Early Transcriptomic and Pathologic Changes of Col8a2 Mutant Fuchs Endothelial Corneal Dystrophy

Abstract

Purpose

The purpose of this study was to characterize the early transcriptomic and pathologic changes of Fuchs endothelial corneal dystrophy (FECD) using the Col8a2^Q455K/Q455K mutant mouse model.

Methods

The Col8a2^Q455K/Q455K mutant mice were divided into the early-stage (≤2-month-old) group and the late-stage (≥8-month-old) group, based on corneal endothelial changes evaluated by slit-lamp microscopy, optical coherence tomography (OCT), and confocal microscopy. The corneal endothelial cells from early-stage mutant and age-matched wild-type (WT) mice were collected for transcriptomic analysis and validated by quantitative PCR and immunofluorescence staining.

Results

The Col8a2^Q455K/Q455K mutant mice showed no observable corneal abnormality before 2 months of age. Morphological changes of the corneal endothelium appeared at 4 months and aggravated continuously with apparent corneal edema. However, when analyzing transcriptomic changes of the corneal endothelium, we found that the early-stage mutant mice exhibited 221 upregulated and 55 downregulated genes compared with age-matched WT mice; these differences were even more pronounced in the late-stage mutant mice. The upregulated genes predominantly enriched three signaling processes, including extracellular matrix (ECM) remodeling (e.g. Lgals3, Timp1, and Mmp3), endoplasmic reticulum (ER) stress (e.g. Hspa5, Dnajb9, and Atf3), and early activation of immune-related pathways (e.g. Icam1, Bpifb1, and C1q). Moreover, the qPCR and immunofluorescence staining further validated changes in gene and protein expressions prior to the morphological abnormalities in the mutant mice.

Conclusions

The Col8a2^Q455K/Q455K mutant mice exhibit aberrant activation of ECM remodeling, ER stress. and immune responses in the corneal endothelium prior to observable pathogenic changes, providing the first in vivo evidence of potential early biomarkers and therapeutic treatments for FECD.

Fuchs endothelial corneal dystrophy (FECD) is a prevalent corneal degenerative disorder characterized by progressive loss of corneal endothelial cells (CECs), thickening of Descemet's membrane (DM), and aberrant deposition of extracellular matrix (ECM). These pathological changes result in corneal edema and visual impairment, establishing FECD as a leading indication for corneal transplantation.^1–3 Epidemiological studies indicate that FECD affects nearly 300 million individuals aged ≥30 years, with prevalence varying across gender, ethnicity, and geographic region. The number of affected individuals is projected to increase to approximately 415 million by 2050.⁴ Although endothelial keratoplasty remains the most effective therapeutic option for FECD, its accessibility is limited by the global shortage of donor corneas.^5,6 This growing burden highlights the urgent need to elucidate the pathogenic mechanisms underlying FECD and to develop safer, more widely accessible alternative therapeutic strategies.

The pathogenesis of FECD is complex and multifactorial, involving aberrant ECM assembly, oxidative stress, mitochondrial DNA damage, unfolded protein response (UPR), and endoplasmic reticulum (ER) stress.^1,7–9 Environmental factors such as smoking and ultraviolet (UV) exposure are implicated in disease progression.^1,7,10 Because of the limited availability of clinical tissues, most studies have focused on end-stage FECD, characterized by extensive endothelial cell loss and confluent guttae. These pathological changes usually manifest at advanced stages, when irreversible corneal edema and visual impairment necessitate transplantation.⁸ Consequently, the limited understanding of the mechanisms involved in the early stages of FECD has restricted its early diagnosis and development of targeted, protective interventions.

Mutations in the COL8A2 gene are recognized as a major genetic determinant of FECD.¹ Type VIII collagen, a key structural component of the corneal DM, may disrupt ECM homeostasis when aberrantly expressed or mutated.¹¹ Abnormal ECM secretion contributes to the formation of guttae, a hallmark of FECD, thereby disturbing the microenvironment of the corneal endothelium.^12,13 Although Col8a2 mutant mice faithfully recapitulate the clinical and histological hallmarks of FECD, the pathogenic mechanisms driving early-stage phenotypic manifestations remain poorly understood due to limited investigation.^14–17

In this study, comprehensive transcriptomic analysis of the corneal endothelium in the Col8a2^Q455K/Q455K mutant mouse model at both early and late disease stages revealed significant molecular changes preceding any clinical or histological changes. These alterations were primarily characterized by ECM remodeling, ER stress, and immune response. Our findings indicate that abnormalities in these signaling pathways emerge at the early stage of FECD. These molecular alterations may serve as early biomarkers and potential therapeutic targets, offering new insights for the early diagnosis and intervention of FECD.

