Identification and analysis of oxidative stress-related genes in endometriosis

Abstract

Background: Early diagnosis and treatment of endometriosis (EM) remain challenging because of the lack of knowledge about EM development. While oxidative stress (OS) has been associated with EM, the link is unclear. We explored OS-related genes (OSRGs) and their role in EM pathogenesis.

Material and methods: We combined two ectopic endometrium (EC) and eutopic endometrium (EU) datasets (GSE11691 and GSE25628) into a dataset for analysis. Bioinformatic analyses were used to identify differentially expressed genes (DEGs), OS-related genes (OSRGs), enriched pathways, competitive endogenous RNA network, and immune cell infiltration. Finally, real time-quantitative polymerase chain reaction (RT-qPCR) and Western blot (WB) were used to validate the expression of key OSRGs in clinical patient samples.

Results: Bioinformatic analysis identified 459 DEGs between EC and EU samples, including 67 OSRGs. A ceRNA network was established, encompassing 28 DE-OSRGs, 32 miRNAs, and 53 lncRNAs. Four key OSRGs (CYP17A1, NR3C1, ENO2, and NGF) were selected from protein-protein interaction network analysis. The RT-qPCR and WB analysis showed that these genes' abnormal changes in RNA and protein levels were consistent with data in public databases. Weighted gene co-expression network analysis identified three immune-related OSRGs (CYP17A1, NR3C1, and NGF) and 20 lncRNAs that may regulate NR3C1 through 10 miRNAs.

Conclusion: The key OSRGs may function via multilayered networks in EM. We provide insights into EM and underscore the potential significance of OSRGs and the immune environment for diagnostic and prognosis evaluation.

Keywords: ectopic endometrium; endometriosis; eutopic endometrium; immune cells; lncRNA; machine learning; oxidative stress.

Figure 1

Schematic diagram of the overall study design.

Figure 2

Visualization of the data processing for GSE11691 and GSE25628, highlighting the effects of background adjustment and quantile normalization. (A) The principal component analysis (PCA) results before correction. (B) The box plots before normalization. (C) The PCA results after correction. (D) The box plots after normalization.

Figure 3

Identification of differentially expressed genes (DEGs) between ectopic endometrium (EC) and eutopic endometrium (EU) samples. (A) Volcano plot representing the distribution of DEGs. (B) Heat map detailing the expression patterns of the DEGs. (C) Venn diagram showing the intersection of 67 DE-OSRGs between DEGs and OSRGs.

Figure 4

Construction of the competitive endogenous RNA (ceRNA) regulation network. (A) Volcano plot representing the distribution of differentially expressed miRNAs (DE-miRNAs). (B) Volcano plot representing the distribution of differentially expressed lncRNAs (DE-lncRNA). (C) ceRNA network encompassing mRNAs, miRNAs, and lncRNAs.

Figure 5

Function enrichment analysis and protein-protein interaction (PPI) network. (A, B) Gene Ontology (GO) of DE-OSRGs. (C, D) Kyoto Encyclopedia of Genes and Genomes (KEGG) of DE-OSRGs. (E) PPI network of DE-OSRGs. (F) The top 20 genes with the highest connectivity.

Figure 6

The development of the machine learning diagnostic models and key OSRGs. (A, B) Among the 31 DE-OSRGs, ACTA2, CYP17A1, ENO2, NGF, NR3C1, SELP, and SPP1 were identified with a connectivity score greater than five by least absolute selection and shrinkage operator (LASSO) algorithm. (C, D) CYP17A1, NR3C1, ENO2, NGF, and CCL2 were pinpointed via the Support vector machine recursive feature elimination (SVM-RFE) algorithm. (E) CYP17A1, NR3C1, ENO2, and NGF were obtained by overlapping both machine learning results as key OSRGs. (F) Receiver operating characteristic (ROC) curve of the four key OSRGs. (G) Expression levels of four key OSRGs were compared in the combined dataset. *P < 0.05, ***P < 0.001, ****P < 0.0001.

Figure 7

The mRNA and protein levels of key OSRGs. (A)The agarose gel electrophoresis of PCR amplification products of four key OSRGs in the indicated samples. GAPDH was used as an endogenous control. (B) RT-qPCR of four key OSRGs. (C) Western blot of four key OSRGs. (D) Quantitative analysis of OSRG proteins from panel (C). β-actin was used as a loading control for western blot analysis. Data are presented as mean ± SD (n = 4). **P < 0.01.

Figure 8

Gene set enrichment analysis (GSEA) results of key OSRG-related signaling pathways. (A) The CYP17A1-related signaling pathways. (B) The NR3C1-related signaling pathways. (C) The ENO2-related signaling pathways. (D) The NGF-related signaling pathways.

Figure 9

Immune infiltration analysis from EC and EU samples. (A) The distribution of 22 distinct immune cell types across each specimen. (B) The proportion of six types of immune cells exhibited significant differences between EC and EU samples. *P < 0.05, **P < 0.01, ***P < 0.001, ns: P > 0.05.

Figure 10

Identification of immune-related module genes via weighted gene co-expression network analysis (WGCNA). (A) The results of cluster analysis indicated the absence of outlier samples. (B) The optimal soft threshold (β) was finally chosen as 11. (C, D) By constructing a co-expression network and setting the minimum number of genes per gene modules to 300, six modules were obtained. (E, F) According to the scoring correlation, the MEblue module was taken as a key module with 2,284 genes. (G) CYP17A1, NR3C1, and NGF were acquired by overlapping key OSRGs with immune-related module genes. (H) The ceRNA regulation network of NR3C1.