Circulating Exosomal miR-20b-5p Is Elevated in Type 2 Diabetes and Could Impair Insulin Action in Human Skeletal Muscle

miRNAs are a class of small noncoding RNAs that function as translational repressors by direct interaction with their target mRNA (1,2). miRNAs function to negatively regulate the abundance of specific proteins and in this way exert control over numerous cellular and biological processes including metabolism (3,4). While miRNAs are transcribed and exert many effects directly in the cell of origin, miRNAs are also secreted and stable miRNAs can be detected in plasma (5). Circulating miRNAs have been detected in most biofluids including blood (serum/plasma), urine, breast milk, and cerebrospinal fluids and are protected from degradation by a variety of mechanisms. A proportion of circulating miRNAs are packaged in extracellular vesicles, such as “exosomes” (50- to 200-nm membrane-coated vesicles) (6–9) that protect RNA cargo from endogenous RNase activity (10). miRNAs can also bind to various proteins including lipoproteins, namely, HDL, or argonaute (AGO) proteins, which are the key components of the RNA-induced silencing complex, to form miRNA-protein complexes for transport (10–12). Exosomes in plasma/serum have been implicated in transfer of miRNA into target cells and thus play a role in cell-cell communication (11–15).

The presence of circulating miRNAs has prompted efforts to identify biomarkers for various pathologies, including cancer, and diseases affecting cardiovascular, neurological, metabolic, and immune function (16–22). Specific circulating miRNAs may be useful biomarkers for diagnoses and management of progressive diseases such as type 2 diabetes (23–25). Despite the fact that patients with type 2 diabetes are characterized by hyperglycemia and elevated HbA_1c levels, these changes in clinical chemistry are only detected once metabolic imbalance has occurred. Defects in multiple tissues controlling glucose homeostasis and insulin sensitivity are often present years prior to diagnosis (26). Given the complex pathophysiology and disease burden of type 2 diabetes, efforts have been focused on identifying circulating miRNAs as novel prognostic biomarkers (27–29).

To date, there is little consensus on the precise nature of circulating miRNA biomarkers in different cohorts of patients with type 2 diabetes, and little is known regarding the functional role(s) of the identified miRNAs in metabolic processes implicated in type 2 diabetes pathogenesis. Here, we determined total serum and exosomal miRNA expression profiles in men with normal glucose tolerance or type 2 diabetes and validated the effects of the differentially abundant miRNAs on metabolism in skeletal muscle.

The study was approved by the ethics committee of Karolinska Institute. Informed written consent was obtained from all volunteers. Twenty men with normal glucose tolerance, 16 men with impaired glucose tolerance, and 21 men with type 2 diabetes were recruited by newspaper advertisement or from a primary health care clinic. The participants with type 2 diabetes, impaired glucose tolerance, and normal glucose tolerance were matched for age and BMI. Clinical characteristics of the study participants are presented in Supplementary Table 1. Blood samples were separated in serum and peripheral blood mononuclear cells.

Extracellular vesicles were obtained from serum with a miRCURY Exosome Isolation Kit – Serum and Plasma (Exiqon, Vedbaek, Denmark) according to the manufacturer’s instructions. Isolated extracellular vesicle samples were analyzed using nanoparticle tracking analysis (NTA). Samples were loaded into the sample chamber of an NS500 unit (NanoSight, Amesbury, U.K.), and five 1-min videos of each sample were recorded (threshold, 6 – multi; blur, auto; and minimum expected particle size, auto). Data analysis was performed with NTA 2.3 software (NanoSight), and the size and concentration of particles included in the extracellular vesicle samples were calculated. An aliquot of isolated exosome-enriched extracellular vesicles from serum was used for NTA, and the remaining isolated extracellular vesicles were used for exosome RNA (exoRNA) extraction.

