Loss of Tmem107 leads to distinctive ocular phenotypes in mouse embryos

We used Tmem107^−/− mouse model to determine the function of Tmem107 in eye development (Christopher et al, 2012). Mutant embryos through all analyzed stages (E10.5–E15.5) exhibited a variety of eye abnormalities (Figs 1A and S1A–F and Table S1) (Video 1 and Video 2). Heterozygous embryos did not exhibit an abnormal phenotype. Two most frequently observed phenotypes included complete loss of an eye (anophthalmia), observed in 33.3% of examined mutants, and abnormally small eye (microphthalmia) with 47% occurrence (Fig 1D). In nine examined cases, both phenotypes were present within the same embryo. In anophthalmic animals, a small area of pigment residue was present, whereas embryos exhibiting microphthalmia often displayed other defects, such as total absence of the lens (aphakia) and/or optic nerve (ON) hypoplasia (Fig 1A). The mice retina with microphthalmic phenotype was smaller and elongated, compared with retina in WT animals. Additional morphometric analyses performed using micro-CT approach revealed decreased retina volume and shortened ON in mutant embryos (Fig 1B and C).

Tmem107 is highly expressed in the retina during eye development

To investigate the role of Tmem107 in the development of specific eye structures, we analyzed in situ expression of Tmem107 using the RNAscope approach during critical stages of eye development (E10–E15). At early stages of optic vesicle outgrowth (E10 and E11) (Fig 2A and B), Tmem107 expression was detected in the bilayered optic cup, where the inner layer represents the presumptive neural retina (NR) and the outer layer will give rise to the retinal pigment epithelium (RPE). Although Tmem107 expression was observed throughout all these structures, the signal was particularly enriched in the presumptive NR (Fig 2A’’’’–F’’’’), whereas the lens placode and RPE layer exhibited lower expression (Fig 2A’–F’ and A’’–F’’). The patterns of Tmem107 expression remained similar during later stages (E12, E13), with the signal located in the anterior and posterior lens epithelium (Fig 2C and D) and in the newly formed cornea (Fig 2C’ and D’). At E13, the formation of a ganglion cell layer (GCL) is associated with lower expression of Tmem107 compared with the rest of NR (Fig 2D’’). Later, at E14 and E15, the neuroblast cell layer (NCL), which contains neuronal progenitors, expresses high levels of Tmem107 (Fig 2E’’’’ and F’’’’), whereas differentiated neurons of GCL exhibit low expression of this gene. Interestingly, Tmem107 mRNA expression in the ciliary marginal zone (CMZ) was very low compared with the rest of the NR (Figs 2F’’ and S2A–D). Thus, our findings indicate that Tmem107 is strongly expressed during early stages of eye development (E10–E15), particularly in the NR.

Key factors in eye development are altered in Tmem107^−/− animals

To gain more insight into molecular changes caused by Tmem107 deletion, we analyzed in situ expression of key proteins involved in eye patterning. Because the phenotypic analysis revealed distinct anomalies affecting certain eye areas including retina, lens, and optic stalk, we further focused on the evaluation of the expression patterns of proteins that are critical for morphogenesis of the retina (PAX6, SOX2), the optic stalk (PAX2), and the lens (SOX1). PAX6 and SOX2 are transcription factors associated with anophthalmia and microphthalmia in humans (Matsushima et al, 2011). Furthermore, SOX1 is a key regulator expressed in the developing lens (Nishiguchi et al, 1998), and PAX2 has previously been linked to the development of optic stalk and closure of optic fissure (Bosze et al, 2021). Because it is well established that all of these transcription factors are expressed at the early stages of eye development, we performed the analyses at E10.5 and E11.5 (Fig 3). We observed altered expression of PAX6 (Fig 3A–A’’ compared with Fig 3B–B’’), PAX2 (Fig 3C–C’’ compared with Fig 3D–D’’) and SOX1 (Fig 3E–E’’ compared with Fig 3F–F’’) in Tmem107^−/− embryos already at E10.5 with more striking differences found at E11.5 (Fig 3G–L). We analyzed the expression of PAX6, PAX2, and SOX1 in microphthalmia mutants because structures expressing these proteins are missing in animals with anophthalmia, which we also see in our mutants. Interestingly, SOX2 expression in microphthalmic Tmem107^−/− mutants at E10.5 was still maintained in NR (Fig S3C–C’’ compared with Fig S3A–A’’) and profound differences were found later at E11.5 (Fig S3F–F’’’ compared with Fig S3D–D’’). In anophthalmic Tmem107^−/− mutants, the expression of SOX2 was reduced in the optic stalk, whereas in the optic cup, it was completely lost at E10.5 and E11.5 (Fig S3B’’ and E’’).

