Corneal endothelial cells (CECs) play a vital role in maintaining corneal clarity and thus visual acuity by regulating the hydration status of the corneal stroma through active pumping¹. Because of the limited proliferative potential of human CECs, the cell number decreases with age, and the repair of corneal endothelial wounds following injury is usually achieved through cell enlargement and migration, rather than cell mitosis². When the CEC count decreases below a threshold of approximately 500 cells/mm², the dehydration status of the corneal stroma cannot be maintained, leading to bullous keratopathy and vision impairment^3,4.

The limited proliferative potential of human CECs has been attributed to several factors, including reduced expression of the epidermal growth factor and its receptor in aging cells⁵, antiproliferative TGFβ2 in the aqueous humor⁶, and contact inhibition^2,7. Although some growth factors, such as basic fibroblast growth factor (bFGF), can increase proliferation in a cultured human corneal endothelium, the culture efficiency remains limited^8,9. Furthermore, CECs may undergo a phenotypic change during ex vivo expansion, resembling epithelial-mesenchymal transition (EMT)^10-13. Endothelial-mesenchymal transition (EnMT) is characterized by cell junction destabilization, apical-basal polarity loss, cytoskeletal rearrangement, alpha smooth muscle actin expression, and type I collagen secretion¹⁴. All of these characteristics may abrogate the normal function of CECs, hampering the use of ex vivo cultured CECs in tissue engineering. Moreover, EnMT has been associated with the pathogenesis of several corneal endothelial diseases, including Fuchs endothelial corneal dystrophy and retrocorneal membrane formation^15,16. Therefore, understanding the mechanism of EnMT may aid in manipulating the EnMT process and facilitate the regeneration of CECs to enable competent function.

In this study, we described a method for isolating bovine CECs from the corneal button. In the primary culture in vitro, the EnMT process, including a phenotypic change, the nuclear translocation of β-catenin, and EMT regulators snail and slug, was observed. We further described a method for demonstrating EnMT in vivo by using a rat corneal endothelium cryoinjury model. Using these 2 models, we demonstrated that marimastat, a broad-spectrum matrix metalloproteinase (MMP) inhibitor, can suppress the EnMT process. The described protocols facilitate the detailed analysis of the EnMT mechanism and the development of strategies for manipulating the EnMT process for further clinical application.