Patient-Derived Immortalized Limbal Epithelial Cells as In Vitro Models of Congenital Aniridia

TL;DR

Researchers created long-lasting lab cell models from patients with congenital aniridia to study the disease and test new treatments.

Contribution

The study introduces patient-derived immortalized limbal epithelial cells as a novel in vitro model for PAX6-associated aniridia.

Findings

Aniridia-derived cell lines showed reduced nuclear PAX6 immunosignals compared to controls.

mRNA sequencing revealed over 14,000 transcripts, including unique subsets in aniridia-derived lines.

Cell lines preserved epithelial characteristics and overall transcriptomic similarity to healthy controls.

Abstract

Purpose: To establish and comprehensively characterize immortalized limbal epithelial cell lines derived from patients with PAX6 haploinsufficiency-associated congenital aniridia, as well as from a healthy donor. These well-defined cell models provide a reliable and reproducible platform for long-term experimental studies, facilitating mechanistic investigations and the development and evaluation of novel therapeutic approaches. Methods: Primary limbal epithelial cells were isolated from biopsies of two patients with distinct PAX6 variants and from a healthy donor. Immortalization was performed by InSCREENex GmbH. The resulting cell lines were characterized using microscopy, BrdU assay, qPCR, flow cytometry, immunocytochemistry, and mRNA sequencing. Results: Immortalized aniridia and control cell lines displayed typical polygonal epithelial morphology and comparable proliferation rates. Total PAX6 mRNA and protein levels were similar among groups; however, nuclear PAX6 immunosignals were significantly reduced in aniridia-derived lines. Expression of ABCG2, TP63, FOSL2, ALDH1A1, and FABP5 showed no significant differences, except for reduced ΔNp63α protein levels in one aniridia line. mRNA sequencing detected more than 14,000 transcripts, including subsets uniquely expressed in control and aniridia-derived lines. Conclusions: Immortalized aniridia limbal epithelial cell lines preserve key epithelial characteristics and overall transcriptomic similarity to controls while exhibiting disease-relevant molecular alterations. These cell lines represent models of PAX6-associated ocular surface disease.

Figures

Full text

1. Introduction

Aniridia-associated keratopathy (AAK) is one of the most prevalent and severe manifestations of congenital aniridia, a rare panocular disorder primarily resulting from PAX6 haploinsufficiency in approximately 90% of cases. PAX6 encodes a transcription factor located on chromosome 11p13 and plays a critical role in ocular development. To date, over 600 different PAX6 pathogenic variants have been identified [1,2,3]. In addition to AAK, congenital aniridia may present with a range of ocular abnormalities, including dry eye disease, iris hypoplasia, cataract, nystagmus, and glaucoma [1]. Among these, AAK is the most frequent and vision-threatening complication, often leading to progressive corneal opacification and blindness despite advanced therapeutic interventions. AAK is characterized by limbal stem cell dysfunction, corneal neovascularization, sporadic goblet cell occurrence, and inflammatory cell infiltration [4]. The underlying mechanisms contributing to the development and progression of AAK remain incompletely understood and are interpreted differently. While some studies primarily attribute AAK to limbal epithelial stem cell (LESC) insufficiency, others emphasize the progressive degradation of the limbal stem cell niche. These processes are likely interrelated and may mutually reinforce each other [5,6]. In addition, chronic wound healing processes and abnormal corneal cell differentiation have been observed during AAK progression [7]. This aberrant differentiation is associated with increased corneal epithelial cell proliferation and heightened susceptibility to oxidative stress [1,7,8]. AAK exhibits varying degrees of severity. According to the classification by Lagali et al. [9], grade 0 is characterized by a clear cornea without vascularization, whereas grade 1 is defined by fine neovascularization extending less than 1 mm beyond the limbus. With disease progression, increasing vascularization and pannus formation are observed, culminating in vascularization of the corneal center at grade 3 or in a fully vascularized and opaque corneal pannus at grade 4. The underlying pathophysiology of AAK is primarily attributed to limbal stem cell deficiency and progressive deterioration of the LESC niche [10]. However, the functional characteristics of LESCs in different PAX6 pathogenic variant types remain poorly understood. Compounding this challenge is the absence of suitable in vitro cell models for basic and translational research.