Moreover, the reduction of SOX1 expression was found in the lens of microphthalmia mutants, whereas PAX2 and PAX6 were down-regulated in the distal part of NR (Fig 3). In summary, our data indicate that TMEM107 is important for early eye patterning in mice and it may establish the proper expression of pivotal players during crucial stages of the optic cup and stalk morphogenesis.

TMEM107 deficiency leads to the failure of neural retina formation in human retinal organoids

Given the striking retinal phenotype observed in Tmem107^−/− animals, we aimed to further test if the human TMEM107 gene is essential for the development of the retina. We used retinal organoids differentiated from hESCs as a model to closely investigate the roles for TMEM107 in human retinal development. Moreover, this approach allowed us to evaluate the direct role of TMEM107 in retinal differentiation without the effects of surrounding or closely associated eye structures and tissues including the brain or surface ectoderm.

We generated TMEM107^−/− hESCs using CRISPR/Cas9 approach (Figs S4 and S5) and differentiated them into retinal organoids using an already published protocol (Kuwahara et al, 2015; Peskova et al, 2020; Celiker et al, 2023). Retinal organoids were analyzed at day 30 (D30, early stage) and day 150 (D150, late stage) of the differentiation process. Early differentiation steps during the retinal organoid formation include the generation of NR epithelium containing progenitors that give rise to photoreceptors (RAX+), ganglion cells (VSX2+), and other cell types of the human retina, whereas the late stage is characterized by photoreceptor maturation (CRX+, RHODOPSIN+) and appearance of inner and outer photoreceptor segments (Burmeister et al, 1996; Furukawa et al, 1997b; Irie et al, 2015).

At D30 of the differentiation process, WT organoids contained NR epithelium, whereas TMEM107^−/− organoids lacked NR epithelium and contained cystic structures (Figs 4A and D and S4A–L). RT–qPCR analysis confirmed that TMEM107^−/− organoids failed to generate NR, as demonstrated by significant down-regulation of genes that are typically expressed in the developing NR structure including RAX, PAX6, SOX2, and VSX2 (Fig 4B, top row) (Furukawa et al, 1997a; Kozmik, 2008; Matsushima et al, 2011; Burmeister et al, 1996).

Interestingly, scanning electron microscopy (SEM) analysis revealed that TMEM107^−/− organoids at D30 lack primary cilia on their surface (Fig 4C). In addition, we confirmed the absence of primary cilia using immunofluorescence staining of primary cilia marker ARL13B (Fig 4D). To reveal whether the absence of TMEM107 leads to the impaired maturation of the retinal organoids and failure to generate photoreceptors and other retinal cell types, we cultured retinal organoids until D150. TMEM107^−/− retinal organoids at the late stage completely failed to generate NR demonstrated by: (I) altered organoid morphology (Fig 4A), (II) lack of photoreceptor outer segments demonstrated by SEM and immunofluorescence staining for ARL13B (Fig 4C and E), (III) significant down-regulation of photoreceptor markers (CRX, RHODOPSIN, RCVRN) and retinal ganglion cell marker (MATH5), as demonstrated by RT–qPCR analysis (Fig 4B, bottom row). Interestingly, Hematoxylin/Eosin staining of retinal organoid cross sections revealed the presence of cysts and Oil Red O lipid staining identified increased the presence of lipids in TMEM107-deficient organoids (Fig S4).

