Genome-wide identification of microRNAs regulating cholesterol and triglyceride homeostasis

Abstract

Genome-wide association studies (GWASs) have linked genes to various pathological traits. However, the potential contribution of regulatory noncoding RNAs, such as microRNAs (miRNAs), to a genetic predisposition to pathological conditions has remained unclear. We leveraged GWAS meta-analysis data from >188,000 individuals to identify 69 miRNAs in physical proximity to single-nucleotide polymorphisms (SNPs) associated with abnormal levels of circulating lipids. Several of these miRNAs (miR-128-1, miR-148a, miR-130b, and miR-301b) control the expression of key proteins involved in cholesterol-lipoprotein trafficking, such as the low-density lipoprotein (LDL) receptor (LDLR) and the ATP-binding cassette A1 (ABCA1) cholesterol transporter. Consistent with human liver expression data and genetic links to abnormal blood lipid levels, overexpression and antisense targeting of miR-128-1 or miR-148a in high-fat diet-fed C57BL/6J and Apoe-null mice resulted in altered hepatic expression of proteins involved in lipid trafficking and metabolism, and in modulated levels of circulating lipoprotein-cholesterol and triglycerides. Taken together, these findings support the notion that altered expression of miRNAs may contribute to abnormal blood lipid levels, predisposing individuals to human cardiometabolic disorders.

Figure 1

Identification of miRNAs located in genomic loci that are enriched for SNPs associated with abnormal circulating total cholesterol, triglyceride, LDL-C and HDL-C levels. (a–c) Locus plots showing associations from Willer et al. with total cholesterol (TC), triglycerides (TG), LDL-C (LDL) and HDL-C (HDL) at the regions harboring miR-128-1 (a), miR-148a (b), and miR-130b and miR-301b (c). Color saturation levels correspond to linkage disequilibrium values (R²) to the top SNP for that particular trait in the 1000 genome EUR panel. Genetic recombination rates, estimated from the 1000 Genomes sequencing project, are plotted in blue. (d,e) Regional association plots for the eQTL analysis (top track) aligned against the Willer et al. joint GWAS analysis (middle track) for total cholesterol (TC) for miR-128-1 (d) and LDL cholesterol for miR-148a (e). Green arrows (bottom track) represent miRNA and protein-coding transcripts as annotated by the GENCODE Project. hsa refers to human (Homo sapiens) miRNA. The data show concordance of SNP peaks and eQTL signals, linking miR-128-1 and miR-148a hepatic expression to SNPs associated with abnormal circulating cholesterol levels.

Figure 2

Expression of LDLR and ABCA1 is regulated by miRNAs located in genomic loci associated with abnormal circulating blood lipid levels. (a–d) Immunoblots of predicted targets of miR-128-1 (a), miR-148a (b), miR-130b (c) and miR-301b (d) that are involved in cholesterol and lipid regulation (LDLR and ABCA1) and energy homeostasis (insulin receptor (InsR) precursor (Pre) and mature (Beta) forms, IRS1, INSIG1, SIRT1, AMPK-α1, AMPK-α2, SIK1, CPT1-α, PGC1-α). Human HepG2 hepatoma cells were transfected with the indicated precursor (pre-miRNA) or antisense (antimiR) oligonucleotides. β-tubulin was used as a loading control. (e) Luciferase activity in HEK293T cells transfected with a luciferase reporter with or without the LDLR 3′ UTR. (f) Luciferase activity in cells co-transfected with the indicated miRNA precursor and a wild-type LDLR 3′ UTR-luciferase reporter or a reporter harboring point mutations in the predicted miRNA target sites. (g,h) Uptake of Dil-LDL by human HepG2 hepatoma cells after treatment with miR-128-1, miR-148a, miR-130b, or miR-301b precursors (g) or antimiRs (h). (i,j) Cholesterol efflux from HepG2 cells loaded with radiolabeled cholesterol after transfection with the indicated precursors (i) or antimiRs (j). PC, precursor control. AC, antimiR control. a.u., arbitrary units. All errors bars represent mean ± s.d. of ≥3 experiments in triplicate. Statistical significance between groups is calculated by unpaired t-test, *P < 0.05, **P < 0.01, ***P < 0.001 compared to cells transfected with empty vector (e), PC (f,g,i) or AC (h,j).

