Dissecting PDE6-Associated Inherited Retinal Dystrophies Using Patient-Derived Retinal Models

Abstract

Inherited retinal dystrophies (IRDs) comprise a diverse group of genetic disorders that frequently result in irreversible vision loss due to photoreceptor dysfunction or degeneration. Among them, retinitis pigmentosa (RP) and achromatopsia (ACHM) are, in some cases, associated with pathogenic variants in PDE6A and PDE6C, respectively, which are key components of the phototransduction cascade. As most of IRDs still lack effective therapies, retinal organoids (ROs) provide a valuable in vitro model for the investigation of disease-associated mechanisms. Here, we generated induced pluripotent stem cell (iPSC)-derived ROs from an RP patient carrying compound heterozygous PDE6A mutations and from a patient with ACHM harboring a homozygous PDE6C mutation, along with their corresponding CRISPR/Cas9-corrected isogenic controls, which, to our knowledge, represent the first patient-derived RO models reported for the PDE6A and PDE6C genes. The mutant PDE6A line exhibited impaired neuroretinal vesicle formation and RO differentiation; however, a subset of RP-derived ROs matured appropriately and retained photoreceptor features. Moreover, the specific isoform expression pattern detected in retinal tissues reflected differences across developmental maturation stages that could influence disease severity. In contrast, the PDE6C_mutant ROs displayed normal structure and maturation, although cGMP hydrolysis within photoreceptors was likely compromised. In both models, CRISPR/Cas9-mediated correction restored the disease-associated phenotype resembling wild-type ROs. Collectively, these findings provide new insights into PDE6-associated pathogenesis, underscore the utility of patient-specific and gene-corrected ROs for elucidating IRD mechanisms, and support gene editing as a promising therapeutic strategy.

1. Introduction

Inherited retinal dystrophies (IRDs) are a large and heterogeneous group of hereditary diseases that often lead to progressive and severe vision loss. Over 400 genes are related to different phenotypes of IRDs, with some of them encoding for proteins essential for phototransduction or other photoreceptor functions [1,2]. Despite the identification of numerous disease-causing genes, therapeutic options for IRDs remain very limited; notably, Luxturna (voretigene neparvovec) is currently the only approved gene therapy for patients with biallelic RPE65 mutations [3]. Retinitis pigmentosa (RP) is the most prevalent IRD, affecting approximately 1 in 3000–4000 individuals worldwide [4]. It is a progressive disease characterized by the primary degeneration of rod photoreceptors, followed by secondary cone loss. The onset and severity of this pathology vary according to the specific pathogenic variant and the gene involved. Both syndromic and non-syndromic RP forms exist, with the latter being more prevalent and associated with over 100 genes, including PDE6A, whose pathogenic variants are inherited in an autosomal recessive manner [2,4].

PDE6A (NM_000440.3; OMIM 180071) encodes the α-subunit of rod-specific cyclic nucleotide phosphodiesterase (PDE6) [4] and is located on chromosome 5q32; it comprises 22 exons and encodes an 860 amino acid protein. PDE6A is specifically expressed in rods and forms an heterotetrameric complex composed of two catalytic subunits (PDE6A, ~99.5 kDa, and PDE6B, ~98.3 kDa) and two identical inhibitory γ-subunits (PDE6G, ~9.6 kDa) [5].

While PDE6A is critical for rod photoreceptor integrity, cones rely on PDE6C (NM_006204.4; OMIM 600827), which encodes the α′-subunit of cone-specific cyclic nucleotide phosphodiesterase. The PDE6C gene is located on chromosome 10q23.33 and includes 22 exons encoding 858 residues of protein [6]. In cone photoreceptors, PDE6 complexes are formed by two catalytic subunits of PDE6C (~99.1 kDa) (α’) and two identical inhibitory subunits of PDE6H(γ) (~9 kDa) (PDE62α’2γ). PDE6C is one of the six known genes causative of achromatopsia (ACHM) [7], inherited as an autosomal recessive trait but characterized by loss of cone photoreceptor function and affecting approximately 1 in 30,000 individuals worldwide [8]. Patients with ACHM present low visual acuity since birth or early infancy, photosensitivity, and poor color discrimination [9].

Both PDE6A and PDE6C share a conserved structure comprising one catalytic and two GAF domains (acronym derived from cyclic guanosine monophosphate (cGMP)-activated PDEs, adenylyl cyclase, and Fh1A): GAF-1 (73–222 aminoacids in PDE6A or 75–224 amino acids in PDE6C), which contains the cGMP binding site and the photoreceptor outer segment (POS) localization signal, and GAF-2 (254–431 amino acids in PDE6A or 256–433 amino acids in PDE6C), where the regulatory subunit γ is bound controlling the catalytic activity of the enzyme [10,11]. The catalytic domain is called cyclic nucleotide phosphodiesterase (PDEase) (483–816 amino acids in PDE6A or 486–819 amino acids in PDE6C) and mediates the light-activated hydrolysis of cGMP (data obtained from UniProt, 2026) [12].