A miRCURY RNA Isolation Kit (Exiqon) was used to extract exoRNA, together with an RNA Spike-in Template Kit (Exiqon), using MS2 RNA (Roche, Basel, Switzerland) as carrier RNA according to the manufacturer’s instructions. Each exoRNA elute was reverse transcribed using the miRCURY LNA Universal RT cDNA Synthesis Kit (Exiqon). For the first screening, human serum/plasma Focus miRNA PCR panels (96-well [V43.AF]) (Exiqon) were used in a quantitative (q)RT-PCR (qRT-PCR)-based approach to determine levels of 179 human miRNAs. The quantitative PCR (qPCR) was performed using a StepOne Plus (Applied Biosystems) with 40 amplification cycles, using the cycling parameters recommended by Exiqon. Raw data were processed using StepOne software version 2.3 (Applied Biosystems) to assign the baseline and threshold for threshold cycle (Ct). For determination of the technical variation between the Exiqon serum/plasma Focus miRNA PCR panel plates, the interplate calibrator (UniSp3) was analyzed. Ct values of the interplate calibrator were analyzed to be highly similar across all samples. The normalization and analysis of the PCR panel plate results were provided by Exiqon GenEx software, version 6. For validation of the results of the PCR panel plate in a second screening, we used a miRCURY LNA miRNA PCR System (Exiqon) to assess the expression level of individual miRNAs. Gene expression levels were quantified using the miRNA-specific LNA PCR primer. Relative expression was calculated using the comparative Ct method.

Satellite cells were isolated from vastus lateralis skeletal muscle as previously described (30). Cell cultures were maintained at 37°C under 7.5% CO₂ as myoblast cells. For measurement of gene expression, myoblast cells were differentiated into myotubes as previously described (31). Myotube cells where fusion and multinucleation were observed at day 10 after initiation of differentiation were used for total RNA extraction, protein detection, and cell assays. Cells were transfected using a double transfection protocol 48 h after differentiation (day 6) and 48 h later (day 8) with 10 nmol/L Ambion miRNA-20b-5p (Thermo Fisher Scientific). Control cells were transfected with a scrambled miRNA mimic (10 nmol/L). Each transfection was performed for 5 h with transfection complexes formed in reduced serum media (OptiMEM; Thermo Fisher Scientific) using Lipofectamine RNAiMAX transfection reagent according to the manufacturer’s protocol. RNA was isolated using an miRNeasy Kit (QIAGEN) at day 10.

HEK293 and HepG2 cells were obtained from ATCC and cultured in high-glucose (4.5 g/L) DMEM supplemented with 10% (vol/vol) FBS. miRNA-20b-5p or miRNA mimic was transfected into those cells using Lipofectamine RNAiMAX transfection reagent according to the manufacture’s protocol, and RNA was isolated using the miRNeasy Kit (QIAGEN).

mRNA content from miR-20b–transfected cells was profiled by hybridizing biotinylated sense strand cDNA to GeneChip Human Gene 2.0 ST arrays (Thermo Fisher Scientific). Sense strand cDNA was synthesized from total RNA and biotin labeled with the GeneChip WT PLUS Reagent Kit (Thermo Fisher Scientific) before being hybridized to arrays. Gene arrays were washed, stained, and scanned as instructed by Affymetrix (Santa Clara, CA). Preprocessing of data was performed using a robust multiarray average with sketch quantile normalization by Expression Console software (Affymetrix). Differential expression of transcripts was determined with a paired class comparison with a univariate test using a random variance model comparing gene expression of control (miRNA mimic–transfected cells) versus miR-20b-5p–transfected cells. Genes with a false discovery rate of <10% were considered to be differentially regulated. The microarray data were submitted to the National Center for Biotechnology Information Gene Expression Omnibus (GEO) and can be found under the GEO series accession number GSE102295.

Gene Set Enrichment Analysis (GSEA) was used to link genes identified in a specific gene group with their occurrence in biological pathways or processes. The rank gene list was further annotated using MSigDBv5.0 downloaded from the Broad Institute (http://www.broadinstitute.org/), which contains curated functional gene sets of various biological states.

Putative target sites were probed in silico by target prediction algorithms (TargetScan). For validation experiments, 500 ng total RNA from cells was reverse transcribed using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific), and qRT-PCR was performed to measure the expression level of six genes using SYBR Green Master Mix Kit (Thermo Fisher Scientific) and StepOnePlus (Bio-Rad) (primer list in Supplementary Table 2). For the stat3 gene only we used TaqMan Assay (Thermo Fisher Scientific). We used the TBP and M2B genes as reference genes, and relative quantification values were calculated using the equation 2^−ΔΔCt.

