Type 2 diabetes in obese individuals is associated with a distinct hepatic miRNA signature.

(A) Overview of all analyzed human transcripts within the GeneChipTM miR4.0 assay. (B) Expression of the 28 regulated miRNAs in T2D compared to ND after adjustment for age, sex, BMI and NAS ranked by fold-change. Data is represented as relative mean to ND ± SD. (C) Pearson’s correlation coefficients (r) for the correlation between microarray miRNA log2 values and metabolic parameters. miRNAs are ranked by their association strength to first HbA1c, glucose, NAS, insulin, triglycerides, age and lastly BMI. (D) Mapping of significant correlations from (C) to metabolic parameters. (E) Venn diagram of all mature miRNA detected by microarray measurement in human and diet-induced obese (DIO) murine liver samples. (F) Pathway enrichment analysis of validated target genes of conserved miRNAs from (E) available in miRTarBase (19). Yellow dots indicate a significant enrichment of target genes in the respective pathway which correlates with the dot size and violet dots a potential implication meaning a non-significant enrichment of target genes in this pathway. Corrected for multiple testing: *q<0.05.

hsa-miRNA-182-5p is a gate keeper for LRP6-dependent regulation of glucose homeostasis and hepatic lipid metabolism.

(A) Expression of miR-182-5p is 2.3-fold upregulated in liver tissue of obese subjects with type 2 diabetes (T2D, HbA1c ≥ 6.5 % or anti-diabetic medication) compared to non-diabetic (ND, HbA1c < 5.7 %) obese subjects in the extended human liver cohort (n=85). (B) Correlation plot of hepatic miR-182-5p expression and its target genes in human liver (green box) and of gene expression with metabolic parameters from blood (red box) as well as with confounders (blue box). Non-tested correlations are indicated by gray squares. (C) Expression of miR- 182-5p target genes in human diabetic liver (See also Figure A.1 C). (D) Expression of target genes after overexpressing miR-182-5p for 48h in HepG2 cells in comparison to a negative control (nc, n=3). (E) Protein abundance of the novel target gene LRP6 is reduced after overexpression of miR- 182-5p for 72h in HepG2 cells (n=3). (F) Overexpression of miR-182-5p in HEK-293 cells decreases luciferase activity after 48h in a luciferase reporter assay for the LRP6 wild type (WT) sequence but not in the mutated seed (n=3). (G) Glucose uptake is significantly reduced (0.82-fold) in HepG2 cells after 48h of miR-182-5p overexpression and acute insulin stimulation for 20 min (n=3). (H) Measurement of pAkt/Akt via Western Blot indicate insulin resistance in HepG2 cells 48h after overexpression of miR-182-5p (0.13-fold).

Data are shown as scatter dot plots with mean ± SD (A, E-H) or correlation matrices (B-D). Without multiple testing correction: ***p<0.001, **p<0.01, *p<0.05 (A, E-H); corrected for multiple testing: ***q<0.001, *q<0.05, significant prior to adjustment: #p<0.05 (B,C,D); Students t-test (A,C,D, E-H) or Pearson’s correlation (B).

Hepatic gene expression of miR-182-5p target genes and correlation to miRNA expression as described by correlation plot (main Figure 2B) and heat map (main Figure 2C).

(A) Serum miR-182-5p levels are not altered between obese human subjects without (ND) or with T2D. (B) Serum miR-182-5p levels are not associated to hepatic miR-182-5p expression in obese subjects. (C) Expression of metabolic miR-182-5p target genes. Correlation analysis between miR-182-5p expression and its target genes LRP6 (D), IRS1 (E), PPARA (F) and PCK1 (G) in obese human liver.

Data are shown as mean ± SD (A), scatter plot (B, D-G) or box-whisker plots with min and max values, expression mean is indicated as cross within the box plots (C). Corrected for multiple testing: ***q<0.001, *q<0.05, significant prior to adjustment for multiple testing: #p<0.05 (B,C,D); Students t-test (A,C) or Pearson’s correlation (B, D-G).

Mimic transfection in HepG2 cells led to a significant 400-fold upregulation of miR-182-5p.

Western blot lanes used for quantification and which are corresponding to figure 2E (A) or figure 2H (B).

Effects of suppressing basal miR-182-5p in HepG2.

(A) Treatment of HepG2 cells with 10 nM miR-182-5p antagomir lead to a 0.4-fold reduction of miR-182-5p expression. (B) The reduction did not lead to a significant change in glucose uptake.

Hepatic miR-182-5p expression in obese mice can be reversed by weight-loss.

(A) Body weight in gram of mice in the weight cycling cohort (n=7 Chow, n=6 HFD, n=8 HC, n=9 YoYo). Groups of obese HFD-fed mice were switched to chow for 12 weeks (HC) to induce body weight loss and re-fed with HFD for another 12 weeks (YoYo). (B) Expression of miR-182-5p in liver of mice undergoing weight cycling. (C) Expression of target genes of miR-182-5p in murine liver compared to the Chow control group (see also Figure S2). (D) Correlation plot of hepatic miR-182-5p expression and its target genes in murine liver (green box) and of gene expression with relevant metabolic parameters (red box). Non-tested correlations are indicated by gray squares. TAG: triacylglycerol.