To corroborate the phenotype of TMEM107^−/− organoids, we used a different loss-of-function approach—the shRNA-mediated knockdown of TMEM107 in human induced pluripotent stem cells (hiPSCs). We generated lentiviral particles containing mCherry reporter and doxycycline (DOX)-inducible expression of shRNA for TMEM107 down-regulation. Upon transduction, puromycin selection, and FACS sorting, hiPSCs and the generated retinal organoids expressed mCherry reporter (Fig S6A and B). DOX was applied from D2 of the differentiation process, and the retinal organoids were harvested and analyzed at D25. We found ∼50% down-regulation of TMEM107 gene expression as determined by RT–qPCR in the presence of DOX (Fig S6C) that led to alterations in primary cilia formation including extremely elongated or very short cilia with expanded bulges in their tip (Fig S6D), cyst formation inside of the organoids (Fig S6B), and failure to form NR structures in retinal organoids (Fig S6E), thus corroborating the results generated using the knock-out approach.

Taken together, our results indicate that the absence of TMEM107 leads to the following: (I) absence of primary cilia on early-stage organoids and outer segments on late stage retinal organoids, (II) the down-regulation of retina-specific genes, (III) the failure to generate the NR structures and cell types in the human retinal organoid model, (IV) the generation of organoid with cysts containing lipids.

Tmem107^−/− animals have primary cilia defects in pigment epithelium and neural retina

Primary cilia defect in different organs in Tmem107^−/− embryos have been previously reported (Cela et al, 2018; Shylo et al, 2020). However, the potential ciliary anomalies in the retina of these animals remain elusive. To test whether Tmem107^−/− eye phenotype in the mouse model is associated with primary cilia defects, we labeled cilia using anti-ARL13B antibody, counted the number of ciliated cells, and measured the length of primary cilia. We analyzed primary cilia in distinct eye structures with a special focus on RPE and NR at developmental stages E10.5, E11.5 or E12.5. Analysis of the NR region revealed a reduced number of primary cilia (RCC–ratio of ciliated cells) at stages E10.5 (Fig 5A–C), E11.5 (Fig 5D–F), and E12.5 (Fig 5G–I) associated with significantly reduced cilia length at E10.5 (Fig 5C’) and E12.5 (Fig 5I’). In addition, the length of cilia was also reduced at E11.5, but not with the statistical significance (Fig 5F’). However, more profound changes were observed in the RPE region, where we found a decreased length and cilia number in all analyzed samples of different developmental stages: E10.5 (Fig 5J–L’), E11.5 (Fig 5M–O’), and E12.5 (Fig 5P–R’). These data suggest that TMEM107 plays a role in cilia biogenesis in vivo and regulation of ciliary length in the developing eye.

Loss of TMEM107 leads to aberrant SHH signaling in retinal cells

The evident association of primary cilia disruption with eye phenotypes in all our TMEM107^−/− models prompted us to proceed with the analysis of the effects of TMEM107 loss at the molecular level. Because the function of the Shh pathway fully relies on the formation of primary cilium (Huangfu & Anderson, 2005), we aimed to closely investigate the effects of TMEM107 loss on this signaling pathway. Retinal organoids do not represent a suitable model to specifically address affected molecular pathways in individual cells, because of the heterogeneity of different cell types they contain, therefore it is challenging to study signaling pathways using an organoid model. We, therefore, used the ARPE-19 cell line, derived from retinal pigmented epithelium, to address the effects of TMEM107 loss on the Shh pathway by generating ARPE-19 TMEM107^−/− and TMEM107^+/− cell lines, using CRISPR/Cas9 technology.

First, we assessed the expression of the crucial components of the Shh pathway GLI1 and PTCH1 using RT–qPCR. Whereas there was no significant difference in PTCH1 expression between WT and mutant cells, we detected ∼15-fold up-regulation of GLI1 expression in TMEM107^−/− cells (Fig 6A), indicating an aberrant activation of the Shh pathway in the absence of TMEM107. To confirm the activation of the Shh pathway and to reveal whether ARPE-19 cells possess the functional Shh pathway, we treated the cells using Shh activator smoothened agonist (SAG). Upon SAG treatment, WT and TMEM107^+/− cells up-regulated the expression of GLI1 and PTCH1 by ∼twofold, but TMEM107^−/− cells failed to up-regulate GLI1 and PTCH1 (Fig 6B and C). Therefore, high GLI1 expression and no response to SAG treatment of TMEM107^−/− cells indicate that the Shh pathway is aberrantly activated in TMEM107^−/− cells and that TMEM107 is critical for Shh signaling.

