Epithelial program-level remodeling in periodontitis reveals coordinated basal-cycling program expansion with suppressed wound-migratory states

Background Periodontitis remodels the gingival epithelium, yet program-level epithelial changes remain incompletely defined. Methods We analyzed an integrated single-cell RNA-seq atlas of oral mucosa (GSE152042, GSE161267, GSE162412 and the COVID-19 Cell Atlas) to delineate keratinocyte states. Marker-guided enrichment and clustering resolved seventeen epithelial subsets that collapsed into five transcriptional programs: basal, cycling, inflammatory, migratory/EMT-like, and wound-activated. Donor-aware pseudobulk testing, trajectory inference, and ligand–receptor modeling (CellChat) was used to map condition-specific remodeling. Gene enrichment and regulatory network analyses were performed with STRING, ReactomePA, and iRegulon. To assess the functional effects of selected candidate regulators, experiments were conducted in OKF6/TERT2 and HGK keratinocytes. GRHL1 and TFAP2C were overexpressed and RELA was silenced, with perturbation efficiency assessed by RT-qPCR and Western blot. Proliferation (CCK-8), colony-formation, and wound-healing assays quantified functional outcomes. Results Periodontitis epithelium was associated with increased inflammatory and basal program activity and reduced wound-activated and migratory program activity. CellChat predicted increased epithelial connectivity toward fibroblasts and immune cells via CXCL/TNF signaling. In vitro, GRHL1 and TFAP2C overexpression and RELA silencing reduced proliferation, colony formation and wound closure. These findings demonstrate effects on keratinocyte behavior but do not directly validate the predicted transcriptional programs or isolate migration from proliferation. Conclusions Periodontitis is associated with inflammatory activation and attenuation of migratory and wound-healing programs in the gingival epithelium. GRHL1, TFAP2C and RELA emerge as candidate regulators of keratinocyte proliferation and migration; their roles in the program shifts observed in vivo require further mechanistic validation.