Extracellular vesicle transfer of miR-1 to adipose tissue modifies lipolytic pathways following resistance exercise

Abstract

Extracellular vesicles (EVs) have emerged as important mediators of intertissue signaling and exercise adaptations. In this human study, we provide evidence that muscle-specific microRNA-1 (miR-1) was transferred to adipose tissue via EVs following an acute bout of resistance exercise. Using a multimodel machine learning automation tool, we discovered muscle primary miR-1 transcript and CD63+ EV count in circulation as top explanatory features for changes in adipose miR-1 levels in response to resistance exercise. RNA-Seq and in-silico prediction of miR-1 target genes identified caveolin 2 (CAV2) and tripartite motif containing 6 (TRIM6) as miR-1 target genes downregulated in the adipose tissue of a subset of participants with the highest increases in miR-1 levels following resistance exercise. Overexpression of miR-1 in differentiated human adipocyte-derived stem cells downregulated these miR-1 targets and enhanced catecholamine-induced lipolysis. These data identify a potential EV-mediated mechanism by which skeletal muscle communicates with adipose tissue and modulates lipolysis via miR-1.

Keywords: Adipose tissue; Metabolism; Muscle biology; Skeletal muscle; Transport.

Figure 1. Acute resistance exercise in humans induces increased production of skeletal muscle miR-1 concurrent with increased miR-1 in EVs and adipose tissue.

(A) Schematic diagram showing timeline of skeletal muscle and adipose tissue biopsies and blood draws relative to the acute resistance exercise bout. Expression of mature miR-1 and primary transcripts of miR-1 (pri-miR-1a, pri-miR-1b) in (B) skeletal muscle (n = 32), (C) serum EVs (mature miR-1 only; n = 31), and (D) adipose tissue relative to baseline levels (BL; denoted by the dotted line; n = 32). Data are expressed with min-to-max box plots and were compared using 1-sample Wilcoxon t tests (skeletal muscle and adipose tissue) or a 1-way ANOVA with Dunnett corrections for multiple comparisons (serum EVs). Serum EV miR-1 outliers were removed using the ROUT method (Q = 5%). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 2. Machine learning identifies CD63 EVs as a major contributor to exercise-induced changes in adipose miR-1.

(A) Schematic diagram outlining the workflow of CLASSify. (B) Nonlinear predictors of the exercise-induced changes in miR-1 levels in adipose tissue identified by 1 or more of the 3 nonparametric models: K-Nearest Neighbors, Neural Network, and Logistic Regression. Participants were divided into low (< 0.5-fold change; n = 7), medium (0.5–2.0-fold change; n = 15), and high (> 2.0-fold change; n = 10) miR-1 based on changes in adipose miR-1 in response to acute exercise. The relationship among exercise-induced changes in adipose miR-1 and (C) muscle miR-1 primary transcript 1b (pri-miR-1b), (D) BMI, (E) CD63 count, and (F) age. Data are expressed with min-to-max box plots. BMI, body mass index; BL, baseline; T-0, 0-minute time point; T-30, 30-minute time point; T-90, 90-minute time point.

Figure 3. RNA-Seq analyses reveal elevated EV uptake pathways in participants with high adipose miR-1 in response to exercise and identifies potential miR-1 target genes.

Expression of (A) miR-1, (B) β-adrenergic receptor 1 (ADR-β1), ADR-β2, and ADR-β3 in adipose tissue in participants with high (n = 6) versus low (n = 6) changes in miR-1. (C) Comparison of gene expression between high and low miR-1 responders. Gene enrichment analyses for upregulated and downregulated genes (D and E) with differences in expression levels between high and low miR-1 responders for selected genes (F–I). Data are expressed with box plots (Data are expressed with box plots displaying the 90th and 10th percentiles at the whiskers). FMNL, formin-like 1; GNLY, granulysin; PEX19, peroxisomal biogenesis factor 19; ACOT2, acyl-CoA thioesterase 2; IDH1, isocitrate dehydrogenase 1; DGAT2, diacylglycerol O-acyltransferase 2; GPAM, glycerol-3-phosphate acyltransferase; MOGAT2, monoacylglycerol O-acyltransferase 2.

Figure 4. miR-1 targets CAV2 and TRIM6 mRNAs to enhance catecholamine-induced lipolysis.

Differences in (A) caveolin 2 (CAV2) and (B) tripartite motif-containing protein 6 (TRIM6) expression between participants with high (n = 10) versus low (n = 7) changes in adipose miR-1 in response to exercise, as well as seed sequence schematic depicting the miR-1 binding affinity for these mRNAs. (C) Efficiency of miR-1 transfection as depicted by fold change in miR-1 relative to scrambled control (Scr). (D) Changes in CAV2 and TRIM6 expression with miR-1 transfection. The abundance of (E) Nonesterified fatty acids (NEFAs) and (F) glycerol in the media of scrambled control (Scr) and miR-1–transfected human adipose-derived stem cells (ADSCs) treated with vehicle (VEH) or epinephrine (EPI). (G) Protein abundance of (H) comparative gene identification-58 (CGI-58), (I) adipose triglyceride lipase (ATGL), (J) phosphorylated hormone-sensitive lipase (p-HSL), and (K) tuberous sclerosis protein 1 (TSC1) in Scr and miR-1–transfected ADSCs treated with VEH or EPI normalized to β-actin. Data are expressed as mean ± SD. Unpaired, 1-tailed Mann-Whitney tests were used to compare differences in CAV2 and TRIM6 expression in panel A and B. 1-tailed Welch’s unpaired t tests with Holm-Šidák’s corrections for multiple comparisons as needed were used to compare gene expression in panels C and D. 2-way ANOVAs with Tukey’s corrections for multiple comparisons were used to compare glycerol, NEFA, and protein levels in panels E, F, and H–K. *P < 0.05; **P < 0.01; ***P < 0.001.