Integrated Bioinformatics Analysis Identifies FCER1G, IL1R2, and BST1 as Diagnostic Biomarkers and Therapeutic...

Sepsis remains difficult to diagnose since blood culture has low sensitivity and current biomarkers lack sufficient specificity. We sought to identify robust diagnostic and therapeutic targets for neonatal sepsis through an integrated bioinformatics and experimental workflow.

The GSE69686 dataset (108 neonates; 48 sepsis, 60 controls collected ≤ 3 days after birth) was analyzed using a multi-step pipeline: WGCNA to identify sepsis-associated modules; limma-based differential expression; feature selection by LASSO and SVM-RFE; immune deconvolution (CIBERSORT); and pathway interrogation (GSEA). Drug enrichment (Enrichr/DSigDB) and molecular docking (AutoDock) supported targetability. Experimental validation employed a THP-1 sepsis-like model (1 μg/mL LPS, 24 h) with qRT-PCR, shRNA-mediated BST1 knockdown, Western blot, ELISA (IL-6, TNF-α, IL-1β), CCK-8, and Annexin V/PI flow cytometry.

Intersecting the outputs of LASSO and SVM-RFE yielded three robust biomarkers-FCER1G, IL1R2, and BST1. These genes are associated with broad shifts in inferred immune infiltration, and GSEA linked their high expression to INFLAMMATORY RESPONSE; BST1 and FCER1G were additionally enriched for TNFα/NF-κB signaling. In THP-1 cells, LPS upregulated all three genes; focusing on BST1, shRNA knockdown partially rescued proliferation (0-72 h), reduced IL-6/TNF-α/IL-1β secretion, and lowered apoptosis relative to sh-NC. Docking analysis suggested artesunate as a candidate modulator with a binding energy of -6.27 kcal/mol to BST1.

Our integrated pipeline nominates FCER1G, IL1R2, and BST1 as biologically coherent biomarkers of sepsis and highlights BST1 as a plausible therapeutic target warranting further investigation. These findings support external validation in multicenter cohorts and mechanistic extension-particularly functional testing of predicted modulators-in primary cells.