Identification of hypoxia- and immune-related biomarkers in patients with ischemic stroke

Abstract

Background: The immune microenvironment and hypoxia play crucial roles in the pathophysiology of ischemic stroke (IS). Hence, in this study, we aimed to identify hypoxia- and immune-related biomarkers in IS.

Methods: The IS microarray dataset GSE16561 was examined to determine differentially expressed genes (DEGs) utilizing bioinformatics-based analysis. The intersection of hypoxia-related genes and DEGs was conducted to identify differentially expressed hypoxia-related genes (DEHRGs). Then, using weighted correlation network analysis (WGCNA), all of the genes in GSE16561 dataset were examined to create a co-expression network, and module-clinical trait correlations were examined for the purpose of examining the genes linked to immune cells. The immune-related DEHRGs were submitted to gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. A protein-protein interaction (PPI) network was constructed by Cytoscape plugin MCODE, in order to extract hub genes. The miRNet was used to predict hub gene-related transcription factors (TFs) and miRNAs. Finally, a diagnostic model was developed by least absolute shrinkage and selection operator (LASSO) logistic regression.

Results: Between the control and IS samples, 4171 DEGs were found. Thereafter, the intersection of hypoxia-related genes and DEGs was conducted to obtain 45 DEHRGs. Ten significantly differentially infiltrated immune cells were found-namely, CD56dim natural killer cells, activated CD8 T cells, activated dendritic cells, activated B cells, central memory CD8 T cells, effector memory CD8 T cells, natural killer cells, gamma delta T cells, plasmacytoid dendritic cells, and neutrophils-between IS and control samples. Subsequently, we identified 27 immune-related DEHRGs through the intersection of DEHRGs and genes in important modules of WGCNA. The immune-related DEHRGs were primarily enriched in response to hypoxia, cellular polysaccharide metabolic process, response to decreased oxygen levels, polysaccharide metabolic process, lipid and atherosclerosis, and HIF-1 signaling pathway H. Using MCODE, FOS, DDIT3, DUSP1, and NFIL3 were found to be hub genes. In the validation cohort and training set, the AUC values of the diagnostic model were 0.9188034 and 0.9395085, respectively.

Conclusion: In brief, we identified and validated four hub genes-FOS, DDIT3, DUSP1, and NFIL3-which might be involved in the pathological development of IS, potentially providing novel perspectives for the diagnosis and treatment of IS.

Keywords: Hub gene; Hypoxia-related genes; Immune cell infiltration; Immune microenvironment; Ischemic stroke.

Graphical abstract

Fig. 1

Identification of DEGs in IS: (a) Volcano plot of DEGs among IS and normal sample; (b) Heatmap of Top 15 up- and down-regulated DEGs between normal and IS sample; and (c) Wayne diagram of DEFRGs.

Fig. 2

Identification of infiltrating immune cells in IS: (a) heatmap of 28 infiltrating immune cells in IS and normal sample, and (b) significantly different abundances of 10 infiltrating immune cells.

Fig. 3

Development of co-expression networks: (a,b) detection of outlier sample detection (a) and clustering tree of samples (b); (c) calculation of the best soft-thresholding power; (d) idenrification of modules based on the co-expression network; and (e) relevance of modules and differential immune cells.

Fig. 4

Identification of DEHRGs and functional enrichment analysis: (a) Wayne diagram of immune-related DEHRGs; (b) GO analysis of immune-related DEHRGs; and (c) KEGG analysis of immune-related DEHRGs.

Fig. 5

Identification of hub genes: (a) PPI network of immune-related DEHRGs; (b) MCODE plug-in identified hub genes; and (c) degree results for the four hub genes.

Fig. 6

Functional similarity of the four hub genes.

Fig. 7

Correlation analysis of the four hub genes.

Fig. 8

Correlation analysis between hub genes and immune infiltrating cells.

Fig. 9

GSEA results related to the function of hub genes: (a) DDIT3; (b) DUSP1; (c) NFIL3; (d) FOS.

Fig. 10

Drug–gene networks constructed by Cytoscape. Blue represents hub genes and green represents targeted drugs. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 11

miRNA–hub gene–TF regulation network constructed using Cytoscape. The light blue rectangle is hub gene, orange oval is miRNA, and purple diamond is TF. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 12

Expression analysis of the four hub genes.

Fig. 13

ROC curve analysis of the four hub genes. AUC, area under the curve.

Fig. 14

Construction and validation of the immune-related genes involved in IS diagnostic model: (a,b) The plot of error plots for 10-fold cross-validation (a) gene coefficients (b) and in LASSO analysis; (c) ROC curve of diagnostic model; (d,e) PR and DCA curves of diagnostic model; and (f) ROC curve of diagnostic model tested in GSE58294 dataset.

Fig. 15

Expression verification of the four hub genes: (a) The expression of FOS, DDIT3, DUSP1, and NFIL3 after MCAO, observed by immunohistochemical staining; and (b) Expression of FOS, DDIT3, DUSP1, and NFIL3 after MCAO, examined by Western blotting; 3 mice/group, Scale bar = 20 μm. The entire data are expressed as means ± SDs. Fold change is the ratio of signal density between the control and MCAO samples. *P < 0.05 represents a statistically significant difference between the two groups. Each experiment was repeated three times.