Figure 3

miR-128-1, miR-148a, miR-130b and miR-301b regulate ABCA1 expression and cholesterol efflux in mouse macrophages. (a) Luciferase activity in HEK293T cells transfected with a luciferase reporter with or without the ABCA1 3′ UTR. (b) Luciferase activity in cells co-transfected with the indicated miRNA precursor and a wild-type ABCA1 3′ UTR-luciferase reporter or a reporter harboring point mutations in the predicted miRNA target sites. (c) Abca1 expression in the mouse macrophage cell line J774A.1 after overexpression (pre-miRNAs) or inhibition (antimiRs) of miR-128-1, miR-148a, miR-130b or miR-301b. β-tubulin was used as a loading control. (d) Cholesterol efflux from J774A.1 cells loaded with radiolabeled cholesterol after transfection with the indicated precursors (left) and antimiRs (right). (e) Abca1 expression in LXR agonist (T0901317; T090)-stimulated mouse peritoneal macrophages transfected with the indicated antimiRs. A scrambled antimiR was used as a control. Hsp90 was used as a loading control. (f) Cholesterol efflux from mouse peritoneal macrophages after transfection of the indicated antimiRs (inh-miRNAs) with or without T0901317 treatment. Cl, scrambled antimiR control. (g) Ago2 PAR-CLIP analysis of mouse BMDM. Genome browser screen shot (mouse mm9 release) of the Abca1 locus is shown. Black shadings on the top line show RNA-seq data (RPKM, reads per kilobase per million mapped reads) from BMDM. Ago2 PAR-CLIP reads, shown in blue, identified five distinct, high-confidence miRNA binding sites. The Abca1 transcript is indicated by blue boxes (the wider box indicates the coding region and the narrower box indicates the 3′ UTR) and the arrows indicate the direction of transcription of the Abca1 gene. RefSeq, reference sequence database. TargetScan 6.2 was used to identify miRNAs that interact with the miRNA binding sites identified by the Ago2 PAR-CLIP analysis. miRNAs identified in this study are labeled in red. MicroRNA expression levels in mouse BMDM are indicated in colored boxes on a log₁₀ scale (red indicates high expression, black indicates low expression). PC, Precursor control. AC, AntimiR control. a.u., arbitrary units. All errors bars represent mean ± s.d. of ≥3 experiments in triplicate. Statistical significance between groups is calculated by unpaired t-test, *P < 0.05, **P < 0.01 compared to cells transfected with the empty vector (a), PC (b,d) or AC (d,f).

Figure 4

Overexpression of miR-128-1 and miR-148a in C57BL/6J mice decreases circulating levels of HDL-C, whereas antisense antagonism of these miRNAs increases LDL-C clearance. (a) Analysis of hepatic Ldlr and Abca1 expression in mice injected with lentivirus expressing miR-128-1 or miR-148a (n = 7 per group). β-tubulin was used as a loading control. (b) Fast protein liquid chromatography (FPLC) analysis of pooled sera from C57BL/6J mice (n = 10) injected with lentivirus expressing miR-128-1 or miR-148a and fed HFD. Control mice were injected with a control lentivirus expressing a scrambled pre-miRNA. (c) Mean area under the curve (AUC) values calculated for HDL-C fractions 33–38 isolated by FPLC from b. (d) Copepod green fluorescent protein (copGFP) expression in the liver of C57BL/6J mice infected with lentiviruses, which encoded the indicated pre-miRNA together with copGFP. (e–h). Hepatic expression of Abca1 (e), Ldlr (f), Irs1 (g), and Cpt1a (h) in C57BL/6J mice fed HFD and overexpressing the indicated miRNAs. n = 10 mice per group. (i,j) LDL clearance after inhibition of miR-128-1 (i) or miR-148a (j) in C57BL/6J mice (n = 5 per group). Control mice were injected with a scrambled LNA. Error bars represent s.e.m. Statistical significance between groups is calculated by unpaired t-test, *P < 0.05, **P < 0.01, ***P < 0.001 compared to mice injected with a control lentivirus (b,c,e–h) or a scrambled LNA control (i,j). a.u., arbitrary units.

