Retinal tissue grown from patient stem cells keeps structure despite lower levels of faulty AIPL1 protein

The retinal neuroepithelium formed following manual dissection of optic vesicle (OV)-like structures after 4 weeks (W) of differentiation (Fig. 1B, W4 and W4' and Supplementary Fig. S1A, W4) and grew in an apically convex manner after that (Fig. 1B, W12, and Supplementary Fig. S1A, W12). The translucent projections, representing presumptive inner segments (ISs), connecting cilia (CC), and nascent OSs, at the apical edge of the ROs started to appear at W20 of differentiation and became abundant by W23 (Fig. 1C and Supplementary Fig. S1A) similarly to previous reports^31,32. Protrusions grew over the subsequent weeks, reaching up to 50 µm in length by W26. Excised OVs formed retinal neuroepithelium with an efficiency of 23.6 ± 1.15% (mean ± SD; N = 3 differentiation experiments, n > 200 OVs) for AIPL1-LCA and 23.3 ± 1.52% and 24.3 ± 2.5% (mean ± SD; N = 3 differentiation experiments, n > 200 OVs) in Control 1 and Control 2, respectively (Supplementary Fig. S1B), as judged by light microscopy. These results correlated to histological examinations. The morphology adopted by W26 was maintained upon prolonged culture to W33 (data not shown) in all genotypes.

We next explored molecular aspects of retinal cell specification in differentiating ROs from the AIPL1-LCA patient hiPSC line and from the two control hiPSC lines (combined as one control because of similar gene expression profiles as described previously³³), by analyzing transcriptome profiles at four different time points (day (D) 25, D60, D88, and D123 corresponding to W4, W9, W13, and W18, respectively). These time points span retinal organoid formation before the appearance of apical protrusions from ROs. During development, multipotent retinal progenitor cells (RPC) are committed to specific lineage (precursors) and eventually give rise to all six neurons of the retina and Müller glia. The employed differentiation protocol recapitulates retinal neurogenesis, capturing the transient progenitor and precursor states before reaching mature cell features.

Then, we checked the established cell type –specific genes in the retina (Fig. 2). Figure 2A depicts the expression pattern for RPC markers, which displayed an upregulation in expression of the eye field transcription factors (PAX6, RAX, SIX3, and LHX2), whereas downregulation of expression was evident for RPC genes that are selectively repressed in developing retinas (ATOH7, ASCL1, and NEUROG2). We next explored retina-specific transcription factors and their co-regulators; Fig. 2, panel B shows the early activation of the orthodenticle homeobox 2 (OTX2) photoreceptor precursor cell marker and its downstream target CRX, which is upregulated during photoreceptor development. We observed the expression of NR2E3 and its transcriptional activator NRL (neural leucine zipper) together with their interacting protein PIAS3 according to the correct transcriptional hierarchy of rod differentiation. Furthermore, we noted the later activation of nuclear hormone receptors, such as retinoic acid receptor-related orphan receptor alpha (RORA), another regulator of retinal development and function. The expression of transcription factors required for development, survival, and specification of cone photoreceptors also increased over time (e.g., cone thyroid hormone receptor B (THRB), associated with cone viability, or RXRG, an S-OPSIN suppressor known to be active in immature cones).

We observed the upregulation of other photoreceptor-specific proteins such as AIPL1 and RECOVERIN by W9 (D60) of differentiation and before NRL expression both in control and AIPL1-LCA ROs, in accordance with previous observations in human retinas^21,35 (Fig. 2C). No change was evident in AIPL1 expression in both developmental schemes, and the single nucleotide missense mutation did not affect alignment quality and quantity (Figs. 2C,D). The expression of mature rod and cone cell markers such as G-protein-coupled receptor RHODOPSIN increased at W18, while we failed to detect cone-specific OPN1MW by W18 (D128) of differentiation (Figs. 2C,D). We observed low levels of OPN1SW expression. We also detected the activation of phototransduction proteins both in rods (PDE6B, PDE6G, PDE6A, GUCA1A, GUCA1B, and CNGB1) and cones (PDE6C, CNGB3, GNAT2, and GNB3) in a similar fashion in control and AIPL1-LCA ROs. The presence of other retinal cell type-specific markers suggested the genesis of horizontal (PROX1), bipolar, amacrine (NEUROD1, NEUROD4) and ganglion cell types (transient expression of POU4F1, POU4F2, ISL1, or increasing expression of THY1) in all ROs (Fig. 2E–I).

