Integrative Analysis of miRNA and mRNA Expression Profiles Associated With Human Atrial Aging

FIGURE 1. miRNA-seq analysis of human atrial samples with SR. (A) Differential expression analysis of 1,482 miRNAs between 60 and 40 years old group. X-axis, miRNA expression was quantified as count per million at log2 scale. Y-axis, fold change (SR60/SR40) at log2 scale. (B) Differential expression analysis of 1,482 miRNAs between 70 years old group and 50 years. (C) Venn diagram showing significant miRNAs from the above two comparisons at nominal P < 0.05. (D) Heatmap and hierarchical structure of seven overlapping miRNAs across 12 human atrial samples.

FIGURE 2. Comparative analysis of seven human AA-associated miRNAs. (A) Correlation analysis of effect size of miRNAs from SR60 vs. SR40, and SR70 vs. SR50. LogFC, fold change (SR60/SR40 or SR70/SR50) at log2 scale. (B) Venn diagram showing two miRNAs overlapping with cardiac aging-related miRNAs [GSE433556, false discovery rate (FDR) < 0.05]. (C,D) Linear regression analysis of miR-23a-5p and miR-34a-5p with age. The miRNA expression was quantified as count per million at log2 scale (Y-axis). For each miRNA, linear tread and 95% confidence interval were shown as purple line and a gray area. (E) Independent qRT-PCR replication of three miRNAs among four groups.

FIGURE 3. mRNA-seq analysis of human atrial samples. (A) Meta-analysis of two mRNA-seq datasets. X-axis, −log10 (P value) calculated from differential analysis between SR60 and SR40. Y-axis, −log10 (P value) calculated from differential analysis between SR70 and SR50. Dots with steel blue color indicate genes (or mRNAs) significantly associated with human AA at FDR < 0.05. (B) Counts of genes with opposite (−+ and +) and consistent effects (− and ++). (C) Heatmap and hierarchical structure of 42 AA-associated genes across 12 human atrial samples. Each row indicated a gene, its Entrez ID or official symbol was shown at right side. (D) Examples of top-ranked genes (CACNA1G and VTN) showing strong linear trends with human age. Pearson Correlation coefficient (r) and the corresponding P value were shown at the top.

FIGURE 4. Identifying downstream genes of AA-associated miRNAs. (A) Flowchart of predicting miRNA-mRNA pairs for human AA by integrating mRNA-seq data with bioinformatics analysis. FDR, false discovery rate. (B) Four types of miRNA-mRNA interactions. In each plot, X-axis represented expression levels (count per million at log2 scale) of a miRNA, and Y-axis represented expression levels of a gene. Labels with “–” or “++” indicated decreased or increased expression in older group compared to younger group. Such effect size were also highlighted by color (color, down-regulation; red, up-regulation). Pearson Correlation coefficient (r) and the corresponding P value were shown at the top.

FIGURE 5. Network analysis of 114 miRNA-mRNA pairs associated with human AA. Each node represented a miRNA (triangle shape) or gene (oval shape). The color of node indicated whether the expression of a miRNA or gene increased (light red) or decreased (light green) in older group. The color of edge indicated the negative (light green) or positive (light green) correlation between miRNA and downstream gene. Edge with sold line indicated direction interaction between miRNA and gene, i.e., gene was a predicted target of miRNA and it negatively correlated with expression of miRNA. Edge with dashed line indicated miRNA-mRNA pairs with indirect correlation, that said, the gene strongly correlated with expression of miRNA (Pcor < 0.05) and AA (FDR < 0.05). The genes in round rectangle were highlighted by gene set enrichment analysis (Supplementary Table S3).