Figure 5

Antisense inhibition of miR-148a in Apoe^−/− mice fed a Western-type diet results in altered circulating lipoprotein/lipid levels. (a) Hepatic miR-148a levels determined by qRT-PCR in Apoe^−/− mice injected with a LNA targeting miR-148a (n = 10 per group). (b) Ldlr and Abca1 expression in liver from Apoe^−/− mice treated with antimiR-148a. (c) Hepatic expression of Ldlr, Abca1, and Cpt1a mRNA after short-term (5 d) antimiR-148a treatment of Apoe^−/− mice fed a Western-type diet (n = 10 mice per group). (d) FPLC analysis of pooled sera from Apoe^−/− mice (n = 10) fed a Western-type diet after treatment with antimiR-148a for 5 d. (e,f) Mean AUC values calculated for the LDL (e) and HDL (f) fractions (17 to 24 and 31 to 37, respectively) isolated by FPLC from d. (g) FPLC analysis of pooled sera from 10 mice after longitudinal (16 weeks) antimiR-148a treatment of Apoe^−/− mice fed a Western-type diet. (h,i) Mean AUC values calculated for the VLDL/LDL (h) and HDL (i) fractions (15 to 24 and 31 to 36, respectively) isolated by FPLC from g. Control mice were injected with a scrambled LNA. AUC, area under the curve. Error bars represent s.e.m. Statistical significance between groups is calculated by unpaired t-test, *P < 0.05, **P < 0.01, ***P < 0.001. a.u., arbitrary units.

Figure 6

Antisense inhibition of miR-128-1 in Apoe^−/− mice fed a Western-type diet results in altered circulating lipoprotein/lipid levels, improved glucose homeostasis/insulin signaling and decreased hepatic steatosis. (a) Hepatic miR-128-1 levels determined by quantitative RT-PCR in Apoe^−/− mice injected with an LNA targeting miR-128-1. n = 10 per group. (b) Immunoblots in liver tissue from Apoe^−/− mice treated with antimiR-128-1 showing expression levels of Ldlr, insulin receptor (InsR) precursor (Pre) and mature (Beta) forms, Irs1, phosphorylated (activated) and total Akt1. β-tubulin was used as a loading control. (c,d) Levels of cholesterol (c) and TAGs in the VLDL fraction (d) isolated by FPLC fractionation from pooled sera of Apoe^−/− mice (n = 10) fed a Western-type diet. Histograms represent the mean AUC values for cholesterol (c, right) and triglyceride (d, right). (e) H&E staining (left) and steatosis quantification (right) in liver of Apoe^−/− mice treated with antimiR-128-1 over 16 weeks. Micrographs from two control LNA–treated (#1 and #2) and two antimiR-128-1 LNA–treated (#3 and #4) mice are shown. Scale bars, 1 mm. (f) IP-GTT performed after 16 weeks of treatment with antimiR-128-1 in Apoe^−/− mice (n = 10 per group). (g) VLDL-TAG secretion in Apoe^−/− mice treated with antimiR-128-1 after inhibition of lipoprotein lipase with Triton WR1339 (n = 10 per group). Error bars represent s.e.m. Statistical significance between groups is calculated by unpaired t-test, *P < 0.05, **P < 0.01, ***P < 0.001. a.u., arbitrary units.