The aberrant Shh signaling in TMEM107^−/− cells could be explained by altered primary cilia formation that we, indeed, also observed in TMEM107^−/− retinal organoids. To test this, we stained ARPE-19 cells for the ciliary marker ARL13B. WT and TMEM107^+/− formed primary cilia, whereas TMEM107^−/− completely failed to form these structures (Fig 6D and E). To test SHH activation, we used the SAG treatment approach as described above, and assessed in situ localization of other critical Shh players SMO and GLI2. Immunofluorescence staining revealed that upon SAG treatment SMO localizes into the primary cilia and GLI2 becomes up-regulated and localized into nuclei of WT and TMEM107^+/− cells. TMEM107^−/− failed to form primary cilia and GLI2 was up-regulated even in the absence of SAG (Fig 6D and E).

Previous studies showed down-regulated SHH in mice with Tmem107 mutation (Christopher et al, 2012; Cela et al, 2018; Shylo et al, 2020). However, because of distinct ciliary transmission of SHH in different organs, both up-regulation and down-regulation have been observed in different tissues of the developing embryo (Burnett et al, 2017). In models that lack primary cilia, an up-regulation of SHH has been observed in the embryonic retina (Burnett et al, 2017). To test how the altered morphology of primary cilia affected Shh signaling in the mouse model, we analyzed the expression of SHH protein and Ptch1 RNA at stage E12.5 (Fig 7). Whereas in WT eyes, SHH localized to discrete regions in the distal part of the eye (Fig 7A’), Tmem107^−/− animals exhibited higher expression of this ligand in proximal and distal parts of the NR and in optic stalk (Fig 7B). The differences were most prominent in the lens and the optic stalk regions (Fig 7B’ and B”). On the contrary, RNAScope analysis of Ptch1 revealed no distinct changes in its expression in the eyes of E12.5 mutant animals (Fig 7D–D”), as compared with WT (Fig 7C–C”). These data demonstrate that although the expression of SHH was elevated, decrease in primary cilia keeps the pathway dysfunctional in mice mutants.

Taken together, our results indicate that the absence of TMEM107 leads to the following: (I) failure to form primary cilia in retinal pigmented epithelial cells, (II) aberrant up-regulation of the Shh pathway demonstrated by the up-regulation of GLI1, GLI2, and GLI3 FL, (III) incapability of TMEM107^−/− cells to respond to SAG treatment, because lack of cilia, and (IV) increased level of SHH ligands in vivo.

Animals

Embryos of mice containing Tmem107^null allele were generated as described previously (Tang et al, 2010; Christopher et al, 2012). The details on the generation of the mutant can be found via the MMRRC strain information page (https://www.mmrrc.org/catalog/sds.php?mmrrc_id=32632). Targeted or gene trap mutations were generated in strain 129/SvEvBrd-derived embryonic stem (ES) cells. The chimeric mice are bred to C57BL/6J albino mice to generate F1 heterozygous animals. All of the mouse husbandry was performed in accordance with Yale’s Institutional Animal Care and Use Committee guidelines.

Immunofluorescent staining on paraffin-embedded sections of mice

Embryonic mouse samples were collected (Table S2) and fixed in 4% paraformaldehyde overnight and then processed for paraffin embedding and sectioned in transversal planes. Before immunofluorescent analysis, sections underwent the xylene/ethanol series for deparaffinization and rehydration. For antigen retrieval, 0.1 M sodium-citrate buffer pH6 with 0.04% Tween was used. Retrieval was performed under 98°C for 25 min.

Sections were incubated with primary antibodies (Table S3) for 1 h at room temperature and subsequently with the following secondary antibodies for 1 h at room temperature: anti-mouse Alexa Fluor 488 (1:200, cat. No. A11001), anti-mouse Alexa Fluor 568 (1:200, cat. No. A11004), anti-rabbit Alexa Fluor 488 (1:200, cat. No. A11008), and anti-rabbit Alexa Fluor 594 (1:200, cat. No. A11037; all Thermo Fisher Scientific). Sections were washed in PBS and coverslipped in Fluoroshield with DAPI (cat. No. F6057; Sigma-Aldrich, Merck). The pictures were taken under a fluorescence microscope Leica DM LB2 (Leica) and the Leica SP8 confocal microscope (Leica). Individual images were assembled into Figures in Adobe Photoshop 5.0. Pictures of wild-type and mutant specimens were taken under the same microscope settings.

