Lab-grown retinal tissue from AIPL1 gene-edited cells shows key molecular signs of LCA4 disease

To establish a human, isogenic photoreceptor model of AIPL1-related LCA, we set out to generate a genotype-matched isogenic iPSC line from a healthy individual using CRISPR/Cas9 technology to promote loss of AIPL1 expression (Figure 1). First, we designed CRISPR/Cas9 guide sequences to target the early exon 1 of AIPL1 downstream of the starting codon (Figure S1A). A Cas9-mediated double-strand break (DSB) in this region is expected to induce indel formation by the non-homologous end joining (NHEJ) repair pathway, resulting in the frameshift of the coding region and consequent gene silencing. Four gRNAs selected by in silico analysis—AIPL1 g1.1, g1.2, g1.3 and g1.4—were designed to direct the Cas9 endonuclease to AIPL1 exon 1 (Figure S1A). The targeting ability of each guide was tested in the HEK293T cell line transfected with AIPL1 gRNA and Cas9-encoding constructs PX458 (Figure S1B). We observed that all guides effectively induced targeted indels in AIPL1 exon 1 as detected by the T7 endonuclease I assay (Figure S1C) with an approximate efficacy of 35% (Figure S1D, p < 0.0001). We selected AIPL1 g1.4 for the generation of AIPL1 knockout (KO) iPSC by simultaneous reprogramming and CRISPR/Cas9 gene editing of healthy human BJ dermal fibroblasts [24] (Figure 1A). The emergence of BJ-derived iPSC colonies was detected as early as 14 days post-reprogramming, which were then isolated and expanded within 28 days (Figure 1B). To screen for AIPL1-edited iPSC, the genotype of single cell-derived iPSC clones was analysed by Sanger sequencing. Of a total of 66 clones, homozygous AIPL1 KO was observed in 11 clones (16.7%), while heterozygous KO was observed in 3 clones (4.5%) and 52 clones (78.8%) were unedited (Figure 1C). We selected one iPSC clone with homozygous AIPL1 KO through a 1 bp insertion at the AIPL1 g1.4 target (Figure 1D), resulting in a missense mutation and frameshift leading to a premature stop codon within exon 2 (p.H18Qfs*24). Immunocytochemistry analysis validated the pluripotency of both AIPL1 wild type (WT) and AIPL1 KO lines, with the detection of OCT4, TRA-1-60, NANOG and SSEA4 markers (Figure S2A), and detection of trilineage potential toward germ layer differentiation (Figure S2B). Finally, we validated the specificity of the AIPL1 KO by sequencing the 10 most likely off-targets predicted in silico with the highest homology to the AIPL1 g1.4 target sequence. No undesirable off-target mutations were detected in these lines (Figure S2C).

Next, we sought to evaluate the phenotypic effect of the CRIPSR/Cas9-introduced AIPL1 premature translation termination mutation. For this, AIPL1 WT and KO iPSC were differentiated into three-dimensional retinal organoids (RO) using a two-dimensional to a three-dimensional protocol for generating iPSC-derived neural retinal vesicles (NRV) [19]. In both cell lines, NRV was observed after 20–25 days of culture. Following mechanical isolation, NRV increased in diameter and developed clear lamination, and around day 170 began displaying brush borders comprising presumptive photoreceptor structures (Figure 2A). Evaluation of AIPL1 protein abundance found consistent expression in the outer nuclear layer of mature WT RO, but not in the AIPL1 KO model (Figure 2B). AIPL1 was detected in rhodopsin-positive rod photoreceptors (Figure 2C) and cone arrestin-positive cone photoreceptors (Figure 2D) as early as week 24 (day 170) in AIPL1 WT RO, but not in AIPL1 KO RO at any time point.