Materials and Methods

Experimental Animals

C57BL/6 wild-type (WT) mice (purchased from Shandong Taike Biotechnology Co., Ltd., China) and Col8a2^Q455K/Q455K homozygous mutant mice on a C57BL/6 background (JAX stock #029749; The Jackson Laboratory, USA) were used in this study. All animals were housed in the Laboratory Animal Center of Shandong First Medical University. All procedures involving animals were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the IACUC guidelines. The animal experimental protocols were approved by the Ethics Committee of the Shandong Eye Institute.

Corneal Evaluation

Slit-lamp biomicroscopy (Topcon, Tokyo, Japan) and optical coherence tomography (OCT) (Optovue, Fremont, CA, USA) were performed to assess corneal transparency and central thickness. Corneal endothelial cell density (ECD) and morphology were evaluated using in vivo confocal microscopy (Heidelberg Engineering, Germany). Images of the central corneal endothelium were acquired with the subject under topical anesthesia, and three high-quality frames per eye were selected for analysis. Cell density was quantified using ImageJ software (NIH, Bethesda, MD, USA). Endothelial cell morphology was further assessed by ZO-1 immunofluorescence staining of whole-mounted corneas. High-resolution images were captured with a Leica Thunder Imager (Leica Microsystems, Wetzlar, Germany), and the endothelial cell area and the percentage of hexagonal cells were quantified using ImageJ software based on manually delineated cell borders. For each sample, at least three random fields from the central cornea were analyzed.

Low-Input RNA Sequencing

Corneal endothelium was harvested from WT and Col8a2 mutant mice at early (≤2 months) and late (≥8 months) stages. For each group, 3 biological replicates were prepared, with each replicate consisting of pooled tissue from 10 mice. This sampling strategy was established based on previously reported transcriptomic protocols and further optimized in our experiments.^10,15,18 Prior to tissue collection, all mice underwent in vivo anterior segment and corneal endothelial examinations to ensure consistent corneal phenotypes across age groups. Samples with RNA integrity numbers (RIN) ≥6 were sent to OE Biotech (Shanghai, China) for library construction and sequencing on the Illumina platform. Differential gene expression analysis was performed using DESeq2 (version 1.22.2), applying a cutoff of |log₂ fold change| ≥1 and p < 0.05. Subsequent bioinformatic analyses were conducted on the OECloud platform (https://cloud.oebiotech.com/task/). All raw and processed RNA sequencing (RNA-seq) data reported in this study have been deposited in the NCBI Gene Expression Omnibus (GEO) database and are publicly accessible through the accession number GSE305104.

Quantitative Real-Time PCR

To validate the transcriptomic findings, selected differentially expressed genes were analyzed by quantitative real-time PCR (qRT-PCR). Total RNA was extracted from isolated mouse corneal endothelium using the FlysisAmp Cells-to-CT 2-Step SYBR Green Kit (Vazyme, Nanjing, China). After confirming RNA concentration and purity, complementary DNA (cDNA) synthesis was performed using the HiScript IV All-in-One Ultra RT SuperMix for qPCR (Vazyme) on a Mastercycler X50s PCR system (Eppendorf, Hamburg, Germany). The qRT-PCR reactions were conducted using ChamQ Universal SYBR qPCR Master Mix (Vazyme) on a QuantStudio 5 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Primers were designed with Primer-BLAST (NCBI) and synthesized by Qingdao Daraoda Biotechnology Co., Ltd. All primer sequences are listed in the Table. Specificity was confirmed by melting curve analysis and agarose gel electrophoresis. β-actin served as the internal control, and relative gene expression levels were calculated using the 2^-ΔΔCt method. Each experimental group included five biological replicates, with each sample analyzed in five technical replicates. Data are presented as mean ± standard error of the mean (SEM), and statistical significance was determined using a two-tailed unpaired Student's t-test with a threshold of P < 0.05.

Table.