Immunofluorescent staining of organoids

For frozen sections staining, organoids were fixed in 4% paraformaldehyde for 30 min at room temperature, washed with PBS, cryopreserved in 30% sucrose (16104; Sigma-Aldrich) in PBS overnight at 4°C, embedded in Tissue-Tek O.C.T. Compound medium (Sakura) and sectioned on a cryostat (CM1850; Leica) with thickness of slides ∼10 μm. For cells grown on coverslips, cells were fixed in 4% paraformaldehyde for 15 min at room temperature. Samples were washed with PBS and blocked in a blocking buffer (PBS with 0.3% Triton X-100 [T8787; Sigma-Aldrich] and 5% normal goat serum [G9023; Sigma-Aldrich]) for 1 h at room temperature in a humidified chamber. The primary antibodies were diluted in antibody diluent (PBS with 0.3% Triton X-100 and 1% bovine serum albumin [A9647; Sigma-Aldrich]) and applied to the sections overnight at 4°C in a humidified chamber. List of primary antibodies is provided in Table S3. The sections were washed with antibody diluent and a secondary antibody (anti-Mouse AlexaFluor 488; Thermo Fisher Scientific, 1:1,000 and anti-rabbit AlexaFluor 594; Thermo Fisher Scientific, 1:1,000) was applied for 1 h at room temperature in a humidified chamber. The nuclei were stained with 1 μg/ml DAPI in PBS for 4 min at room temperature and the sections were mounted using Fluoromount Aqueous Mounting Medium (F4680; Sigma-Aldrich). Sections were scanned using Zeiss LSM 880 confocal microscope and files were processed using ZEN Blue Edition software (Zeiss).

In situ gene expression analysis using RNAScope technique

The embryos were fixed in 4% PFA for 24–36 h and proceeded for dehydration in an ethanol series. After embedding in paraffin, 5-μm transverse sections were obtained. The sections were deparaffinized in xylene and dehydrated in 100% ethanol. To detect mRNA expression of Tmem107 (805841; ACD Bio) and Ptch1 (476231; ACD Bio) in mouse embryos, we used the RNAScope R Multiplex Fluorescent v2 Assay kit (Cat. No. 323 110; ACD Bio) for formalin-fixed paraffin-embedded tissues according to the manufacturer’s instructions. All reactions were performed at 40°C in the HybEZTM II Oven (ACD Bio). Visualization of hybridized probes was done using the TSA-Plus Cyanine 3 system (NEL744001KT; Perkin-Elmer), according to the manufacturer’s protocol. DAPI (cat. n. 323 108; ACD Bio) was used for nuclear staining. Pictures were taken with a fluorescence microscope Leica DM LB2 (Leica) and the Leica SP8 confocal microscope (Leica).

CRISPR/Cas9-mediated gene editing

Cells were transfected with a PX458 vector containing gRNA sequences for TMEM107 knockout. The pSpCas9(BB)-2A-GFP (PX458) was a gift from Feng Zhang (Plasmid #48138; Addgene; http://n2t.net/addgene:48138; RRID:Addgene_48138). gRNA was designed using Benchling [Biology Software] (2021), retrieved from https://benchling.com, to target the NcoI restriction site (Fig S7A) in the first exon of TMEM107 (FWD: CACCGAAGCCCTGAGACCCGGCCCA, REV: aaacTGGGCCGGGTCTCAGGGCTTC). 48 h after transfection, GFP-positive cells were bulk FACS-sorted using BD FACS Aria. Then, the cells were transfected for the second time and individual clones were generated using single-cell FACS sorting. Selected clones were screened by PCR using the following primers flanking the first TMEM107 exon (FWD: CTAGATTTGTCGGCTTGCGG, REV: GCAGGGTAGGACTAGGACCA). PCR products were digested using the NcoI restriction enzyme (Fig S7B and C). Clones that lacked the NcoI restriction site were selected for the final next-generation sequencing screen using the Illumina platform (Malcikova et al, 2015) and compared with the wild-type sequence (Figs S8 and S9).