Data are shown as mean ± SD (A,B) or correlation matrices (C,D). Corrected for multiple testing: ***q<0.001, ***q<0.01, *q<0.05, significant prior to adjustment: #p<0.05 (C, D); One-Way ANOVA (B) or Pearson’s correlation (C,D).

Expression and phenotypic association of conserved metabolic microRNAs and target genes in murine liver during weight cycling with a comparison between predicted murine and human regulatory gene networks.

(A) Weight cycling in mice over 24 weeks induces changes in body weight that are associated with alterations in plasma markers and hepatic fat content (n=7 Chow, n=6 HFD, n=8 HC, n=9 YoYo). (B) Only mmu-miR-149-5p follows the expected expression pattern of overexpression in HFD and reversal during weight cycling as observed in human T2D, whereby the other three conserved miRNAs are reduced after HFD feeding. The miRNAs miR-149-5p and miR-330-3p are expressed from conserved genes, whereby mmu-miR-1962 and mmu-miR-7056-5p constitute orthologues to human hsa-miR-485-5p and hsa-miR-7847-3p respectively based on seed matches. Shared sequences are indicated in red. (C) Hepatic expression of Lrp6, Irs1 and Pck1 is significantly altered by weight cycling in obese mice. (D) Mmu-miR-182-5p shows the strongest positive associations to hepatic triacylglycerol (TAG), plasma leptin and insulin, body weight and cholesterol in the weight cycling cohort. mmu-miR-149-5p is further associated to serum TAG, which is also observed for human hepatic hsa-miR-149-5p expression. (E) Direct comparison between murine and human miRNA-target gene networks reveals a highly conserved and connected network with every metabolic miRNA potentially targeting several genes. The expression values of the murine network are based on qPCR measurements from the weight cycling cohort and the expression values of the human network are based on miRNA microarray and target gene qPCR measurements of n=40 individuals.

Data are shown as scatter dot plots with mean ± SD (A-C), correlation matrices (D) or interaction network (E). Corrected for multiple testing: ***q<0.001, **q<0.01, *q<0.05, significant prior to adjustment: #p<0.05 (A-D); One-Way ANOVA (A-C) or Pearson’s correlation (D).

In vivo overexpression of miR-182-5p elevates fasting insulin levels and hepatic fat content and reduces LRP6 protein levels.

(A) Mice were challenged with HFD for four weeks prior to the first injection of 1 mg miRNA mimic per kg body weight and sacrificed after seven days (n=8 per group). A second injection was performed after 3.5 days. Body composition was examined at days −1 and 7 by NMR. Glucose and insulin levels were determined from blood at days −1, 5 and 7. Glucose tolerance was evaluated at day 5. (B) Hepatic miR-182-5p expression was 584-fold upregulated at day 7. (C) Fat mass tended to be increased in the miR-182-5p injected group but (D) body weight was not different between control and miR-182-5p treated mice. (E) Glucose tolerance and (F) fasting glucose were not different between both groups. (G) Fasting insulin levels and (H) hepatic triglyceride content were increased 2.25-fold and 2.19-fold in miR-182-5p treated mice, respectively. (I) Comparable mRNA, but (J) 0.52-fold diminished LRP6 protein levels in liver samples of miR-182-5p treated mice.

Data are shown as mean ± SD (B-H). *p<0.05, **p<0.01; Student’s t-test (B-H).

Additional phenotypic characteristics of mice overexpressing of miR-182-5p in liver

(A) Detection of miR-182-5p after tail vain injection in spleen and heart is significantly increased. (B) Body weight gain and (C) area under the curve (AUC) of the glucose tolerance test (GTT) did not differ between control and miR-182-5p treated mice (n=8 per group). (D) HE liver staining of two representative control and mimic treated animals. Mice overexpressing miR-182-5p show more lipid accumulations. Scale bar represents 100 µm. (E) miR-182-5p target gene expression was not changed after one week of acute miRNA overexpression, but Lrp6, Irs1, Foxo1 and G6pc showed a strong tendency to reduced expression. Data are shown as mean ± SD (B) or scatter dot plots with mean ± SD (A,C,E).

Dynamic regulation of insulin resistance and hepatic lipogenesis by miR-182-5p

Upregulation of hsa-miR-182-5p in human obese diabetic liver simultaneously decreases the metabolic pathways of beta oxidation and stimulates lipogenesis. LRP6 (highlighted) is the main target gene and consistently altered in humans, cell culture and mouse. miR-182-5p is induced by long-term feeding of high-fat diet in mice and reversed by weight loss. Identified target genes are consequently altered inversely to miRNA-182-5p expression. To prove direct effects, liver-specific upregulation of miR-182-5p in metabolically challenged mice caused a significant reduction of LRP6 which is accompanied by increased hepatic fat accumulation and increased fasting insulin levels.