Single-Cell Transcriptomics Shows Cellular Heterogeneity, Intercellular Communication, and Extracellular Matrix Remodeling in Corneal Fibrosis In Vivo

Alkali burns cause rapid corneal remodeling, which is characterized by fibrosis, neovascularization, and immune infiltration; however, the cellular dynamics driving these responses remain poorly understood. Here, we present the first single-cell transcriptomic atlas of the rabbit cornea on day 14 postalkali injury, capturing the cellular and molecular architecture of stromal fibrosis. Single-cell RNA sequencing of naïve and injured corneas identified 14 transcriptionally distinct populations, including keratocytes, EMT-like epithelial cells, immune infiltrates, and heterogeneous fibroblast subtypes. Injury induced the expansion of ECM-remodeling fibroblasts and immune cells, accompanied by widespread transcriptional reprogramming. Stromal subclustering revealed four fibroblast states—quiescent, activated, proliferative, and progenitor-like—each with distinct gene signatures and functions. Pseudotime analysis revealed a trajectory from proliferative fibroblasts to ECM-secreting myofibroblasts, marked by stage-specific activation of mitotic, matrix, and contractile programs. Ligand–receptor inference via CellChat identified coordinated signaling through the TGFß, PDGF, VEGF, and FGF pathways, alongside lesser-known axes such as NAMPT and NECTIN, implicating stromal–vascular crosstalk and metabolic stress in fibrotic progression. This study provides a comprehensive cellular framework for alkali-induced corneal fibrosis and identifies actionable targets for antifibrotic and antiangiogenic intervention in ocular surface disease. Overall design: This study employed a rabbit model of alkali-induced corneal fibrosis, where six New Zealand White rabbits underwent controlled corneal wounding, followed by longitudinal clinical imaging and intraocular pressure monitoring. Corneal tissues were collected 14 days post-injury and processed for histology, immunofluorescence, and scRNA-seq. scRNA-seq libraries were prepared using the 10x Genomics Chromium platform and sequenced on an Illumina NovaSeq. Raw data were processed using Cell Ranger, and downstream analyses were performed with Seurat, Harmony, Slingshot, tradeSeq, and CellChat for clustering, batch correction, trajectory inference, cell–cell communication, and gene expression dynamics. Functional enrichment was conducted via clusterProfiler using human orthologs.