Particulate Matter Exposure and Diabetic Kidney Dysfunction: Insights from Integrated Transcriptomic and Bioinformatics Analyses
Abstract
Exposure to ambient particulate matter (PM) has been linked to renal dysfunction, particularly in diabetic populations, but the underlying mechanisms remain unclear. We performed bidirectional Mendelian randomization to assess causal relationships between PM exposure and estimated glomerular filtration rate (eGFR), integrated transcriptomic datasets to identify PM-related genes overlapping with diabetic kidney disease (DKD) differentially expressed genes, and applied machine learning approaches to select key feature genes and construct diagnostic models. Single-cell and spatial transcriptomic analyses were used to characterize cell-type and region-specific expression patterns, while in silico knockout analysis explored potential functional associations. PM2.5–10 exposure was causally associated with decreased eGFR, particularly among individuals with diabetes, with no evidence of reverse causality. Transcriptomic integration identified 168 shared PM-DKD genes enriched in inflammatory, immune, and metabolic pathways, including AGE-RAGE, IL-17, TNF, and PI3K-Akt signaling. Seven feature genes (AVPI1, DUSP1, FOSB, JUNB, PDK2, TPPP3, and VIM) showed good diagnostic performance across training and external validation cohorts, and machine learning models and nomogram analyses demonstrated consistent predictive performance. Single-cell and spatial transcriptomic analyses revealed distinct cell-type and region-specific expression patterns, with VIM enriched in interstitial and fibrotic regions, TPPP3 mainly detected in podocytes, and other genes distributed across tubular or immune cell populations. In silico knockout analysis suggested potential associations of these genes with mitochondrial metabolism, oxidative stress, tubular function, and inflammatory processes. Database-based therapeutic exploration identified VIM as a potential candidate target, with sanguinarine showing favorable predicted binding affinity. Collectively, these findings suggest that PM2.5–10 exposure may contribute to DKD susceptibility through inflammatory, metabolic, and oxidative stress-related mechanisms, and provide candidate molecular markers for further investigation.