Western blot analysis was performed as previously described (32). Protein content of the supernatants was determined by BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL). Membranes were stained with Ponceau S to confirm equal loading of samples and quality control for the transfer procedure. Membranes were incubated with primary antibodies directed to glycogen synthase (number 3893; Cell Signaling Technology), phosphorylated (phospho)–glycogen synthase (3891; Cell Signaling Technology), AKT (9272; Cell Signaling Technology), phospho-AKT (Thr³⁰⁸) (4056; Cell Signaling Technology), signal transducer and activator of transcription (STAT)3 (9139; Cell Signaling Technology), and GAPDH (sc-25778; Santa Cruz Biotechnology, Dallas, TX). Membranes were incubated with species-appropriate horseradish peroxidase–conjugated secondary antibody before proteins were visualized by enhanced chemiluminescence (Amersham ECL Western Blotting Detection Reagent, GE Healthcare Life Sciences, Little Chalfont, U.K.). When appropriate, protein content was quantified by densitometry (Quantity One; Bio-Rad). All quantifications were performed using a positive control to control for intergel variability.

The luciferase reporter clone having the AKTIP 3′ untranslated region (UTR) (HmiT088513-MT05) was purchased from GeneCopoeia (Rockville, MD). This clone included two predicted miR-20b-5p target sites in AKTIP 3′UTR (Fig. 3H). Target search in microRNA.org (http://www.microrna.org) was used for prediction of miR-20b-5p target sites. Predicted miR-20b-5p binding sites were mutated using the QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA). Oligoprimers used for mutagenesis of the AKTIP 3′UTR were 5′-GATGGTGAATCTGGTGCACCATCCTGAAACCTGCTAGACTCTGGCCTAG-3′, 5′-CTAGGCCAGAGTCTAGCAGGTTTCAGGATGGTGCACCAGATTCACCATC-3′, 5′-GAGAGCAGGTTCCATAGCTCACCTGCGATAAGTGGAAGATCATTTGAATCTC-3′, and 5′-GAGATTCAAATGATCTTCCACTTATCGCAGGTGAGCTATGGAACCTGCTCTC-3′. Cells were seeded in 96-well plates 24 h before experimentation. 3′UTR promoter plasmids (100 ng/well) was transfected into HEK293 cells in 96-well plates using the transfection reagent Lipofectamine 2000 (Life Technologies) with 10 nmol/L miRNA mimic for miR-20b-5p. Control cells were transfected with appropriate scrambled miRNA mimic. After 24 h, the cell culture medium was collected and processed for luciferase assay using the Secrete-Pair Luminescence Assay Kit (GeneCopoeia). Assays were read in the CLARIOstar (BMG LABTECH) and normalized with secreted alkaline phosphatase signals.

Insulin-stimulated glucose incorporation into glycogen was determined as previously described (30).

Data are presented as mean ± SEM. Differences were analyzed using either paired or unpaired Student t test as appropriate. Relationships were evaluated by computation of Pearson correlation coefficients. Significance was set at P < 0.05.

We used a commercial exosome isolation kit to isolate exosome-enriched extracellular vesicles rapidly from serum to determine miRNA expression profile of exosomes. Differential ultracentrifugation, the current gold standard of isolation, was compared with the commercial isolation kit. While both methods yielded similar results, the reduced requirement of serum for isolation dictated the use of the commercial kit. The resultant fractions were analyzed by immunoblotting for known exosome markers and major protein components of HDL particles in serum (Supplementary Fig. 1). This showed that exosome marker proteins, ALIX and CD9, were more enriched in the isolated fraction compared with original serum, accompanied by a significant reduction in the HDL marker protein, apolipoprotein A1 (APOA1). Next, size distribution and concentration of extracellular vesicles in serum from men with normal glucose tolerance, or type 2 diabetes, were analyzed by NTA. Patients with type 2 diabetes had altered serum lipid levels (Supplementary Table 1), which may influence the presence of extracellular vesicles, including exosomes, which contain lipids. However, particle size (Fig. 1A) and particle concentration (Fig. 1B) of the extracellular vesicle samples were unaltered between men with normal glucose tolerance and men with type 2 diabetes.

