mRNA Sequencing of Limbal Epithelial Cells and mRNA/miRNA Profiling of Limbal Stromal Cells in PAX6-Related Congenital Aniridia

Abstract

The dysfunction of limbal epithelial cells (LECs) and limbal stromal cells (LSCs) in congenital aniridia remains incompletely understood. We aimed to analyze mRNA expression profiles of primary human LECs and LSCs, as well as microRNA (miRNA) expression in LSCs, from patients with congenital aniridia (AN-LECs and AN-LSCs). mRNA sequencing of primary human LECs and mRNA and miRNA sequencing of LSCs were performed from patients with aniridia and healthy controls. Gene ontology and pathway analyses were used to evaluate biological processes, cellular components, and molecular functions. Selected deregulated mRNAs and miRNAs were validated by quantitative real-time PCR (RT-qPCR). A total of 188 differentially expressed genes (DEGs) were identified in AN-LECs, and 3001 DEGs in AN-LSCs. In AN-LECs, the top hub genes were associated with inflammatory and interferon-related responses. In contrast, AN-LSCs showed predominant deregulation of mitochondrial and metabolic genes. Pathway analysis revealed involvement of inflammation-related pathways in AN-LECs and metabolic pathways in AN-LSCs. Additionally, 48 deregulated miRNAs were identified in AN-LSCs. This study provides comprehensive mRNA profiles of LECs and LSCs and miRNA profiles of LSCs in congenital aniridia. The findings emphasize the importance of LSC influence and offer insights into molecular mechanisms underlying aniridia-associated keratopathy (AAK), supporting future research and potential therapeutic target identification.

1. Introduction

Congenital aniridia is a rare panocular disease, predominantly caused by PAX6 haploinsufficiency [1]. During embryonic eye development, PAX6 functions as a key transcription factor, playing an essential role in multiple ocular tissues [2]. To date, over 600 distinct mutations in the PAX6 gene have been identified [3]. For patients, PAX6 haploinsufficiency results in abnormalities affecting various structures of the eye. Beyond the characteristic partial or complete absence of the iris, these include impaired Schlemm’s canal, often leading to glaucoma, and altered meibomian glands, which can disrupt the tear film, causing dry eye disease. The composition of the tear fluid in affected individuals exhibits elevated interleukin levels and other inflammatory proteins [4,5,6]. Additionally, conjunctival epithelial cells in aniridia patients show altered mRNA and miRNA expression compared to healthy individuals [7], a difference that is also reflected at the protein level [8]. To better understand the regulatory mechanisms underlying the development of aniridia-associated keratopathy (AAK), cellular and animal models are being developed to facilitate research aimed at preserving vision in affected individuals for as long as possible [9,10,11,12].

The limbal epithelial stem cells (LESCs) residing in the limbal niche differentiate into transient amplifying cells, which subsequently develop into mature corneal epithelial cells forming the outermost corneal layer. It is well established that, in addition to limbal stem cell deficiency, progressive degradation of the limbal stem cell niche occurs, potentially contributing to AAK progression [13]. While most studies have focused on limbal epithelial cells (LECs), the limbal stem cell niche also contains additional cell populations, including mesenchymal stem cells (MSCs), the progenitors of stromal cells [14,15]. PAX6 plays a crucial role in maintaining homeostasis of the limbal stem cell niche. Reduced PAX6 expression in LECs may contribute to the degradation of this niche, as observed in AAK. Altered PAX6 expression may also impair epithelial cell differentiation and is indirectly linked to inflammatory processes. To date, the role of limbal stromal cells (LSCs) has been investigated only to a limited extent. These cells support corneal regeneration during wound healing and exhibit immunomodulatory, anti-inflammatory, and angiogenic properties [16,17,18,19,20]. It is therefore plausible that LSCs play key roles in preserving the microenvironment necessary for stem cell niche maintenance. Understanding the role of both LECs and stromal cells may provide deeper insights into the pathophysiology of AAK and aid in the development of new therapeutic strategies. These cells are rarely studied in aniridia research. Beyond cell and animal models, mRNA and miRNA sequencing of patient-derived samples can help identify deregulated genes and miRNAs that, alongside PAX6, may significantly contribute to the development of AAK. The findings of this study provide a comprehensive overview of deregulated mRNAs in limbal epithelial cells and both mRNAs and miRNAs in limbal stromal cells from aniridia patients. These dysregulated molecules may play a crucial role in AAK progression and could serve as potential targets for novel therapeutic strategies.

2. Materials and Methods

2.1. Tissue Collection and Processing

Detailed demographic data of all patients and donors are presented in Supplementary Tables S1 and S2. Limbal biopsies from aniridia patients were collected during surgery, with a tissue size of 1–2 mm³. Biopsies from healthy donors were obtained from corneal donors at the Klaus Faber Center for Corneal Diseases, including the Lions Eye Bank.

The biopsies were enzymatically digested overnight at 37 °C using Collagenase A (Roche Pharma AG, Basel, Switzerland) in Keratinocyte Growth Medium 3 (KGM3) (cat. No. C-20021, PromoCell GmbH, Heidelberg, Germany), supplemented with the provided supplement mix and CaCl_2.