The only exception to the overall similarity in expression profiles between the control and AIPL1-LCA ROs was in NEUROD6 (Fig. 2G), a helix-loop-helix transcription factor involved in GABAergic amacrine cell subtype specification³⁶. We failed to observe NEUROD6 expression in AIPL1-LCA ROs.

Taken together, these data suggest the activation of the molecular networks required for retinal cell specification, which follows the chronological cell birth sequence of the native retina for both healthy and AIPL1-LCA ROs.

Differential expression (DE) analyses between the time points, within each group separately, revealed at least 2-fold levels of alterations in 3794 genes, with 2182 genes common between groups (Supplementary Fig.). 797 genes and 815 genes showed response to time either in control or patient derived ROs respectively (Supplementary Fig.). The gene ontology (GO) terms of intersecting genes (Supplementary Fig.) reflect a summary of GO terms belonging to clusters having a similar developmental pattern in both datasets (Supplementary Fig.). GO identified enrichment in genes associated with mitotic cell division, signal release from the synapse, and visual perception suggesting that the main steps of commitment toward retinal cell fates occur in both genotypes. The number of DE genes with respect to time unique to each group indicated differences between both development schemes even though the core development is very comparable. The genes unique to the control group (797) are enriched in neuronal connection-formation related GO terms (Supplementary Fig.). AIPL1-LCA mutant-specific genes belonged to two GO terms; one related to antigen possession for immune responses, and the other related to microtubule functioning (Supplementary Fig.). The transcriptomic differences between the two groups could be used to direct appropriate downstream experiments. S3A S3A S3B S3C,D S3B S3B

Moreover, clustering of all the time-responsive DE genes revealed 12 clusters, (C1-C12), with some differences between the two groups (Supplementary Fig.). Four clusters showed gene expression changes that are likely caused by AIPL1 mutation in patient organoids. C1 genes are turned off earlier in the control, whereas their down regulation during maturation is slower in the patient-derived retinal organoid samples. Genes in C7 exhibit increasing expression with time in the patient organoids but not appreciably in controls. Expression of C8 genes peaks at D88 in the control group; however, their expression continues to be high at D123 in the patient group. C9 genes are showing increasing expression with developmental time in control group, but not in the patient organoids. S3C

Although the differences in gene expression profiles had no apparent effect on normal differentiation of retina, especially on the formation of the OS, DE genes unique to either group showed substantial disruption on time matching compared to common DE genes indicated by RV coefficients 0.69 and 0.95, respectively (Supplementary Fig.). Furthermore, the captured meaningful variation (related to the developmental time) by PCA of common DE genes was not much higher than that of all expressed genes (less than 9%, 26.69%, and 17.82%, respectively) (Supplementary Fig.and). S4B,C S2A S4A