Exosomes are known to carry noncoding RNAs (6), such as miRNA. We determined the miRNA expression profile of exoRNA derived from serum from men with normal glucose tolerance or type 2 diabetes. For the initial screening, exoRNA was extracted from four men with normal glucose tolerance and four men with type 2 diabetes, and miRNA expression was profiled using a qPCR panel. Four men from each group were randomly selected, still matching for age and BMI. In this first screen, six exosomal miRNAs (miR-20b-5p, miR-532-3p, miR-150-5p, miR-502-3p, miR-363-3p, and miR-30d-3p) were identified to be up- or downregulated among the 179 miRNAs included in the qPCR panel (P < 0.05) (Table 1). Investigating the miRNA expression profiles of total serum RNA obtained from the same individuals showed larger interindividual variation and did not reveal any significant differences of miRNAs between the men with normal glucose tolerance and men with type 2 diabetes (data not shown), suggesting exosomal rather than serum-derived miRNAs are altered in type 2 diabetes. Next, we validated the results of the qPCR panel using individual miRNA assays in exoRNA isolated from a larger cohort. This analysis confirmed that expression of both miR-20b-5p and miR-150-5p was increased in serum exosome-enriched extracellular vesicles from men with type 2 diabetes (Table 1). For determination of whether changes in miRNA content of extracellular vesicles are present in subjects with a high risk of developing diabetes, miRNA expression in extracellular vesicle RNA was determined in individuals with impaired glucose tolerance (N = 16) (Supplementary Table 1). Size distribution and concentration of extracellular vesicles analyzed by NTA in serum from men with impaired glucose tolerance were not different compared with subjects with diabetes or normal glucose tolerance (data not shown). The relative expression of miR-20b-5p was 1.30 ± 0.3 (P = 0.63) and miR-150-5p 1.15 ± 0.11 (P = 0.68) in subjects with impaired glucose tolerance compared with subjects with normal glucose tolerance. Although both miR-20b-5p and miR-150-5p content were slightly increased in subjects with impaired glucose tolerance, this was not significant.

miRNAs have been proposed as progression biomarkers in various diseases (19,33). Thus, we determined whether expression of exosome miRNAs correlated with clinical parameters in the study cohorts (Table 2). Exosome-enriched extracellular vesicle content of miR-150-5p was not correlated with clinical features of the study cohorts (data not shown), while content of miR-20b-5p correlated with 2-h glucose, as well as with the percent fat mass, in the men with normal glucose tolerance (P < 0.05) (Table 2). Interestingly, these correlations were not significant in either the cohort with impaired glucose tolerance or the cohort with type 2 diabetes (r = 0.12, P = 0.61). We noted an inverse, but nonsignificant, correlation between miR-150-5p and HOMA of insulin resistance in the men with normal glucose tolerance (r = −0.446, P = 0.091).

Of the two miRNAs that were identified as more highly expressed in serum exosome-enriched extracellular vesicles from men with type 2 diabetes, miR-20b-5p displayed the larger fold change in the qPCR panel data (Table 1), and miR-20b-5p content in serum exosome-enriched extracellular vesicles from men with normal glucose tolerance correlated with 2-h glucose values (Table 2). Thus, miR-20b-5p was transfected into three different human cell types to evaluate miR-20b-5p effects on gene expression. miR-20b-5p was overexpressed in HEK293 (a kidney-derived cell line), HepG2 (a liver-derived cell line), and primary human skeletal muscle cells. miR-20b-5p expression increased significantly in all three cell types (data not shown). STAT3 is a direct target of miR-20b-5p in MCF-7 breast cancer cells (34) (Supplementary Fig. 2A); thus, we determined the effect miR-20b-5p expression on mRNA and protein content of STAT3 in transfected cells. While STAT3 mRNA was decreased by miR-20b-5p transfection in both HepG2 and human skeletal muscle cells, STAT3 mRNA in HEK293 was not affected (Fig. 2A). Furthermore, STAT3 protein was reduced by miR-20b-5p transfection only in human skeletal muscle cells (Fig. 2B).

We next evaluated the mRNA expression profile in human skeletal muscle cells after miR-20b-5p transfection. GSEA, followed by miR-20b-5p overrepresentation analysis, identified key cellular functions for each gene category (Table 3 and Supplementary Fig. 3). Fourteen gene sets with a P value of < 0.05 and a false discovery rate of <0.05 were considered significant. Analysis revealed that five gene sets were downregulated by miR-20b-5p, all of which regulated immune response or immune function. Pathway analysis identified interferon-α and interferon-γ response, tumor necrosis factor-α, interleukin (IL)2-STAT5, and IL6–janus kinase (JAK)–STAT3 signaling pathways. Nine gene sets were upregulated in cells overexpressing miR-20b-5p (Table 3). The upregulated pathways include several metabolic pathways such as cholesterol homeostasis (P < 0.001), fatty acid metabolism (false discovery rate: q = 0.077), and mammalian target of rapamycin (mTOR [also known as mechanistic target of rapamycin]) signaling pathway (P < 0.001).

