Retinal Organoids derived from hiPSCs of an AIPL1-LCA Patient Maintain Cytoarchitecture despite Reduced levels of Mutant AIPL1

Aryl hydrocarbon receptor-interacting protein-like 1 (AIPL1) is a photoreceptor-specific chaperone that stabilizes the effector enzyme of phototransduction, cGMP phosphodiesterase 6 (PDE6). Mutations in the AIPL1 gene cause a severe inherited retinal dystrophy, Leber congenital amaurosis type 4 (LCA4), that manifests as the loss of vision during the first year of life. In this study, we generated three-dimensional (3D) retinal organoids (ROs) from human induced pluripotent stem cells (hiPSCs) derived from an LCA4 patient carrying a Cys89Arg mutation in AIPL1. This study aimed to (i) explore whether the patient hiPSC-derived ROs recapitulate LCA4 disease phenotype, and (ii) generate a clinically relevant resource to investigate the molecular mechanism of disease and safely test novel therapies for LCA4 in vitro. We demonstrate reduced levels of the mutant AIPL1 and PDE6 proteins in patient organoids, corroborating the findings in animal models; however, patient-derived organoids maintained retinal cell cytoarchitecture despite significantly reduced levels of AIPL1.


Supplemental experimental procedures
. List of used antibodies Table S2. List of TaqMan® Gene Expression ID Assays used for qRT-PCR

LCA Patient Clinical Phenotype
The patient was diagnosed with LCA one year after birth based on clinical evaluation due to congenital rotary nystagmus, reduced vision, and strabismus since birth. The patient was the first child of an endogamic couple from the northwest of Spain. The patient also then presented bilateral keratoconus at 12 years of age. In the setting of the corneal ectasia, the patient suffered three episodes of acute corneal hydrops due to disruption of Descemet's membrane. The ophthalmic data were obtained at 27 years old. Visual acuity was reduced to light perception.
The central cornea at both eyes has a pronounced cone shape, hydrops, and small leucomas.
Keratoconus-associated Munson's sign was also detected on the lower eyelid. Fundus showed vascular attenuation, bone-spicule pigmentation in the mid-periphery and papillar gliosis. The Asper LCA microarray chip identified a previously known homozygous variant NM_014336.4:c.265T>C; p.(Cys89Arg) in AIPL1 1 . Further Sanger sequencing showed correct segregation in both parents.

RNA extraction and Taqman Assay
Total RNAs were extracted from 15-20 ROs per time point using RNeasy kit (Quiagen) and RNA yields and quality were checked with a NanoDrop spectrophotometer (Thermo Fischer

Strand-specific RNA-seq and data analysis
Retinal organoids (15-20) from Control 1, Control 2 and AIPL1-LCA from 3-5 independent differentiation experiments were collected at indicated time points during the differentiation.
Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany), and treated with DNase I to remove any genomic DNA contamination.
Directional RNA-seq and primary data analysis were performed as previously described 2 . Quality control, sequence alignment, transcript and gene-level quantification of primary RNA-seq data was done via an established bioinformatics pipeline 3, 4 except the whole genome alignment and by excluding non-coding RNAs from target transcriptome. All other bioinformatics analyses were performed in the R statistical environment. Genes were kept for further analysis only if there were ≥5 count per millions (CPM) in all replicates of at least one group. The data were subjected to TMM normalization by edgeR v3.10.5, and PCA and Pearson correlation were performed with normalized log2 CPM values. Co-inertia analysis (CIA) was performed with R package called "made4". All the dimension reduction plots are custom. The organoid transcriptome data from the two control lines was comparable by PCA and therefore combined for DE analysis against the patient organoids. DE analysis was performed using limma v3.24.15, and genes having ≥ 2-fold change at least between 2 time points in one of the genotyped transcriptomes and a false discovery rate (FDR) ≤ 0.01 were considered to be significantly differentially expressed. Gene expression clustering was performed on gene-wise Z-scores using Affinity Propagation (AP) (apcluster v1.4.3 package 5 with k-means option where k=12 by choosing corSimMat function as similarity measure. To determine the number k, PCA projections of genes by using their z-scores were used. Since the sample projections shows time as the primary source of variation same would be the true for the gene projections however the number of the genes is incomparably higher than the sample number, one axis would not be enough to explain the time. Therefore, the angles between x-axis and the imaginary vector from origin to each gene projected coordinate were sorted after converting them to radians. Sorted angle transformations then divided to four groups (one per quadrant).
Then resulting 4 heatmaps were integrated via eyeballing, and the optimal number of clusters became visible. GO enrichment analysis was performed using clusterProfiler v3.2.14 6 . To reduce redundancy associated with GO analysis, we performed semantic similarity comparisons of the enriched GO terms using GOSemSim v2.0.4 7 and semantic similarities clustered with AP. The term with the highest level in each cluster is taken as the enriched GO term. Heatmaps 4 of GO term gene lists were generated using the negDistMat order of AP on the log2 CPM values. All the heatmaps are the products of customized ggplots. Figure S1. Differentiation of control hiPSC lines toward ROs. A. Phase contrast micrographs of the differentiation stages of Control 1 ROs: hiPSCs, the floating aggregates after treating the hiPSCs with dispase (W2), aggregates plated on Matrigel GFR-coated plates reach a typical morphology by week 4 (W4), are dissected manually and are grown in suspension thereafter. The typical transparent NR domain (*) is formed by W12. At W20 the projections (arrowhead) begin to emerge at the apical edge of organoids. The projections are more abundant at W23. At W26 the organoids have reached 1-1,5 mm in diameter and show dense translucent projections at the apical edge. Inset shows higher magnification of the apical edge with protrusions. B. Efficiency of neuroepithelium formation in ROs. Data were analyzed by One way ANOVA test. The differences between groups were non-significant (n.s.); data are mean  s.e.m¸ N=3 independent experiments with >288 ROs each.