To address this gap, human in vitro models are essential for studying the pathological mechanisms of AAK. These models enable comparison between limbal epithelial cells (LECs) from aniridia patients and healthy individuals, thereby facilitating the identification of therapeutic targets relevant to the in vivo condition. Previously, PAX6^+/−^ LECs generated by CRISPR/Cas9-mediated editing have served as a surrogate model of PAX6 haploinsufficiency [11]. Additional models include patient-derived induced pluripotent stem cells (iPSCs) and primary LECs subjected to PAX6 knockdown via siRNA techniques [12,13].

Due to the rarity of congenital aniridia, access to primary patient-derived cells for research purposes is limited. Therefore, the aim of this study was to establish and comprehensively characterize immortalized limbal epithelial cell lines derived from patients with PAX6 haploinsufficiency-associated congenital aniridia, as well as from a healthy donor. These well-defined cell models provide a reliable and reproducible platform for long-term experimental studies, facilitating mechanistic investigations and the development and evaluation of novel therapeutic approaches.

2. Materials and Methods

2.1. Tissue Collection and Cell Immortalization

Limbal epithelial cells selected for immortalization were derived from two patients with congenital aniridia, each carrying distinct PAX6 pathogenic variants. Both patients provided written informed consent for biopsy collection after receiving detailed information. Corneal donors or their legal representatives provided written consent for the use of corneal tissue for research purposes. Comprehensive demographic data for these two patient donors and one healthy donor are provided in Table 1. Limbal epithelial cells from aniridia patients (AN-pLEC) were obtained during planned surgical procedures, such as corneal transplantation or cataract surgery, via biopsies measuring approximately 1–2 mm^3^. The healthy control sample (pLEC) was collected from a corneoscleral donor rim provided by the Klaus Faber Center for Corneal Diseases, including the Lions Eye Bank Saar-Lor-Lux, Trier/Westpfalz.

Biopsies obtained from both aniridia patients and healthy corneoscleral donor rim were subjected to overnight enzymatic digestion at 37 °C using collagenase A (Roche Pharma AG, Basel, Switzerland) and cultured in Keratinocyte Growth Medium 3 (KGM3; Cat. No. C-20021, PromoCell GmbH, Heidelberg, Germany), supplemented with the manufacturer’s supplement mix and CaCl_2_, as previously described [14]. To separate limbal epithelial cells (pLEC) from limbal stromal cells (pLSC), the resulting cell suspension was filtered through a Flowmi™ cell strainer (40 µm pore size; SP Bel-Art Scienceware, Wayne, NJ, USA). Epithelial cells retained in the strainer were washed with pre-warmed Trypsin-EDTA (0.05% trypsin/0.02% EDTA) to obtain a single-cell suspension and collected into a centrifuge tube. After the addition of DMEM/F12 containing 5% FCS, the suspension was centrifuged, and the cell pellet was resuspended in KGM3 medium. Cells were seeded into a single well of a six-well plate, with medium changes every other day. LECs reached approximately 80% confluence within 5–7 days.

Following expansion, cells were cryopreserved in liquid nitrogen for future use. Immortalization of the cell lines was conducted by InSCREENex GmbH (Braunschweig, Germany) [15] using a defined combination of transgenes (Table 2).

The resulting immortalized cell lines are hereafter referred to as AN1-iLEC, AN2-iLEC, and ctrl-iLEC. Following thawing, the immortalized cell lines were cultured in 6-well plates coated with epithelial cell coating solution (InSCREENex GmbH) and maintained in KGM3 medium at 37 °C in a humidified atmosphere with 5% CO_2_. The medium was changed every third day until cells reached approximately 80% confluence. To ensure that any observed differences were not attributable to variation in passage number, all experiments involving the two aniridia-derived cell lines and the control cell line were performed using passages with minimal differences.

2.2. Cell Proliferation Assay Using BrdU ELISA

The proliferation rate of the established cell lines was evaluated using the 5-bromo-2′-deoxyuridine (BrdU) Cell Proliferation ELISA, colorimetric assay (Cat. No. 11647229001; Merck KGaA, Darmstadt, Germany). Ctrl-iLEC, AN1-iLEC, and AN2-iLEC were seeded in 96-well plates at a density of 15,000 cells/cm^2^ in 100 µL of KGM3 medium per well, in triplicate. Following a 48-h incubation period, BrdU incorporation was assessed according to the manufacturer’s protocol. BrdU incorporation was assessed using cells from eight independent passages per cell line group, with each experiment conducted in triplicate as technical replicates. All analyses were performed using passages 20 to 46 across all cell lines.

