Introduction: Antibody-mediated rejection (ABMR) significantly threatens kidney allograft stability and the long-term survival of recipients. This study aims to determine if differential RNA expression can detect ABMR in kidney allografts and to understand the biological significance of these differences between stable and ABMR allografts.
Method: RNA transcriptome data were downloaded from the Gene Expression Omnibus database, including samples with definite labels of either ABMR or no rejection (NR). Differential RNA expression was identified using Empirical Bayes Estimation with a threshold of false discovery rate <0.001 and |log2 fold change| ≥ 2. Cibersort was employed for immune cell infiltration assays in the kidney allograft based on these differential expression signatures (DEGs). Additionally, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was conducted based on these differential expression signatures. Natural Killer Cell-related signatures were used to establish diagnostic classifiers via an artificial neural network (ANN), and their performance in the test set was evaluated.
Results: A total of 325 differentially expressed RNA signatures (DEGs) were identified with the above mentioned threshold. Cibersort analysis revealed distinct immune cell infiltration patterns between ABMR and NR samples. ABMR samples showed higher infiltration of memory B cells, plasma cells, neutrophils, M1 macrophages, activated NK cells, and resting NK cells. In contrast, NR samples were enriched in CD4+ Tregs, γδ T cells, naïve CD4+ T cells, resting dendritic cells, eosinophils, M0 and M2 macrophages, resting mast cells, and monocytes infiltration. KEGG analysis indicated these DEGs were significantly enriched in cytokine-cytokine receptor interaction, NK cell-mediated cytotoxicity, and cell adhesion molecules processes. Particularly, HLA-E, GZMB, PRFI, ICAM1, ICAM2, KLRD1, SH2D1B, and PRF1 in the NK cell-mediated cytotoxicity pathway were up-regulated in ABMR samples. The ANN classifiers based on these signatures involved in the NK cell-mediated cytotoxicity process demonstrated an ability to identify ABMR samples with 87.88% sensitivity, 77.27% specificity, a Kappa of 0.5658, and a P-value (Acc>NIR) of 0.03819.
Conclusion: NK cell-mediated cytotoxicity signatures were up-regulated in ABMR allografts and were effective in identifying ABMR samples when using the ANN classifier. This study establishes a molecular basis for diagnosing ABMR and elucidates the biomedical significance of these molecules in the ABMR pathological process. Further exploration of NK cell involvement in ABMR could enhance our understanding of its underlying mechanisms.