shRNA-mediated gene knockdown

DNA sequences for shRNA-mediated TMEM107 down-regulation were designed using a web tool (Gu et al, 2014). Generated sequences: shRNA-TMEM107-F: CCGGGACATTAAGACTTATATAAtaacctgacccattaTTATATAAGTCTTAATGTCTTTTTG, shRNA-TMEM107-R: AATTCAAAAAGACATTAAGACTTATATAAtaatgggtcaggttaTTATATAAGTCTTAATGTC were cloned into a lentiviral vector containing DOX-inducible U6 promoter and TetRep-P2A-Puro-P2A-mCherry (Eshtad et al, 2016) (vector kindly provided by Mikael Altun). Lentiviral particles were generated as described previously (Barta et al, 2016; Peskova et al, 2019, 2020) using pMD2.G (#12259; Addgene) and psPAX2 (#12260; Addgene) (gift from Didier Trono, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland). We generated non-clonal hiPSC lines using fluorescent activated cell sorting of mCherry-positive hiPSCs. After transduction, at least 5,000 mCherry-positive hiPSCs were sorted using BD FACSAria (BD Biosciences). Generated cell lines were then propagated in the presence of puromycin (1 μg/ml). DOX 0.5 μg/ml was applied from day (D)2 to D25 of the differentiation of retinal organoids.

Analysis of primary cilia length

Ciliary axonemes were visualized using ARL13B immunostaining as described above. Imaging was performed with a Leica SP8 laser scanning microscope using the 63× oil objective, and image acquisition was performed using LAS X software (Leica). Image data were acquired as Z-stacks of images and a 0.3-μm distance separated neighboring Z-sections. Cilia length measurements were done using the Imaris Cell module (Imaris v9.5; Oxford Instruments) using the methodology of fitting an ellipsoid into the surface analysis to obtain ellipsoid axis length. Statistical analyses were performed using a two-tailed t test.

Generation of retinal organoids

Retinal organoids were generated from human embryonic stem cells (hESCs) (H9 cell line, WiCell Bank) or hiPSCs (M8 cell line, [Peskova et al, 2020]), according to the protocol described elsewhere (Kuwahara et al, 2015; Peskova et al, 2020) with small modifications. Briefly, cells in E8 medium were seeded into a U-shaped, cell-repellent 96-well plate (5,000 cells/well) (Item No: 650970; Cellstar, Greiner) in the presence of 20 μM ROCK inhibitor (Y-27632; STEMCELL Technologies). After 48 h (Day 0 of the differentiation process) the culture medium was changed to gfCDM (Kuwahara et al, 2015) (45% IMDM [Item No: 21980-032; Gibco], 45% Hams F12 [F12, Item No: 21765-029; Gibco], 10% KnockOut Serum Replacement [Item No: 10828-028; Gibco], 1% chemically defined lipid concentrate [Cat. No. 11905-031; Gibco], 1× penicillin–streptomycin solution [Item No: XC-A4122; Biosera], 10 μM β-mercaptoethanol [Item No: M3148; Sigma-Aldrich]). On day 6, recombinant human BMP4 (Peprotech) was added to the culture to the final concentration 2.2 nM and then the medium was changed every third day. On day 18 of the differentiation process, gfCDM was changed to a NR medium (DMEM/F12 [Item No: 21331-020; Gibco], 1% N-2 supplement [Item No: 17502001; Gibco], 1% GlutaMAX supplement [Item No: 35050061; Gibco], 10% FBS [Item No: FB-1101; Biosera], 0.5 μM retinoic acid [Item No: R2625; Sigma-Aldrich], 0.1 mM taurine [Item No: T8691; Sigma-Aldrich], 1× penicillin–streptomycin solution [Item No: XC-A4122; Biosera]).