We also studied the differentiation progress of AIPL1-LCA hiPSCs by immunofluorescence. We examined the genesis of retinal cell types by following the expression of a panel of major retinal cell type markers spanning the 27 weeks of differentiation (Supplementary Fig. S5). hiPSC-derived ROs followed the temporal sequence of expression of human retinal cells²⁴, namely ganglion cells followed by photoreceptor progenitors, amacrine, and horizontal cell types. Towards the end of the differentiation process, photoreceptors mature and are then followed by the appearance of bipolar cells and Müller glia. We detected ganglion cells by NEUN staining at W9, followed by the OTX2 and CRX early photoreceptor progenitor markers, which were restricted toward the presumptive outer nuclear layer (ONL) on the apical border at later stages. We detected the expression of NRL, the earliest rod-specific marker, at W13 in 33,3 ± 5% of the ROs (16 ROs analyzed, n = 3 differentiation experiments). The AP2 pan amacrine cell marker began to be expressed before W10, while CALRETININ positive immunostaining, specific for the type II subtype, appeared later at W13. We observed the CALBINDIN horizontal cell marker basally at W13 followed by an increase in the number of cells and the segregation to the presumptive intermediate layer. We also observed fluorescence signal at the apical surface of the ROs, which we presume to be cones, as CALBINDIN can also label cone cells^37,38. Müller glia exhibited a weak cellular retinaldehyde-binding protein (CRALBP) staining by W17 but high expression at W21, with the characteristic extended feet across the organoid. The bipolar cells were born last and were not detected until W25 (not shown).

Overall, the time course of retinal cell genesis in the absence of functional AIPL1 followed the conserved sequence of emergence of vertebrate retinal cells, with ganglion cells being born first, followed by photoreceptor precursors, horizontal, amacrine, and later by Müller glia, bipolar cells, and maturing photoreceptors. We observed the organization of retinal cell types along the apico-basal axis in agreement with their distribution in the native retina.

Overall, the ultrastructural examination revealed similar mature features found in AIPL1-LCA and control ROs of the same differentiation stage (Supplementary Fig. S6), also showing similarities to previous reports^31,32.

To investigate the development of cilia, we stained for the ARL13B ciliary marker in both ROs, observing staining at the apical surface. The rod-specific TRANSDUCIN-α subunit and RetGC (not shown) phototransduction proteins were also expressed in the presumptive ONL-like layer positively staining the ISs and less intensely the cell bodies (inset).

Thus, we did not detect any significant differences with regards to the expression of photoreceptor-specific markers and their localization between AIPL1-LCA and control ROs.

In contrast, AIPL1-LCA ROs displayed only weak staining at W21 and W27 under the same experimental conditions. The AIPL1 signal was significantly reduced between patient and Control 1 ROs at W27 (Fig. 5B). We verified the expression of AIPL1 by immunoblotting, and in line with immunofluorescence data, we found a dramatic reduction in expression of AIPL1 in patient-derived ROs as compared with control ROs collected at W27 of differentiation (Supplementary Fig. S7A) and in W33-old ROs (Fig. 5D and Supplementary Fig. S7B).

Animal models had shown that AIPL1 deficiency causes destabilization of PDE6 by preventing its assembly, which leads to rapid protein degradation⁸. In the native retina, PDE6 complex is synthesized and assembled in the cytosol of photoreceptor IS, and subsequently traffics to the OS, the site of phototransduction cascade⁴⁰. We performed immunofluorescence to detect PDE6α in ROs and detected positive staining in a portion of control ROs (26,6 ± 5,7% N = 17 ROs, n = 3 differentiation experiments) in the ISs. None of the AIPL1-LCA ROs displayed PDE6α signal at the apical surface at the same differentiation stage (Fig. 5C). We corroborated the reduced levels of PDE6α by immunoblotting W27 and W33 ROs (Supplementary Fig. S7A and Fig. 5D). We also assayed for PDE6 heteromer assembly using a previously characterized monoclonal antibody, ROS-1, that recognizes a fully assembled PDE6 complex. Our pull-down with ROS-1 in both AIPL1-LCA and control ROs revealed the presence of assembled PDE6 (Fig. 5E and Supplementary Fig. S7C).

Thus, in line with the AIPL1 animal model data^8,14,41, we observed reduced levels of the AIPL1 primary target protein PDE6α, in AIPL1-LCA RO photoreceptors. Of note, the reduction in AIPL1 and PDE6α protein levels were not caused by a reduction in their mRNA levels (Supplementary Fig. S8).