2.3. Transmission Electron Microscopy (TEM)

Control and aniridia-derived cell lines were seeded on either tissue culture-treated plastic coverslips or glass coverslips pre-coated with epithelial cell coating solution (InSCREENex GmbH, Braunschweig, Germany) and cultured in 6-well plates to form monolayers. Upon reaching ~90% confluency, cells were fixed overnight at 4 °C in 2% (w/v) paraformaldehyde (PFA) and 2.5% (v/v) glutaraldehyde in PBS (pH 7.4) to preserve ultrastructural integrity. The fixed samples were rinsed with 100 mM cacodylate buffer and post-fixed in 1% (w/v) osmium tetroxide (OsO_4_) in 100 mM cacodylate buffer for 1 h at 4 °C to enhance membrane contrast. After several washes in cacodylate and sodium maleate buffers, samples were stained en bloc with 2% (w/v) uranyl acetate in 50 mM Na-maleate buffer for 3 h at 4 °C to improve the electron density of cellular components. Subsequently, the samples were dehydrated through a graded ethanol series (50%, 70%, 80%, 90%, and 99%), followed by a 20-min incubation in 100% acetone at room temperature. Infiltration was performed using graded mixtures of acetone and Epon resin (3:1, 1:1, 1:3 v/v) for 3 h each at room temperature on a rotating wheel. Finally, coverslips were incubated in 100% Epon overnight, followed by polymerization at 60 °C for 48 h.

2.4. RT-qPCR Analysis of Congenital Aniridia-Associated Markers

To validate the identity of the established cell lines, we analyzed mRNA expression levels of PAX6, as well as additional direct PAX6 target genes, including ATP-binding cassette sub-family G member 2 (ABCG2), tumor protein p63 (TP63), and FOS-like antigen 2 (FOSL2). We also examined aldehyde dehydrogenase 1 family member A1 (ALDH1A1) and fatty acid-binding protein 5 (FABP5), which have been associated with congenital aniridia. Detailed primer information is provided in Supplemental Table S1.

After reaching confluency in 6-well plates, AN1-iLEC, AN2-iLEC, and ctrl-iLEC cells were lysed using 300 µL of SKP buffer (included in the RNA/DNA/Protein Purification Plus Micro Kit, Cat. No. 47700, Norgen, Thorold, ON, Canada) supplemented with 3 µL β-mercaptoethanol. RNA and protein were isolated according to the manufacturer’s instructions. To prevent genomic DNA contamination, total RNA was purified using a gDNA removal column, followed by optional on-column DNase digestion. The isolated RNA was stored at −80 °C until further use. RNA concentration and purity were assessed using a UV/VIS spectrophotometer (Analytic Jena AG, Jena, Germany).

For RT-qPCR, 500 ng of total RNA was reverse-transcribed into cDNA using the OneTaq RT-PCR Kit (New England Biolabs GmbH, Frankfurt am Main, Germany). Quantitative PCR was performed with ACEq DNA SYBR Green Master Mix (Vazyme Biotech, Nanjing, China) on a QuantStudio 5 Thermocycler (Applied Biosystems, Waltham, MA, USA). Glucuronidase beta (GUSB) and TATA-box binding protein (TBP) served as reference genes and were amplified under identical conditions to the target genes. RT-qPCR was conducted on six independent passages per cell line group.

2.5. Flow Cytometry of AN-iLECs and Ctrl-iLECs

Flow cytometry (FC) analysis was conducted on AN1-iLEC, AN2-iLEC, and ctrl-iLEC cell lines to assess the expression of PAX6, ABCG2, ΔNp63α, FOSL2, ALDH1A1 and FABP5. Cells from eight independent passages (Supplementary Table S2) were harvested using Trypsin-EDTA solution (Cat. No. T3924, Merck KGaA, Darmstadt, Germany).

With the exception of ABCG2, all antibodies targeted intracellular proteins; therefore, cells were fixed and permeabilized using the Transcription Factor Staining Buffer Set (Fix-Perm; Cat. No. 130-122-981, Miltenyi Biotec B.V. & Co. KG, Bergisch Gladbach, Germany), according to the manufacturer’s instructions. PAX6 detection was performed using a phycoerythrin (PE)-conjugated antibody, with Fix-Perm-treated but unstained cells serving as the negative control. All other primary antibodies were unconjugated, and detection was achieved using either fluorescein isothiocyanate (FITC)- or allophycocyanin (APC)-conjugated secondary antibodies. Secondary antibody-only controls were included for each antibody. A comprehensive list of all primary and secondary antibodies used is provided in Supplementary Table S3.

