Constructing a seventeen-gene signature model for non-obstructive azoospermia based on integrated transcriptome analyses and WGCNA

Abstract

Abstract Background Non-obstructive azoospermia (NOA) affects approximately 1% of the male population worldwide. The underlying mechanism and gene transcription remain unclear. This study aims to explore the potential pathogenesis for the detection and management of NOA. Methods Based on four microarray datasets from the Gene Expression Omnibus database, integrated analysis and weighted correlation network analysis (WGCNA) were used to obtain the intersected common differentially expressed genes (DESs). Differential signaling pathways were identified via GO and GSVA-KEGG analyses. We constructed a seventeen-gene signature model using least absolute shrinkage and selection operation (LASSO) regression, and validated its efficacy in another two GEO datasets. Three patients with NOA and three patients with obstructive azoospermia were recruited. The mRNA levels of seven key genes were measured in testicular samples, and the gene expression profile was evaluated in the Human Protein Atlas (HPA) database. Results In total, 388 upregulated and 795 downregulated common DEGs were identified between the NOA and control groups. ATPase activity, tubulin binding, microtubule binding, and metabolism- and immune-associated signaling pathways were significantly enriched. A seventeen-gene signature predictive model was constructed, and receiver operating characteristic (ROC) analysis showed that the area under the curve (AUC) values were 1.000 (training group), 0.901 (testing group), and 0.940 (validation set). The AUCs of seven key genes (REC8, CPS1, DHX57, RRS1, GSTA4, SI, and COX7B) were all > 0.8 in both the testing group and the validation set. The qRT-PCR results showed that consistent with the sequencing data, the mRNA levels of RRS1, GSTA4, and COX7B were upregulated, while CPS1, DHX57, and SI were downregulated in NOA. Four genes (CPS1, DHX57, RRS1, and SI) showed significant differences. Expression data from the HPA database showed the localization characteristics and trajectories of seven key genes in spermatogenic cells, Sertoli cells, and Leydig cells. Conclusions Our findings suggest a novel seventeen-gene signature model with a favorable predictive power, and identify seven key genes with potential as NOA-associated marker genes. Our study provides a new perspective for exploring the underlying pathological mechanism in male infertility.

Keywords

• Non-obstructive azoospermia
• Sperm
• Integrated analysis
• Weighted correlation network analysis
• LASSO regression