In the current study, a total of 24,636 genes were detected in CECs by RNA-Seq, and 2,366 genes were identified as DEGs in FECD (1,092 upregulated and 1,274 downregulated genes). PCA revealed the presence of two visual groups: control and FECD. GO analysis indicated enrichment of the extracellular structure organization, ECM organization, responses to oxidative stress, and the apoptotic signaling pathway. Consistent with this, the reactome pathway analysis revealed a dysregulation of ECM-related pathways.
Late-onset FECD, the common form of FECD, typically appears after 50 years of age, whereas early-onset FECD is a very rare disease and shows a clinically different phenotype [2, 3]. A mutation in COL8A2 has been identified as a cause of early-onset FECD [16], while late-onset FECD shows an autosomal dominant pattern of inheritance, although sporadic cases are often seen in the clinical setting [3, 17]. Genetic linkage analysis of large families with FECD has identified multiple potential chromosomal loci associated with FECD [18–22], and four genetic mutations, TCF8 [22], SLC4A11 [23, 24], LOXHD1 [25], and AGBL1 [26], have been proposed as causes of FECD. However, these genetic mutations were rarely found in other cohorts [17]. For instance, we reported that the single nucleotide polymorphisms (SNPs) in TCF8, LOXHD1, and AGBL1 showed no heterogeneity in 36 FECD cases, while three nonsense mutations were detected in SLCA411 [27]. Therefore, the identification of other causative genetic factors is anticipated for the majority of late-onset FECD cases [28].
In 2010, Baratz and colleagues reported that several non-coding SNPs, including rs613872 around the transcription factor 4 (TCF4) gene on chromosome 18, show a strong association with FECD [29]. The same research group subsequently reported that 79% of the patients with FECD harbored an expansion of CTG trinucleotide repeat ≥ 50, whereas only 3% of non-FECD control subjects harbored this CTG expansion [30]. The high prevalence of the CTG expansion in FECD has been confirmed in multiple ethnic cohorts, with the prevalence depending on ethnicity [27, 31–37]. Following those discoveries, several disease mechanisms induced by CTG repeat expansion have been proposed: 1) dysregulation of TCF4 transcripts [12, 36, 38, 39]; 2) RNA-mediated toxicity [40–43]; 3) repeat-associated non-AUG dependent (RAN) translation [28, 44]; and 4) somatic instability of CTG repeat expansion [45]. Although these hypothetical mechanisms have been actively investigated, inspired by the high prevalence of the CTG repeat expansion, the mechanism of FECD in cases that do not harbor the repeat expansion remains unclear. One unanswered question is whether FECD without the repeat expansion has an independent causative genetic basis that does not involve TCF4 or whether FECD with and without the repeat expansion shares the same basis. This question motivated our present RNA-Seq analysis of the multiple aspects of RNA biology to understand the molecular dysregulation inducing FECD.
In this study, we identified 1,706 protein-coding DEGs, including 696 upregulated and 1,010 downregulated genes, from a total of 2,366 DEGs. Our enrichment analysis demonstrated the involvement of ECM organization, ECM-receptor interactions, and the endoplasmic reticulum lumen in the corneal endothelial transcriptome, and oxidative stress in FECD. The reduced vision associated with FECD arises from the formation of fibrous excrescences (clinically called guttae) and thickening of Descemet’s membrane [4–7]. Indeed, guttae have recently been removed by Descemet’s membrane stripping for the improvement of vision [46–52].
The enrichment of pathways related to ECM in this current study is consistent with the clinical finding that excessive production of ECM plays an important role in vision. The endoplasmic reticulum of the CECs in the FECD subjects was morphologically changed and further associated with an upregulation of markers of the unfolded protein response (UPR). Engler and colleagues proposed that the UPR plays an important role in the mechanism of FECD [53]. Consistent with this, we showed an accumulation of unfolded proteins in the corneal endothelium of 21 independent FECD subjects [54]. Our subsequent study, using a cell model established from FECD subjects, showed that TGF-β signaling induced a chronic overloading of ECM proteins into the endoplasmic reticulum, with a resulting triggering of the intrinsic apoptotic pathway through the UPR [55].
In addition, the current findings showed that oxidative stress and the p53 signaling pathway were related to FECD. Many reports suggest an involvement of oxidative stress as a canonical cause of disease pathology [56–60]. For instance, the corneal endothelium in eyes with FECD is susceptible to oxidative DNA damage, which in turn leads to p53-mediated apoptosis, which may play a role in the cell death process [57]. Taken together, our current enrichment analysis findings support several of the proposed potential mechanisms underlying FECD. In the future, researchers can utilize RNA-Seq to generate data regarding gene expression related to identified pathways for further elucidation of the molecular mechanism of FECD.
In the early stage of FECD, the corneal endothelium maintains a polygonal cell morphology, but it shows a drop in cell density and the formation of sporadic guttae in the corneal center [2, 3]. By contrast, in the severe stage, the CECs lose their polygonal shape and are transformed into fibroblastic cells [61]. Therefore, the DEGs observed here were induced by two processes: 1) the primary alteration of genes due to FECD and 2) a secondary alteration induced by the wound-healing process due to severe cell death. In the current study, we obtained samples from patients with relatively early-stage FECD, so their CECs presumably still had a polygonal morphology. In a future study, comparing the DEGs between early-stage and severe-stage subjects could be informative to illustrate the primary or secondary alterations of gene expression.
A key limitation of our study is the lack of analysis of CTG trinucleotide repeat expansion in TCF4, as this repeat expansion has been viewed as the most likely potential cause of FECD, accounting for 20–80% of occurrences [27, 31–37]. Only one report has investigated DEGs in patients with and without the repeat expansion [12]. RNA-Seq using each of three batches of samples in that study showed upregulation of 28 genes and downregulation of 11 genes in patients with the repeat expansion compared to patients without the repeat expansion, but no significantly enriched GO terms were found. Repeating this analysis in a larger number of samples in different cohorts would be worthwhile, as it could provide insights into whether FECD with and without the repeat expansion shares a common genetic cause.
In conclusion, we have generated an RNA-Seq dataset from patients with FECD. Enrichment analysis identified multiple ECM-related pathways that are consistent with the FECD clinical hallmarks of the formation of guttae and the thickened fibrous Descemet’s membrane. The findings also support our previous hypothetical proposal that excessive production of ECM plays a central role in the pathophysiology of FECD through cell death induced by ECM changes and promotion of the UPR. Modulation of ECM dysregulation might be a potential therapeutic modality to counteract guttae formation and CEC death.