Primer Sequences and Product Sizes Used for qRT-PCR

Gene	Forward Primer (5′→3′)	Reverse Primer (5′→3′)	Product Size (bp)
Lgals3	CGCATGCTGATCACAATCATG	GTCCTGCTTCGTGTTACACACAA	150
Timp1	AAGTCCCAGAACCGCAGTGA	ACAGCCAGCACTATAGGTCTTTGA	150
Dpt	TGGATCGTGAGTGGCAATTTT	TGGTTGTTGCTCCTCGCATA	150
Mmp3	AGATCGATGCTGCCATTTCTAATA	CCCTTGAGTCAACACCTGGAA	150
Smoc2	CAATGATGACGGCACCTACAGT	TGCGGCACCTTTTCATTTATT	150
Kera	CTGCAGCACCTTCACCTTGAT	CGGTGGCTTGATTTCATTCC	150
Col23a1	CCACCAGGCCTTATTGGGTT	CCCCTTCTCTCCACGTTCAC	126
Hspa5	CTCCGGCGTGAGGTAGAAAA	AGAGCGGAACAGGTCCATGT	150
Dnajb9	AAAAATAAAAGCCCTGATGCTGAA	AGGACTCCCATTGCCTCTTTG	150
Atf3	TGCCAAGTGTCGAAACAAGAA	CGGTGCAGGTTGAGCATGTA	150
Calr	ACATCATGTTTGGTCCGGACAT	CTGGCCGCACAATCAGTGT	150
Derl3	GGGCAACTCGGTTGTCACA	GGGTCTTCCTGAGGGTCATCTA	153
Bpifb1	CCCCCACATTGTGCTGAAC	AGCCGGTTGTCTTCTGTGAAG	150
Ctss	AATCGGACATTGCCTGACACT	GGATATCAGCTTCCCCGTTTT	150
Icam1	CATGGGAATGTCACCAGGAAT	CCTGATCTTTCTCTGGCGGTTA	150
C1qc	CCAGCTTCTGCGACCACAT	AACCAGAGAAGACGCTGTTGGA	150
C1qa	ACACGGGTCGCTTCATCTGT	GCCCCTTGTTGTTGGTGTTAG	150
C1qb	GGTCATTCGCTTCGAAAAGG	CGGCCACGAACGAGATTC	150
Ccl8	GAGAAGCTGACTGGGCCAGAT	ACCCTGCTTGGTCTGGAAAA	150
Actb	ACGGCCAGGTCATCACTATTG	AGAGGTCTTTACGGATGTCAACGT	150

Immunofluorescence Staining of Whole-Mount and Cryosectioned Corneas

Both whole-mount and cryosectioned corneas were used for immunostaining. For whole-mount corneas, mouse eyeballs were fixed in 4% paraformaldehyde (PFA; Solarbio, Beijing, China), corneas were dissected, permeabilized with 0.1% Triton X-100 (Solarbio), and blocked with 5% bovine serum albumin (BSA; Solarbio). Primary antibodies were incubated overnight at 4°C, followed by a 1-hour incubation with secondary antibodies at room temperature (UElandy, Hangzhou, China). Nuclear counterstaining was performed with DAPI (Solarbio). For cryosectioned corneas, eyeballs were embedded in optimum cutting temperature compound (Sakura, Torrance, CA, USA), frozen at −80°C, and sectioned at a thickness of 7 µm. Staining procedures were the same as for whole-mount corneas. All primary antibodies were diluted 1:200 and included anti-ZO-1 (Invitrogen; 33-9100), anti-Na⁺/K⁺-ATPase (Millipore; 05-369), anti-Mmp3 (Abcam; ab52915), anti-Timp1 (Proteintech; 26847-1-AP), anti-Galectin-3 (Proteintech; 14979-1-AP), anti-Laminin (Invitrogen; PA1-16730), anti-C1qc (Proteintech; 16889-1-AP), anti-C1qa (Proteintech; 85719-4-RR), anti-C1qb (Invitrogen; PA5-102737), anti-Icam1/CD54 (Proteintech; 16174-1-AP), anti-Lplunc1 (Invitrogen; MA5-17124), anti-GRP78 (Abcam; ab21685), anti-Atf3 (Invitrogen; PA5-101089), and anti-Dnajb9 (Abcam; ab118282). Images were acquired using a Leica Thunder Imager (Leica Microsystems, Wetzlar, Germany) and fluorescence intensity was quantified using ImageJ software. Each group included three biological replicates.