The PDE6 complex is a key component of the phototransduction cascade in rod and cone photoreceptors [13]. Its primary function is to selectively hydrolyze cGMP into guanosine monophosphate (GMP). The decrease in cGMP levels triggers the closure of cyclic-nucleotide-gated (CNG) cation channels, thereby limiting Ca²⁺ and Na⁺ influx, which leads to membrane hyperpolarization and ultimately initiates signal transmission to secondary neurons. Conversely, excessive accumulation of cGMP can result in the persistent opening of CNG channels, causing continuous and pathological Ca²⁺ entry, which may culminate in photoreceptor cell degeneration [14,15].

In the past decade, induced pluripotent stem cell (iPSC)-derived retinal organoids (ROs) have emerged as a powerful model to study retinal development and disease mechanisms [16]. These three-dimensional structures are capable of recapitulating human retinas until the final stages and express all the retinal cell types and neuronal retinal markers [17]. Disease modeling using iPSC-derived ROs has provided critical insights into the molecular and cellular mechanisms underlying IRDs. For instance, Gao et al., 2020, generated ROs from an RP harboring a homozygous mutation in PDE6B, successfully modeling relevant pathological features in vitro [17]. Strikingly, PDE6A- and PDE6C-associated retinal dystrophies have not yet been modeled using this approach.

In the present study, we aimed to generate and characterize iPSC-ROs from a patient with RP carrying compound heterozygous PDE6A mutations, as well as ROs from iPSCs obtained from a patient with ACHM harboring a homozygous PDE6C mutation. In parallel, we used established isogenic models to analyze genotype–phenotype correlations and to investigate the molecular mechanisms underlying PDE6A- and PDE6C-associated pathogenesis. Notably, while the mutant PDE6A line exhibited impaired neuroretinal vesicle formation, a subset of RP-derived ROs matured appropriately and preserved functional features. In the case of the PDE6C-associated achromatopsia model, the mutant ROs exhibited normal structural maturation, although the homozygous PDE6C pathogenic variant likely impaired efficient cGMP hydrolysis within the photoreceptors. Together, those RO models offer a promising and robust approach for investigating the pathogenic mechanisms underlying IRDs.

2. Materials and Methods

2.1. Clinical Data

Both patients were clinically diagnosed at the Institut de Microcirurgia Ocular (Barcelona, Spain) based on standard ophthalmic evaluations. Patients with a suspected inherited retinal disease underwent multimodal imaging (retinography, autofluorescence retinography and optical coherence tomography), and both patients also underwent visual field testing. Our first patient was diagnosed with retinitis pigmentosa because of his symptoms and clinical findings. He reported nyctalopia and peripheral visual field limitation and a fundus examination revealed the typical findings of RP: mid-peripheral pigment disturbances with some scattered bone-spicule pigmentation and sparing of the macular area, waxy pallor of the optic disk, and attenuation of retinal vessels. Autofluorescence retinography showed patchy hypoautofluorescence at mid-periphery and central perimacular hyperautofluorescence in both eyes. Optical coherence tomography showed preserved external retinal layers at the foveal area, with perifoveal retinal atrophy. Visual field testing revealed peripheral perimetric constriction, with preserved visual field in the central 5° in both eyes. The second patient was diagnosed with achromatopsia due to the presence of low visual acuity, color vision deficiency and nystagmus since early childhood. Fundus examination revealed central macular pigmentary atrophy, no perimacular flecks, and no other signs of peripheral retinal involvement. Autofluorescence retinography showed marked hypoautofluorescence in the macular area, with no signs of peripheral involvement. Visual field testing confirmed the presence of a central scotoma in both eyes, but no involvement of the peripheral visual field. Optical coherence tomography showed foveal atrophy in both eyes.

2.2. CRISPR/Cas9 Gene Editing and iPSC Culture

The generation of FRIMOi001-A (PDE6A_mutant) and FRIMOi007-A (PDE6C_mutant) human iPSC lines was described in Riera et al., 2019, and Domingo-Prim et al., 2019, respectively [18,19]. Isogenic iPSC lines were established through CRISPR/Cas9-mediated correction as described in Siles et al., 2025, and Siles et al., 2023, respectively [20,21].

Briefly, sgRNAs and ssODNs were designed using the Invitrogen TrueDesign Genome Editor (Thermo Fisher Scientific, Waltham, MA, USA) and are the following: PDE6A sgRNA 5′-3′TCAGTCTTGACTCAATTTCT, PDE6A ssODN 5′-3′GTCTCTGATTCCTCCTTTCTTTCTTTTTCAGTCTTTGACTCAATTTCTCGGCTGGTCTGTCTTAAATCCTGACACCTAT, PDE6C sgRNA 5′–3′ AATTGTGGTAAGTGACAGCT, and PDE6C ssODN 5′–3′ TACCAGATGGATGTACACTGTGAGGAAAGGGTATCGAGCTGTCACTTACCACAATTGGCGGCATGGGTTCAACG. For gene editing, 1 × 10⁵ iPSCs were electroporated (Neon, Thermo Fisher Scientific) with two pulses of 20 ms at 1200 V for transfection with 10 pmol sgRNA, 15 pmol ssODN, and 10 pmol High-Fidelity (HiFi) SpCas9 protein (Thermo Fisher Scientific) as ribonucleotide particles. The iPSCs were seeded onto Matrigel-coated dishes and cultured in StemFlex medium supplemented with 10 μM ROCK inhibitor (Merck, Bedford, MA, USA), 10 μM of the HDR activator L755507 (Merck), and 0.5 μM of NHEJ inhibitor M3814 (Selleckchem, Houston, TX, USA) for 24 h. Individual clones were picked and cultured individually in 96-multiwell plates and Sanger-sequenced (Macrogen, Madrid, Spain) to assess gene editing outcomes. The iPSCs were routinely maintained on Matrigel-coated dishes (Merck) in Stem Flex Medium (Thermo Fisher Scientific) and were renewed with fresh medium every two days. Passages were performed with 0.5 mM EDTA (Invitrogen, Carlsbad, CA, USA).

