Bioinformatics analysis of potential common pathogenic mechanism for carotid atherosclerosis and Parkinson's disease

Abstract

Background: Cerebrovascular disease (CVD) related to atherosclerosis and Parkinson's disease (PD) are two prevalent neurological disorders. They share common risk factors and frequently occur together. The aim of this study is to investigate the association between atherosclerosis and PD using genetic databases to gain a comprehensive understanding of underlying biological mechanisms.

Methods: The gene expression profiles of atherosclerosis (GSE28829 and GSE100927) and PD (GSE7621 and GSE49036) were downloaded from the Gene Expression Omnibus (GEO) database. After identifying the common differentially expressed genes (DEGs) for these two disorders, we constructed protein-protein interaction (PPI) networks and functional modules, and further identified hub genes using Least Absolute Shrinkage and Selection Operator (LASSO) regression. The diagnostic effectiveness of these hub genes was evaluated using Receiver Operator Characteristic Curve (ROC) analysis. Furthermore, we used single sample gene set enrichment analysis (ssGSEA) to analyze immune cell infiltration and explored the association of the identified hub genes with infiltrating immune cells through Spearman's rank correlation analysis in R software.

Results: A total of 50 shared DEGs, with 36 up-regulated and 14 down-regulated genes, were identified through the intersection of DEGs of atherosclerosis and PD. Using LASSO regression, we identified six hub genes, namely C1QB, CD53, LY96, P2RX7, C3, and TNFSF13B, in the lambda.min model, and CD14, C1QB, CD53, P2RX7, C3, and TNFSF13B in the lambda.1se model. ROC analysis confirmed that both models had good diagnostic efficiency for atherosclerosis datasets GSE28829 (lambda.min AUC = 0.99, lambda.1se AUC = 0.986) and GSE100927 (lambda.min AUC = 0.922, lambda.1se AUC = 0.933), as well as for PD datasets GSE7621 (lambda.min AUC = 0.924, lambda.1se AUC = 0.944) and GSE49036 (lambda.min AUC = 0.894, lambda.1se AUC = 0.881). Furthermore, we found that activated B cells, effector memory CD8 + T cells, and macrophages were the shared correlated types of immune cells in both atherosclerosis and PD.

Conclusion: This study provided new sights into shared molecular mechanisms between these two disorders. These common hub genes and infiltrating immune cells offer promising clues for further experimental studies to explore the common pathogenesis of these disorders.

Keywords: Parkinson’s disease; atherosclerosis; bioinformatics analysis; hub genes; immune cells infiltration.

FIGURE 1

Overall flowchart of this study: two atherosclerosis datasets and two Parkinson’s datasets were analyzed in this study to identify potential hub genes and types of infiltrating immune cells.

FIGURE 2

Differentially expressed genes (DEGs) analysis of individual datasets: Hierarchical clustering heatmap (A), PCA plot (B), and volcano plot (C) of atherosclerosis dataset GSE28829; Hierarchical clustering heatmap (D), PCA plot (E), and volcano plot (F) of PD dataset GSE49036.

FIGURE 3

Protein-protein interaction (PPI) network of common DEGs: (A) PPI network with 28 nodes and 68 edges; (B) the GO enrichment result of these 28 common DEGs; (C) the KEGG enrichment result of these 28 common DEGs.

FIGURE 4

Functional module of common DEGs: (A) PPI network of the key functional module with 19 nodes and 58 edges; (B) the GO enrichment result of these 19 common DEGs; (C) the KEGG enrichment result of these 19 common DEGs.

FIGURE 5

Least absolute shrinkage and selection operator (LASSO) regression model based on 19 common DEGs: (A) LASSO coefficient profiles of the 19 prognostic DEGs; (B) cross-validation to select the optimal tuning parameter (λ); (C) the GO enrichment result of the common hub genes; (D) the KEGG enrichment result of the common hub genes.

FIGURE 6

Validating the diagnostic efficacy of hub genes: Wilcoxon test (A) and ROC curve (B) of the diagnostic efficacy of hub genes in atherosclerosis dataset GSE28829; Wilcoxon test (C) and ROC curve (D) of the diagnostic efficacy of hub genes in PD dataset GSE49036. (E) The ROC curves of the diagnostic efficacy of individual hub genes in GSE28829 dataset; (F) the ROC curves of the diagnostic efficacy of individual hub genes in GSE49036 dataset.

FIGURE 7

Immune cell infiltration of individual datasets: (A) boxplot of normalized ssGSEA scores of 28 types of immune cells in atherosclerosis dataset GSE28829; (B) boxplot of normalized ssGSEA scores of 28 types of immune cells in PD dataset GSE49036. *P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

FIGURE 8

Least absolute shrinkage and selection operator (LASSO) regression model based on seven common infiltrating immune cells: (A) LASSO coefficient profiles of the seven prognostic infiltrating immune cells; (B) cross-validation to select the optimal tuning parameter (λ); Wilcoxon test (C) and ROC curve (D) of the diagnostic efficacy of key common infiltrating immune cells in atherosclerosis dataset GSE28829; Wilcoxon test (E) and ROC curve (F) of the diagnostic efficacy of key common infiltrating immune cells in PD dataset GSE49036.

FIGURE 9

Correlation between hub genes and immune cells: (A) Heatmap of the correlation between hub genes and immune cells in atherosclerosis dataset GSE28829; (B) heatmap of the correlation between hub genes and immune cells in PD dataset GSE49036. *P < 0.05, ^**P < 0.01.