Additional RO characterisation found no observable differences between AIPL1 WT and KO RO for the expression of photoreceptor recoverin or photoreceptor cilia (ARL13B) during RO maturation (Figure S3A), or in the expression of both L/M opsin and S opsin (Figure S3B). Finally, ONL thickness (Figure 2E) and cell apoptosis (Figure 2F) were measured and quantified in the ONL of AIPL1 WT and KO RO with no significant difference found between model types, indicating loss of AIPL1 had minimal impact on RO development and maturation. This was confirmed at the transcriptional level, where comparable expression patterns were observed for markers of photoreceptor development (CRX, NRL) and function (recoverin (RCVRN), rhodopsin (RHO), cone opsins (OPN1SW, OPN1MW/LW) and phosphodiesterase subunits (PDE6A, PDE6B, PDE6G, PDE6C, PDE6H)) as well as markers of retinal cell types, specifically retinal ganglion cells (BRN3B, ISLET1), bipolar cells (PKC-A), Müller glia (CRALBP) and amacrine and horizontal cells (PROX-1) (Figure S4). Moreover, AIPL1 transcription was retained in KO organoids, first detected in both WT and KO RO at 8 weeks with increased expression during RO maturation (Figure S4).

Downstream effects of AIPL1 protein loss were investigated in the AIPL1 WT and KO RO. Quantitative gene expression analysis was carried out for both AIPL1 and PDE6 subunits (Figure 3A), where expression was significantly increased over time, but no significant differences were measured between the WT and KO RO. Interestingly, a downward trend in AIPL1 expression was detected at each age examined in the AIPL1 KO compared to WT, but this did not reach significance. In addition, key proteins were measured in the whole RO for both WT and KO (Figure 3B). AIPL1 was consistently detected in WT RO but below the limit of detection in KO RO. Similarly, PDE6α was significantly reduced in KO RO (p = 0.032) compared to WT RO. Recoverin was used as a control for photoreceptor cells, and no difference was observed in recoverin levels between WT and KO RO, confirming that alterations in AIPL1 and PDE6α protein levels were unrelated to RO cellular makeup. The spatial location of rod-specific PDE6 enzymes (PDE6α and PDE6β) was also examined, where they were found to be abundant above the outer limiting membrane in WT but not in KO RO (Figure 3C). PDE6 enzymes function to hydrolyse cGMP within the retina, and the loss of these enzymes should lead to cGMP accumulation. When whole RO was quantified for cGMP concentrations, a significant increase (p = 0.041) was found in the AIPL1 KO model compared to WT (Figure 3D). Together, these data suggest AIPL1 KO RO manifest early molecular features of the LCA4 disease phenotype, including downstream effects on post-transcriptional levels of PDE6 and cGMP levels, and would be a suitable model to examine disease progression and treatments.

Human embryonic kidney 293T (HEK293T) cells (American Type Culture Collection; ATCC, Manassas, VA, EUA) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies, Carlsbad, CA, USA) supplemented with 10% (v/v) foetal bovine serum (FBS; Gibco, Loughborough, UK), 2 mM L-glutamine and 1% (v/v) penicillin–streptomycin (Gibco). Cells were routinely passaged using Trypsin 0.05%. Induced pluripotent stem cells (iPSC) were maintained on Geltrex-coated 6 well plates in mTeSR Plus media (Stemcell, Cambridge, UK). Cells were routinely passaged using Versene (Gibco). Cells were maintained at 37 °C in a humidified atmosphere of 5% CO₂.

CRISPR/Cas9 guide RNAs (gRNAs) were designed in silico to target the AIPL1 exon 1 using Benchling CRISPR design software (https://www.benchling.com↗). The top four gRNA candidates (Table S1) with optimal on-target efficacy and specificity were cloned into BbsI sites of pSpCas9(BB)-2A-GFP (PX458, Addgene plasmid #48138) vector encoding CBh-driven Cas9 nuclease and U6-driven gRNA. To validate AIPL1-targeted CRISPR/Cas9 sequences, HEK293T cells were transfected with PX458 encoding AIPL1-targeting gRNAs by Lipofectamine 2000 (Thermo Fisher Scientific, Loughborough, UK) according to manufacturer instructions. CRISPR/Cas9-mediated indel formation at the AIPL1 target sequence was assessed by T7 endonuclease I (T7EI) assay for detection of DNA heteroduplexes according to Ran et al. 2013 [45]. Primers used for cloning AIPL1 gRNAs and for amplification of AIPL1 exon 1 for T7EI assay are represented in Table S2.

