Succinylation-annotated genes in AMI: multi-omics and single-cell prioritization of ASGR2 and NPL

Figures

• 

  Figure 1. Workflow diagram.

• 

  Figure 2. Identification of key modules and functional enrichment analysis in AMI. (A) Selection of the soft-thresholding power for WGCNA based on both the scale-free topology fit index and mean connectivity. β = 14 was selected because it was the lowest power at which the scale-free topology fit index exceeded 0.8 while maintaining acceptable network connectivity. (B) Module-trait relationship analysis identified the pink module as most significantly correlated with AMI. (C) Venn diagram intersection of succinylation-annotated genes (n = 565) with AMI-associated genes (n = 797) yielded 18 candidate genes for further analysis. (D,E) Functional enrichment of succinylation-annotated AMI genes revealed.

• 

  Figure 3. Machine learning-based identification and validation of hub genes in AMI. (A) Performance evaluation of 107 algorithm combinations derived from 11 machine learning models. The Stepglm[both] + plsRglm and Stepglm[backward] + plsRglm models achieved the highest mean AUC values, establishing them as optimal diagnostic classifiers. (B,C) ROC curve analysis of the 8 candidate hub genes in (B) the training cohort and (C) the independent validation cohort. ASGR2 and NPL demonstrated superior diagnostic performance in both cohorts. (D,E) Violin plots illustrating the expression distribution of the 8 hub genes in (D) AMI patients vs. (E) healthy controls. ASGR2 and NPL showed the most significant differential expression.

• 

  Figure 4. Immune infiltration landscape and hub gene correlations in AMI. (A) Heatmap displaying the relative abundance of 22 immune cell types across all samples. (B) Comparative analysis revealed significantly elevated infiltration levels of monocytes and resting dendritic cells in AMI patients vs. controls. (C) Correlation matrix illustrating associations between the 8 hub genes and 22 immune cell types. (D) Focused analysis identified three hub genes (ASGR2, GLUL, NPL) showing particularly strong correlations with monocyte infiltration (|r|>0.3, FDR < 0.001).

• 

  Figure 5. Single-cell localization of hub genes in AMI. (A) Unsupervised clustering of all cells identified 22 distinct clusters (umap visualization). (B) Annotation of the 22 clusters revealed 7 major cell types, including monocytes, NK cells, and T cells. (C) Illustrating the cell-type-specific distribution of hub genes, with ASGR2 and NPL predominantly localized to monocytes. (D) Quantifying hub gene expression levels across the 7 cell types. ASGR2 and NPL exhibited higher expression in monocytes compared to other cell types.

• 

  Figure 6. Single-cell resolution of hub gene localization in monocyte subsets. (A) Unsupervised clustering of monocytes identified 15 transcriptionally distinct subpopulations. (B) Annotation using canonical markers (CD14, FCGR3A) classified monocytes into three functional subsets. (C) Hub gene expression mapping revealed ASGR2 was predominantly localized to classical monocytes, while NPL showed pan-monocyte expression. (D) Visualization of the proportion of hub gene expression in monocytes.

• 

  Figure 7. Hub gene expression dynamics in PR and NPR. (A) Comparing the expression profiles of Classical, Intermediate, and Non-classical monocytes between PR and NPR cohorts. (B) The ratio of the three monocytes is shown by histogram. (C) Subset-specific hub gene localization. (D) Visualization of the ratio of PR and NPR hub gene expression.

• 

  Figure 8. (A–D) Exploratory pseudotime analysis of monocytes suggested a continuum of transcriptional states spanning classical, intermediate, and non-classical monocytes. Plaque rupture and non-plaque rupture samples showed different distributions of monocyte states along the inferred pseudotime axis. (E,F) Dynamic expression patterns of NPL and ASGR2 along the inferred pseudotime axis. (G,H) Heatmaps of pseudotime-associated genes, illustrating genes with dynamic changes across the inferred monocyte-state continuum, including NPL and ASGR2.

• 

  Figure 9. ELISA results. Comparison of ELISA levels of NPL and ASGR2 among the Normal, NPR, and PR groups. Pairwise comparisons were performed using two-sided unpaired Wilcoxon rank-sum tests, followed by Benjamini-Hochberg correction for multiple comparisons across all six pairwise tests. FDR-adjusted P values are shown.