2.3. Characterization of Human iPSCs

Mutant and corrected clones were characterized to ensure genomic stability and proper differentiation potential.

For lineage differentiation, the cells were seeded into Matrigel-coated plates, and when the desired confluence was reached, the cells were cultured in differentiation medium using the Human Pluripotent Stem Cell Functional Identification Kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s instructions. Karyotype analysis was performed on twenty G-banded metaphase cells at 300-band resolution (Reference Laboratory S.A., Barcelona, Spain).

2.4. Differentiation of iPSC into Retinal Organoid

ROs were generated using two distinct 2D-3D differentiation protocols. In order to study the rod-specific protein PDE6A, the ROs were produced following the protocol described by Sanjurjo-Soriano et al. 2022 [22]. For the cone-specific studies of PDE6C, ROs were obtained using the protocol described by Gonzalez-Cordero et al. 2017 [23]. Briefly, the ROs were generated from the iPSCs. To minimize variability between differentiations and lines, we carefully standardized the initial seeding conditions: all differentiations were initiated at 90% confluence, using the same cell density for each iPSC line, ensuring consistent starting conditions across all experiments. This resulted in the formation of neuroretinal vesicles (NRVs), which were manually isolated between weeks 4 and 7 and cultured individually in low-attachment plates under free-floating conditions. Organoid maturation was achieved through stage-specific supplementation with fetal bovine serum (FBS), taurine, N-2 supplement, B-27 supplement, and retinoic acid (RA), with protocol-specific differences in the timing of supplementation and proneural induction steps. Between three and four independent RO differentiations were carried out.

2.5. Quantification of NRVs, Non-Retinal Regions, and Retinal Organoids

Structures were classified as NRVs or non-retinal regions based on predefined morphological criteria [23]. NRVs were identified as well-organized, phase-bright vesicular structures with a defined border and characteristic neuroepithelial-like morphology. In contrast, non-retinal regions were defined as irregularly shaped, poorly organized structures lacking a defined border and typical NRV morphology.

NRVs and non-retinal regions were quantified by manual counting from culture images acquired at each time point. For each cell line and time point, 6–10 randomly selected fields per differentiation were analyzed. Three independent differentiations were included for each of the two mutant lines and the corrected line, and two independent differentiations for the control line.

The total number of structures per field was recorded, and the proportion of each category was calculated. For each cell line and time point, data were expressed as the percentage of NRVs and non-retinal regions relative to the total number of structures (NRVs + non-retinal aggregates), which was set to 100%. The results were represented graphically, with NRVs shown in gray and non-retinal regions in white. All the analyses were performed using consistent criteria across conditions.

The size of the NRVs was measured using ImageJ software 1.53k (NIH, Bethesda, MD, USA). For each vesicle, the longest axis (diagonal) was traced and recorded as the diameter. Measurements were performed using consistent criteria across all conditions. A total of 20–100 NRVs per line per time point were measured, pooled from 6 to 10 randomly selected fields across 2–3 independent differentiations.

ROs exhibiting POS or lacking POS were quantified by manual counting from culture images acquired at day 245 of differentiation. Thirty organoids per cell line were analyzed. The data were expressed as percentages of POS-positive (POS+) and POS-negative (POS−) organoids (total set to 100%). The results were plotted with POS+ organoids shown in gray and POS− organoids in white. Analyses were performed using consistent criteria across all conditions.

2.6. Organoid Fixation and Cryosectioning

The ROs were rinsed twice with phosphate-buffered saline (PBS, Thermo Fisher Scientific) and fixed in 4% paraformaldehyde (PFA, Thermo Fisher Scientific) for 15 min at 4 °C. After fixation, the organoids were washed three times with PBS, and incubated in 30% sucrose (in PBS) at 4 °C. On the following day, the ROs were pre-stained with trypan blue (20 min) to aid visualization during cryosectioning. Embedding molds were embedded in O.C.T. mounting media (VWR International, Radnor, PA, USA), placed on dry ice, and stored at −80 °C. Cryosections were prepared at 14 µm thickness using a cryostat.

2.7. Immunostaining and Quantitative Analysis

The iPSCs were fixed in 4% PFA in PBS for 15 min at 4 °C and washed twice with PBS. The RO cryosections and fixed iPSCs were permeabilized with 0.25% Triton X-100 (Merck) (15 min, room temperature, RT), and washed and blocked with 5% FBS and 0.5% Tween-20 (Thermo Fisher Scientific) in PBS for 1 h at RT. Primary antibodies (Table S1) were incubated overnight at 4 °C in the same blocking solution. The day after, the samples were washed twice with PBS and incubated with secondary antibodies (Table S1) for 1–2 h at RT and counterstained with DAPI (Thermo Fisher Scientific). Immunofluorescence imaging was acquired on a Zeiss Axiovert microscope equipped with an Axiocam 503 mono camera (Carl Zeiss Inc., Jena, Germany), and images were processed using ImageJ software 1.53k (NIH, Bethesda, MD, USA).