AIPL1 KO isogenic iPSC were generated from control BJ dermal fibroblast line (ATCC CRL-2522) by simultaneous reprogramming and CRISPR/Cas9 knockout according to Howden et al. 2015 [24]. Briefly, BJ fibroblasts were nucleofected with reprogramming plasmids—pCXLE-hOct3/4-shp53-F (Addgene plasmid #27077), pCXLE-hSK (Addgene plasmid # 27078), pCXLE-hUL (Addgene plasmid #27080) and pSimple miR302/367 (Addgene plasmid #98748)—and PX458 constructs encoding AIPL1-targeted CRISPR/Cas9 sequences. Single-cell derived iPSC colonies were manually picked at approximately 25–28 days post-nucleofection and expanded for genotyping analysis. Genomic DNA from each iPSC clone was extracted using the QuickExtract reagent (Lucigen, Middleton, WI, USA) solution and screening of AIPL1 KO clones was performed by Sanger sequencing. Off-target analysis was performed by in silico prediction of the top 10 regions with highest homology to the AIPL1-targeting sgRNA using Benchling software (https://www.benchling.com↗) and Sanger sequencing to evaluate the presence of undesirable mutations caused by CRISPR/Cas9 off-target activity. Retention of iPSC potential for trilineage differentiation following CRISPR/Cas9 treatment was confirmed using the STEMdiff™ Trilineage Differentiation Kit (Stemcell Technologies, Cambridge, UK) as per manufacturer’s instructions with RNA assessed for Oct4 (pluripotency), Pax6, Nestin and Otx2 (Ectoderm), T and Cdx2 (Mesoderm), and AFP, Sox17 and Gata6 (Endoderm) to confirm differentiation to three germ lineages.

Directed differentiation of iPSCs into 3D RO was carried out as previously described [19,21]. iPSCs were dissociated with Versene (Gibco, Loughborough, UK), clumps were collected, washed twice with PBS, resuspended in culture medium, and seeded onto Geltrex (Gibco) coated 6-well plates. iPSC colonies were grown until 90–95% confluent, at which point they were cultured in Essential 6TM media (Gibco) for 2 days (Day 1 and 2 of differentiation) then neural induction media (Advanced DMEM/F12 (1:1), 1% non-essential amino acids (NEAA), 1% N2 supplement, 1% GlutaMAX, 1% antibiotic–antimycotic (All Gibco)) until development of neuro-retinal vesicles (NRVs), typically between day 28 and 42. NRVs were manually excised using 21G needles/scalpel blades and grown in low-binding 96-well plates (96-well Nunc Sphera Round Bottom Plates, Thermo Fisher Scientific) in retinal differentiation media (DMEM/F12 nutrient mix (3:1 ratio; Gibco), 10% foetal bovine serum (FBS; Gibco), 2% B27 supplement (without vitamin A; Gibco), 100 μM taurine (Tocris, Abingdon, UK), 2 mM GlutaMAX (Gibco, Loughborough, UK) and 100 U/mL penicillin/streptomycin (Gibco, Loughborough, UK)), with media changes every 2 days. At day 70, cultures were supplemented with 1 μM retinoic acid, and ROs were transferred into low-binding 24-well plates. At day 84, the cultures were supplemented with 1% N2 and RA lowered to 0.5 μM. At day 100, B27 and RA were removed from the medium.

ROs were fixed in 4% PFA, 5% sucrose in PBS for 30 min at 4 °C then dehydrated for 1 h each in 6.25%, 12.5% and 25% sucrose:PBS at 4 °C, ROs were embedded in Tissue-Tek OCT compound and stored at −80 °C until sectioned. The amount of 7 µm cryosections was mounted on SuperFrost PlusTM slides (Thermo Fisher Scientific) and stored at −20 °C. Slides were stained for ICC by first incubating in block buffer (10% donkey serum (Sigma-Aldrich, Gillingham, UK), 0.01% Triton-X (Sigma-Aldrich, Gillingham, UK) in PBS) for 1 hr at room temperature (RT), then incubating with a primary antibody for 1 h (RT; antibody-specific dilutions;). Slides were washed 3 times with PBS and incubated with species-specific secondary antibodies for 45 min (1:1000;). If required, slides were then incubated with stains () for 20 min before further incubation with 4,6-diamidino-2-phenylindole (DAPI; 2 mg/mL) (Invitrogen, Loughborough, UK) in PBS for 5 min, then washed 3 times with PBS. Slides were dried at RT and mounted in fluorescence mounting media (Dako; Agilent, CA, USA). Slides used for TUNEL assay (In Situ Cell Death Detection Kit; Merck) were stained as per kit guidelines, then counterstained with DAPI and mounted as per the above protocol. All images were acquired using a Leica Stellaris 5 laser-scanning confocal microscope. Images were prepared using Adobe Photoshop, Image J and Adobe Illustrator CS6. Table S3 Table S3 Table S3

