Identification of biomarkers related to copper metabolism in patients with pulmonary arterial hypertension

Abstract

Background: The pathogenesis of pulmonary arterial hypertension (PAH) and associated biomarkers remain to be studied. Copper metabolism is an emerging metabolic research direction in many diseases, but its role in PAH is still unclear.

Methods: PAH-related datasets were downloaded from the Gene Expression Omnibus database, and 2067 copper metabolism-related genes (CMGs) were obtained from the GeneCards database. Differential expression analysis and the Venn algorithm were used to acquire the differentially expressed CMGs (DE-CMGs). DE-CMGs were then used for the coexpression network construction to screen candidate key genes associated with PAH. Furthermore, the predictive performance of the model was verified by receiver operating characteristic (ROC) analysis, and genes with area under the curve (AUC) values greater than 0.8 were selected as diagnostic genes. Then support vector machine, least absolute shrinkage and selection operator regression, and Venn diagrams were applied to detect biomarkers. Moreover, gene set enrichment analysis was performed to explore the function of the biomarkers, and immune-related analyses were utilized to study the infiltration of immune cells. The drug-gene interaction database was used to predict potential therapeutic drugs for PAH using the biomarkers. Biomarkers expression in clinical samples was verified by real-time quantitative PCR.

Results: Four biomarkers (DDIT3, NFKBIA, OSM, and PTGER4) were screened. The ROC analysis showed that the 4 biomarkers performed well (AUCs > 0.7). The high expression groups for the 4 biomarkers were enriched in protein activity-related pathways including protein export, spliceosome and proteasome. Furthermore, 8 immune cell types were significantly different between the two groups, including naive B cells, memory B cells, and resting memory CD4 T cells. Afterward, a gene-drug network was constructed. This network illustrated that STREPTOZOCIN, IBUPROFEN, and CELECOXIB were shared by the PTGER4 and DDIT3. Finally, the results of RT-qPCR in clinical samples further confirmed the results of the public database for the expression of NFKBIA and OSM.

Conclusion: In conclusion, four biomarkers (DDIT3, NFKBIA, OSM, and PTGER4) with considerable diagnostic values were identified, and a gene-drug network was further constructed. The results of this study may have significant implications for the development of new diagnostic biomarkers and actionable targets to expand treatment options for PAH patients.

Keywords: Biomarkers; Copper metabolism-related genes; Pulmonary arterial hypertension.

Fig. 1

Identification of 85 DE-CMGs and their enrichment analysis. A 814 DEGs including 258 up-regulated (red dots) and 556 down-regulated (green dots) genes from the GSE33463 dataset in the volcano map. B Venn diagram to detect 85 DE-CMGs. C Heatmap of the expression of the top 100 DEGs. D The top 27 GO terms included 10 biological process (BP) terms, 7 cellular component (CC) terms, and 10 molecular function (MF) terms of the DE-CMGs. E Top 20 KEGG pathways of the DE-CMGs

Fig. 2

Construction of WGCNA to identify DE-CMG modules in the GSE33463 dataset. A Cluster dendrogram of module eigengenes to detect outlier samples. B Dendrogram of all expressed genes in the PAH and control samples clustered based on a dissimilarity measure (1‐TOM). C Analysis of the scale-free topology fit index and the mean connectivity for various soft-threshold powers (β) for the genes. D Hierarchical clustering tree based on the topological overlap dissimilarity (1-TOM). E Heatmap of the module-trait relationships. The corresponding P values are also annotated

Fig. 3

Identification and validation of four biomarkers. A Heatmap of the expression of 28 module DE-CMGs in PAH and control samples from the training set. B The accuracy and error of estimate generation for the SVM‐RFE algorithm in the training set. (C) Candidate genes selected by the LASSO regression model. D Four biomarkers detected by Venn diagram. ROC curves of the prognostic values of the four biomarkers in the training (E), testing (F), and GSE113439 (G) sets

Fig. 4

Ten module DE-CMGs with AUC values > 0.8

Fig. 5

Top 5 KEGG pathways of the four biomarkers. Top 5 KEGG pathways of DDIT3 (A), NFKBIA (B), OSM (C), and PTGER4 (D) by GSEA enrichment

Fig. 6

Evaluation of immune cell infiltration using the CIBERSORT algorithm. A A stacked bar plot of the proportions of 22 immune cell types in control and PAH samples from the training based on the CIBERSORT algorithm. B Vioplot of 22 immune cell contents in the control and PAH samples from the training set. C Correlations between the 4 biomarkers and 22 immune cell types. D Correlation analysis between the expression level of DDIT3 and abundance of neutrophils (left), and the expression level of OSM and abundance of resting memory CD4 T cells (right) in the training set

Fig. 7

Wilcoxon’s test. Wilcoxon’s test of the expression of OSM and DDIT3, and the abundance of memory-resting CD4 T cells and neutrophils between the PAH and control samples

Fig. 8

Drug–gene interaction diagram. The red square indicates the four biomarkers, and the green circle indicates the drugs

Fig. 9

The expression of biomarkers (NFKBIA (A), OSM (B), DDIT3 (C), and PTGER4 (D)) in clinical PBMC samples detected by RT-qPCR. NP indicates normal peopole. **p value < 0.01