Quantitative analysis was performed by defining regions of interest corresponding to the inner segments/outer segments (IS/OS), outer nuclear layer (ONL), and inner nuclear layer (INL), based on morphological features and the layer borders delineated by dashed lines in the figures. Measurements were performed on at least three independent samples per staining, ensuring consistency across experimental conditions. Mean fluorescence intensity was measured for each region using identical acquisition and analysis settings across all experimental conditions.

2.8. RT-PCR and qPCR

Total RNA from 1 to 2 ROs or from iPSC samples was isolated using TRIzol (Thermo Fisher Scientific) following the manufacturer’s instructions. RNA concentration was determined using a Qubit 3.0 fluorometer. Each RNA sample consisted of 1–2 pooled organoids representing one biological replicate from independent differentiations. First-strand cDNA was synthesized from 0.25 µg of total RNA with the Transcriptor First Stand cDNA Synthesis Kit (Roche Diagnostics, Basel, Switzerland). Commercial retinal cDNA acquired from a 77-year-old human donor was sourced from BioChain Institute Inc. (Newark, CA, USA). Relative quantification was performed using TaqMan Gene Expression master mix in QuantStudio 3 (Thermo Fisher Scientific). Three PDE6A TaqMan probes were used to detect the region spanning exons 4–5 (Hs00959619_m1), the exon 2–3 boundary (Hs00959617_g1), and the exon 9–10 junction (Hs00959624_m1). These probes are referred to throughout the text as Probes A, B, and C, respectively. Gene expression values were normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH, Hs02758991_g1) using the ΔΔCt (delta Ct) method.

2.9. Retinal Organoid Protein Quantification and Cyclic GMP Measurement

RO protein was extracted with 0.1 M HCl and homogenized by vortex agitation. The samples were incubated during 20 min at 4 °C, then centrifuged for 10 min, 1000× g, and the supernatant was quantified using the Pierce Detergent Compatible Bradford Assay Kit (Thermo Fisher Scientific), following the manufacturer’s instructions. The same supernatant was then used for cGMP quantification using the cGMP Direct Immunoassay Kit (Abcam, Cambridge, UK), following the manufacturer’s instructions. Briefly, the assay is based on a competitive ELISA in which cGMP extracted from an RO competes with an HRP-conjugated cGMP tracer for binding to a cGMP-specific antibody. Upon the addition of a TMB substrate, HRP catalyzes its oxidation, generating a colorimetric signal quantified by measuring absorbance at 450 nm. Therefore, the OD450 nm intensity is directly proportional to the amount of cGMP-HRP and inversely proportional to the cGMP levels in the RO. The samples were acetylated prior to absorbance measurement using the Varioskan™ LUX multimode microplate reader (Thermo Fisher Scientific). cGMP concentrations were calculated from a standard curve generated using known cGMP standards, with background subtraction applied. cGMP levels were normalized to total protein content. The data are expressed as mean values ± SD from one experiment including three independent samples, each derived from two independent differentiations and measured in technical triplicates. Each sample contained at least three ROs.

2.10. DynaMut2 Prediction

DynaMut2 was employed to evaluate the structural and dynamic effects of the identified pathogenic variants in PDE6A and PDE6C (using the DynaMut2 web server, version 2.0).

2.11. Statistical Analysis

Statistical analysis was performed using the non-parametric Kruskal–Wallis test to compare patients, isogenic controls, and control lines with Prism 9.3.1 (Graph Pad Software, La Jolla, CA, USA). A statistically significant difference was defined as p < 0.05 (ns p > 0.05). The results were expressed as mean ± SD.

3. Results

3.1. Modeling PDE6 Mutations Using Patient-Derived Retinal Organoids and Isogenic Cell Lines

Our first patient was a 51-year-old male diagnosed of retinitis pigmentosa at the age of 24, presenting with nyctalopia and peripheral visual field loss (see Section 2.1 for detailed clinical characterization). The patient has shown a very slow disease progression over the past 10 years, with a visual field defect that is mostly unchanged, mild worsening of the hypoautofluorescence pattern, and best corrected visual acuity of 20/25 in both eyes, similar to the findings 10 years prior (Figure 1A).

In 2016, the RP patient was genetically diagnosed at our laboratory as a compound heterozygote carrier of two pathogenic variants in PDE6A (NM_000440.3): c.305G>A (p.Arg102His) [24] and c.1268delT (p.Leu423Ter) [25] (Figure 1B). The c.305G>A variant (rs750539462) is a missense mutation where Arginine has changed to Histidine in position 102 of the PDE6A protein. The p.Arg102His variant affects a highly conserved residue across different species (phyloP: 7.89 [−19.0, 11.0]) (Figure 1C). Bioinformatics analysis suggested that this change is likely pathogenic, and DynaMut2 predicted that its effect destabilized the folding of the PDE6A protein (Gibbs Free Energy (ΔΔG), −1.16 kcal/mol), due to the loss of interactions (Figure 1D). In turn, the c.1268delT variant is a thymine deletion that alters the reading frame and introduces a premature stop codon at the end of the GAF2 domain, located in the middle of the protein (Figure 1B). The establishment and characterization of the iPSC line derived from this patient, referred to as FRIMOi001-A, was carried out by Riera et al. 2019 [18].

