Identification of markers correlating with mitochondrial function in myocardial infarction by bioinformatics

Abstract

Background: Myocardial infarction (MI), one of the most serious cardiovascular diseases, is also affected by altered mitochondrial metabolism and immune status, but their crosstalk is poorly understood. In this paper, we use bioinformatics to explore key targets associated with mitochondrial metabolic function in MI.

Methods: The datasets (GSE775, GSE183272 and GSE236374) were from National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) in conjunction with mitochondrial gene data that were downloaded from the MitoCarta 3.0 database. Differentially expressed genes (DEGs) in the dataset were screened by ClusterGVis, Weighted Gene Co-Expression Network Analysis (WGCNA) and GEO2R, and functional enrichment was performed by Gene Set Enrichment Analysis (GSEA) and Kyoto Encyclopedia of Genomes (KEGG). Then mitochondria-associated DEGs (MitoDEGs) were obtained. Protein-protein interaction (PPI) networks were constructed to identify central MitoDEGs that are strongly associated with MI. The Cytoscape and miRWalk databases were then used to predict the transcription factors and target miRNAs of the central MitoDEG, respectively. Finally, the mouse model has been established to demonstrate the expression of MitoDEGs and their association with cardiac function.

Results: MitoDEGs in MI were mainly involved in mitochondrial function and adenosine triphosphate (ATP) synthesis pathways. The 10 MI-related hub MitoDEGs were then obtained by eight different algorithms. Immunoassays showed a significant increase in monocyte macrophage and T cell infiltration. According to animal experiments, the expression trends of the four hub MitoDEGs (Aco2, Atp5a1, Ndufs3, and Ndufv1) were verified to be consistent with the bioinformatics results.

Conclusion: Our study identified key genes (Aco2, Atp5a1, Ndufs3, and Ndufv1) associated with mitochondrial function in myocardial infarction.

Fig 1. Flowchart of the multistep screening strategy on bioinformatics data.

Fig 2. Differential gene analysis and functional enrichment of the GSE775 dataset over time using the ClusterGVis package.

C1-C6: The left side shows a line plot of the expression trend of DEGs in each cluster of the all samples. The left side of the rectangle shows the serial number of clusters, and the number of genes within each cluster (cluster size) is shown below the line plot. In the middle is a heat map of the clusters of DEGs in the control and MI samples. The right side shows the pathways that were significantly enriched for DEGs in each cluster after KEGG enrichment analysis.

Fig 3. Construction of weighted co-expression network related datasets and identification of related key modules.

A. Network topology analysis for various soft thresholds(β); B. Clustering of block feature genes. A cut line (0.4) was selected for the module dendrogram and a number of modules were merged based on the estimated similarity of the module feature genes; C. Gene dendrograms were obtained by average chained hierarchical clustering. The colored rows below the dendrogram show the module assignments determined by the dynamic tree-cutting method; D. The heatmap shows the Topology Overlap Matrix (TOM) values between the proteins of the modular network divided with the dynamic method. Yellow represents low topological overlap matrix values and red represents high topological overlap matrix values; E. WGCNA Module Adjacency Heat Map; F-G. Scatter plots of the degree of Cox regression and P-value in the dataset. x-axis represents the degree of regression and y-axis represents gene significance. Each circle represents a gene.

Fig 4. MitoDEGs in DCM.

A. Venn diagrams showed the number of Hub MitoDEGs; B. PPI network. The circles in the PPI network represent proteins and the lines represent interactions between proteins. The red depth indicates the level of importance; C-E. The enriched BP, GO, CC terms of up-regulated DEGs in Hub MitoDEGs; F. KEGG pathway enrichment results in Hub MitoDEGs.

Fig 5. MitoDEGs in MI; PPI network analysis and hub MitoDEGs identification.

Hub MitoDEGs-TFs-miRNAs regulatory network; A. Biology module from cytoscape’s MCODE plug-in; B. Hub gene identification using eight Cytohubba plugins (MCC, Closeness, MNC, Degree, EPC, Betweenness, Radiality and Stress) in Cytoscape; C. Upset Venn diagram applied to demonstrate the predicted hub genes; D. TF–hub MitoDEGs regulatory network: the red squares represent hub MitoDEGs, and the yellow dots represent transcription factors; E. miRNA–hub MitoDEGs regulatory network: the red squares represent hub MitoDEGs, and the yellow dots represent miRNA.

Fig 6

Infiltration of immune cell types compared between the MI and CON; A. The violin plot of the immune cell proportions; B. Stacked bar chart of the immune cell; C. Heatmap of the proportions of immune cell types; D. The correlation matrix of immune cell proportions.

Fig 7. Confirmation of hub MitoDEGs expression and association with cardiac function in MI mice.

A. Representative ultrasound images of the infarcted mouse; B. Hub MitoDEGs mRNA expression of CON and MI mice; C. Protein levels of Hub MitoDEGs by western blotting; D. Correlations between Aco2, Atp5a1, Ndufs3, Ndufv1 mRNA levels and cardiac functional parameters in CON and MI mice; E. Quantitative analysis of western blotting in cardiac tissues, Numbers represent 6 independent samples in each group. All data are expressed as the means ± SD. *P<0.05, **P<0.01, ***P<0.001 vs the CON group.