Data Analysis

All data were analyzed using GraphPad Prism 9.0 (GraphPad Software, San Diego, CA, USA). Statistical tests were chosen based on data distribution and variable type. Data are presented as mean ± SEM. Group comparisons were performed using unpaired Student's t-tests or 2-way analysis of variance (ANOVA), as appropriate. A P value < 0.05 was considered statistically significant. All experiments were independently performed in at least three biological replicates.

Results

Progressive Corneal Endothelial Degeneration in Col8a2 Mutant Mice

To delineate the early pathological alterations preceding corneal endothelial decompensation, we systematically examined the corneal phenotypes of Col8a2 mutant mice across 2 to 8 months of age to determine the timepoint of distinguishing early from late disease stages. Slit-lamp and optical coherence tomography imaging revealed that Col8a2 mutant mice began to exhibit mild corneal haze at 6 months of age compared with WT mice, which progressed to marked corneal edema and opacity by 8 months (Figs. 1A, 1B). Confocal microscopy showed that the morphology of the corneal endothelium in Col8a2 mutant mice was normal at 2 months but developed visible abnormalities such as guttae-like deposits and decreased cell density by 4 months, which progressively expanded by 8 months (Figs. 1C, 1D). Accordingly, mutant mice aged 2 months or younger and lacking visible macroscopic abnormalities were designated as the early-stage group, whereas those aged 8 months or older presenting with evident lesions were assigned to the late-stage group. In addition, the endothelial ZO-1 staining identified a slight expansion of cell area and hexagonality reduction in early-stage mutant mice, whereas these changes reached statistical significance in the late-stage mutant mice (Figs. 1E, 1F, 1G). The endothelial Na⁺/K⁺-ATPase staining demonstrated that the endothelial pump function was impaired only in late-stage, but not in early-stage mutant mice (Fig. 1H).

Figure 1.

Corneal endothelial cell changes of Col8a2 mutant mice. (A) Representative slit-lamp and the anterior segment OCT images of wild-type (WT) mice and Col8a2 mutant mice (mutant) from 2 to 8 months of age. (B) Quantification of central corneal thickness measured by OCT (n = 5). (C) Representative images of corneal endothelium captured by confocal microscopy. (D) Quantification of corneal endothelial cell density (n = 3). (E) Representative images of ZO-1 immunofluorescence staining in whole-mount corneas from WT and mutant mice at early and late stages. (F, G) Quantification of endothelial cell morphology, including average cell area (F) and the percentage of hexagonal cells (G) (n = 3). (H) Representative images of Na⁺/K⁺-ATPase immunofluorescence staining of whole-mount corneas. Statistical analysis was performed using unpaired 2-tailed Student's t-tests for comparisons in B and D), or 2-way ANOVA followed by Tukey's post hoc test for comparisons in F and G. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. Scale bar = confocal images, 50 µm; staining images = 25 µm.

Transcriptomic Alterations of Col8a2 Mutant Mouse Corneal Endothelium

To explore the initial transcriptional alterations of FECD in Col8a2 mutant mice, we performed RNA sequencing of corneal endothelium from the early- and late-stage mutants, as well as their age-matched WT controls (Fig. 2A). Principal component analysis (PCA) revealed distinct separations between mutant and WT samples at both stages (Fig. 2B). Differential expression analysis identified 276 differentially expressed genes (DEGs) in the early-stage mutants (221 upregulated and 55 downregulated) and 1163 DEGs in the late-stage mutants (555 upregulated and 608 downregulated) when compared with the WT controls (Fig. 2C). Volcano plots highlighted the significant upregulation of gene expression (Fig. 2D). Gene Ontology (GO) analysis further demonstrated significant positive enrichment of ECM- and immune-related pathways, whereas pathways associated with structural homeostasis, cell proliferation, and transcriptional regulation were negatively enriched (Figs. 2E, 2F). Among the 151 DEGs shared between the early- and late-stage mutants, 132 genes were significantly upregulated at the early-stage mutants (Fig. 2G). GO analysis of the 132 genes confirmed their functional enrichment in ECM remodeling, ER stress, and immune-related processes (Fig. 2H). These findings suggest that multiple pathogenic pathways have been activated in Col8a2 mutant mice prior to the FECD onset.

Figure 2.

