Identification and immuno-infiltration analysis of cuproptosis regulators in human spermatogenic dysfunction.

Introduction

Cuproptosis seems to promote the progression of diverse diseases. Hence, we explored the cuproptosis regulators in human spermatogenic dysfunction (SD), analyzed the condition of immune cell infiltration, and constructed a predictive model.

Methods

Two microarray datasets (GSE4797 and GSE45885) related to male infertility (MI) patients with SD were downloaded from the Gene Expression Omnibus (GEO) database. We utilized the GSE4797 dataset to obtain differentially expressed cuproptosis-related genes (deCRGs) between SD and normal controls. The correlation between deCRGs and immune cell infiltration status was analyzed. We also explored the molecular clusters of CRGs and the status of immune cell infiltration. Notably, weighted gene co-expression network analysis (WGCNA) was used to identify the cluster-specific differentially expressed genes (DEGs). Moreso, gene set variation analysis (GSVA) was performed to annotate the enriched genes. Subsequently, we selected an optimal machine-learning model from four models. Finally, nomograms, calibration curves, decision curve analysis (DCA), and the GSE45885 dataset were utilized to verify the predictions' accuracy.

Results

Among SD and normal controls, we confirmed that there are deCRGs and activated immune responses. Through the GSE4797 dataset, we obtained 11 deCRGs. ATP7A, ATP7B, SLC31A1, FDX1, PDHA1, PDHB, GLS, CDKN2A, DBT, and GCSH were highly expressed in testicular tissues with SD, whereas LIAS was lowly expressed. Additionally, two clusters were identified in SD. Immune-infiltration analysis showed the existing heterogeneity of immunity at these two clusters. Cuproptosis-related molecular Cluster2 was marked by enhanced expressions of ATP7A, SLC31A1, PDHA1, PDHB, CDKN2A, DBT, and higher proportions of resting memory CD4+ T cells. Furthermore, an eXtreme Gradient Boosting (XGB) model based on 5-gene was built, which showed superior performance on the external validation dataset GSE45885 (AUC = 0.812). Therefore, the combined nomogram, calibration curve, and DCA results demonstrated the accuracy of predicting SD.

Conclusion

Our study preliminarily illustrates the relationship between SD and cuproptosis. Moreover, a bright predictive model was developed.