We next focused on validating the genes that were downregulated after miR-20b-5p transfection. We identified six genes (expression fold change >2.5, P values >0.005) as possible direct miR-20b-5p target candidates based on a target scan identification of putative miR-20b-5p target sites (Table 4). These six genes were cytochrome b reductase 1 (CYBRD1), TBC1 domain family member 2 (TBC1D2), AKT (also known as PKB) interacting protein (AKTIP), RNase/angiogenin inhibitor 1(RNH1), glycoprotein integral membrane 1 (GINM1), and muscle cofilin 2 (CFL2). For confirmation of the microarray data of these six genes, individual qRT-PCR analysis was performed, and reduced expression of all targets, with the exception of RNH1, was validated (Table 4).

We determined the protein abundance and insulin-stimulated phosphorylation of proteins involved in glycogen synthesis. Total protein content of glycogen synthase was reduced in skeletal muscle cells expressing miR-20b-5p, and levels were unaffected by 1 h exposure to insulin (Fig. 3A). In control myotubes, insulin exposure reduced inactive phospho–glycogen synthase as expected. In myotubes expressing miR-20b-5p, phospho–glycogen synthase was reduced under basal conditions, reflecting the reduced abundance of glycogen synthase. Furthermore, in miR-20b-5p–transfected cells, insulin did not alter total phospho–glycogen synthase content (Fig. 3B). The ratio of the inactive form of phospho-GSK3 to total GSK3 abundance was increased in myotubes overexpressing miR-20b-5p, and this ratio was increased in response to insulin, although not to the same extent as in control cells (Fig. 3C). AKT is an upstream regulator of GSK3 in the insulin signaling pathway. We found that AKTIP was downregulated by miR-20b-5p transfection (Table 4). Insulin increased AKT phosphorylation in control cells; however, this effect was attenuated by miR-20b-5p overexpression (Fig. 3D) (AKT phospho-Thr³⁰⁸). Similar results were noted for AKT phospho-Ser⁴⁷³ (data not shown), whereas total AKT protein was unaltered. GSEA identified that miR-20b-5p overexpression reduced STAT3 signaling pathway (Table 4). Total STAT3 protein content was decreased by miR-20b-5p transfection (Fig. 3F).

For further validation of whether miR-20b-5p is directly involved in the regulation of AKTIP, a protein identified to interact with AKT1 and enhance its phosphorylation of both regulatory sites, we constructed luciferase reporter assays for the AKTIP 3′UTR region containing the predicted miR-20b-5p target sites (Supplementary Fig. 2B and Fig. 3H). After overexpression of miR-20b-5p, luciferase activity for AKTIP 3′UTR was decreased (Supplementary Fig. 4), whereas mutagenesis of the predicted target sites of miR-20b-5p in the AKTIP 3′UTR abolished the effects of miR-20b-5p overexpression on luciferase activity (Fig. 3H).

We determined whether miR-20b-5p transfection had direct effects on glucose or lipid metabolism. Palmitate oxidation, as well as basal and insulin-mediated glucose uptake, was unaltered in human muscle cells transfected with miR-20b-5p (data not shown). In contrast, miR-20b-5p transfection increased basal (1.2-fold) (P < 0.05) and absolute insulin-stimulated glucose incorporation into glycogen (Fig. 4A). However, the insulin-stimulated increment above basal was reduced in cells expressing miR-20b-5p, indicating that overexpression of miR-20b-5p attenuated the insulin response with respect to glucose metabolism (Fig. 4B).