Transcriptomic profiling of corneal endothelium in early- and late-stage Col8a2 mutant mice. (A) Schematic diagram illustrating the grouping and sampling process for low-input RNA sequencing. (B) PCA showed distinct clustering between mutant and WT samples at both early and late stages. (C) Bar graph summarizing the number of DEGs in the early and late stages. (D) Volcano plots showing expression profiles of DEGs in early and late stages. (E, F) GO enrichment analysis of DEGs at the early and late stages, with red indicating upregulated genes and blue indicating downregulated genes. The top 10 enriched GO terms (ranked by the adjusted P value) are shown. (G) Venn diagram showing the overlapping number of DEGs between the early and late stages. (H) GO enrichment analysis of the 132 overlapping genes at the early stage. For each GO category (Biological Process, Molecular Function, and Cellular Component), the top 10 enriched terms ranked by the adjusted P value are shown.

ECM Remodeling of Corneal Endothelium in Early Col8a2 Mutant Mice

Abnormal accumulation of ECM in the corneal endothelium represents the main characteristics of FECD in patients and animal models.¹ To investigate whether ECM abnormalities constitute an initiating event in the pathogenesis of Col8a2 mutant mice, we performed gene set enrichment analysis (GSEA) on the early-stage transcriptome. The analysis revealed significant enrichment of ECM-related terms, including collagen-containing ECM (GO:0062023), fibronectin binding (GO:0001968), collagen binding (GO:0005518), and laminin binding (GO:0043236; Fig. 3A). These findings were consistent with the GO enrichment results of 132 overlapping genes at the early stage (Fig. 3B). A heatmap of enriched genes demonstrated early upregulation of multiple ECM regulators, such as Lgals3 (Galectin-3), Mmp3, Timp1, Col23a1, Smoc2, Dpt, and Kera (Fig. 3C). The expression patterns were validated by qRT-PCR with statistically significant differences (Fig. 3D). Furthermore, immunofluorescence staining revealed stronger signals of Lgals3, MMP3, and Timp1 in the corneal endothelium of early-stage mutant mice (Figs. 3E–G), and quantitative fluorescence analysis is shown in Supplementary Figure S2. Laminin immunofluorescence staining revealed the irregular and intensified deposition in both early- and late-stage mutant mouse corneas, whereas WT corneas showed a uniform band-like pattern (Fig. 3H, Supplementary Fig. S1A), and quantitative analysis is presented in Supplementary Figure S1C. Together, these findings demonstrate that Col8a2 mutant mice develop early ECM remodeling prior to the observed clinical features.

Figure 3.

ECM remodeling in the early-stage Col8a2 mutant mouse corneal endothelium. (A) GSEA of whole-transcriptome data revealed significant enrichment of ECM-related pathways in early-stage mutants (normalized enrichment score [NES] > 2 and false discovery rate [FDR] < 0.05). Representative terms among the top 60 enriched categories are shown. (B) GO enrichment analysis of 132 overlapping genes upregulated in early-stage mutants. Shown are ECM-related terms selected from the top 10 enriched categories in each GO domain, with representative enriched genes indicated. (C) Heatmap of the top 20 differentially expressed genes enriched in ECM-related GO pathways. (D) The qRT-PCR validation of selected ECM-related genes from C in CECs of early-stage mutant mice. Statistical analysis was performed using unpaired two-tailed Student's t-tests. *P < 0.05; **P < 0.01; ***P < 0.001 (n = 5). (E–H) Immunofluorescence staining of corneal endothelium from early-stage mutant and WT: E Lgals3 in whole-mount cornea; F Timp1 in whole-mount cornea; G schematic illustration (left) showing the anatomic layers of the cornea, including the epithelium, stroma, Descemet's membrane (DM), and endothelium. The area outlined by the red dashed box corresponds to the region shown in the right panel, which presents immunofluorescence staining of MMP3 (red) in corneal cryosections of wild-type and mutant mice. The stroma, DM, and corneal endothelium have been clearly indicated in the figure. White arrowheads indicate the corneal endothelial layer. Dashed lines delineate the DM. H Laminin in whole-mount cornea. Scale bar = 25 µm.