Flow cytometry was performed using a CytoFLEX flow cytometer (Beckman Coulter GmbH, Krefeld, Germany). Cell debris was excluded via forward and side scatter (FSC/SSC) gating, and single cells were identified using FSC-A vs. FSC-H gating. Fluorescence intensities were analyzed using CytExpert software version 2.6 and visualized as histograms. Data were collected over a five-month period across different passages. To account for inter-assay variability, the mean fluorescence intensity (MFI) of healthy ctrl-iLECs was used as an internal reference. Relative fluorescence intensity for each test sample (healthy or aniridia-derived) was normalized using the following formula: Relative Fluorescence Intensity = MFI (test sample)/MFI (internal control) [16].

2.6. Western Blot Analysis

PAX6 protein expression in aniridia and control cell lines was assessed by Western blot. For each sample, 30 µg of total protein was extracted from cell lysates using the RNA/DNA/Protein Purification Plus Micro Kit (Cat. No. 47700; Norgen, Thorold, Canada). Primary limbal epithelial cells served as a positive control. Samples were boiled for 5 min at 95 °C and separated on NuPAGE™ 4–12% bis-tris precast gels (Thermo Fisher Scientific GmbH, Dreieich, Germany). Proteins were then transferred to nitrocellulose membranes using the Trans-Blot Turbo Transfer System (Bio-Rad, Hercules, CA, USA). Membranes were incubated with a rabbit polyclonal anti-PAX6 primary antibody (Cat. No. AB2237; Sigma-Aldrich Chemie gmbH, Taufkirchen, Germany) diluted at 1:1000 in WesternFroxx anti-rabbit HRP solution containing a blocking reagent and the secondary antibody (BioFroxx GmbH, Einhausen, Germany). Protein bands were visualized using Western Lightning Plus-ECL chemiluminescence reagent (PerkinElmer, Waltham, MA, USA). Protein bands were visualized using the iBright Imaging System (Invitrogen, Waltham, MA, USA). Total protein normalization (TPN) was performed using the No-Stain™ Protein Labeling Reagent (Cat. No. A44717; Invitrogen). Membranes were incubated in the prepared No-Stain labeling solution for 10 min and imaged with the iBright™ FL1500 Imaging System (Invitrogen) to assess total protein content for normalization.

2.7. Immunofluorescence Microscopy

For immunolabelling of control and aniridia-derived limbal epithelial cell lines, glass coverslips were first coated with epithelial cell coating solution (InSCREENex GmbH, Braunschweig, Germany). Cells were seeded in 6-well plates containing the coated coverslips and cultured in KGM3 medium. Upon reaching ~80% confluence, cells were gently washed with phosphate-buffered saline (PBS) and subsequently fixed in 4% paraformaldehyde for 30 min at room temperature (RT). Following fixation, cells were permeabilized using ice-cold pure methanol for 5 min. Non-specific binding was blocked by incubating the samples in a blocking buffer containing 5% fetal calf serum (FCS) and 0.3% Triton X-100 in PBS for 2 h at RT. Primary antibody incubation was carried out overnight at 4 °C using rabbit anti-PAX6 antibody (1:1000; AB2237, Merck, Darmstadt, Germany) and mouse anti-Cytokeratin 12 antibody (1:1000; sc-515882, Santa Cruz Biotechnology, Inc., Dallas, TX, USA) diluted in incubation buffer (1% FCS and 0.1% Triton X-100 in PBS). The next day, coverslips were washed three times with PBS to remove unbound primary antibody and incubated for 2 h at RT with Alexa Fluor 568-conjugated anti-rabbit secondary antibody (1:1000 dilution; A10042, Thermo Fisher Scientific), Alexa Fluor 488-conjugated anti-mouse secondary antibody (1:1000 dilution; A21200, Thermo Fisher Scientific) and the nuclear stain NucSpot^®^ Live 650 (Ex/Em: 655/681 nm; 1:3000 dilution; Cat. No. 40082-T, Biotium, Fremont, CA, USA) in incubation buffer. After a final PBS wash, coverslips were mounted using Fluoromount (Cat. No. F4680, Merck, Darmstadt, Germany). Fluorescent images were captured using a Nikon A1R confocal laser scanning microscope (Nikon Europe B.V., Amstelveen, The Netherlands) with a 60×/1.40 NA oil immersion objective and processed using NIS Elements software (version AR 3.2, Düsseldorf, Germany) [17,18].