The second patient included in this study was a 42-year-old male diagnosed with achromatopsia, characterized by reduced visual acuity, color vision deficiency, and nystagmus since early childhood (see Section 2.1 for full clinical details). Clinical findings have remained stable over time (Figure 1E).

The genetic diagnosis performed in our laboratory revealed that the ACHM patient carried the homozygous missense mutation c.1670G>A (p.Arg557Gln) in PDE6C (Figure 1F) [19]. Arginine 557 is located in the catalytic domain and shows high conservation across several vertebrate species (Figure 1G). It interacts with Glutamate 454 (Glu454) and Asparagine 458 (Asn458) located in the other catalytic subunit of PDE6C. The change in Gibbs Free Energy (ΔΔG) of the folding due to the mutation predicted by DynaMut2 was −0.77 kcal/mol, which suggested a destabilizing effect on protein structure (Figure 1H). The generation of this iPSC line, referred to as FRIMOi007-A, was previously described by Domingo-Prim et al. 2019 [19].

As the RP case had an autosomal recessive condition, the correction of a single pathogenic PDE6A variant is sufficient to revert the disease-associated phenotype and generate an isogenic control. The FRIMOi001-A line, hereafter referred to as PDE6A_mutant, was edited by correcting the c.1268delT single nucleotide deletion using CRISPR/Cas9 technology [20] and is referred to as PDE6A_corrected line. Both iPSC lines retained pluripotency, differentiated into the three embryonic germ layers, and maintained normal karyotypes after gene editing (Figure S1).

In the ACHM case, also an autosomal recessive disorder, the FRIMOi007-A iPSC line, hereafter referred to as PDE6C_mutant, was edited using CRISPR/Cas9 technology [21]. Two distinct corrected clones were generated: one with both alleles corrected, PDE6C_corr/Homo, and a heterozygous clone, PDE6C_corr/Het, in which only one allele was reverted.

3.2. PDE6A Patient-Specific Mutations Impair Early Retinal Organoid Development

In recent years, numerous studies have employed ROs as the model to investigate retinal diseases [17,26]. In order to study the PDE6A-associated phenotype, we followed a protocol described by Sanjurjo-Soriano et al. 2022 [22], that generates a higher proportion of rod photoreceptors and is more suitable for investigating the rod-specific protein PDE6A (Figure 2A (upper panel)).

During the first weeks of differentiation, clusters of lightly pigmented RPE cells start to emerge and subsequently give rise to optic vesicle-like structures containing putative neuroretinal vesicles (NRVs) characterized by the retinal neuroepithelium. Importantly, in PDE6A_mutant cultures, higher amounts of non-retinal regions exhibiting forebrain-like neuroepithelial morphology were found compared to wild-type. Conversely, the corrected line showed a similar behavior to the control line (Figure 2A (lower panel)). The NRVs and non-retinal regions are indicated by yellow and white arrows, respectively, in Figure 2A. Accordingly to these observations, the quantification of NRVs and non-retinal regions revealed an increased proportion of non-retinal structures in PDE6A_mutant cultures compared to the control and corrected lines (Figure 2B).

To exclude the possibility of intrinsic variability of the selected patient-derived colony, another iPSC colony from the patient (PDE6A_mutant.2) was included in the analysis, showing the same results found in the initial PDE6A_mutant line (Figure 2A,B).

The NRVs increased in size over time in all examined lines. However, the quantification of the diameter length (µm) showed that mutant lines exhibited significantly smaller NRVs at all analyzed time points (days 15, 20, 25 and 30), compared with the corrected line (Figure 2C). Notably, the NRVs expressed retinal progenitor markers indicating appropriate specification toward retinal lineages in all analyzed lines: PAX6 (a marker of early retinal cell fate), MITF (a transcription factor involved in early retinal/optic-vesicle specification), SOX2 (a neural progenitor marker), and ZO-1 (an epithelial tight junction marker) (Figure S2A,B). Additionally, to confirm the absence of retinal cell fate impairment, mutant and corrected iPSC lines were successfully differentiated into mature retinal pigment epithelium (RPE) cells following an established protocol [27] (Figure S2C).

Between days 28 and 30 of differentiation, NRVs from all lines were manually dissected from the differentiation cultures and grown in suspension. By contrast, in the mutant condition, some NRVs required an additional 1–2 weeks before dissection, indicating a delayed developmental progression. Later, POS, defined as pronounced brush-like protrusions emerging from the apical border of the neuroepithelium [22,23], started to appear as a distinctive feature of advanced RO maturation, and a sign that the ROs had reached a relevant functional stage (Figure 2D). Accordingly, after 245 days of differentiation, both the control and corrected lines efficiently generated mature ROs, the majority of which exhibited well-defined POS (yellow arrows in Figure 2D) and properly organized retinal layers. Conversely, a subset of mutant ROs (approximately two-thirds) lacked discernible POS structures (red arrow in Figure 2D), whereas the remaining one-third successfully developed POS (yellow arrows in Figure 2D), as quantified in Figure S3A, and were subsequently characterized. The genetic identification of the derived ROs was confirmed, demonstrating the presence of the pathogenic PDE6A variants in the patient-derived ROs, whereas the isogenic control showed the correction of the c.1268delT variant (Figure S3B).

