss of inner cells by hypoxia and/or inanition [28], [53]. The most affected cell types are retinal ganglion cells (RGCs), which develop in the inner core of the RO. Besides the location, studies have shown that these cells are progressively lost while the photoreceptor maturates, leading to cellular disorganization of the inner core [28], [54]. Although RGCs do not directly perform phototransduction, they are essential for the retina function and response to light stimuli, altered in aging-related diseases such as AD and PD [55], [56], [57]. Therefore, preserving both RGCs and photoreceptors is essential and desirable in ROs.

There are several proposed forms of stimulating or inducing vascularization under study, such as implantation on a highly vascularized tissue (Fig. 2 A), spontaneous vascularization by mesodermal progenitors/endothelial cells/blood vessel organoids co-culturing (Fig. 2B), bioprinting (Fig. 2 C), and in vitro BBB resembling using microfluidics (Fig. 2E) [58].

Between these strategies, organ-in-a-chip is particularly interesting since it recapitulates the RPE-photoreceptor contact, enabling metabolic support to the Ros. Moreover, this alternative maintains photoreceptors and RPE functionality in vitro [62]. This functionality is important, especially considering the phototransduction process that generates a high energy consumption. Although studies already described RO responses to light stimulus [63], some characteristics are essential to better mimic diseases related to photoreceptors, such as RP.

The ROs derived from iPSC/ESC has successfully developed precursor and mature photoreceptor cells, preserving both the inner segment (IS) and the OS, and possess a light-evoked response capacity besides retinol accumulation and glycolytic signature - a metabolically active cascade that resembles the phototransduction process [23], [28], [64], [65], [66], [67]. Functional photoreceptors play an essential role in RP studies since the disease-associated mutations lead to dysfunctional phototransduction, resulting in progressive cell death due to calcium accumulation and elevated cGMP levels [6].

Several genes are involved in RP, leading to several morphological, molecular, and/or cellular alterations in photoreceptor and RPE stratification. The RP disease progression could be classified into early, middle, and late stages [5]. The retina presents an initial rod cell death in the early stages, mainly due to mutations. The cone cells tend to die after rods, and in the late stage, most photoreceptors are lost. The activated microglia migration into the outer retina occurs, possibly phagocytizing the remained photoreceptors and extending the degeneration area [68], [69]. Active gliosis increases in late RP stages, and the loss of synaptic contacts are related to extensive cell death (Fig. 3 A).

The hiPSCs differentiation into RO has significantly advanced the understanding of RP's cellular and molecular basis. It is now possible to study the patient's cells non-invasively and identify gene mutations in RO, besides looking at early and late RP stages. Furthermore, in vitro RP modeling personalizes the therapy focusing on drug delivery and gene modifications aiming to identify options for clinical trials [70], [71], [72]. Focusing on mimicking RP, an in vitro 3D retinal model should present i) photoreceptors with consistent morphology composed of IS/OS/discs, ii) phototransduction process, iii) cell-cell contact to proceed with the electrochemical signaling. In this regard, both electrical and chemical synapses and the respective receptors should be present in RO, preferably responding to light stimulus. Several works revealed their essentiality in a dark-adapted environment [73], [74] and the development of visual circuitries [75], [76], [77].

In this way, previous studies improved functional photoreceptors formations and synapses. Kaya and collaborators developed a protocol [78] to enhance photoreceptor maturation, mainly rods, by adding 9-cis, instead of all-trans RA, in RO cultures [79]. The addition of thyroid hormones has improved the cone subtypes specifications [80], and the FGF1 and DHA factors provide the mature photoreceptors with longer cilia [81].

Some authors have described the RP patient-hiPSC-derived RO formation with distinct related genes, as discussed in [82]. Mature and immature photoreceptors showed no remarkable differences from healthy patient-RO in the early stages [83]. During ROs maturation, differences in morphology, electrophysiology, cytokine release, and metabolic response were observed [71], [83], [84]. In this model, the apoptosis pattern (peak in D150) seems to be increased before the photoreceptor cell death (D180), leading to late-onset consequences such as synapse alterations (D190), disorganization in retina cilia (D230) and loss of connections between photoreceptor and rod-bipolar cells (D260). Moreover, remarkable metabolic differences are observed in ROs after D230, mainly in hydrolysis alteration and cGMP increase [71], [84], [85] (Fig. 3B).

Another strategy is directly reprogramming hiPSC into RPCs in 2D, favoring the observations of cell-cell connections [86], [87]. Indeed, both 2D and 3D cell culturing strategies permit to readout effects of drug and gene therapies in the phototransduction cascade signaling, aiming to develop treatments to prevent or delay cell death in RP patients (Fig. 3 C). Nevertheless, the photoreceptor development does not seem concurrent. Displaced photoreceptors are observed, which could be assuaged in RO co-cultured with RPE, as discussed in [28]. These epithelial cells may help to obtain a more reliable in vitro model, restoring the subretinal niche functionality provided by the RPE layer around ROs [88]. Another possibility would be cell therapy, seeking to replace the affected cells. Clinical trials use hESC or autologous hiPSC-derived RPE cells as suspensions or sheets under the retina in patients with AMD [89]. In this context, particularly interesting is the gene editing of the autologous iPSCs according to the gene mutation observed in ROs. Afterward, corrected cells could be transplanted to the host as a potential therapy (Fig. 3 C).