Mitochondrial metabolic remodeling and multi-omics profiling identify plasma biomarkers of myocardial infarction

Figures

• 

  Fig. 1. Overview of the experimental workflows employed for proteomic and transcriptomic analyses. Heart tissue samples, including infarct, border, and remote regions, were collected from sham operated and MI groups. RNA was extracted from each tissue region, labeled with cyanine 3 dye, hybridized to a custom ovine gene microarray and scanned with an Agilent G2505B 2 dye scanner. Plasma samples were obtained at 24 and 48 hour post-surgery. Proteomic sample preparation was performed using an automated workflow on the GANUS-3 workstation. Both tissue and plasma proteomes were analysed using DIA-based mass spectrometry to enable biomarker discovery.

• 

  Fig. 2. The macroscopic characteristics of an infarcted sheep heart, corresponding cardiac tissue stained with 2,3,5-triphenyltetrazolium chloride (TTC) and histological sections stained with hematoxylin and eosin (H&E). (A) Sheep heart three days post-MI, with arrows indicating the ligation site and the sources of samples from different regions (infarct, border, and remote). (B) Cross-section of the left ventricle with sampled regions. The transmural extent of the infarction is evident by a paler hue, with tissue samples systematically gathered from distinct zones: the infarct zone (indicated by the green dashed line), and the border zone (blue dashed line). The remote sample was collected from a distant part of the ventricular region as indicated in (A). (C) Technical replicates of hematoxylin & eosin-stained micrographs of sheep myocardial tissue from infarct, border (as indicated in (B)) and sham regions captured at 20× magnification (scale bar = 100 μm). Black arrows indicate noticeable myocardial necrosis is observed in the infarct core on day 3 post MI, and less visible in the salvageable border zone. (For interpretation of the references to colour in this figure legend, the…

• 

  Fig. 3. Functional network analysis of 98 downregulated mitochondrial proteins identified in the infarct region relative to both the border and remote regions (from a total of 102 significantly dysregulated proteins; see Supplementary File 1). Networks were constructed from KEGG (Kyoto Encyclopedia of Genes and Genomes)/Reactome pathways as well as Gene Ontology (GO) terms. Functionally related terms were clustered according to shared protein composition using the kappa score. Each cluster is represented by a distinct colour, with only the most significant term within each cluster displayed as the representative label. The size of each node reflects the statistical significance of the corresponding enrichment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Functional network analysis of 98 downregulated mitochondrial proteins identified in the infarct region relative to both the border and remote regions (from a total of 102 significantly dysregulated proteins; see Supplementary File 1). Networks were constructed from KEGG (Kyoto Encyclopedia of Genes and Genomes)/Reactome pathways as well as Gene…

• 

  Fig. 4. Comparison of mitochondrial protein in infarct versus remote, infarct versus border and border versus remote (A) DIA-MS analysis was used to compare a targeted set of mitochondrial proteins in infarct, border, and remote regions. Histogram bars indicate relative differences indicated based on proteomics data between the regions (infarct versus remote region (navy blue), infarct versus border region (pink) and border versus remote region (azure blue)). (B) RNA extracted from infarct, border, and remote tissues was analysed via microarrays, and the quantified mRNAs obtained through transcriptomics were compared across the three regions. (C) Analysis of variance comparing infarct versus remote regions (navy blue) and border versus remote regions (azure blue) as well as infarct versus remote regions (navy blue) and infarct versus border regions (pink) shows a significant reduction in total mitochondrial proteins associated with tricarboxylic acid cycle (ANOVA P < 0.0001, P < 0.0004) (E) Oxidative Phosphorylation (ANOVA P < 0.0001, P < 0.0003). (D) Proteomics-based pathway visualization using Pathview [27] illustrating KEGG maps of the Tricarboxylic acid cycle (F) and oxidative…

• 

  Fig. 5. Integrative analysis of protein, mRNA and miRNA expression profiles for gene regulatory network inference (A–B). Overlap between 30 upstream regulators identified using the Upstream Regulator Analysis tool of IPA from proteins that changed upon Infarction (blue) and (A) 1621 differentially expressed genes (red), (B) 7488 human miRNA gene targets (black) inferred for downregulated miRNAs in adolescent Infarct [16]. Out of 7488 miRNAs, 26 overlapped with upstream regulators. (C) Overlap between 26 genes identified (blue) in (B) and 7109 orthologous sheep genes (black) identified 12 overlapping upstream regulators with miRNA binding sites. (D) Identification of four upstream regulators (that include three transcription regulators: PPARGC1A, ZEB1, MYC and one cytokine: IL6) because of consensus between proteomics, mRNA and miRNA analyses. The Venn diagram shows an overlap between 30 upstream regulators predicted from proteomics data (blue), 1621 genes measured using microarray (red) and 12 miRNA target genes (black) identified in (C). The inserts with volcano plots highlight proteins annotated as regulated by the consensus upstream regulators in IPA. (E) A combined network map for 79…

• 

  Fig. 6. Differential expression analysis of plasma proteins in an ovine model of myocardial infarction. (A–B) Volcano plots showing the false discovery rate (FDR)-corrected p-values from a t-test comparing protein abundance at 24 h post-MI versus sham (A) and 48 h post-MI versus sham (B), plotted against log₂ fold change values (calculated from peak areas). Proteins significantly upregulated are highlighted in orange, and those significantly downregulated are highlighted in green (p < 0.05 after FDR correction and fold change ≥2). (C) Venn diagram illustrating differentially expressed plasma proteins at 24 h and 48 h post-surgery, highlighting the overlap between time points. The shared set includes established plasma cardiac biomarkers such as cardiac troponin I, creatine kinase, and fatty acid binding protein 3. (D) The comparison of 67 proteins at two time points revealed that the direction of changes was consistent for 51 proteins and opposite for 16 proteins at both the 24-hour and 48-hour time points. Established plasma cardiac biomarkers are highlighted using their respective gene symbols. (E) Pathway enrichment analysis of the 67 biomarker panel proteins was conducted using the…

• 

  Fig. 7. Validation of plasma biomarker panel candidates for ovine model of myocardial infarction. (A) A Venn diagram showing all mitochondrial proteins shared among datasets. Mitochondrial proteins associated with mRNA profiles (infarct vs. remote), proteomic analyses (infarct vs. remote), and plasma proteomics (24 h and 48 h) were identified using Mitocarta 3.0 [26]. Four proteins, namely SUCLG1, MDH2, HADHB, and HADHA, were found to be shared across all datasets. (B–C) A dual Y-axis horizontal bar plot illustrates the direction and abundance of common mitochondrial proteins in the plasma dataset at the 24-hour and 48-hour time points (B). This is presented in conjunction with the expression profiles of mRNA and their corresponding proteins from the transcriptomics and proteomics datasets, specifically infarct vs remote conditions (C).