Those PDE6A_mutant ROs which developed POSs exhibited similar structure and morphology in comparison to the control and isogenic ROs. The expression of Rhodopsin (RHO), the visual pigment characteristic of rod photoreceptors; guanine nucleotide-binding protein G(t) subunit alpha-1 (GNAT1), which encodes the α-subunit of rod transducin involved in phototransduction; and arrestin-3 (ARR3), a marker specifically associated with cone photoreceptors was also confirmed (Figure 2E). Similarly, PDE6A expression in the mutant ROs was localized to the rod outer segments, corresponding to the outermost region of the ROs, as observed in the isogenic control ROs (Figure 2F). This localization was supported by fluorescence intensity quantification (Figure 2G).

The PDE6 complex functions as the primary enzyme responsible for hydrolyzing cGMP into GMP in photoreceptors. When this reaction proceeds correctly, cytosolic cGMP levels decrease, leading to the closure of cyclic-nucleotide-gated (CNG) cation channels. Impairment of the PDE6 complex disrupts cGMP hydrolysis, leading to cGMP accumulation, which can trigger cellular stress and ultimately photoreceptor degeneration [14,15]. This pathway is particularly relevant for assessing PDE6A functionality in rod photoreceptors.

cGMP levels were quantified using the direct immunoassay kit, following the manufacturer’s instructions (see Section 2.9 for the detailed method and experimental principles; Figure 3A). Quantification was performed in one experiment including three independent samples (each containing at least three ROs), derived from two independent differentiations and measured in technical triplicates. In this regard, we analyzed the concentration of cGMP in the PDE6A_mutant ROs, and surprisingly the cGMP levels were comparable to that observed in the control and corrected ROs (Figure 3B). Furthermore, the cGMP distribution was examined by immunofluorescence, revealing proper localization within the POSs across all analyzed organoids, indicating a functional cGMP pathway in the PDE6A_mutant ROs (Figure 3C). Importantly, these findings were supported by quantitative fluorescence analysis (Figure 3D).

3.3. Analysis of PDE6A Isoforms Revealed Developmental Stage-Dependent Differences

According to Ensembl and UCSC annotations, the PDE6A gene produces three protein-coding isoforms (

Table 1. Summary of PDE6A protein-coding isoforms, including transcript IDs, coding sequence (CDS) length, number of exons, predicted missing exons, protein length, and probe detection.

Name	Transcript ID	CDS Length (bp)	No. of Exons	Missing Exons	Protein Length (aa)	Probe Detection
PDE6A-1	ENST00000255266.10; NM_000440.3	2583	22	-	860	A, B, C
PDE6A-2	ENST00000613228.1	2340	20	2 and 3	779	A, C
PDE6A-3	ENST00000890426.1	2439	21	10	812	A, B

). The canonical full-length transcript, designated as PDE6A-1, contains all 22 coding exons. The two remaining isoforms, PDE6A-2 (skipping exons 2 and 3) and PDE6A-3 (skipping exon 10), correspond to alternative splice isoforms that encode shorter protein products (Figure 3D).

Previous work from our laboratory showed that ROs at around day 350 of differentiation exhibited a gene expression profile closely resembling that of the human adult retina (77-year-old donor) for specific IRD-related genes [28]. Based on this, the expression of PDE6A isoforms was examined in wild-type ROs at two earlier maturation stages (days 125 and 210) and compared with the human adult retina. Quantitative PCR (qPCR) was performed using three PDE6A TaqMan probes. Notably, these TaqMan probes are not fully isoform-specific, and the detected signals represent the combined contribution of the isoforms containing the targeted regions rather than individual transcript levels. Specifically, Probe A (spanning exons 4 and 5) detects all three PDE6A isoforms, Probe B (exon 2–3 boundary) detects isoforms 1 and 3, and Probe C (exon 9–10 junction) detects isoforms 1 and 2 (Table 1; Figure 3E).

Relative mRNA analysis revealed that the total PDE6A mRNA was more than tenfold higher in the human adult retina than in the D125 and D250 ROs, as assessed using Probe A (which detects all isoforms) (Figure 3F). The signals detected by Probes B and C, which target subsets of PDE6A isoforms, were also higher in adult retina compared with ROs, but to a lesser extent than with Probe A. The smaller relative increase observed with Probes B and C suggests that specific isoforms may not scale proportionally with total PDE6A expression. Together, these results indicate the differential regulation of PDE6A transcript regions during development, potentially reflecting changes in isoform composition, pointing to higher levels of alternative splice isoforms in the adult retina.

We next examined the relationship between the observed isoform expression pattern and the genotype of the RP patient carrying two PDE6A variants. One variant affects exon 1, potentially impacting all three isoforms. The second variant is a thymine deletion in exon 10 that introduces a premature stop codon. As exon 10 is not included in isoform 3, this isoform is therefore expected to remain expressed from the unaffected allele (Figure 3G). Considering the differential impact of the patient’s variants across PDE6A isoforms and the stage-dependent expression pattern observed, these results suggest a stronger phenotypic effect of the variants at earlier developmental stages, with a reduced impact at later stages, when PDE6A-3 expression increases.