Therefore, AIPL1-LCA ROs exhibited a reduction in the levels of AIPL1 protein and decreased PDE6α expression, whereas other examined photoreceptor markers remained unaltered. Altogether, our findings show that despite reduced levels of AIPL1 and PDE6, the assembly of PDE6 heteromer is not affected in AIPL1-LCA ROs.

The cell lines employed in this study include the hiPSC cell line (LCA-FiPS4F1)²⁹ from the LCA patient homozygous for the mutation in AIPL1, p.Cys89Arg (c.265 T > C) and two control hiPSC cell lines: Ctrl2-FiPS5F2 (Control 1)⁵⁰ and Ctrl1-FiPS4F1 (Control 2), were subjected to the same differentiation conditions in parallel. All cell lines are deposited in the Spanish National Stem Cell Bank (www.isciii.es/QueHacemos/Servicios/BIOBANCOS/BNLC/↗).

hiPSCs were cultured in plates coated with BD Matrigel^TM, human embryonic stem cell-qualified Basement Membrane Matrix (Corning) using mTeSR1 medium. Passages were performed using Dispase (STEMCELL Technologies, #07913), every 5–7 days at 1:6–1:10 split ratio. Quality controls such as mycoplasma contamination, karyotype verification, and cell line authentication were regularly performed.

The hiPSC lines were induced to differentiate toward 3D neural retina following the protocol published by Zhong et al.²⁴. The hiPSC were grown to 60–80% confluency and then lifted with dispase or EDTA, pipetted up and down and transferred to ultra low attachment (ULA) plates for 1 week in neural-induction medium (NIM) (DMEM/F12 (1:1), 1x N2 supplement (Invitrogen), 1X MEM non-essential amino acids (NEAA), 2 µg/ml heparin (Sigma)). The floating aggregates were then transferred to plates coated with BD, Matrigel^TM-Growth Factor Reduced (GFR) (Corning) and left to grow for 2 weeks. The OV-like structures were excised with the needle and cultured individually in ULA 96 well plates in neural retina medium (NRM, DMEM:DMEM/F12 (2:1), 1x B27 (Invitrogen), 1X NEAA, and 1% antibiotic–antimycotic). At day 42 of culture, the NRM was supplemented with 10% FBS (Gibco), Taurine (100 µM), all-trans retinoic acid (1 µM). From day 93 the retinoic acid was reduced to 0,5 µM.

hiPSC-ROs were fixed in PFA 4% for 30 min at room temperature (RT). Thereafter the ROs were introduced into 10%, 20% and 30% sucrose successively O/N at 4 °C and included into OCT. Histological sections, 10 µm thick, were performed using Microm HM 505E cryostat. Antibodies used are described in Supplementary table. DAPI (40,6-diamidino-2-phenylindole) was used for nuclear counterstaining. Slides were mounted with Vectashield (Vector Laboratories). The images were taken on Leica SP8 confocal microscope with HC PL APO CS2 63×/1.40 OIL and HC PL APO CS2 40×/1.30 OIL objectives and processed by Leica LAS AF, Photoshop CS2 (Adobe), and Image J software. All presented images labelled as Control are from Control 1. No differences in retinal cell type expression and distribution were observed between Control 1 and Control 2. S1

hiPSC-ROs were fixed in 4% paraformaldehyde, 2,5% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7,2–7,4) for 1,5 h and washed with the same buffer. Then, the samples were post-fixed with 2% osmium, rinsed, dehydrated and embedded in Durcupan resin (Fluka, Sigma-Aldrich, St. Louis, USA). Serial semi-thin sections (1.5 µm) were cut with an ultramicrotome Ultracut UC-6 (Leica, Heidelberg, Germany), mounted onto slides and stained with 1% toluidine blue in 1% Borax and analyzed under a light microscope. Ultrathin sections (0.07–0.09 µm) were prepared with the Ultracut and stained with lead citrate. Finally, photomicrographs were obtained under a transmission electron microscope FEI Tecnai G2 Spirit (FEI Europe, Eindhoven, Netherlands) using a digital camera Morada (Olympus Soft Image Solutions GmbH, Münster, Germany). Sample processing was performed by CIPF Electron Microscopy Core Facility.

