Multi-omics identification of aging-related diagnostic genes and therapeutic targets in renal fibrosis.

BACKGROUND

Renal fibrosis is a pathological hallmark of progressive chronic kidney disease, partly driven by aging-related molecular alterations. However, clinically relevant diagnostic biomarkers and therapeutic targets remain limited.

METHODS

We integrated four GEO datasets (GSE76882, GSE65326, GSE22459, GSE7392) to identify aging-related differentially expressed genes (DE-ARGs) in renal fibrosis. Functional enrichment, GSEA, and GSVA were performed to explore biological pathways. A total of 108 machine learning models were constructed to identify optimal diagnostic signatures. Immune infiltration (CIBERSORT), single-cell RNA sequencing (GSE269062), drug-target enrichment, molecular docking, and Mendelian randomization (MR) were used to evaluate key genes. Experimental validation was conducted using RT-PCR and immunohistochemistry in the UUO-induced renal fibrosis mouse model.

RESULTS

We identified 84 DE-ARGs enriched in inflammatory, senescence-related, and transcriptional pathways. A 7-gene diagnostic model (MMP7, AGR2, RASSF5, AHR, HIF1A, AXL, PTTG1), constructed using Random Forest and Elastic Net, showed strong predictive performance (AUC = 0.790-0.917). These genes were significantly associated with T cells and macrophages. Single-cell analysis revealed cell-type specificity, such as MMP7 in collecting duct cells and AHR in macrophages. Drug-target analysis identified 1-(5-deoxypentofuranosyl)-5-fluoropyrimidine-2,(1H,3H)-dione as a potential multitarget agent. Molecular docking confirmed stable binding. RT-PCR and IHC confirmed significant upregulation of model genes, especially LCN2. MR further supported LCN2 as a causal plasma protein for renal fibrosis.

CONCLUSIONS

We identified and validated a robust aging-related diagnostic gene signature for renal fibrosis using integrated bioinformatics and experimental approaches, offering novel insights and actionable targets for future intervention.