To separate LECs from LSCs, the cell suspension was passed through a Flowmi ™ cell strainer (Bel-Art SP Scienceware, Wayne, NJ, USA) with a 40 µm pore size. The stromal cells in the flow-through were cultured in DMEM/F12 medium supplemented with 5% fetal calf serum (FCS). Initially, LSCs were seeded in a six-well plate. Upon reaching confluence, they were passaged into a T25 flask and subsequently expanded in a T75 culture flask. For experiments, cells at passages 4–8 were used.

The epithelial cells retained in the cell strainer were washed with pre-warmed Trypsin-EDTA (0.05% trypsin/0.02% EDTA solution) to generate a single-cell suspension and rinse the epithelial cell fraction into a centrifuge tube. The resulting cell suspension was supplemented with DMEM/F12 + 5% FCS, centrifuged, and the cell pellet was resuspended in KGM3 medium. The cells were then seeded into one well of a six-well plate. The medium was changed every other day, and limbal epithelial cells (LECs) reached ~80% confluency after approximately 5 to 7 days.

Cells from five patients were available for mRNA sequencing of LECs, while samples from eight patients were used for mRNA and miRNA sequencing of LSCs. LECs derived from patients with aniridia generally exhibit lower proliferative capacity and yield fewer cells compared with stromal cells. Because sufficient amounts of mRNA from the same samples were required for post-sequencing validation experiments, only LECs from five patients could be included in the present study. In contrast, stromal cells typically generated higher cell numbers in culture, allowing the inclusion of a larger cohort for this cell type. All samples were processed individually and without pooling for all subsequent analysis.

2.2. RNA Isolation and Quality Control

After reaching confluency, AN-LECs and LECs from healthy subjects were lysed using 300 µL SKP Buffer from the RNA/DNA/Protein Purification Plus Micro Kit (Norgen, Thorold, ON, Canada, cat. no. 47700), with the addition of 3 µL β-mercaptoethanol. RNA and protein purification were carried out according to the manufacturer’s protocol. For AN-LSCs and LSCs cultured in T75 flasks, cells were harvested using trypsin/EDTA, and the resulting cell pellet was lysed with 300 µL SKP Buffer containing 3 µL β-mercaptoethanol, then stored at −80 °C until RNA isolation. RNA extraction from LSCs was performed using the RNA Isolation Kit (Norgen, cat. no. 48500) following the manufacturer’s instructions. For quality control, the total RNA concentration was measured using a UV/VIS spectrophotometer (Analytik Jena AG, Jena, Germany).

2.3. Whole Transcriptome and miNome Sequencing and Data Analysis

Transcriptome profiling was performed on LECs obtained from five individuals with congenital aniridia and five unaffected controls, as well as on LECs derived from eight patients with congenital aniridia and eight healthy donors. These analyses were conducted as separate experimental sets. RNA-Seq libraries were generated using the MGIEasy rRNA Depletion Kit in combination with the Universal Library Prep Set (MGI Tech, Shenzhen, China), following the manufacturer’s instructions. Sequencing was carried out on a DNBSEQ-G400RS platform at the Sequencing Unit of the Core Facility for Molecular Single Cell and Particle Analysis at Saarland University, applying a paired-end read length of 100 bp. RNA-Seq data processing and quantification were performed using the mRNA workflow implemented in snakePipes [21]. Reads were aligned to the human reference genome GRCh38 at the gene level using STAR [22], and gene-level read counts were generated with FeatureCounts [23]. Data quality was assessed using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/; accessed on 9 January 2025) and summarized with mulitQC to verify sequencing integrity [24]. Raw count data were subjected to variance stabilizing transformation 8VST using DESeq2 [25] and the transformed dataset was used for clustering analyses. Differential expression analysis was conducted with DESeq2 based on raw count values, and log fold changes were calculated using the Benjamini–Hochberg procedure, applying a false discovery rate (FDR) cutoff of 0.05.

For small RNA profiling, libraries were prepared using the MGIEasy Small RNA Library Prep Kit (MGI Tech, Shenzhen, China) in accordance with the manufacturer’s guidelines. Sequencing was conducted on the same DNBSEQ-G400RS system using a 50 bp single-end configuration. The resulting FASTQ files were processed with miRmaster 2.0 pipeline [26], which performs adapter trimming, read collapsing, and miRNA quantification based on miRbase version 22.1 [27]. miRNA expression levels were normalized as reads per million mapped miRNAs (rpmmm) to allow comparison between samples.

The sequencing datasets generated and analyzed during this study have been deposited in the Dryad repository: http://datadryad.org/stash/share/C5ffmIX9GQ2EGGkhY27yZ6mqQwbeJPKu5mWg596auE8, accessed on 26 February 2025.

2.4. Identification of microRNA-mRNA Interactions

In order to find potential target genes of deregulated miRNAs of AN-LSCs, the online tool miRTargetLink (https://ccb-compute.cs.uni-saarland.de/mirtargetlink2, accessed on 9 January 2025), was used in conjunction with strong validated targets based on miRTarbase (https://mirtarbase.cuhk.edu.cn, accessed on 9 January 2025).

2.5. Protein–Protein Interaction (PPI) Network and Hub Gene Identification

The PPI network of differentially expressed genes (DEGs) in AN-LECs and AN-LSCs was constructed separately using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING-db) (https://string-db.org, accessed on 9 January 2025). Cytoscape (version 3.10.3) was utilized to visualize the PPI interaction networks and analyze the relationships between the identified genes.