Cell cultures and their treatments

ARPE-19 (ATCC - American Type Culture Collection) cells were cultured in media Knockout DMEM (Invitrogen, Life Technologies Ltd.), containing 10% FBS, 1× GlutaMAX (Invitrogen, Life Technologies Ltd.), 1× MEM nonessential amino acid solution (Invitrogen, Life Technologies Ltd.), 1× penicillin/streptomycin (PAA), and 10 μM β-mercaptoethanol (Sigma-Aldrich). The cells were incubated at 37°C/5% CO₂ and regularly passaged using trypsin (Sigma-Aldrich). For the small-molecule agonist of the Shh-Smo pathway (SAG) treatment, confluent ARPE-19 cells were cultured 24 h in the presence of 1 μM SAG (Tocris) without FBS and harvested or fixed for downstream processing. H9 hESCs and M8 hiPSCs were cultured in Essential 8 Medium (Invitrogen, Life Technologies Ltd.) on cell culture dishes coated with vitronectin and regularly passaged using 0.5 mM EDTA in PBS.

Gene expression analyses by RT–qPCR

ARPE-19 cells were washed with PBS and harvested into 1 ml of RNA Blue Reagent (an analog of Trizol) (Top-Bio). For retinal organoid analysis, at least eight organoids were washed with PBS and homogenized using a 1 ml insulin syringe in 1 ml RNA Blue Reagent. RNA was isolated using chloroform extraction, and then reverse-transcribed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). The RT product was amplified by real-time PCR (Roche LightCycler 480 PCR instrument) using PowerUp SYBR Green Master Mix (Applied Biosystems TM). Primer sequences are shown in Table S4. All experiments were performed using at least three independent replicates.

Micro-CT analyses

Before micro-CT scan the samples of E15.5 mouse embryos were dehydrated in the alcohol grades (30%, 50%, 70%, 80%, and 90%). The samples were then stained in the solution of 1% iodine in 90% MeOH to enhance the contrast of soft tissue. For the purpose of motion stabilization during the scan, the embryos were embedded in 1% agarose gel in a 2-ml Eppendorf tube.

The micro-CT scanning was performed using laboratory system GE phoenix v|tome|x L 240 (Waygate Technologies), equipped with a 180 kV/15W maximum power nanofocus X-ray tube and high-contrast flat panel detector dynamic 41|100 (dimension - 4,048 × 4,048 pixels, with pixel size of 100 μm). The measurements were carried out in an air-conditioned cabinet (21°C). The tomographic reconstruction was performed using the software GE phoenix datos|x 2.0 (Waygate Technologies, Germany). The measurements were acquired with the following settings: acceleration voltage 60 kV, X-ray tube current 200 μA, exposition time was 900 ms, 2,200 images were taken over the 360° rotation, and voxel resolution of 3.6 μm.

The manual segmentation of the eyes was carried out using Avizo 9.5 software (Thermo Fisher Scientific). All 3D visualizations, measurements, and specialized analyses were performed in VG Studio MAX 2022 software (Volume Graphics GmbH).

The wall thickness analysis was performed in the VG Studio software. The analysis is based on fitting the spheres in the 3D space into the defined object of the eye. The larger the space in between the walls of the object in 3D can fit a larger sphere. The quantification of this analysis is showing the count of the voxels (voxel = 3D pixel) with the appointed wall thickness.

The optic stalk length was measured with the polyline tool. This tool enables the length measurement of the not-straight objects by placing the individual points defining the line alongside the shape of the measured object. The volumes of the segmented eyes were measured in the VG Studio software.

SEM

SEM was performed as described previously (Peskova et al, 2020). Briefly, the organoids were fixed with 3% glutaraldehyde in a 0.1 M cacodylate buffer, then washed three times with 0.1 M cacodylate buffer, dehydrated using ascending ethanol grade (30, 50, 70, 80, 90, 96, and 100%), and dried in a critical point dryer (CPD 030; BAL-TEC Inc.) using liquid carbon dioxide. The dried organoids were sputtered with gold in a sputter coater (SCD 040; Balzers Union Limited) and observed in a scanning electron microscope (VEGA TS 5136 XM; Tescan Orsay Holding) using a secondary emission detector and a 20-kV acceleration voltage.

Statistical analysis

All results are expressed as means ± SD of three replicates for each condition. For the testing of normal distribution, Shapiro–Wilk test was used and data were further compared by either paired in case of RT–qPCR and micro-CT analysis or nonparametric unpaired two-tailed t test in case of primary cilia length measurements and Western blot analyses. Differences were considered to be significant at P < 0.05.