Integrated multi-omics and causal inference framework with experimental validation reveals key drivers of air pollution–induced acute kidney injury

Background: Air pollution has emerged as a significant risk factor for acute kidney injury (AKI), yet the molecular mechanisms underlying this association remain poorly defined. This study aimed to elucidate the nephrotoxic effects of representative air pollutants and identify molecular targets involved in pollutant-induced AKI. Methods: We developed a multi-layered computational and experimental framework integrating omics-based target prediction, network toxicology, machine learning, Mendelian randomization (MR), single-cell profiling, molecular docking with dynamic simulations, and analysis of pollutant-exposed model. Nine representative air pollutants were selected based on environmental relevance and suspected nephrotoxicity. A diagnostic gene signature was constructed using multiple machine learning algorithms, and key targets were evaluated through transcriptome-wide MR. Pollutant-protein interactions were assessed using molecular docking and dynamics simulations. Single-cell data and in vivo transcriptomes from pollutant-exposed models were used to construct a pollutant-target-celltype network. Finally, experimental validation was performed using in vitro exposure of mouse proximal tubular cells. Results: Nephrotoxicity predictions revealed substantial heterogeneity among pollutants, with carbon monoxide, benzene, and ozone exhibiting the highest toxic potential. A total of 49 overlapping genes were identified and found to be enriched in pathways related to inflammation and oxidative stress. A 38-gene diagnostic model demonstrated strong predictive performance across datasets, highlighting a set of core targets potentially involved in both the pathogenesis and prognosis of air pollution–induced AKI. Transcriptome-wide MR analysis further prioritized five genes— NPPA, TGIF1, IL18, CRLS1, and KLF2, —with significant causal associations with AKI. Single-cell transcriptomic profiling revealed that proximal tubular, immune, and endothelial cells are particularly susceptible to pollutant-induced injury. Molecular docking and dynamic simulations identified high-affinity pollutant–protein interactions. In vitro experiments showed that exposure of mouse proximal tubular cells to PM 2.5 and benzene reduced cell viability, induced apoptosis, and significantly dysregulated key genes, providing experimental support for computational predictions. Conclusion: This study provides novel mechanistic insights into air pollution–induced AKI by identifying key genes, pathways, and susceptible renal cell types. The integrative framework combining multi-omics, causal inference, and experimental validation establishes a robust foundation for future translational research and therapeutic development targeting environmentally driven kidney injury.