Machine Learning Reveals Ets2 as a Novel Target for Membranous Nephropathy Treatment and Its Role in Immune Infiltration.

Background: Membranous nephropathy (MN) is a common pathological phenotype for adult nephrotic syndrome (NS). The occurrence of MN is increasing across China, but diagnostic methods for MN still rely on kidney biopsy and PLA2R and THSD7A detection in plasma and kidney tissue, and there has been no new biomarker for MN discovered since 2014. Immune infiltration status in MN patients suffers from the dearth of associated studies. In the present study, we aimed to find new bio-markers for MN and evaluate the role of immune cells infiltration in MN pathology.

Methods: We downloaded MN expression profile from the Gene Expression Omnibus database and used R-project to screen differentially expressed genes (DEGs) and performed functional correlation analysis. Least absolute shrinkage and selection operator (LASSO) logistic regression and Radom Forest algorithms were used to screen and verify the bio-markers of MN. Finally, CIBERSORT was used to evaluate the infiltration of immune cells in MN tissues.

Results: A total of 463 DEGs were screened from the MN tissue in this study. ETS2 was identified as bio-marker for MN. The CIBERSORT results showed that there were statistical differences in monocytes, plasma cells, regulatory T cells, and memory B cells. In addition, ETS2 was positively related to monocytes, M1 phase macrophages, and neutrophils and negatively correlated to plasma cells, CD4+ T memory cells, M2 macrophages, CD8+ T cells, memory B cells, and resting mast cells.

Conclusion: (1) Machine learning algorithms reveals Ets2 as a novel target for membranous nephropathy patients. (2) Immune infiltration plays an important part in membranous nephropathy. (3) Ets2 expression is related to immune cells infiltration.