Decoding mitochondrial stress genes in DCM: towards precision diagnosis and therapy

Abstract

Background: Mitochondrial oxidative stress (ROS) is a crucial factor in the pathogenesis of dilated cardiomyopathy (DCM). Despite its significance, robust biomarkers for assessing its role remain scarce. This study investigates ROS mechanisms in DCM and identifies associated biomarkers, offering fresh insights into diagnosis and treatment.

Methods: We sourced transcriptomic data from the GEO database and mitochondrial oxidative stress-related genes from GeneCards. Using consensus clustering, we identified 64 genes associated with mitochondrial oxidative stress in DCM and further isolated five hub genes through protein-protein interaction and machine learning techniques. These genes were analyzed for functions related to immunity, drug sensitivity, and single-cell localization. Concurrently, we collected blood samples from DCM patients to validate the hub genes' expression.

Results: The study identified five hub genes related to mitochondrial oxidative stress: VCL, ABCB1, JAK2, KDR, and NGF. Expression analysis revealed high levels of VCL, ABCB1, KDR, and NGF in the non-failing (NF) group, while JAK2 was elevated in the DCM group (p < 0.05). Diagnostic efficacy, measured by area under the curve (AUC), was significant for VCL (76.4), ABCB1 (80.1), JAK2 (68.2), KDR (78.1), and NGF (71.8). Moreover, several drugs were identified as potential regulators of these hub genes, including Topotecan, CDK9_5576, Acetalax, Afatinib, and GSK591. Notably, VCL showed increased expression in DCM patient blood samples, consistent with transcriptomic and single-cell findings.

Conclusion: This research highlights key genes associated with mitochondrial oxidative stress-VCL, ABCB1, JAK2, KDR, NGF-that show differential expression in DCM and myocardial infarction. These findings underscore their diagnostic potential and pave the way for new therapeutic strategies.

Keywords: Bioinformatics; Diagnosis; Dilated cardiomyopathy; Immune infiltration; Mitochondria; Oxidative stress.

Fig. 1

Identification of differential oxidative stress genes in DCM. A NF and DCM were subjected to a reduction in dimension through principal component analysis. B Differential gene volcanoes in GSE141910. C Difference analysis heat map in GSE141910. D Consistency clustering results for K = 2. E Consistency clustering CDF of K = 2 to 8. F Three-dimensionality reduction results (UMAP, TSNE and PCA) demonstrate differences in sample clustering. G Wayne diagram of the intersection of mitochondrial oxidation genes and differential genes. H Cluster1 and cluster2 difference gene volcano map. I Cluster1 and Cluster2 differential gene heat maps

Fig. 2

Intersection Gene Enrichment Analysis. A Wayne diagram of the intersection of different mitochondrial oxidation genes and different cluster difference genes. B Intersection gene GO enrichment molecular function bubble diagram. C Intersection gene GO enrichment biological process bubble diagram. D Intersection gene GO enrichment analysis cell composition bubble diagram. E Bubble plot of intersection gene GSEA enrichment analysis. F GO enrichment analysis network diagram

Fig. 3

Multiple machine learning to further screen key genes. A Protein interaction network diagram of the intersection gene string in string database. B MCODE algorithm obtains protein interaction subnetworks. C Chord diagram of gene correlation of the subnetwork. D LASSO regression coefficient screening plot. E LASSO regression gene locus plot. F Random forest algorithm gene importance ranking. G LASSO gene and random forest gene intersection Venn diagram

Fig. 4

Immunoinfiltration analysis in GSE141910 dataset. A Difference of immune cells between DCM and NF group. B Heat map of correlation between Hub gene and immune cells in DCM samples

Fig. 5

Differential expression of key genes. A Differential expression of key genes between NF and DCM groups. B Differential expression of key genes between cluster 1 and cluster 2. C RT-qPCR results for key genes. D ROC curve of key genes in predicting disease or not in a sample

Fig. 6

Drug susceptibility analysis of key genes. A Difference in drug susceptibility between DCM and NF group. B Drug susceptibility differences between cluster 1 and cluster 2. C Scatter plot of correlations between key genes and differential drugs

Fig. 7

Single cell localisation of key genes in GSE121893 dataset. A UMAP results for different samples of single cells. B UMAP maps of cells from different clusters. C UMAP map of cell type annotation. D UMAP of cells at different ages. E Heat maps of key gene expression in different cell types. F Bubble plot of key gene expression in different cell types