Hub genes were identified using the cytoHubba plug-in in Cytoscape, with the Maximal Clique Centrality (MCC) algorithm applied to select the top 10 hub genes. Additionally, the interaction networks of the identified hub genes were further analyzed to assess their potential roles in AAK-related molecular pathways.

2.6. Pathway Analysis and Gene Ontology (GO) Classification Analysis

To visualize the differentially expressed genes in AN-LECs and AN-LSCs, volcano plots were generated using GraphPad Prism software (Version 10.4.0), allowing for the identification of significantly upregulated and downregulated genes. To confirm alterations in biological pathways and perform GO classification analysis, the online tool GeneTrail 2 was utilized. The statistical analysis was conducted using an over-representation analysis, focusing on three GO categories, biological processes (BP), cellular components (CC), and molecular functions (MF), along with Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis to identify relevant molecular pathways. To further illustrate the biological variance between samples, heatmaps were generated using ClustVis 2.0 (https://biit.cs.ut.ee/clustvis/, accessed on 9 January 2025), providing an overview of gene expression patterns and clustering across different conditions.

2.7. RT-qPCR Validation of Deregulated mRNAs in AN-LECs and AN-LSCs and miRNAs in AN-LSCs

From the top 20 upregulated and downregulated genes identified in AN-LECs and AN-LSCs, a total of 10 candidate genes were selected for RT-qPCR validation. For miRNA validation, five candidates were chosen from the top 40 deregulated miRNAs. The selection criteria for both mRNAs and miRNAs included genes with high, low, and medium fold change to ensure a representative validation set.

To confirm the mRNA and miRNA expression data obtained from sequencing, RT-qPCR was performed on the same sample set. For mRNA validation, complementary DNA (cDNA) was synthesized from 500 ng total RNA using the One Taq RT-PCR Kit (New England Biolabs GmbH, Frankfurt am Main, Germany). qPCR was conducted using the ACEq DNA SYBR Green Master Mix (Vazyme Biotech, Nanjing, China) on a QuantStudio 5 PCR Thermocycler (Applied Biosystems, Waltham, MA, USA).

To ensure accurate normalization, Glucuronidase Beta (GUSB) and TATA-box binding protein (TBP) were used as reference genes, and all reactions were run under identical conditions as the target genes.

For miRNA validation, cDNA synthesis was performed using the miRCURY LNA miRNA RT Kit, followed by qPCR in duplicate using the miRCURY LNA miRNA SYBR Green PCR Kit (Qiagen GmbH, Hilden, Germany), both according to the manufacturer’s protocol. miR-103a-3p and snRNA RNU6 (Qiagen, Hilden, Germany) were used as endogenous controls for normalization. Relative expression levels (fold change) were calculated using the ΔΔCt method. The primers used for mRNA and miRNA validation are listed in Supplementary Table S3.

3. Results

3.1. Sequencing of Limbal Epithelial and Limbal Stromal Cells in Patients with Congenital Aniridia

3.1.1. mRNA Sequencing Profile of Limbal Epithelial Cells in Patients with Aniridia

RNA sequencing was performed on cultured LECs from aniridia patients (AN-LECs) and healthy limbal epithelial cells as controls. A total of 38,744 RNAs were detected, and after applying a cutoff threshold (fold change [FC] < −2.0 or >2.0, t-test adjusted p < 0.05), 188 DEGs were identified. Among these, 162 mRNAs were downregulated, and 26 mRNAs were upregulated, as illustrated in the volcano plot (Figure 1a).

Among the lowest expressed mRNAs in AN-LECs compared to healthy controls were KRT12, NRN1, CXCL10, MUC22, CXCL5, CXCL8, VNN2, and CXCL11. In contrast, SIX3, IL20RA, FOXE1, BRINP1, CRABP2, and PEX6 were among the highest expressed mRNAs in aniridia-derived LECs.

Table 1. The 20 most down- and upregulated mRNAs in (a) limbal epithelial cells (AN-LECs) and (b) limbal stromal cells (AN-LSCs) from patients with aniridia, ordered by increasing fold changes.

