LCN2 drives ferroptosis-associated ischemia–reperfusion injury after renal transplantation: integrated machine learning and in vivo validation

Renal ischemia-reperfusion injury (IRI) remains a critical obstacle to optimal renal transplant outcomes, driving acute graft dysfunction and long-term allograft failure. While ferroptosis-an iron-dependent form of cell death-has been linked to IRI pathogenesis, the role of lipocalin-2 (LCN2), a regulator of iron homeostasis and inflammation, in transplant-related renal IRI remains uncharacterized. Six murine IRI transcriptomic datasets (83 samples) were integrated using weighted gene co-expression network analysis (WGCNA) and differential expression profiling to screen for IRI-associated hub genes. Findings were validated in two human transplant cohorts (212 samples) via 113 machine learning algorithms, including logistic regression, random forest, and ensemble models. Single-cell RNA sequencing (GSE237429) was used to map gene expression to specific renal cell populations, while a murine warm IRI model evaluated the effects of LCN2 inhibition (ZINC00640089) on tubular injury, ferroptosis markers (MDA, GSH, Fe²⁺), and inflammatory cytokines (IL-6, TNF-α) across mild (50-minute) and severe (80-minute) ischemia subgroups. WGCNA identified 36 hub genes, with LCN2 emerging as a key node in ferroptosis and immune regulation pathways. A six-gene machine learning model, including LCN2, CLU, and SOX9, demonstrated robust predictive accuracy for IRI (AUC = 0.93). Single-cell analysis revealed elevated LCN2 expression in neutrophils and macrophages in IRI kidneys, correlated with increased immune cell infiltration. In vivo, LCN2 inhibition significantly reduced severe ischemia-induced tubular injury, suppressed lipid peroxidation (MDA), restored glutathione levels (GSH), and alleviated iron overload (Fe²⁺) and reactive oxygen species (ROS). Systemic inflammation was mitigated, with IL-6 and TNF-α levels significantly reduced. This study establishes LCN2 as a pivotal mediator of ferroptosis and immune dysregulation in transplant IRI. A machine learning-driven multi-omics approach provides a novel diagnostic framework, while the inhibition of LCN2 is shown to alleviate IRI-induced tissue damage in these models. These findings highlight the utility of integrative analytics in uncovering biological targets and offer new therapeutic avenues for improving kidney transplant outcomes.