Immunofluorescence Microscopy Analysis of PAX6 and FABP5 Signal Intensities

Image analysis of the confocal images was performed using ImageJ software (Fiji, version 1.54p). To ensure consistency across samples, confocal image acquisition was conducted using the “re-use camera settings” function in the NIS Elements software, thereby maintaining identical imaging parameters for both control and aniridia-derived cell lines. High-resolution confocal images were captured, and regions of interest (ROIs) were manually drawn around individual LECs using the freehand selection tool in ImageJ. ROIs were defined based on the visible borders of single cells. Nuclear regions within the same cells were identified using the nuclear stain NucSpot^®^ 650 (far-red channel), and the same ROIs were applied to both the PAX6 (568 nm) and nuclear staining (650 nm) channels. The mean gray value for each ROI was measured in the PAX6 channel (red, laser 568) and normalized to the corresponding nuclear stain signal (far-red, laser 650) to account for potential cell-to-cell variability. This normalization enabled relative quantification of PAX6 signal intensity per cell. Normalized fluorescence intensities were then averaged, and the data were presented as mean ± standard deviation.

2.8. Whole Transcriptome Sequencing and Data Analysis

Whole transcriptome sequencing was performed by the Sequencing Unit of the Core Facility for Molecular Single Cell and Particle Analysis at Saarland University using a DNBSEQ-G400RS platform (MGI Tech, Shenzhen, China) with a 100 bp paired-end sequencing strategy. RNA-Seq libraries were generated using the MGIEasy rRNA Depletion Kit and the MGIEasy Universal Library Prep Set, following the manufacturer’s protocols. Read alignment and quantification were carried out using the mRNA module of the snakePipes workflow suite [19].

Sequence alignment to the GRCh38 human reference genome was conducted using the STAR aligner [20]. Gene-level quantification was subsequently performed with FeatureCounts [21]. To ensure high-quality data, sequencing reads underwent quality control using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/; URL accessed on 13 January 2025) and summary reports were generated with MultiQC [22]

Raw read counts were normalized using the variance stabilizing transformation (VST) function from the DESeq2 package [23]. The resulting transformed matrix was then used for Uniform Manifold Approximation and Projection (UMAP) visualization and unsupervised clustering of gene expression profiles.

2.9. Statistical Analysis

Fold changes in mRNA expression levels were calculated using Microsoft Excel (Microsoft, Redmond, WA, USA). Normality of data distribution was assessed using the Shapiro–Wilk test. For normally distributed data, one-way ANOVA followed by Dunn’s multiple comparisons test was applied. For data not meeting the assumptions of normality (as determined by the Shapiro–Wilk test), the non-parametric Kruskal–Wallis test followed by Dunn’s post hoc test was used. Statistical significance was defined as a p-value < 0.05. All data analyses and graphical representations were performed using GraphPad Prism software (version 10.4.2, GraphPad Software, San Diego, CA, USA). Box plots illustrating the distribution of PAX6 immunosignal intensity were generated using OriginPro 2025 (OriginLab Corporation, Northampton, MA, USA).

3. Results

3.1. Cell Morphology

Immortalized cell lines derived from a healthy donor (ctrl-iLEC) and from patients with aniridia (AN1-iLEC and AN2-iLEC), cultured in KGM3 medium, exhibited a typical polygonal epithelial morphology with well-defined cell borders under phase-contrast microscopy (Figure 1A). The cell morphology remained consistent across higher passage numbers, as shown in Supplementary Figure S1. To further assess cellular morphology at the ultrastructural level, TEM was performed. No discernible differences were observed in the overall cellular architecture or organelle structures between the control and aniridia-derived limbal epithelial cell lines (Figure 1B).

3.2. Cell Proliferation Assay Using BrdU ELISA

To evaluate proliferative capacity, a BrdU-based proliferation assay was performed using different passages of ctrl-iLEC (n = 8), AN1-iLEC (n = 7), and AN2-iLEC (n = 8). Each passage was considered an independent experiment and analyzed in triplicate as technical replicates. Experiments were conducted using passages 20 to 46 across all cell lines. The proliferation rates of AN1-iLEC and AN2-iLEC did not differ significantly from that of ctrl-iLEC (p = 0.892 and p = 0.754, respectively) (Figure 2).