(a) Deregulated mRNAs in AN-LECs			(b) Deregulated mRNAs in AN-LSCs
Gene Symbol	Fold Change	Adjusted p-Value	Gene Symbol	Fold Change	Adjusted p-Value
KRT12	−895.20	0.0006	IVL	−31.52	<0.0001
NRN1	−181.38	<0.0001	THSD7A	−24.63	<0.0001
CXCL10	−59.25	<0.0001	BEX1	−20.80	0.0007
KRT81	−58.52	0.0001	TMPRSS11D	−20.29	0.0006
RSAD2	−42.95	<0.0001	PSG11	−19.34	<0.0001
EYA4	−31.95	0.0360	PSG7	−18.72	0.0025
IL36G	−28.52	<0.0001	TMPRSS11A	−18.67	0.0035
MUC22	−27.06	0.0077	MUC5AC	−17.26	0.0115
CLEC7A	−26.01	0.0136	CSF3	−15.01	0.0002
CXCL5	−25.86	<0.0001	PAH	−13.87	0.0060
SYN3	−24.62	0.0295	TSBP1	−13.68	0.0027
ITGB3	−20.88	0.0043	ELOA2	−13.08	0.0005
SPP1	−19.19	0.0001	IGSF23	−13.05	0.0037
ZBP1	−19.05	0.0323	NUTM2E	−12.99	0.0006
IFI44L	−17.29	0.0052	SLC26A8	−12.94	0.0011
BST2	−16.52	0.0004	GSTT2	−12.69	0.0387
CMPK2	−16.12	<0.0001	CPHXL	−12.36	0.0055
CXCL8	−14.95	<0.0001	HAL	−12.31	0.0091
HEPHL1	−14.93	0.0016	FGL2	−11.74	<0.0001
VNN2	−14.54	0.0466	NTF4	−11.08	0.0076
PIF1	2.21	0.0094	C2CD4D	9.97	0.0467
ADGRL2	2.32	<0.0001	IRX6	9.98	0.0340
ATP8B3	3.06	0.0496	KCND3	10.18	0.0035
ANKRD29	3.13	0.0202	PTH1R	10.32	<0.0001
TINCR7	3.15	0.0232	SHE	10.47	<0.0001
CRABP2	3.24	0.0228	PLA2G2A	11.79	0.0093
PCDH7	3.28	0.0001	RSPO1	11.91	0.0041
KCNQ5	3.31	0.0192	RELN	12.06	0.0007
GATA3	3.63	0.0050	WNT2	12.42	0.0053
SAMD5	3.85	0.0094	CYP24A1	12.93	0.0110
ESR1	3.91	0.0029	DIO3	14.03	0.0108
NRGN	4.21	0.0001	SERTAD4	14.21	0.0021
PEG10	4.61	0.0052	SEC14L5	16.11	0.0080
HOXA1	6.57	0.0040	TNXB	16.63	0.0002
BRINP1	7.19	0.0025	CLEC3B	18.03	0.0084
PDE1A	9.08	0.0383	GPR88	18.37	0.0021
FOXE1	10.24	0.0001	ADRA2A	19.66	0.0044
IL20RA	10.32	0.0029	APOD	20.40	<0.0001
GABRP	21.44	0.0003	PIK3R6	25.98	0.0047
SIX3	33.47	0.0054	PCSK9	29.70	<0.0001

a and Figure 1b present the top 20 upregulated and downregulated mRNAs in LECs from patients with congenital aniridia.

3.1.2. mRNA Sequencing Profile of Limbal Stromal Cells in Patients with Aniridia

A total of 3001 deregulated RNAs were identified in cultured AN-LSCs compared to healthy LSCs, applying a fold change threshold of <−2.0 or >2.0 and an adjusted p-value of p < 0.05. Among these, 2670 RNAs were downregulated, while 331 RNAs were upregulated, as depicted in the volcano plot (Figure 1c).

The heatmap (Figure 1d) presents the top 20 upregulated and downregulated mRNAs in AN-LSCs. Among the most downregulated mRNAs were IVL, MUC5AC, CSF3, ELOA2, IGFS23, SLC26A8, CPHXL, and FGL2, whereas IRX6, SHE, PLA2GA2, RELN, WNT2, SEC14L5, TNXB, APOD, and PSCK9 were among the most upregulated mRNAs in AN-LSCs.

3.1.3. MicroRNA Sequencing Profile of Limbal Stromal Cells in Patients with Aniridia

In this study, we also analyzed miRNA expression in LSCs of aniridia patients using small RNA sequencing. Applying the previously mentioned criteria, we identified a total of 540 miRNAs, among which 26 miRNAs were downregulated, and 22 miRNAs were upregulated in AN-LSCs compared to healthy controls (Figure 1e).

In contrast to a previous study, where only one differentially expressed miRNA was identified in LECs from aniridia patients [28], our findings reveal a broader deregulation of miRNAs in LSCs, highlighting their potential role in AAK.

Table 2. The 20 most down- and upregulated miRNAs in aniridia limbal stromal cells (AN-LSCs), ordered by increasing fold changes.

Downregulated miRNAs in AN-LSCs			Upregulated miRNAs in AN-LSCs
miRNA	Fold Change	Adjusted p-Value	miRNA	Fold Change	Adjusted p-Value
hsa-miR-146a-5p	−8.051	0.032	hsa-miR-493-3p	2.634	0.032
hsa-miR-137-3p	−7.796	0.043	hsa-miR-758-3p	2.648	0.043
hsa-miR-3651	−6.652	0.032	hsa-miR-1185-2-3p	2.697	0.032
hsa-miR-29b-3p	−5.996	0.032	hsa-miR-127-5p	2.791	0.043
hsa-miR-301b-3p	−5.706	0.043	hsa-miR-6877-5p	2.942	0.044
hsa-miR-1246	−5.005	0.032	hsa-miR-487b-3p	2.946	0.032
hsa-miR-4516	−4.801	0.043	hsa-miR-1185-1-3p	3.097	0.032
hsa-miR-3182	−4.554	0.043	hsa-miR-299-5p	3.138	0.035
hsa-miR-4301	−4.440	0.032	hsa-miR-326	3.200	0.043
hsa-miR-3960	−3.790	0.032	hsa-miR-485-5p	3.218	0.034
hsa-miR-9-5p	−3.761	0.032	hsa-miR-381-5p	3.216	0.032
hsa-miR-4454	−3.561	0.037	hsa-miR-134-3p	3.296	0.032
hsa-miR-218-5p	−3.552	0.032	hsa-miR-127-3p	3.427	0.032
hsa-miR-137-5p	−3.337	0.043	hsa-miR-370-3p	3.521	0.032
hsa-miR-1275	−3.275	0.032	hsa-miR-485-3p	3.619	0.034
hsa-miR-12136	−2.924	0.032	hsa-miR-494-5p	3.680	0.043
hsa-miR-106a-5p	−2.813	0.043	hsa-miR-493-5p	4.090	0.032
hsa-miR-29a-5p	−2.729	0.044	hsa-miR-323b-3p	4.170	0.032
hsa-miR-452-5p	−2.609	0.035	hsa-miR-433-3p	4.213	0.043
hsa-miR-193a-3p	−2.585	0.043	hsa-miR-1247-5p	7.400	0.034