3.4. iPSC-Derived Retinal Organoids from ACHM Patient Exhibit Disrupted PDE6C Localization and cGMP Dysregulation

In order to analyze the impact of the homozygous pathogenic variant in PDE6C, the iPSC line derived from the ACHM patient, the isogenic controls and the wild-type lines were differentiated into ROs following the method described in Gonzalez-Cordero et al. 2017 [23] (Figure 4A (upper panel)), which generated ROs with an enriched population of cones. No differences were detected during the early stages of the differentiation process between the studied lines (Figure 4A (lower panel)). Moreover, all iPSC-derived ROs matured well, and exhibited similar morphology on day 245, with POS properly formed (as quantified in Figure S4A) and a well-defined laminar retinal structure (Figure 4A (lower panel)). Also, the presence or not of the PDE6C mutation was confirmed in the mutant and in the different isogenic ROs (Figure S4B).

By day 245 of differentiation, the mutant ROs showed protein expression comparable to their corrected counterparts for several key photoreceptor markers, such as rhodopsin (RHO), opsin (Opsin, L/M), arrestin-3 (ARR3), and PDE6H (phosphodiesterase 6H) (Figure 4B,C).

Importantly, some differences were found in the PDE6C expression pattern between the cell lines, as assessed by immunofluorescence (Figure 4D). Specifically, PDE6C exhibited a diffuse and spread distribution through the inner and outer segments of the photoreceptor (IS/OS) cells and through the multiple retinal layers (ONL and INL) in the mutant ROs, while in the isogenic ROs it was concentrated in the IS/OS region, as in the wild-type controls (Figure 4D). Notably, this observation was supported by quantitative fluorescence analysis, which confirmed the differential spatial distribution of PDE6C in the mutant versus control ROs (Figure 4E). This result suggests a possible mislocalization of this protein due to the homozygous pathogenic variant in PDE6C.

To further analyze if this defect in PDE6C could trigger photoreceptor dysfunction, we next assessed the cGMP distribution and concentration in RO photoreceptors. In this case, the PDE6C complex is the key enzyme responsible for hydrolyzing cGMP into GMP in cone photoreceptors. First, to investigate this pathway, cGMP distribution was examined by immunofluorescence using a specific anti-cGMP antibody in all the analyzed ROs. In the wild-type and isogenic ROs, cGMP was located appropriately within the POS. In contrast, the mutant ROs displayed a cGMP signal extending throughout the ONL and the INL, accompanied by a pronounced punctate staining pattern, consistent with cGMP accumulation (Figure 4F). Quantitative fluorescence analysis was performed to confirm these visual observations, revealing a clear difference in cGMP distribution between the mutant and both the control and isogenic ROs (Figure 4G).

cGMP levels in the ROs were quantified using the direct immunoassay kit (Figure 3A), based on one experiment including three independent samples derived from two independent differentiations and measured in technical triplicates (three ROs per sample). In line with the immunofluorescence results, the mutant ROs exhibited higher cGMP concentrations than the wild-type ROs. Notably, the cGMP levels in the isogenic control ROs were comparable to those of the wild-type ROs, indicating functional restoration following gene editing (Figure 4H).

Overall, these results suggest that the homozygous missense mutation in PDE6C, located in the catalytic domain, probably affects the stability of the protein and the hydrolysis of cGMP in cone photoreceptors. Consequently, elevated cGMP levels could maintain CNG channels in an open state, thereby driving continuous and pathological Ca²⁺ influx that may ultimately lead to photoreceptor degeneration.

4. Discussion

Despite the identification of numerous pathogenic genes associated with inherited retinal dystrophies, effective therapies for most retinal degenerations remain limited. In recent years, disease modeling has emerged as a powerful approach for investigating human retinal development and disease mechanisms. ROs recapitulate key aspects of human retinal architecture through the final stages of development and exhibit all major retinal cell types, expressing appropriate neuronal markers. In this work, two IRDs, retinitis pigmentosa and achromatopsia, were modeled using iPSC-derived ROs from patients carrying mutations in PDE6A and PDE6C, respectively. In parallel, phenotypic rescue was assessed using isogenic control lines generated through CRISPR/Cas9-mediated genome editing, allowing for the direct evaluation of mutation-specific effects.

The patient-specific PDE6A mutations were associated with altered early stages of retinal organoid development, as evidenced by impaired neuroretinal vesicle formation and reduced vesicle growth. Given that PDE6A expression is restricted to rod photoreceptors at later developmental stages, the mechanism underlying this early phenotype remains unresolved and may involve indirect effects on neuroepithelial organization. Although RO differentiation was initiated in all the studied lines, the mutant one exhibited an increased proportion of non-retinal, forebrain-like regions during early differentiation. Importantly, this phenotype was consistently observed in a second patient-derived iPSC clone and was rescued upon CRISPR/Cas9-mediated correction, demonstrating that the observed defects were mutation-dependent rather than attributable to intrinsic properties of the patient iPSCs. Despite these early developmental alterations in mutant lines, both the mutant and control lines expressed key retinal progenitor markers and retained the capacity to differentiate into mature RPE cells. These findings indicate that PDE6A mutations are associated with early developmental differences in RO organization.

