Abstract
Background Tertiary lymphoid organs (TLOs) are ectopic lymphoid structures that arise in chronically inflamed tissues, including renal allografts. While increasingly recognized in kidney transplant biopsies, the molecular features, clinical significance, and prognostic implications of TLOs remain poorly defined. This study aimed to characterize TLO-related immune landscapes and investigate their association with allograft rejection and outcomes using a multi-omics approach. Methods We integrated bulk transcriptomic datasets from multiple kidney transplant cohorts to derive TLO scores using ssGSEA and evaluated their associations with rejection, Banff histology, and graft survival. Nonnegative matrix factorization (NMF) was employed to identify TLO-based molecular subtypes. Multiple machine learning algorithms were integrated to construct prognostic and diagnostic models. Single-cell RNA sequencing was used to explore the cellular sources and differentiation dynamics of hub TLO-related genes (TRGs). In vivo validation was performed in a rat model of allograft rejection. Results Elevated TLO transcriptional activity was significantly associated with acute rejection and inferior graft survival across independent cohorts. TLO scores correlated positively with Banff lesion severity and the chronic allograft damage index. NMF clustering revealed a rejection-prone molecular subtype with increased immune infiltration and pro-inflammatory signaling. Diagnostic and prognostic models incorporating TLO-related features exhibited strong predictive performance (AUCs > 0.8). Four hub TRGs-CXCL9, CXCL11, CD40, and SH2D1A-were subsequently identified. Single-cell analysis demonstrated dynamic expression of these genes in endothelial, B, and T cell lineages, particularly during transitions toward key TLO-forming immune and stromal phenotypes. Experimental validation confirmed upregulation of hub TRGs in rejecting rat allografts. Conclusion TLOs represent immunologically active structures that shape intragraft immune architecture and contribute to rejection and chronic injury in kidney transplantation. Our study provides the first comprehensive multi-omics framework linking TLO features with clinical outcomes and identifies TRG-based biomarkers with diagnostic and prognostic utility. These findings support the integration of TLO-informed tools into transplant immunomonitoring and therapeutic decision-making.