Endoplasmic Reticulum Stress of Corneal Endothelium in Early Col8a2 Mutant Mice

ER stress has been reported as a predominant factor in corneal endothelial injury.¹⁴ To determine whether ER stress is activated at the early stage of FECD, we analyzed ER stress-related transcriptional signatures in the Col8a2 mutant mice. In the preceding GSEA analysis, we observed upregulation of ER stress-related pathways, including the ER unfolded protein response (GO:0030968), ER lumen (GO:0005788), and misfolded protein binding (GO:0051787), which was consistent with the enrichment results of 132 overlapping genes upregulated at the early stages (Figs. 4A, 4B). A heatmap of genes enriched in the ER stress-related pathways revealed significant upregulation of several regulators in early-stage mutant mice, including Hspa5 (GRP78/BiP), Calr (Calreticulin), Dnajb9 (ERdj4), Derl3 (Derlin-3), and Atf3 (Fig. 4C). The expression patterns were validated by qRT-PCR with statistically significant differences (Fig. 4D). Furthermore, immunofluorescence staining revealed the irregular and intensified deposition of Hspa5 in both early- and late-stage mutant mouse corneas, whereas WT corneas showed a uniform band-like pattern (Fig. 4E, Supplementary Fig. S1B), and quantitative analysis is presented in Supplementary Figure S1D. Cryosectioned corneas staining confirmed elevated protein levels of Dnajb9 and Atf3 in early-stage mutants (Figs. 4F, 4G), and quantitative fluorescence analysis is shown in Supplementary Figure S2. Collectively, these findings demonstrate that ER stress is already evident in Col8a2 mutant mice prior to the onset of clinical phenotypes.

Figure 4.

ER stress in the early-stage Col8a2 mutant mouse corneal endothelium. (A) GSEA of whole-transcriptome data revealed significant enrichment of ER stress-related pathways in early-stage mutants (normalized enrichment score [NES] > 2 and false discovery rate [FDR] < 0.05). Representative terms among the top 60 enriched categories are shown. (B) GO enrichment analysis of 132 genes upregulated in early-stage mutants. Shown are ER stress-related terms selected from the top 10 enriched categories in each GO domain, with representative enriched genes indicated. (C) Heatmap of differentially expressed genes enriched in ER stress-related GO pathways. (D) The qRT-PCR validation of selected ER stress-related genes from C in CECs of early-stage mutant mice (n = 5). Statistical analysis was performed using unpaired two-tailed Student's t-tests. *P < 0.05; **P < 0.01; ***P < 0.001. (E–G) Immunofluorescence staining of corneal endothelium from early-stage mutant and WT: E Hspa5 in whole-mount cornea; F Dnajb9 in cryosectioned corneas; and G Atf3 staining in cryosectioned corneas. The stroma, Descemet's membrane (DM), and corneal endothelium have been clearly indicated in the figure. White arrowheads indicate corneal endothelial cells. Scale bar = 25 µm.

Immune Response of Corneal Endothelium in Early Col8a2 Mutant Mice

Previous studies have demonstrated that complement pathways are activated in the aqueous humor of patients with FECD.¹⁶ To investigate whether immune activation contributes to early pathogenic changes in Col8a2 mutant mice, we examined immune-related transcriptional signatures at the early stage. The preceding GSEA analysis also revealed significant positive enrichment of multiple immune-related pathways in early-stage mutant mice including the major histocompatibility complex (MHC) class I peptide loading complex (GO:0042824), antigen processing, and presentation of exogenous peptide antigen via MHC class II (GO:0019886), CCR chemokine receptor binding (GO:0048020), and neutrophil chemotaxis (GO:0030593; Fig. 5A). GO enrichment analysis of 132 overlapping genes upregulated in early stages showed results consistent with GSEA. These pathways included immune system process, innate immune response, and antigen presentation pathways involving both MHC class I and II molecules (Fig. 5B). A heatmap of genes enriched in these pathways revealed marked upregulation of key immune-related genes, such as Lgals3, Bpifb1, Ctss, Icam1, C1qc, C1qa, C1qb, and Ccl8 (Fig. 5C). These transcriptomic findings were further validated by qRT-PCR (Fig. 5D). Immunofluorescence analysis showed stronger staining of C1q (C1qa, C1qb, and C1qc), Icam1 and Bbifb1 in the corneal endothelium of early-stage mutants (Figs. 5E–G), and quantitative analysis is presented in Supplementary Figure S2. These results collectively suggest that Col8a2 mutation triggers immune response activation in the corneal endothelium preceding clinical manifestation.

Figure 5.