and Figure 1f present the top 20 upregulated and downregulated miRNAs in LSCs from aniridia patients. Among these, miR-146a-5p was identified as the most downregulated miRNA, with a fold change of −8.015, while miR-1247-5p was the most upregulated miRNA, with an FC of 7.40. These findings suggest potential regulatory mechanisms involving miRNAs in stromal cell dysfunction in aniridia.

3.1.4. Target Genes of Deregulated microRNAs in AN-LSCs

The target genes of the top 40 deregulated miRNAs were identified using miRTargetLink (https://ccb-compute.cs.uni-saarland.de/mirtargetlink2, accessed on 9 January 2025). Only potential targets with strong experimental evidence were considered for further analysis.

The list of target genes associated with the deregulated miRNAs in AN-LSCs is provided in Supplementary Table S4a.

3.2. Protein–Protein Interaction (PPI) Network and Hub Gene Selection & Analysis

3.2.1. Protein–Protein Interaction Network in AN-LECs

To identify the functional relationships between deregulated genes in AN-LECs, potential PPIs were analyzed using the STRING database with default settings and a minimum interaction score of 0.4. The visualization of the PPI network and cluster formation derived from STRING was performed using Cytoscape (Figure 2a).

Analysis of the PPI network revealed the presence of two main clusters. The first cluster consisted of 38 nodes, with genes associated with interferon alpha/beta signaling, ISG15-protein conjugation, and RIG-I-like receptor signaling. The second cluster comprised 20 nodes, primarily linked to the chemokine signaling pathway. These findings suggest that deregulated genes in AN-LECs may play a role in immune response activation, inflammation, and epithelial barrier function, highlighting key molecular mechanisms contributing to AAK.

3.2.2. Hub Gene Selection and Analysis in AN-LECs

To identify genes with high connectivity within the PPI network, which are likely to play a crucial role in network regulation, we performed multiple correlation clustering (MCC) using the cytoHubba plugin in Cytoscape. Based on this analysis, the top 10 hub genes were selected, including IFIT1, IFIT2, IFIT3, ISG15, IRF7, IFIH1, DDX58, MX1, RSAD2, and OAS1. These hub genes are primarily involved in antiviral defense mechanisms, immune response regulation, and interferon signaling pathways, suggesting their potential role in AN-LEC dysfunction and AAK progression. The hub gene network and its interactions are illustrated in Figure 2b.

3.2.3. Protein–Protein Interaction Network in AN-LSCs

Since the number of identifiers for calculating PPI in the STRING db is limited to 2000, we reduced the criteria for analyzing fold change [FC] < −2.0 or >2.0, t-test adjusted p < 0.05, to fold change [FC] < −2.0 or >2.0, t-test adjusted p < 0.01 in order to be able to perform the analysis with a number of 1957 DEGs (Figure 2c). In contrast to the PPI network observed in LECs, where distinct clusters were identified, no cluster formation was detected in LSCs.

3.2.4. Hub Gene Selection and Analysis in AN-LSCs

The top 10 hub genes identified in AN-LSCs were MT-CO1, MT-CO2, MT-ND1, MT-ND2, MT-ND4, MT-ND4L, MT-ND5, MT-ND6, NDUFB8, and NDUFV2. These genes play key roles in mitochondrial function and oxidative phosphorylation, indicating a potential link between mitochondrial dysfunction and limbal stromal cell impairment in AAK.

The hub gene network and its interactions are visualized in Figure 2d, highlighting their central role in AN-LSC pathophysiology and their potential relevance as therapeutic targets.

3.3. Pathway Analysis and Gene Ontology Classification Analysis

3.3.1. Pathway Analysis and GO Classification Analysis in AN-LECs

Pathway analysis and GO classification of deregulated RNAs in AN-LECs were conducted using the GeneTrail platform. The KEGG database identified 21 enriched pathways, including cytokine-cytokine receptor interaction, IL-17 signaling pathway, TNF signaling pathway, and chemokine signaling pathway (

Table 3. Significantly enriched pathways in limbal epithelial cells (AN-LECs) and limbal stromal cells (AN-LSCs) of patients with congenital aniridia, relative to healthy limbal epithelial cells.

