S100A8 is associated with transitional epithelial dysfunction in sepsis-induced acute lung injury

Sepsis-induced acute lung injury remains a major cause of mortality, largely because of persistent inflammatory damage and impaired alveolar epithelial regeneration. However, the regulatory mechanisms that drive epithelial repair failure in sepsis-induced lung injury remain incompletely understood. This study aimed to identify key epithelial cell states and regulatory molecules involved in sepsis-induced acute lung injury and to explore potential diagnostic and therapeutic strategies. Single-cell RNA sequencing data from septic acute lung injury and control lung tissues were integrated and analyzed to characterize epithelial cell populations, cell–cell communication, and disease-associated transcriptional changes. Differential expression and enrichment analyses were performed to identify dysregulated pathways. A consensus multi-algorithm machine learning framework was used to screen diagnostic genes, and their predictive performance was evaluated in independent validation cohorts. In vivo experiments were performed using a cecal ligation and puncture mouse model, the effects of paquinimod-mediated modulation of S100A8/A9-associated inflammatory signaling were assessed by histological analysis, inflammatory cytokine measurement, immunostaining, and survival analysis. Single-cell analysis revealed a marked expansion of Krt8-positive transitional epithelial cells in sepsis-induced acute lung injury. S100A8 was aberrantly upregulated within this transitional epithelial population, suggesting a disrupted epithelial repair program. Pharmacological inhibition of S100A8 reduced the accumulation of Krt8-positive transitional epithelial cells, attenuated lung inflammatory injury, decreased inflammatory cytokine release, and partially improved survival in septic mice. Machine learning analysis identified six core diagnostic genes, including ARL6IP1, CEBPA, PDK4, CAT, NPC2, and TSPAN2, which showed moderate discriminatory performance in independent validation datasets. This study identifies an S100A8-dependent regulatory axis associated with transitional epithelial dysfunction in sepsis-induced acute lung injury. These findings suggest that S100A8 inhibition may represent a potential therapeutic strategy, while the machine learning-derived gene signature may provide useful diagnostic value for sepsis-associated lung injury.