Decoding the air pollutant-psoriasis axis: A multi-layered systems toxicology investigation

Emerging epidemiological evidence suggests that chronic exposure to air pollutants may contribute to the development and exacerbation of immune-mediated diseases, including psoriasis. However, the molecular mechanisms underlying this association remain unclear, in part due to the limitations of traditional toxicology, which often focuses on single chemicals and single targets.

We hypothesized that air pollutants may disrupt immune homeostasis in psoriasis by engaging key inflammatory pathways and immune targets. To test this, we first compiled psoriasis-related and pollutant-associated genes from public databases. Their intersection was used for network toxicology analyses, including protein-protein interaction and pathway enrichment. Subsequently, we constructed 113 machine learning classifiers using five GEO transcriptomic datasets to identify a robust diagnostic gene signature, which was then used for downstream molecular docking simulations between seven structurally defined pollutants and immune gene targets.

We identified 51 overlapping genes enriched in IL-17, JAK-STAT, and cytokine signaling pathways. A 12-gene model (e.g., S100A9, IL4R, CCL20) demonstrated strong performance across three validation datasets (AUCs 0.911-0.966). Molecular docking showed that toluene displayed high binding affinities (< -5.0 kcal/mol) with AHR, JAK1, and NOS2, indicating potential immune regulatory disruption.

Our integrative framework combining network toxicology and machine learning provides novel insights into the immunotoxic mechanisms linking air pollution and psoriasis, highlighting candidate biomarkers and supporting future environmental health interventions.