Control 1, Control 2 and AIPL1-LCA ROs collected at W27 (N = 2) and Control 1 and AIPL1-LCA ROs collected at W33 (N = 2) were lysed in RIPA buffer (R0278 Sigma) containing a protease inhibitor coctail (Roche), and total protein was quantified using a Bradford Reagent protein assay (B6916 Sigma-Aldrich). Protein lysates were denatured by 1X SDS Sample Buffer. The resulting samples were incubated at 95 °C for 5 min. Protein samples (20–30 µg) were then separated on TGX Stain-FreeTM Gels (BioRad), and electroblotted onto a PVDF membrane (Trans-Blot® TurboTM Transfer Pack/ Bio-Rad). Membranes were cut horizontally (at ladder bands corresponding to MW 72 kDa, and 34 kDa) into 3 strips, incubated in blocking buffer (5% nonfat dried milk diluted in TBS + 0,1% Tween) for 1 h at room temperature, washed three times in TBS + 0.1% Tween for 5 min and each strip incubated with corresponding primary antibody (AIPL1, 1:10.000, RECOVERIN 1:1000 and PDE6α 1:1000) in blocking buffer overnight at 4 °C. Thereafter, blots were washed three times in TBS + 0.1% Tween and incubated with secondary HRP-conjugated antibody in blocking buffer for 45 min at RT. Blots were washed another five times and protein bands were visualized using SuperSignal West Pico PLUS (Thermo Scientific) on X-ray films (AIPL1, RECOVERIN, PDE6α) and G.BOX Chemi XX6 (SYNGENE) system (PDE6α).

To analyze effect of AIPL1 C89R mutation on PDE6 complex assembly, immunoprecipitation (IP) was performed with antibody against ROS-1. This was followed by immunoblotting (IB) with PDE6α antibody. 25 Weeks retinal organoids derived from controls and AIPL1-LCA patient were lysed in 200 µl of lysis buffer (50 mM Tris.Cl-pH 7.4, 300 Mm NaCl, 5 Mm EDTA and 0.02% Sodium Azide) containing Protease inhibitor pellet, 10 mM of iodoacetamide and 1% triton X-100. After centrifugation at the speed of 14000 × g for 30 sec at 4 °C; the supernatant (T) was incubated with 5 µg of antibody (ROS-1) for 2 h at 4 °C followed by incubation with 10 µl of A/G agarose beads for 3 h at 4 °C. The Unbound proteins (U) were removed by centrifugation at 14000 × g for 30 sec at 4 °C. The beads were washed three times with 1 ml of the wash buffer (0.1% triton x-100, 50 mM Tris.Cl-pH 7.4, 300 Mm NaCl, 5 Mm EDTA and 0.02% Sodium Azide) and a final wash in 1 ml of 1 × PBS. After the wash, the samples were eluted to a final concentration of 1X SDS sample buffer (4x Tris-Cl/SDS pH6.8, 30% glycerol, 10% SDS, 0.6 M DTT and 0.012% bromophenol and β-mercaptoethanol). The samples were boiled for 5 min. 10 µL of each sample were loaded on SDS PAGE gel and then immunoblotted using PDE6α antibody diluted in 1:1 ratio of blocking buffer (Licor) and 1X PBST (1:1000). The blot was imaged using Li-cor imaging system. This experiment was repeated three times with different retinal organoids samples each time.

The study was approved by the corresponding Ethics Committee CAEC (Comité Etico Autonomico de Estudios Clínicos), Valencia, Spain. Informed consent was signed by the involved human subjects. All procedures were done in accordance with institutional guidelines and regulations.

Supplementary Information.