3.3. PAX6 and Its Direct Target Genes

Congenital aniridia is most commonly caused by mutations in the PAX6 gene, a key transcription factor involved in ocular development [24]. PAX6 mRNA and total protein levels did not differ significantly between AN1-iLEC/AN2-iLEC and ctrl-iLEC based on qPCR (Figure 3A) and flow cytometry (Figure 4) (p ≥ 0.427). Western blotting did not demonstrate any significant variation in PAX6 protein level among AN1-iLEC/AN2-iLEC, and Ctrl-iLEC (p ≥ 0.853) (Supplementary Figure S2).

To gain further insight, mRNA sequencing was performed on immortalized cell lines derived from primary limbal epithelial cells of aniridia patients and a healthy donor. As only one biological replicate per sample was analyzed, statistical comparison was not feasible. Across the three samples, 14,846 mRNAs were identified in AN1-iLEC, 14,644 in AN2-iLEC, and 14,668 in ctrl-iLEC. Notably, 412 mRNAs were uniquely expressed in AN1-iLEC and 344 in AN2-iLEC. Venn diagram analysis using the Venny tool (https://bioinfogp.cnb.es/tools/venny/index.html; URL accessed on 30 January 2025) revealed 13,700 mRNAs commonly detected in all three samples (Figure 3B). These 13,700 shared transcripts were used for downstream analysis. Expression values from AN1-iLEC and AN2-iLEC were averaged to ctrl-iLEC to calculate fold-change ratios. Genes with a fold-change <0.5 or >2.0 were selected, resulting in a subset of 2250 differentially expressed mRNAs. This subset was visualized using a hierarchical clustering heatmap generated via ClustVis (https://biit.cs.ut.ee/clustvis/; URL accessed on 31 January 2025) (Figure 3C). A complete list of these transcripts is available in Supplementary Table S4.

Known direct target genes of PAX6 include ABCG2, TP63, and FOSL2, while ALDH1A1 and FABP5 have been associated with aniridia pathogenesis [25]. Interestingly, previous studies have reported no significant alterations in the mRNA expression levels of these genes in pLECs derived from aniridia patients [25,26]. To evaluate whether these findings extend to the immortalized cell models, we assessed the mRNA expression of these genes in aniridia-derived and control limbal epithelial cell lines. Quantitative PCR analysis revealed no significant differences in mRNA levels between aniridia cell lines (AN1-iLEC and AN2-iLEC) and the control cell line (ctrl-iLEC) for any of the target genes examined (p ≥ 0.138) (Figure 3A). Protein-level analysis by flow cytometry corroborated the transcript-level findings, showing no significant differences in protein expression between aniridia and control cell lines for most markers (p ≥ 0.084). The only exception was a significant reduction in ΔNp63α protein level in AN2-iLEC compared to ctrl-iLEC (p = 0.045) (Figure 4).

Since PAX6 mRNA and total protein levels did not differ significantly between groups based on sequencing, flow cytometry, and Western blot analysis, we further examined PAX6 protein localization and intensity using immunofluorescence microscopy. Both AN1-iLEC and AN2-iLEC demonstrated a marked reduction in normalized nuclear PAX6 protein expression compared to ctrl-iLEC (p < 0.0001), indicating a potential functional impairment of PAX6 in the aniridia cell lines (Figure 5A–C). Immunolabeling of PAX6 together with cytokeratin KRT12, which labels intermediate filaments, and a nuclear stain revealed strong colocalization of PAX6 and nuclear signals in the healthy control cell line. In contrast, a slight reduction in PAX6–nuclear colocalization was observed in the AN1-iLEC and AN2-iLEC cell lines, suggesting a redistribution of PAX6 toward a more diffuse cytoplasmic localization (Figure 5D).

