Blood Milieu in Acute Myocardial Infarction Reprograms Human Macrophages for Trauma Repair

Figures

• 

  Figure 1. Acute myocardial infarction (AMI) serum exposure elicits a distinct transcriptional signature in primary human macrophages. A) Human cardiac AMI tissue stained for von Willebrand Factor (brown) and haematoxilin (blue). 10× magnification of the healthy and ischemic region within the same heart. Scalebar is 100 µm. B) Schematic study design. C) Differentially expressed genes (DEGs) in AMI‐mac versus Ctrl‐mac (n = 47 AMI, n = 20 ctrl). D) Transcriptome‐based principal component analysis (PCA) plot of AMI‐mac and Ctrl‐mac. E) Area under the receiver operating characteristics curve (ROC AUC) and accuracy (Mean ± SD) in support vector machine (SVM), random forest (RF), and XGBoost (XGB) classification (n = 47 AMI, n = 20 ctrl) using transcriptomics data. F) Differential expression levels of 36 common feature selected genes in AMI‐mac versus Ctrl‐mac.

• 

  Figure 2. Acute myocardial infarction (AMI) and control serum exposed macrophages are functionally different. A) High content screening images with image cytometry data graphs for phagocytosis, lipid uptake, apoptosis, hypertrophy#, morphology#, actin stress#, granularity#, and inflammasome## assays in AMI‐mac versus Ctrl‐mac in 10×, #40×, or ##20× magnification. Dotted areas are enlarged in inserts. n = 50 AMI, n = 20 ctrl. Scalebar is 20 µm. B) Macrophage cytokine secretion after 24 h AMI or control serum exposure. n = 50 AMI, n = 20 ctrl. C) Partial least squares discriminant analysis (PLS‐DA) based on Macroscreen data. D) Area under the receiver operating characteristics curve (ROC AUC) and accuracy (Mean ± SD) in support vector machine (SVM), random forest (RF), and XGBoost (XGB) classification using Macroscreen data. E) SHapley Additive Explanations (SHAP) calculated feature importances of Macroscreen functionalities to the classification model RF. *p < 0.05, **p < 0.01, ***p < 0.001.

• 

  Figure 3. Acute myocardial infarction (AMI) serum exposure induces immune activation and metabolic programs, while cell cycling is suppressed. A) Pearson correlation coefficients (PCC) between AMI‐mac and Ctrl‐mac weighted gene co‐expression network analysis (WGCNA) modules [module size] and Macroscreen functionalities. p‐adj.: * < 0.05, ** < 0.01, *** < 0.001. PCC numbers (p‐adj.) are added for AMI. B) Gene ontology (GO) term interaction network of AMI‐correlating WGCNA modules M2, M5, M7, M14, and M17. C) Nuclear counts (mean ± SEM) and 10× high content screening images of Hoechst‐labeled nuclei of AMI‐mac and Ctrl‐mac. n = 50 AMI, n = 20 ctrl. ***p < 0.001. Scalebar is 50 µm. D) Significantly overlapping differentially expressed genes (DEGs) between AMI‐correlating WGCNA module genes and AMI‐associated DEGs in independent peripheral blood mononuclear cell (PBMC) mRNA dataset GSE59867 (hypergeometric testing; p = 1.62 × 10−26).

• 

  Figure 4. Acute myocardial infarction (AMI) serum reprogramming of macrophages associates with clinical progression. (A) Pearson correlation coefficient (PCCs) between AMI‐specific weighted gene co‐expression network analysis (WGCNA) modules [module size] and infarct size (INFSIZE), ejection fraction (EF), and end‐diastolic volume (EDV). *p < 0.05. B) Expression heatmap of AMI‐specific WGCNA module genes in GSE59867 (AMI patients with versus without subsequent heart failure, HF). Similar up‐ (red sidebar) or down‐ (green sidebar) regulation between the two datasets is indicated. C) Principal component analysis (PCA) plot of gene expression data in GSE59867 (HF versus non‐HF peripheral blood mononuclear cells (PBMCs)). D) Regulatory network showing the top 1% drivers of WGCNA modules (color‐coded) that correlated with cardiac performance 4 months after AMI. E) Overlap between prostaglandin E2 (PGE2) downregulated and POSTEMI‐Large upregulated genes (hypergeometric testing; p = 9.36 × 10−13). F) Log2 fold changes for the POSTEMI‐Large top‐ranked differentially expressed genes (DEGs; p < 0.001).