Regarding the mature mutant ROs, a subset of them did not properly develop POSs, whereas the rest successfully generated them, exhibiting normal photoreceptor morphology, PDE6A expression and localization, and cGMP levels, compared to the controls. Notably, the presence of functionally competent ROs within the mutant population may help explain the clinical stability observed in the patient during adulthood.

To further contextualize the phenotypic variability observed in the mature ROs, and in light of the early developmental alterations observed in the PDE6A_mutant organoids, we examined the developmental regulation of the PDE6A transcript isoforms. The analysis revealed shifts in isoform expression during organoid maturation, with alternative splice isoforms (non-canonical) becoming more prominent at later maturation stages. In the RP cell line analyzed here, the compound heterozygous variants affecting exons 1 and 10 are predicted to impact multiple PDE6A isoforms, whereas the skipping of exon 10 in the PDE6A-3 isoform may preserve residual protein function. These findings may suggest that PDE6A mutations primarily affect early retinal development, consistent with the defects in NRV formation and limited vesicle growth. In contrast, the partial preservation observed at later stages in a subset of mature mutant ROs may reflect developmental changes in PDE6A isoform expression and could contribute to phenotypic variability. Consistent with this, we did not observe a marked accumulation of cGMP in the PDE6A_mutant ROs, in contrast to the animal models of complete PDE6A loss, in which retinal cGMP levels are significantly elevated in affected retinas and photoreceptor degeneration is linked to cGMP dysregulation [29]. One plausible explanation for this difference is that residual PDE6A activity, potentially supported by exon-skipping transcripts and partial function of the missense allele, is sufficient in this system to maintain cGMP homeostasis despite other aspects of dysfunction. These considerations suggest that variant-dependent residual function and isoform expression patterns can influence both morphological and biochemical phenotypes in human retinal organoids.

Importantly, the potential for residual PDE6A function underscores the relevance of developmental timing and isoform-specific expression for therapeutic interventions, as strategies targeting canonical versus alternative isoforms may have differential efficacy depending on the stage of the disease.

In the PDE6C-associated achromatopsia RO model, the mutant and control cell lines were differentiated efficiently into mature ROs and expressed comparable levels of rod- and cone-specific markers. However, the mutant ROs exhibited a pronounced mislocalization of PDE6C, with the protein abnormally distributed throughout multiple retinal layers rather than being restricted to the inner and outer segments, as observed in the corrected and wild-type ROs. Functionally, this mislocalization was associated with elevated and abnormally distributed cGMP levels, consistent with the reduced catalytic efficiency of PDE6C. Prolonged cGMP accumulation is expected to maintain CNG channels in an open state, promoting pathological Ca²⁺ influx and thereby creating a toxic intracellular environment that may predispose cones to degeneration [30].

Importantly, the correction of the PDE6C mutation in the isogenic clones restored proper protein localization and normalized cGMP distribution, closely resembling the wild-type organoids. Notably, the rescue of PDE6C function was achieved upon monoallelic correction, consistent with the autosomal recessive inheritance of achromatopsia and indicating that a single functional PDE6C allele is sufficient to sustain cone photoreceptor function.

Previous studies using PDE6A- and PDE6C-mutant animal models have reported therapeutic benefits following gene replacement. Notably, in canine models of PDE6A-linked retinitis pigmentosa, gene supplementation using AAV vectors partially restored retinal structure and function [31,32]. More recently, subretinal delivery of AAV8.hPDE6A was evaluated in patients with PDE6A-associated RP [33,34], although no functional improvements were detected within the first year of follow-up. These findings suggest that isoform selection and the timing of intervention may critically influence therapeutic efficacy. For PDE6C, a spontaneous non-human primate achromatopsia model carrying a missense mutation recapitulates key features of ACHM, including photophobia-like behavior and macular abnormalities [35]. Although the mutant PDE6C protein was expressed and correctly localized, it failed to hydrolyze cGMP. Notably, recent AAV-mediated gene supplementation partially restored cone function in this model [36], highlighting the translational potential of gene-based approaches for PDE6C-associated disease.

Despite advances made in animal models, patient-derived ROs provide a complementary and indispensable platform for translational research. This approach preserves the complete genetic background of each patient, and the use of isogenic controls enables the direct attribution of observed phenotypes to specific pathogenic variants. Furthermore, standardized protocols for RO generation are increasingly available. Notably, the successful phenotypic rescue observed in the isogenic control lines underscores the promise of gene editing strategies as potential therapeutic approaches for these patients.

In conclusion, our findings directly link the disease-associated phenotype to specific PDE6A or PDE6C variants, underscoring the value of retinal organoids to elucidate disease mechanisms and advancing personalized therapeutic strategies for inherited retinal dystrophies.

5. Conclusions

The functional impact of PDE6A mutations appears to be modulated by developmental stage-dependent transcript usage, with early retinal development being primarily affected and later maturation-associated isoform diversity potentially mitigating phenotypic severity and contributing to variability among organoids. In contrast, PDE6C mutations disrupt PDE6C localization and lead to abnormal cGMP accumulation in cones, whereas monoallelic correction restores proper localization and function, supporting the potential of gene-targeted therapies for achromatopsia. Collectively, these findings demonstrate that patient-derived retinal organoids combined with isogenic controls provide a powerful platform to directly link disease phenotypes to specific pathogenic variants and to guide personalized therapeutic strategies for inherited retinal dystrophies.