Identification and analysis of novel endometriosis biomarkers via integrative bioinformatics

Abstract

Endometriosis is a gynecological disease prevalent in women of reproductive age, and it is characterized by the ectopic presence and growth of the eutopic endometrium. The pathophysiology and diagnostic biomarkers of endometriosis have not yet been comprehensively determined. To discover molecular markers and pathways underlying the pathogenesis of endometriosis, we identified differentially expressed genes (DEGs) in three Gene Expression Omnibus microarray datasets (GSE11691, GSE23339, and GSE7305) and performed gene set enrichment analysis (GSEA) and protein-protein interaction (PPI) network analyses. We also validated the identified genes via immunohistochemical analysis of tissues obtained from patients with endometriosis or healthy volunteers. A total of 118 DEGs (79 upregulated and 39 downregulated) were detected in each dataset with a lower (fold change) FC cutoff (log2|FC| > 1), and 17 DEGs (11 upregulated and six downregulated) with a higher FC cutoff (log2|FC| > 2). KEGG and GO functional analyses revealed enrichment of signaling pathways associated with inflammation, complement activation, cell adhesion, and extracellular matrix in endometriotic tissues. Upregulation of seven genes (C7, CFH, FZD7, LY96, PDLIM3, PTGIS, and WISP2) out of 17 was validated via comparison with external gene sets, and protein expression of four genes (LY96, PDLIM3, PTGIS, and WISP2) was further analyzed by immunohistochemistry and western blot analysis. Based on these results, we suggest that TLR4/NF-κB and Wnt/frizzled signaling pathways, as well as estrogen receptors, regulate the progression of endometriosis. These pathways may be therapeutic and diagnostic targets for endometriosis.

Keywords: LY96; PDLIM3; PTGIS; TLR4/NF-κB; Wnt/frizzled; endometriosis; estrogen receptor.

Figures 1

Identification of DEGs using integrative bioinformatics analysis. (A) The distribution of transcriptome obtained from GSE11691, GSE23339, and GSE7305 were shown. (B) The volcano plots show the DEGs from GSE11691, GSE23339, and GSE7305. Red indicates relative upregulated genes with p-value < 0.05 and Log2|FC| > 2, orange indicates relative upregulated genes with p-value < 0.05 and 1 < Log2|FC| ≤ 2, blue indicates relative downregulated genes with p < 0.05 and Log2|FC| > 2, and light cyan indicates relative downregulated genes with p < 0.05 and 1 < Log2|FC| ≤ 2. (C, D) Venn diagrams of upregulated and downregulated genes DEGs from three indicated datasets. The cutoff values of Log2|FC| were set to 1 (C) and 2 (D).

Figure 2

Analysis of the enriched pathways and interaction networks of DEGs in endometriosis. (A) KEGG and GO pathway analysis of DEGs from GSE11691, GSE23339, and GSE7305 datasets with Log2|FC| > 1 were performed with JEPETTO plugin of Cytoscape, and scatter plot were constructed with q-value and XD-score as x- and y-axis, respectively. Red character indicates the pathways with q-value < 0.1 and black character indicates the pathways with 0.1 < q-value < 0.25. (B, C) Protein–protein interaction network of DEGs from three datasets with Log2|FC| > 1 were analyzed in Cytoscape with plugins of STRING (B) and GeneMANIA (C). Red character indicates genes enriched in endometriosis and blue character indicates genes enriched in normal.

Figure 3

Expression and correlation of 22 DEGs with Log2|FC| > 2. (A) Heat maps show the expression levels of 22 DEGs in datasets, GSE11691, GSE23339, and GSE7305. (B) Gene networks display correlations of 22 DEGs in the normal and endometrial tissues of each dataset. The depth of edges indicates absolute Rho ranging from 0.5 to 1. The color of edges indicates co-expression ranging from -1 (red) to 1 (blue) by Spearman’s Rho.

Figure 4

Identification of pathway networks by GSEA. (A, B) Common pathway categories identified by GSEA with gene set database of Hallmark (A) and KEGG (B) are presented. (C) GSEA was performed to obtain enriched GO-terms and visualized using Enrichment Map plugin of Cytoscape. The size of each node indicates the size of gene set. Red and blue indicate the node enriched in endometriosis and normal tissue, respectively.

Figure 5

The protein expression of LY96, PDLIM3, PTGIS, and WISP2 in endometriosis tissue. (A–D) The protein expression of (A) LY96, (B) PDLIM3, (C) PTGIS, and (D) WISP2 was measured by IHC analysis of endometrial tissues from healthy volunteers or endometriosis patients. The representative IHC images are shown in upper panel (bar = 100 μm). The graphs in lower panel represent IHC score. **p < 0.01 and ***p < 0.001.

Figure 6

The schematic illustration of the hypothesized signaling network in endometriosis. Signaling interaction between six identified DEGs including C7, CFH, FZD7, LY96, PDLIM3, and PTGIS, and three enriched pathways including TLR4/NF-κB, Wnt/frizzled, and estrogen receptors were hypothesized and schematically illustrated. Solid line indicates direct interaction, and dashed line indicates indirect and/or proposed interaction.