Immune response in the early-stage Col8a2 mutant mouse corneal endothelium. (A) GSEA of whole-transcriptome data revealed significant enrichment of immune-related pathways in early-stage mutants (normalized enrichment score [NES] > 2 and false discovery rate [FDR] < 0.05). Representative terms among the top 60 enriched categories are shown. (B) GO enrichment analysis of 132 genes upregulated in early-stage mutants. Shown are immune-related terms selected from the top 10 enriched categories in each GO domain. (C) Heatmap of the top 20 differentially expressed genes enriched in immune-related GO pathways. (D) The qRT-PCR validation of selected immune-related genes from C in CECs of early-stage mutant mice (n = 5). Statistical analysis was performed using unpaired two-tailed Student's t-tests. *P < 0.05; **P < 0.01; ***P < 0.001. (E–G) Immunofluorescence staining of corneal endothelium from early-stage mutant and WT: E C1q (C1qa, C1qb, and C1qc) in cryosectioned corneas; F Icam1 in cryosectioned corneas; and G Bpifb1 staining in cryosectioned corneas. The stroma, Descemet's membrane (DM), and corneal endothelium have been clearly indicated in the figure. White arrowheads indicate corneal endothelial cells. Scale bar = 25 µm.

Comparisons of Functional Gene Expression Between Early and Late Col8a2 Mutant Mice

To further evaluate functional alterations in the corneal endothelium, we analyzed classical corneal endothelial genes associated with cell adhesion and junctions, pump function, mitochondrial function, and transcription factors (Fig. 6). At the early stage, only mild transcriptional alterations were observed, including downregulation of Acta2 and Ndufs5, upregulation of Atp1a1, Pink1, and Car2. At the same time, transcription factors such as Myc, Xbp1, Snai1, and Snai2 were upregulated. In contrast, transcriptional dysregulation was enhanced in late-stage mutant mice. Adhesion- and junction-related genes (Tjp1, Cldn10, Ocln, Cdh1, Cdh2, and Cd44) as well as pump function-related genes (Atp1a1, Atp1b1, Car2, and Car12) were significantly altered, along with mitochondrial regulators (Prkn, Sod2, Slc25a5, Ndufs5, Sdhb, and Cox4i1) and transcription factors (Klf4, Myc, E2f1, Xbp1, Nfe2l2, Snai1, and Snai2). Collectively, these results indicate that transcriptional alterations of functional genes in the corneal endothelium are predominantly manifested at the late stage.

Figure 6.

Expression of corneal endothelial functional genes in Col8a2 mutant mice. Heatmap showing representative genes associated with cell adhesion and junctions, pump function, mitochondrial function, and transcriptional regulation in early- and late-stage Col8a2 mutant mice compared with wild-type controls. Heatmap values represent normalized FPKM. Statistical significance of differential expression was determined by DESeq2, with significant genes indicated by asterisks. *P < 0.05; **P < 0.01; ***P < 0.001.

In summary, these findings delineate the pathogenic progression of FECD in Col8a2 mutant mice (Fig. 7). The early stage is characterized by aberrant ECM remodeling, ER stress, and immune dysregulation as predominant molecular events, whereas the late stage exhibits typical corneal endothelial structural disruption and cell loss. This model provides an integrative framework for understanding the early pathological mechanisms underlying FECD.

Figure 7.

Schematic model of FECD progression in Col8a2 mutant mice. This model illustrates FECD progression: at the early stage, transcriptomic analysis reveals abnormal activation of ECM remodeling, ER stress, and immune pathways before overt morphological changes; at the late stage, CECs develop guttae-like excrescences, disrupted morphology, and progressive loss, representing the typical clinical features of FECD.

Discussion

FECD is a leading cause of corneal endothelial decompensation, but the molecular basis of early disease remains largely unexplored.^1,5 Using the Col8a2^Q455K/Q455K mouse model, we identified three key molecular events occurring in the early stages of FECD: ECM remodeling, activation of the ER stress response, and aberrant immune responses. Notably, these alterations emerged before the onset of classical clinical and histopathological features, and persisted through the late stage of the disease. Unlike previous studies that mainly focused on the late stages of the disease, this study overcomes the limitation of temporal sampling by concentrating on the phase preceding the appearance of clinically visible lesions in FECD, thereby filling an important knowledge gap and providing mechanistic insights for identifying early biomarkers and therapeutic targets.