(a) Enriched Pathways of AN-LECs, Observed by the KEGG Database
Name	Expected Number of Genes	Observed Number of Genes	Adjusted p-Value
Cytokine-cytokine receptor interaction	0.97	16	4.32 × 10−13
IL-17 signaling	0.31	11	6.68 × 10−13
NOD-like receptor signaling	0.59	13	1.30 × 10−12
TNF-signaling	0.37	10	3.09 × 10−12
Viral protein interaction with cytokine and cytokine receptor	0.32	9	8.80 × 10−10
NF-kappa signaling	0.33	9	9.37 × 10−10
RIG-I-like receptor signaling	0.23	8	1.09 × 10−9
Toll-like receptor signaling	0.35	9	1.09 × 10−9
Chemokine signaling	0.62	10	8.14 × 10−9
MAPK signaling pathway	0.97	7	2.22 × 10−4
Cellular senescence	0.52	5	6.00 × 10−4
Focal adhesion	0.66	5	1.43 × 10−3
Cell adhesion molecules (CAMs)	0.47	4	3.47 × 10−3
Tight junction	0.55	4	5.30 × 10−3
TGF-beta signaling pathway	0.31	3	7.69 × 10−3
FoxO signaling pathway	0.43	3	1.59 × 10−2
Jak-STAT signaling pathway	0.5.3	3	2.56 × 10−2
Retinol metabolism	0.21	2	3.02 × 10−2
Adherens junction	0.23	2	3.31 × 10−2
P53 signaling pathway	0.23	2	3.36 × 10−2
Calcium signaling pathway	0.64	3	3.56 × 10−2
(b) enriched pathways of AN-LSCs, observed by the KEGG database
Name	Expected number of genes	Observed number of genes	Adjusted p-value
Metabolic pathways	57.86	110	3.84 × 10−8
Axon guidance	7.31	18	7.53 × 10−3
MAPK signaling pathway	11.88	25	7.53 × 10−3
Calcium signaling pathway	7.79	17	2.52 × 10−2
PI3K-Akt signaling pathway	14.26	25	3.70 × 10−2
Cell adhesion molecules (CAMs)	5.85	13	4.05 × 10−2
Inflammatory mediator regulation of TRP channels	4.04	10	4.49 × 10−2

a), highlighting the involvement of inflammatory and immune response mechanisms in AN-LECs.

For GO classification, the PANTHER online tool (Protein ANalysis THrough Evolutionary Relationships, (https://www.pantherdb.org/tools/, accessed on 12 December 2024) was used to categorize the 188 deregulated RNAs in AN-LECs. The results showed that in terms of biological processes, deregulated RNAs were primarily associated with the immune response, indicating an active inflammatory environment. Within cellular components, most genes were linked to the cytoplasm, suggesting altered intracellular signaling and protein interactions. The analysis of molecular functions revealed that protein binding was the most significantly enriched function, pointing toward potential disruptions in protein–protein interactions and regulatory pathways.

An overview of the gene ontology terms is presented in Figure 3, providing insight into the functional impact of deregulated RNAs in AN-LECs and their potential role in the progression of aniridia-associated keratopathy.

3.3.2. Pathway Analysis and GO Classification Analysis in AN-LSCs

The KEGG database identified only seven enriched pathways for AN-LSCs, despite the fact that the number of deregulated RNAs (3001) was considerably higher than in AN-LECs (188 deregulated RNAs). The enriched pathways included metabolic pathways, axon guidance, and inflammatory mediator regulation of TRP channels, indicating potential disruptions in cell metabolism, neuronal signaling, and inflammation-related mechanisms. The complete list of enriched pathways is presented in Table 3b.

For GO classification, the 3001 deregulated RNAs were analyzed using the PANTHER online tool. In the category of biological processes, the most significantly enriched terms were related to immune responses, suggesting an active inflammatory state in AN-LSCs. Within cellular components, the majority of deregulated genes were associated with protein-containing complexes, highlighting alterations in protein interactions and structural organization. Interestingly, in the molecular functions category, only a single enriched GO term was identified: olfactory receptor activity, which may indicate potential involvement in G-protein-coupled receptor signaling.

These findings suggest that while AN-LSCs exhibit a broad range of deregulated RNAs, their functional impact appears to be highly specific, particularly in pathways related to immune regulation, metabolic balance, and signaling mechanisms. The overview of GO terms and enriched pathways is illustrated in Figure 3, providing insights into the molecular mechanisms underlying AN-LSC dysfunction in AAK.

3.3.3. Pathway Analysis of Target Genes of Deregulated miRNAs in AN-LSCs

A total of 359 target genes were identified for the top 40 deregulated miRNAs in AN-LSCs using the GeneTrail online software tool. Analysis of these target genes in the KEGG database revealed several enriched pathways, with the top 10 including PI3K-Akt signaling, MAPK signaling, FoxO signaling, and TNF signaling pathways.

Additional enriched pathways included Apoptosis and RIG-I-like receptor signaling, suggesting that miRNA dysregulation in AN-LSCs may influence cell survival, inflammatory responses, and innate immune activation. The complete list of enriched pathways associated with the target genes of deregulated miRNAs is provided in Supplementary Table S4b.

3.4. RT-qPCR Validation of Deregulated mRNAs in AN-LECs and AN-LSCs and miRNAs in AN-LSCs

We selected genes for validation from the top 40 deregulated mRNAs in AN-LECs and AN-LSCs. Although PAX6 was not identified as deregulated in the RNA sequencing data, it was included for validation in both AN-LECs and AN-LSCs.

