Specific circulating miRNAs are associated with plasma lipids in a healthy American cohort

Abstract

Low-density lipoprotein cholesterol (LDL-c) is both a therapeutic target and a risk factor for cardiovascular disease (CVD). MicroRNA (miRNA) has been shown to regulate cholesterol homeostasis, and miRNA in blood circulation has been linked to hypercholesterolemia. However, few studies to date have associated miRNA with phenotypes like LDL-c in a healthy population. To this end, we analyzed circulating miRNA in relation to LDL-c in a healthy cohort of 353 participants using two separate bioinformatic approaches. The first approach found that miR-15b-5p and miR-16-5p were upregulated in individuals with at-risk levels of LDL-c. The second approach identified two miRNA clusters, one that positively and a second that negatively correlated with LDL-c. Included in the cluster that positively correlated with LDL-c were miR-15b-5p and miR-16-5p, as well as other miRNA from the miR-15/107, miR-30, and let-7 families. Cross-species analyses suggested that several miRNAs that associated with LDL-c are conserved between mice and humans. Finally, we examined the influence of diet on circulating miRNA. Our results robustly linked circulating miRNA with LDL-c, suggesting that miRNA could be used as biomarkers for hypercholesterolemia or targets for developing cholesterol-lowering drugs.NEW & NOTEWORTHY This study explored the association between circulating microRNA (miRNA) and low-density lipoprotein cholesterol (LDL-c) in a healthy population of 353 participants. Two miRNAs, miR-15b-5p and miR-16-5p, were upregulated in individuals with at-risk LDL-c levels. Several miRNA clusters were positively and negatively correlated with LDL-c and are known to target mRNA involved in lipid metabolism. The study also investigated the influence of diet on circulating miRNA, suggesting potential biomarkers for hypercholesterolemia.

Keywords: LDL-c; diet; miRNA; microRNA.

Graphical abstract

Figure 1.

A: differential expression analysis in limma of 861 miRNAs between participants with clinically elevated LDL-c (>130 mg/dL) and optimal LDL-c (<100 mg/dL) found two miRNAs (miR-16-5p and miR-15b-5p) that were upregulated in participants with elevated LDL-c. Delta represents difference in miRNA expression, where left of volcano plot (negative values) shows upregulation with clinically elevated LDL-c and right of volcano plot (positive values) shows upregulation with optimal LDL-c. Adjusted P values were extracted from limma and –log transformed. B: heatmap of Spearman’s correlation analysis between module eigenvalues (left vertical axis) and cardiometabolic phenotypes (top horizontal axis) which identified the magenta module and the cyan module to have significantly strong, opposing relationships with LDL-c. Scale to the right of the heatmap represents Spearman’s rho coefficient. *<0.05, **<0.01, ***<0.0005. LDL-c, low-density lipoprotein cholesterol; miRNA, MicroRNA.

Figure 2.

Spearman’s correlation found significant relationships between LDL-c (x-axis) and magenta module eigenvalues (A, y-axis); a newly associated miRNA in the magenta module, miR-15b-5p (B, y-axis); and a previously identified miRNA in the magenta module, miR-27a-3p (C, y-axis). Magenta module eigenvalues (D, y-axis), miR-15b-5p (E), and miR-27a-3p (F) were upregulated in participants with clinically elevated LDL-c (>130 mg/dL) compared to those with optimal LDL-c (<100 mg/dL). miRNA expression in the units of Cyclic Loess normalized log2CPM. Unpaired t test used to compare expression sets between LDL-c status. G: network visualizations of WGCNA were constructed in Cytoscape using measurements from distance topological overlap matrix (TOM). **<0.01, ***<0.0005, ****<0.0001. LDL-c, low-density lipoprotein cholesterol; miRNA, MicroRNA; WGCNA, weighted gene co-expression network analysis.

Figure 3.

Spearman’s correlation found significant relationships between LDL-c (x-axis) and cyan module eigenvalues (A, y-axis); a newly associated miRNA in the cyan module, miR-874-3p (B, y-axis); and a previously identified miRNA in the cyan module, miR-320b (C, y-axis). Cyan module eigenvalues (D, y-axis), miR-874-3p (E), and miR-320b (F) were downregulated in participants with clinically elevated LDL-c (>130 mg/dL) compared with those with optimal LDL-c (<100 mg/dL). miRNA expression in the units of Cyclic Loess normalized log2CPM. Unpaired t test used to compare expression sets between LDL-c status. G: network visualizations of WGCNA were constructed in Cytoscape using measurements from distance topological overlap matrix (TOM). *<0.05, **<0.01. LDL-c, low-density lipoprotein cholesterol; miRNA, MicroRNA; WGCNA, weighted gene co-expression network analysis.

Figure 4.

Spearman’s rho coefficients (bottom horizontal axis) were quantified between miRNA (left vertical axis) from the magenta (A) and cyan (B) modules and LDL-c. In humans, fasting plasma LDL-c was correlated with circulating miRNA expression, while in DO mice exposed to dietary treatments, fasting plasma LDL-c was correlated with hepatic miRNA expression.