ONL thickness was calculated using ImageJ by taking 10 unique measurements from 3 DAPI-stained RO. TUNEL positive cells were quantified using low magnification images where ONL length was calculated using ImageJ and positive cells were manually counted. Statistical significance was determined by unpaired Student’s t test, where * denotes a p value < 0.05.

RNA from iPSC and ROs was extracted using the PicoPure RNA Isolation Kit (Thermo Fisher Scientific). cDNA synthesis was performed using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). 2× GoTaq Green master mix (Promega, Southampton, UK) was used for DNA amplification by PCR with standard cycling conditions for semi-quantitative PCRs. Real-time PCR reactions were set up with 2× LabTaq Green Hi Rox Master Mix (Labtech, East Sussex, UK) and validated primers (0.25 pmol/uL; sequences detailed in Table S4) and run on an Applied Biosystems QuantStudioTM 6 Flex real-time PCR system. Gene expression levels were normalised to CRX as a consistently expressed PR-specific reference gene to negate differences in size and cellular makeup between samples. Three biological replicates per sample were used to calculate averages and standard deviation, and statistical significance was determined by one-way ANOVA where *, **, *** denote a p value ≤ 0.05, ≤0.01 and ≤0.005, respectively.

Protein was isolated from 3–6 pooled RO by first incubating in RIPA buffer for 15 min on ice, briefly sonicating, then centrifuging at 4 °C for 10 min at 13 k rpm. Protein concentration was quantified by Pierce BCA assay (Thermo Fisher Scientific) as per manufacturer’s instructions. Protein concentrations were normalised by addition of RIPA buffer, then mixed 1:1 with 2× SDS PAGE buffer and heated at 95 °C for 5 min to facilitate protein denaturation. Samples were run on precast 4–20% gradient gels (Bio-Rad, Watford, UK) at 120 V for 2 h, or until the dye reached the end of the gel. The proteins were transferred to a nitrocellulose membrane by wet transfer, at 90 V for 1.5 h, using standard protocols. Membranes were blocked for 1 h at room temperature in 5% skim milk powder in PBS-T, then incubated overnight at 4 °C in primary antibody (Table S3). The following day membranes were incubated in species-specific HRP-conjugated secondary antibody for 1 h, then developed using Clarity MAX Western ECL (Bio-Rad, Watford, UK) for 5 min. Membranes were imaged using a ChemiDoc MP Imaging System (Bio-Rad, Watford, UK). Images were formatted using Image Lab (Bio-Rad, Watford, UK), which was also used to calculate band density. Three independent protein isolates were used per condition and statistical significance was calculated by unpaired Students t test, where * denotes a p value < 0.05.

cGMP concentrations were calculated for individual whole RO by ELISA (Cambridge Biosciences, Cambridge, UK) according to the manufacturer’s instructions. In brief, RO were washed with PBS and then incubated in 100 μL of 0.1 M HCl for 20 min at RT and mechanically homogenised. The samples were centrifuged at 1000× g for 10 min then the supernatant was combined with 200 μL of ELISA buffer. An amount of 50 μL of this sample was used per well. Absorbance was measured at a wavelength of 420 nm. Values given are normalised to volume to give cGMP concentration per the whole RO, and statistical significance is calculated by unpaired Students t test, where * denotes a p value < 0.05.

The data presented in this study are contained within the article and. supplementary material