Decoding dry eye disease based on bioinformatics and in vitro experimental: the role of immune responses and natural product intervention.

BACKGROUND: Dry eye disease (DED), a prevalent ocular condition, has seen rising incidence rates. Aberrant inflammation and immune dysregulation are key pathogenic factors in DED. However, the underlying mechanisms linking autoimmunity to DED therapy remain incompletely understood. Herbal medicine's potential in treating DED warrants further exploration due to rendering monotherapy insufficient for clinical needs. METHODS: DED-related datasets and genes were initially identified from the GEO database and other public repositories. After retrieving the data, differential gene expression analysis and functional enrichment analysis were performed. Subsequently, a protein-protein interaction network was constructed to identify core modules and hub genes. To enhance the reliability and diagnostic value of these findings, the hub genes were further validated using independent external datasets and refined through machine learning algorithms (LASSO and SVM-RFE) to extract characteristic genes. These characteristic genes were then used for reverse prediction of potential Traditional Chinese Medicine natural products, leading to the construction of a gene-natural products network. Following this, molecular docking was employed to screen for promising natural products based on binding affinity. In parallel, immune cell infiltration was estimated using CIBERSORTx, and single-cell RNA sequencing data were analyzed to elucidate cell-type-specific expression patterns of the characteristic genes. Finally, the therapeutic potential of the top-predicted natural product was validated using an in vitro cellular model of DED. RESULTS: GSE44101 was selected as the base dataset. Differential gene analysis identified 1089 differential genes and 3504 DED-related genes from five databases, with 235 overlapping genes. Enrichment analysis linked DED to Cytokine-Cytokine receptor interaction, PI3K-Akt, IL-17, and JAK-STAT signaling pathways. The PPI network yielded 20 hub genes, validated and refined using other GEO datasets and machine learning, resulting in 10 characteristic genes: CDK1, CCNA2, CXCL13, CCR1, FEN1, CCR7, SELL, RAD51, CXCL1, and KIF11. Genistein was identified as the key TCM natural product. Molecular docking revealed stable interactions between Genistein and CDK1, IL-1β, CDC20, and CCNA2, with CDC20 showing the highest stability. CIBERSORT analysis demonstrated associations between characteristic genes and specific immune cell infiltration in DED. Single-cell RNA sequencing localized their expression to key corneal cell types, with IL-1β predominantly expressed in macrophages. Cell experiments confirmed that Genistein at a concentration of 12.5 μmol/L could protect human corneal epithelial cells from NaCl-induced hyperosmotic damage and TNF-α-induced immune injury, reducing the cell detachment rate and decreasing the area of cell death loss. Genistein downregulated the expression of CDK1, IL-1β, and CCNA2 at the RNA and protein levels, while upregulating the expression of CDC20. CONCLUSIONS: Genistein demonstrates therapeutic potential for DED, likely through regulating CDK1, IL-1β, CDC20, and CCNA2 to exert anti-inflammatory, anti-oxidative stress, and anti-apoptotic effects.