CECs are a specialized basement membrane-associated cell type that synthesize and secrete various ECM components constituting DM.^19,20 In FECD, studies have demonstrated increased deposition of collagen types IV and VIII, fibronectin and laminin within the DM.^21–24 In the present study, we observed that ECM-associated pathways were significantly activated in the corneal endothelium of Col8a2 mutant mice at the early disease stage, preceding any apparent macroscopic abnormalities. Specifically, we observed upregulation of ECM regulatory genes, including Lgals3 and the coordinated elevation of Mmp3 and Timp1, which are involved in ECM metabolism.^25,26 These findings are highly consistent with observations in FECD patient tissues, which exhibit DM thickening, aberrant laminin accumulation, and dysregulated expression of matrix metalloproteinases and their inhibitors.^18,27–29 Beyond the role as a structural support, the DM constitutes a critical microenvironment for CEC function, with its biochemical composition and mechanical properties influencing cellular adhesion, polarity, differentiation, and survival.^30–32 Therefore, aberrant ECM composition in the early stages of FECD may create a “toxic microenvironment” that disrupts CEC function and induces apoptosis, thereby accelerating disease progression. Together, these findings demonstrate that ECM imbalance arises early in FECD and may represent a pathogenic driver, offering potential avenues for early diagnosis and targeted intervention.

Previous studies have demonstrated that in FECD, the rough endoplasmic reticulum (RER) and ER stress markers, such as Hspa5 (Grp78) and Chop, are upregulated in the corneal endothelium.¹⁴ We also found key genes of ER stress, such as Hspa5, Dnajb9, and Atf3, were significantly elevated in early-stage mutant mice. The sustained elevation of Hspa5, a key sensor of UPR, reflects chronic ER stress.³³ Dnajb9, part of the ERAD system, helps remove misfolded proteins like mutant collagens, but excessive stress can overwhelm this pathway and worsen proteotoxicity.³⁴ Atf3, a stress-responsive transcription factor, is also upregulated and regulates UPR, oxidative stress, and inflammation.^35,36 Sustained activation of the UPR indicates an attempt by the cell to alleviate ER burden and restore proteostasis; however, if compensation fails, this process may eventually trigger apoptosis.³⁷ Further studies have shown that the interplay between the ER and mitochondria plays a crucial role in the pathogenesis of FECD.^38–40 Overall, our results suggest that early ER stress acts as a triggering event that, in concert with mitochondrial dysfunction, oxidative stress, and ECM overload, drives FECD progression.

During the early progression of FECD, immune-related pathways, such as antigen presentation, chemokine release, and complement activation were markedly upregulated in mutant mice, providing in vivo evidence of immune pathway activation at pre-symptomatic stages. Key immune regulators, including Bpifb1, Icam1, and all C1q subunits, were significantly increased in early-stage Col8a2 mutants.^41–44 Lgals3 was likewise elevated at this stage, consistent with the remodeling of the immune microenvironment.⁴⁵ Late-stage FECD showed immune cell infiltration and enhanced immune activation, suggesting that chronic inflammation may contribute to disease progression.^10,46–48 Clinical studies have detected activation of both the classical and alternative complement pathways in the aqueous humor of patients with FECD, supporting a role for complement in pathogenesis.¹⁶ The coordinated upregulation of all C1q subunits at an early stage indicates activation of the classical complement pathway and innate immunity in the corneal endothelium before visible inflammatory changes occur. These findings suggest that early immune dysregulation may play a pivotal role in the onset and progression of FECD, rather than representing a secondary consequence at the late stage.

Our study still has several limitations. We analyzed two discrete timepoints representing early and late disease stages; however, future studies using finer temporal resolution are necessary to fully characterize disease evolution. Moreover, although our omics data highlight ECM remodeling, ER stress, and immune pathway dysregulation as core molecular events, our validation was limited to mouse tissues and lacked a mechanistic intervention or confirmation in human FECD samples. Subsequent research should integrate in vitro functional assays and human tissue validation to elucidate the causal roles of key genes and pathways, thereby enhancing clinical applicability.

Through differential gene expression analysis, our study not only confirmed previously reported biomarkers and therapeutic targets identified in late-stage FECD, but also uncovered several potential early molecular biomarkers, particularly within ECM remodeling, ER stress, and immune signaling pathways. Based on these early molecular markers, future studies integrating genetic testing, transcriptomic profiling, and molecular subtyping may facilitate the development of early risk prediction and precision intervention systems for FECD.