4. Discussion

In this study, we present newly established immortalized limbal epithelial cell lines derived from patients with congenital aniridia, along with a corresponding control cell line. To date, more than 600 distinct mutations in the PAX6 gene have been identified [27], contributing to the wide variability in clinical presentation and severity of AAK. Although the exact mechanisms underlying AAK remain incompletely understood, current evidence suggests that limbal stem cell deficiency and disruption of the limbal stem cell niche play central roles in disease pathogenesis [5,10]. Given the rarity of aniridia and the limited availability and lifespan of pLECs from affected individuals, there is a pressing need for robust and reproducible in vitro models to study the molecular mechanisms of AAK and to identify new therapeutic targets. Several models have been developed to address this need, including siRNA-based knockdown systems [13], CRISPR/Cas9-mediated PAX6 genome editing in limbal epithelial cell lines [11], patient-derived induced pluripotent stem cells (iPSCs) [12], and organoid systems [28]. In addition, various animal models have contributed to advancing our understanding of AAK pathogenesis [29].

The aim of our study was to establish immortalized cell lines from aniridia-derived primary limbal epithelial cells. pLECs derived from patients with congenital aniridia carrying different pathogenic PAX6 variants, as well as pLECs from a healthy donor, were used to generate cell lines and to distinguish disease-related alterations from changes associated with the immortalization process (Table 1). In the aniridia subjects whose pLECs were immortalized, a C-terminal extension (CTE) mutation was identified in AN1-iLEC, while a PAX6 deletion (nonsense-mediated RNA decay) was confirmed in AN2-iLEC. CTE variants account for approximately 5% and PAX6 deletions for about 25% of classical congenital aniridia cases [27]. Thus, these newly generated cell lines serve as a valuable model for relatively rare forms of congenital aniridia and may facilitate deeper insights into the underlying pathomechanisms. The immortalized cell lines were generated using a defined transgene combination designed to preserve the original phenotype, as described previously [15]. For the immortalization process, a third-generation self-inactivating lentiviral vector was employed, in which gene expression was driven by an internal SV40 promoter. Certain vectors also contained a neomycin resistance cassette, expressed from a bicistronic transcript via the poliovirus internal ribosomal entry site (IRES). This applied to vectors encoding the following genes: Bmi1, HCV core protein, E6, E7, FOS, MYC, ID1, ID2 ID3, Zfp42, NANOG and SV40 (Table 2) [30]. It should be emphasized that some of these genes may be directly or indirectly linked to PAX6 and aniridia, although the expression of most of them has not yet been analyzed in detail. Furthermore, because immortalization inherently alters cell-cycle dynamics, and cell-cycle dysregulation is also a feature of PAX6-related congenital aniridia [30], these factors should be considered as potential limitations when working with cell lines carrying pathogenic PAX6 variants. Nevertheless, comparing mRNA expression profiles between immortalized aniridia limbal epithelial cell lines (AN-iLECs) and primary aniridia-derived limbal epithelial cells (AN-pLECs) [31] for PAX6, ABCG2, TP63, ALDH1A1, FABP5, and FOSL2, the descriptive comparison of fold-change values revealed largely similar expression patterns (Table S4 ). These findings indicate that AN1-iLEC and AN2-iLEC remain robust in vitro models of limbal epithelial cell cultures from patients with congenital aniridia.

Consistent with earlier reports, the immortalized cell lines exhibited significantly extended proliferative capacity compared to primary LECs, which typically can only be passaged 4–5 times [25,30]. Notably, the proliferation rate of the immortalized cells remained stable across both low and high passage numbers (up to passage 35), indicating robust cellular viability and consistent growth characteristics. Morphologically, both the ctrl-iLEC and aniridia-derived AN-iLEC immortalized cells closely resembled primary LECs, as confirmed by phase-contrast and transmission electron microscopy. The morphological similarity between primary LECs and primary aniridia LECs was previously described by Schlötzer-Schrehardt et al. [5,10]. These findings suggest that the immortalized cell lines also retained the characteristic epithelial morphology observed in previous studies of primary control and aniridia LECs [25].

Importantly, PAX6 haploinsufficiency does not necessarily present as a measurable reduction in PAX6 mRNA or protein levels [25,26]. This was also true for both of our AN-iLEC lines, which showed no significant differences in PAX6 mRNA or protein expression compared with ctrl-iLEC, as assessed by qPCR and flow cytometry. Instead, PAX6 deficiency has been associated with aberrant subcellular localization, most notably a shift from the nucleus to the cytoplasm in suprabasal and superficial epithelial layers [10,25,26]. In our immortalized aniridia cell lines, the reduced nuclear PAX6 protein levels may indicate a PAX6 mislocalization pattern, thereby further validating the fidelity of AN-iLEC as an in vitro model of AAK.