(a) For AN-LECs, the downregulated mRNAs selected for validation were PAX6, KRT12, and VNN2, which showed fold changes of −1.56, −50,000, and −6.62, respectively, in qPCR validation (n = 5). The upregulated mRNAs selected for validation included SIX3, BRINP1, and ANKRD29, with fold changes of 2.12, 8.15, and 3.63, respectively.

(b) For AN-LSCs (n = 7), we selected PAX6, IVL, and MUC5AC as downregulated mRNAs, PLA2G2A, RELN, and PCSK9 as upregulated mRNAs. The qPCR results showed fold changes consistently with sequencing data: PAX6 = −4.45, IVL = −50.0, MUC5AC = −12.5, PLA2G2A = 2.90, RELN = 5.02, and PCSK9 = 205.29.

(c) The validated miRNAs showed the following fold changes for downregulated miRNAs: miR-146a-5p (−8.015) and miR-29b-3p (−5.996).

For upregulated miRNAs, the fold changes were miR-326 (3.20), miR-127-3p (12.279), and miR-493-5p (19.025).

Overall, the fold changes obtained from RT-qPCR closely aligned with those from RNA sequencing, confirming the reliability of the sequencing data Figure 4a–c.

4. Discussion

Patients with congenital aniridia often develop aniridia-associated keratopathy. According to Lagali et al., AAK is classified into five grades, ranging from an intact limbal border without vessel intrusion to a thickened, opaque, and vascularized cornea, which severely impairs vision [29]. During AAK progression, the limbal stem cell niche exhibits structural abnormalities, including disruptions in the palisades of Vogt [13]. The current hypothesis suggests that AAK develops due to limbal epithelial stem cell insufficiency, reduced differentiation capacity, and degradation of the limbal stem cell niche microenvironment, affecting both cell maintenance and function [13,30]. In recent years, increasing attention has been given to the role of stromal limbal cells, also referred to as mesenchymal cells of the limbus. In addition to melanocytes, nerve cells, immune cells, and vascular cells, these stromal cells play a crucial immunomodulatory role and contribute to the overall maintenance of the stem cell niche [31]. To our knowledge, RNA expression changes in limbal epithelial cells and RNA and miRNA expression in stromal cells of patients with aniridia have not yet been investigated. In this study, we present our findings on cultured limbal epithelial and stromal cells from patients with congenital aniridia, followed by RNA sequencing of LECs and both RNA and miRNA sequencing of LSCs.

4.1. mRNA Expression in Limbal Epithelial and Limbal Stromal Cells of Patients with Aniridia

In AN-LECs, we observed a significantly lower number of differentially expressed genes (188) and a distinct clustering of inflammation-related genes in the protein–protein interaction network analysis (Figure 2a). Hub gene analysis further revealed that all identified hub genes belong to the Interferon (IFN)-related gene (IRG) family (Figure 2c). These genes, including IFIT1, IFIT2, IFIT3, ISG15, IRF7, IFIH1, DDX58, MX1, RSAD2, and OAS1, were significantly downregulated in AN-LECs compared to control cells. Previous studies have demonstrated that increased expression of IRGs, such as ISG15 and IFIT2, is associated with tumor cell growth and apoptosis [32,33,34]. The reduced expression of IRGs in LECs of aniridia patients may serve as a counter-regulatory mechanism to mitigate the increased susceptibility to apoptosis, which has been previously observed in PAX6 ^+/− mouse model [11,35].

The most significantly enriched pathways in AN-LECs are related to inflammation and immune response. Among the most enriched pathways are cytokine-cytokine receptor interaction, IL-17 signaling, TNF signaling, and NF-kappa B signaling, which, in addition to their roles in immune regulation, are crucial for cell growth and survival [36,37]. Previous studies have confirmed that inflammation plays a key role in AAK [5,6,7]. A particularly notable finding is the enrichment of the retinoic acid-inducible gene I (RIG-I)-like receptor signaling pathway, which is also involved in immune response activation. RIG-I genes can induce Type I interferons, which, in turn, regulate the expression of interferon-related genes (IRGs) [38]. Additionally, all-trans retinoic acid has been shown to enhance anti-inflammatory responses in lipopolysaccharide-treated RAW 264.7 cells [39]. Retinol metabolism, which is also enriched in AN-LECs, has previously been found to be altered in conjunctival epithelial cells of aniridia patients [7,8], suggesting a broader impact on ocular surface homeostasis.

The most surprising finding of this study was the discrepancy in the number of differentially expressed RNAs between AN-LECs and AN-LSCs. To refine the analysis of deregulated RNAs using STRING, we applied a more stringent p-value threshold (0.01 instead of 0.05). This adjustment reduced the number of analyzed deregulated genes in AN-LSCs from 3001 to 1957, which remains almost ten times higher than the 188 DEGs identified in AN-LECs. Moreover, this large number of deregulated RNAs did not show a clear cluster in the PPI network analysis of the string database (Figure 2b). One possible explanation is that hub gene analysis primarily identified mitochondrial genes that are directly linked to the respiratory chain (Figure 1d). Mitochondria, which are critical for cellular energy metabolism, play a key role in β-oxidation of fatty acids and overall metabolic homeostasis.

