Identification of Endoplasmic Reticulum Stress-Related Biomarkers in Coronary Artery Disease

Abstract

Coronary artery disease (CAD) is caused by atherosclerotic lesions in the coronary vessels. Endoplasmic reticulum stress (ERS) acts in cardiovascular disease, and its role in CAD is not clear. A total of 13 differentially expressed ERS-related genes (DEERSRGs) in CAD were identified. Functional enrichment analysis demonstrated the DEERSRGs were mainly enriched in endoplasmic reticulum (ER)-related pathways. Then, eight genes (RCN2, HRC, DERL2, RNF183, CRH, TMED2, PPP1R15A, and IL1A) were authenticated as ERS-related biomarkers in CAD by least absolute shrinkage and selection operator (LASSO). The receiver operating characteristic (ROC) analysis showed that the LASSO logistic model constructed based on biomarkers had a better diagnostic effect, which was confirmed by the ANN and GSE23561 datasets. Also, ROC results showed that seven of the eight biomarkers had better diagnostic effects. The nomogram model had good predictive power, and biomarkers were mostly enriched in pathways associated with CAD. The biomarkers were significantly associated with 10 immune cells, and RCN2, DERL2, TMED2, and RNF183 were negatively correlated with most chemokines. Eight biomarkers had significant correlations with both immunoinhibitors and immunostimulators. In addition, eight biomarkers were significantly different in both CAD and control samples, CRH and HRC were upregulated in CAD. The quantitative reverse transcription-polymerase chain reaction (qRT-PCR) showed that RCN2, HRC, DERL2, CRH, and IL1A were consistent with the bioinformatics analysis. RCN2, HRC, DERL2, RNF183, CRH, TMED2, PPP1R15A, and IL1A were identified as biomarkers of CAD. Functional enrichment analysis and immunoassays for biomarkers provide new ideas for the treatment of CAD.

Figure 1

Identification of DEERSRGs and functional enrichment analysis. (a) Volcano plot of DEGs between CAD and controls. The horizontal axis is logFC, and the vertical axis is -log10 (adj. p. al). Downregulated genes are shown in blue, upregulated genes in red, and genes with insignificant differences in gray. (b) Heatmap for the top 20 DEGs between CAD and control samples. Red indicates high expression, and blue indicates low expression. (c) Venn diagram of differentially expressed ER stress genes. (d) GO enrichment results. Sankey plots of GO entries and corresponding genes are shown on the left, and bubble plots of enrichment results are shown on the right. (e) KEGG enrichment results. Sankey plots of KEGG entries and corresponding genes are shown on the left, and bubble plots of enrichment results are shown on the right.

Figure 2

Identification of biomarkers. (a) The horizontal axis is the log lambda value, and the vertical axis is the regression coefficient. Different genes in different colors show the change of regression coefficient with the change of log lambda. (b) The horizontal axis is the log lambda value, and the vertical axis is the partial likelihood deviation value. The red dots are the variation of the partial likelihood deviation as the log lambda value is changed. The left dashed line is the position of lambda min, and the right dashed line is the position of lambda lse. (c, d) LASSO logistic model assessment. From left to right: confusion matrix heat map and ROC curve. The first grid on the upper left of the confusion matrix heat map is true positive result, the lower right is true negative result, and the upper right and lower left are false positive and false negative results. AUC, area under the curve.

Figure 3

Construction of a diagnostic model using the biomarkers. (a) Validation of the diagnostic model by plotting ROC curves at one to 5 folds. (b) Artificial neural network. (c) ROC curves in the validation set. (d) ROC curves of biomarkers.

Figure 4

Nomogram construction and the diagnostic value evaluation. (a) The visible nomogram for diagnosing CAD. (b) The calibration plot for internal validation of the nomogram. (c) The DCA curves of the nomogram. (d) Clinical impact curves of the nomogram model.

Figure 5

Immune-related analysis of eight DEGs. (a) CIBERSORT Stack Bar Chart. The vertical axis is a relative percentage, each column is 1 sample, and different colors distinguish 22 kinds of immune cells. (b) Differential immune cell between CAD and control sample. (c) Correlation between biomarkers and 22 kinds of immune cells. The vertical axis shows eight biomarkers, and the horizontal axis shows 22 immune cells. (d) Correlation between biomarkers and chemokines. Chemokines on the horizontal axis and biomarkers on the vertical axis. (e) The relationship between biomarker and immunosuppressants. Immunosuppressants on the horizontal axis and biomarkers on the vertical axis. (f) Correlation between biomarkers and immune activators. Immunosuppressants on the horizontal axis and biomarkers on the vertical axis. In (c–f), blue represents negative correlation, red represents positive correlation, and the number in the grid is the correlation coefficient cor.n|cor|. The larger the value, the higher the correlation. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001.

Figure 6

The expression levels of biomarkers in the CAD and control groups. (a) Training set; (b) qRT-PCR. ns, not significant; ^∗p < 0.05; ^∗∗∗p < 0.001.