Integrative transcriptomic analysis identifies CCL22-associated immune signatures in air pollution-related atopic dermatitis.

BACKGROUND: Air pollution has been associated with the development and exacerbation of atopic dermatitis (AD), but the molecular signatures connecting pollutant-related targets with AD-associated immune dysregulation remain incompletely characterized. METHODS: We applied an integrated systems toxicology and transcriptomic framework to prioritize candidate pollutant-related immune signatures in AD. Pollutant-associated targets were intersected with high-confidence AD-related genes, followed by protein-protein interaction analysis, GO/KEGG enrichment, machine learning, immune infiltration analysis, single-cell transcriptomics, in silico CCL22 perturbation, 1-fluoro-2, 4-dinitrobenzene (DNFB)-induced AD-like mouse validation, and exploratory molecular docking. RESULTS: Shared pollutant-AD targets were mainly enriched in cytokine activity, chemokine signaling, pattern-recognition receptor activity, IL-17 signaling, cytokine-cytokine receptor interaction, and Toll-like receptor-related inflammatory pathways. A machine learning framework based on 15 algorithms and 175 predictive combinations identified plsRglm + AdaBoost as the optimal model, with an AUC of 0.963 in the training cohort and AUCs of 1.000, 0.909, and 0.966 in three validation cohorts. The model identified a pollutant-prioritized AD signature including CCL22, CCL5, CSF2, F2RL1, HRH4, ICAM1, IFNG, IL10, IL17A, and IL18. CCL22 was upregulated in AD samples and mainly localized to dendritic cells and macrophages. In silico CCL22 perturbation was associated with extracellular matrix and stromal remodeling programs, while DNFB-induced AD-like dermatitis confirmed increased CCL22 expression in lesional skin. CONCLUSION: These findings identify CCL22-associated immune and stromal remodeling signatures as candidate molecular features of air pollution-related AD and generate testable hypotheses for future controlled exposure studies.