ABCG2 and tumor protein 63 (TP63) are putative stem cell markers predominantly expressed in the basal epithelial cells of the limbus [32]. Consistent with previous findings in primary aniridia limbal epithelial cells (AN-pLECs) [26], our immortalized aniridia-derived cell lines showed no significant differences in ABCG2 and TP63 mRNA expression compared to control cells, as determined by qPCR. However, flow cytometric analysis revealed a modest reduction in ΔNp63α expression in the AN2-iLEC line relative to the ctrl-iLEC. This finding indicates a slight decrease in stemness in AN2-iLEC, whereas no such change was observed in AN1-iLEC. It is likely that the underlying CTE pathogenic variant and the PAX6 deletion play a decisive role in these differences, although further evaluation is required.

Alterations in the retinoic acid (RA) signaling pathway have been previously reported in conjunctival cells from aniridia patients as well as in siRNA-based PAX6 knockdown models of limbal epithelial cells [33,34,35,36]. In those studies, two RA pathway genes, ALDH1A1 and FABP5, were downregulated, suggesting a direct regulatory role of PAX6 on their expression. In contrast, our results did not replicate these findings, as neither ALDH1A1 nor FABP5 expression was significantly altered in our aniridia cell lines. This observation is consistent with transcriptomic data from primary aniridia limbal epithelial cells, which also showed no changes in these genes [31]. Although PAX6 is known to function as an inducer of RA signaling [37], these discrepancies may reflect a dosage-dependent effect, in which subtle reductions in PAX6 levels are insufficient to elicit measurable changes in RA pathway gene expression at the mRNA or protein level under our experimental conditions. Another gene reported to be deregulated in limbal epithelial cells from aniridia patients is FOSL2, a transcription factor and direct downstream target of PAX6 that has been linked to corneal opacity [38]. Previous studies demonstrated that FOSL2 expression decreases following siRNA-mediated PAX6 knockdown [38]. In contrast, in our study, FOSL2 protein levels remained unchanged in both AN1-iLEC and AN2-iLEC cell lines compared to ctrl-iLEC. These findings suggest that, similar to ALDH1A1 and FABP5, FOSL2 expression may require a critical threshold of PAX6 reduction before a measurable decrease becomes evident.

To better understand the development of AAK and to advance treatment strategies, appropriate cell models are essential, as access to patient-derived primary cells is very limited. The different models currently available provide complementary opportunities for research. Their AAK-associated gene expression profiles are summarized in Table S4. The PAX6^+/−^ aniridia cell line, generated by introducing a nonsense mutation associated with AAK via CRISPR/Cas9, offers a stable model in which the direct consequences of this mutation can be investigated under standardized conditions [11]. By contrast, the siRNA–PAX6 model relies on primary limbal epithelial cells, but greater biological variability is expected with this approach [13]. Induced pluripotent stem cells (iPSCs) derived from aniridia patients represent another valuable model for investigating disease mechanisms [12], although their use is technically demanding and relatively costly. A major advantage of the field is the ability to compare findings across these diverse models.

It is important to note that limbal stem cell deficiency and the deterioration of the limbal niche in AAK are not solely driven by intrinsic defects in limbal epithelial cells. Rather, AAK is believed to result from complex interactions between limbal epithelial cells and other niche components, such as mesenchymal stem cells and stromal support cells [10]. The extended proliferative capacity and preservation of key phenotypic traits make the AN-iLEC and ctrl-iLEC lines valuable tools for future research. These cell lines offer a promising platform for co-culture models and for further investigations into the cellular and metabolic mechanisms underlying AAK pathogenesis.

5. Conclusions

The immortalized AN-iLEC lines exhibit typical epithelial morphology and comparable global gene expression to control cells, but also display key molecular alterations, including reduced nuclear PAX6 and ΔNp63α expression. These disease-relevant features highlight the value of AN-iLECs as a robust and accessible in vitro tool for investigating PAX6-related ocular surface disease. They provide a valuable platform for advancing our understanding of limbal epithelial stem cell biology and may support the development of targeted therapeutic strategies for the treatment or prevention of AAK.

6. Limitations

Due to the rarity of congenital aniridia, the availability of donor material suitable for the generation of cell lines is limited. Consequently, a direct comparison of cells obtained using different methodologies before and after immortalization was not feasible.