Notably, mitochondrial disorders have been implicated in Leber’s hereditary optic neuropathy, where mutations in MT-ND1, MT-ND4, and MT-ND6, or defects in NDUFS1-4 genes, can result in neurodegenerative diseases with optic atrophy [40]. Mitochondrial dysfunction is also known to induce oxidative stress, a phenomenon previously described in stromal cells of aniridia patients [41,42]. This may explain why so many DEGs were identified without a clear functional cluster, as mitochondrial impairment can affect multiple cellular pathways. Furthermore, the most enriched pathway in AN-LSCs was metabolic pathways (Table 3b), further reinforcing the role of mitochondrial metabolism in stromal cell dysfunction in aniridia-associated keratopathy. Future studies should include analyses of protein expression for deregulated genes identified by RNA sequencing, as well as functional assays to clarify their biological significance.

4.2. MicroRNA Expression in Limbal Stromal Cells of Patients with Aniridia

The evaluation of deregulated miRNAs is particularly important, as miRNAs are known to regulate gene expression at the posttranscriptional level.

In conjunctival epithelial cells from aniridia patients, several deregulated miRNAs associated with vascularization and wound healing have been identified, with miR-204-5p being the most significantly affected [7]. Another study reported reduced miR-204-5p levels in the PAX6 sey/+ mouse model, leading to an increase in angiogenic factors such as VEGF and ANGPT1, suggesting a role in vascular remodeling [43]. In AN-LECs, miR-138-5p was found to be the most upregulated miRNA, strongly influencing pathways such as FOXC1, Wnt, focal adhesion, IL-17, and JAK-STAT signaling [28]. In the current study, several deregulated miRNAs in AN-LSCs were identified that may play a critical role in the development of AAK. The most downregulated miRNA, miR-146a-5p, exhibited a fold change of −8.015. This miRNA is involved in inflammatory regulation by reducing the release of pro-inflammatory cytokines, including NF-κB and IL-6 [44,45]. Interestingly, miR-146a-5p was found to be significantly upregulated in peripheral blood mononuclear cells of patients with acute myocardial infarction, a condition commonly associated with acute inflammatory cell infiltration [44]. In HUVECs (human umbilical vein endothelial cells), overexpression of miR-146a-5p reduced LPS-induced pro-inflammatory cytokine expression, whereas in HFLS (human fibroblast-like synovial cells), its expression led to increased cytokine levels [45,46]. This suggests that the effect of miR-146a-5p on inflammation may be cell-type dependent, warranting further investigation into its specific role in AN-LSCs. These findings align with the Gene Ontology analysis, which highlights significant alterations in immune system processes (47.64%) and immune response (31.89%) (Figure 3).

Another notable observation is that the most upregulated miRNA in AN-LSCs is miR-1247-5p, with a fold change of 7.4. MiR-1247-5p has been found to be upregulated in patients with metastatic uveal melanoma [47] and, in contrast, has been shown to suppress tumor growth in hepatocellular carcinoma [48]. Additionally, in breast cancer, miR-1247-5p expression is decreased, where it has been reported to inhibit proliferation and induce apoptosis by suppressing the Wnt/β-catenin signaling pathway [49]. Target genes of miR-1247-5p, including SOX9 and MAP3K9, were identified (Supplementary Table S4). Both SOX9 and MAP3K9 mRNAs were found to be downregulated in AN-LSCs, although without statistical significance.

The biological role of miR-1247-5p in limbal stromal cells in AAK remains unclear and warrants further investigation. It is essential to explore whether similar mechanisms are involved in limbal stromal cells of aniridia patients and how miRNA dysregulation may contribute to disruptions in the stem cell niche. In addition, the functional impact of deregulated miRNAs should be examined in future studies through analysis of their target genes and corresponding protein expression.

5. Conclusions

This study, to our knowledge, is the first to demonstrate differences in RNA expression between cultured limbal epithelial and stromal cells and to characterize the deregulated miRNA expression profile in patients with aniridia.

RNA sequencing in AN-LECs revealed significant alterations in inflammation-related signaling pathways, reflected in hub genes belonging to the group of interferon-stimulated genes. These changes suggest an abnormal response to inflammatory stimuli, which may contribute to the inflammatory component of aniridia-associated keratopathy. A similar pattern was observed in AN-LSCs, where immune response pathways were also affected, alongside widespread changes in metabolic signaling pathways compared to healthy controls. The role of deregulated miRNAs in these processes and their potential as therapeutic targets remain open questions for future research.

These findings also raise the question of how reduced PAX6 protein expression influences the extensive RNA and miRNA deregulation observed in aniridia cells. Further studies are necessary to clarify the mechanistic link between PAX6 haploinsufficiency, miRNA dysregulation, and the progression of AAK.

6. Limitations of the Study

The primary limitations of this study include the relative low sample number, as congenital aniridia is a rare disease. Moreover, the use of older cornea bank donors as controls, necessitated by ethical constraints, represents a limitation of this study and may have influenced the comparability between the older control group and the younger congenital aniridia group. Additionally, the analyzed samples were cultured cells, which may have undergone changes compared to the original biopsies during the culturing process. However, the use of cultured cells allowed us to perform both sequencing and RT-qPCR validation on the same samples, ensuring consistency in the data. Despite these limitations, our findings provide a valuable foundation for future studies to further investigate the molecular mechanisms underlying aniridia-associated keratopathy and to gain deeper insights into its pathogenesis and potential therapeutic targets.