We found that RNA abundance of miR-20b-5p and miR-150-5p is increased in exosome-enriched extracellular vesicles prepared from serum of patients with type 2 diabetes, whereas total serum miRNA expression of these miRNA species is not significantly altered compared with glucose-tolerant men. To study the effects of miR-20b-5p on gene expression, signal transduction, and metabolism, we overexpressed miR-20b-5p miRNA in three different human cell types. We found that miR-20b-5p overexpression in primary human muscle cells suppressed expression of AKTIP, STAT3, and glycogen synthase and impaired insulin signaling in primary human skeletal muscle cells. Thus, peripheral expression of miR-20b-5p alters expression of genes involved in pathways related to immune function and impairs glucose metabolism.

miRNAs are present in biological fluids, including blood, urine, breast milk, and cerebrospinal fluids, and recent efforts have been focused on delineating the expression profiles of miRNAs for use as disease biomarkers (33,35). While serum and plasma are easily obtained by minimally invasive methods and offer sufficient volume for analysis, the presence of cell debris, proteins, and protein complexes makes any analysis of miRNA profiles in biofluids technically challenging. miRNAs in serum and plasma are found in several contexts, including protein complexes such as Ago2-miRNA, exosomes, microvesicles, and lipid-protein complexes such as HDL-miRNA (36). While miRNAs in protein complexes or extracellular vesicles are quite stable in blood, different miRNA carriers appear to have different functions in cells (9). In comparison of serum and plasma prepared from the same blood sample, higher miRNA concentration was noted in serum compared with the corresponding plasma, and the miRNA spectrum between serum and plasma differed (37). In the current study, we focused on miRNA isolated from serum exosome-enriched extracellular vesicles, as exosomes have been functionally identified as mediators of cell-to-cell miRNA transfer. The particle characteristics of exosomes isolated from men with normal glucose tolerance or type 2 diabetes in this study were similar. A limitation of the current study is that although subjects were matched for age and BMI, a number of the subjects with type 2 diabetes were on antidiabetes medication and as a group had a higher level of use of other medications. Thus, we are unable to exclude potential effects of medication on the miRNA profiles of exosome-enriched extracellular vesicles. While total serum miRNA species did not differ between the cohorts, possibly reflecting the more heterogeneous composition of total serum miRNA, we found two miRNA species that were increased in exosome-enriched extracellular vesicles from serum of men with type 2 diabetes. Neither miR-20b-5p nor miR-150-5p was significantly elevated in serum-derived exosome-enriched extracellular vesicles from individuals with impaired glucose tolerance, which could indicate that these miRNAs reflect a more advanced disease stage. Thus, miRNA profiling of functional units such as exosomes may increase the probability of identifying disease-relevant biomarkers.

miR-20b-5p and miR-150-5p content was increased in serum-derived exosome-enriched extracellular vesicles from men with type 2 diabetes. Circulating miR-150-5p in plasma has been proposed as a potential biomarker for acute myeloid leukemia (38) and has been implicated in the promotion of angiogenesis by microvesicle-mediated transfer of miR-150-5p (39). Notably, miR-150-5p is specifically upregulated in skeletal muscle from diabetic Goto-Kakizaki rats (40). Whether the elevation in miR-150-5p directly contributes to the insulin resistant phenotype in Goto-Kakizaki rats, or secondarily to impaired metabolic homeostasis, is not known. In the current study, we did not observe any correlations between miR-150-5p exosome content and type 2 diabetes–related metabolic traits. miR-20b-5p belongs to the miR-17 family and is part of the larger family of highly similar miRNAs, including miR-106a-363, miR-17-192, and miR-106b-25 cluster (41), that modulate VEGF expression by targeting HIF-1α and STAT3 in MCF-7 breast cancer cells (34). While these miRNAs have been associated with metabolic disorders and type 2 diabetes, a detailed understanding of the physiological functions of miR-150-5p or miR-20b-5p is warranted.

Type 2 diabetes is a multifactorial metabolic disease affecting numerous tissues, including liver, skeletal muscle, adipose tissue, pancreas, and brain. To explore possible physiological functions of miR-20b-5p, we transfected miR-20b-5p into human liver, kidney, and skeletal muscle cells and assessed STAT3 protein abundance. STAT3 protein has previously been reported to be a direct target of miR-20b-5p (34). We noted that the miR-20b-5p overexpression led to the greatest reduction in STAT3 protein abundance in skeletal muscle cells. Whether this is due to tissue-specific differences or properties of the cultures (primary muscle cells as opposed to immortalized cell lines for kidney and liver cells) remains to be further investigated. mRNA expression of genes relevant for several pathways implicated in immune function was altered in myotubes overexpressing miR-20b-5p. Given the growing appreciation that insulin resistance and type 2 diabetes are associated with chronic low-grade inflammation, and previous findings that miR-20b-5p may play a role in the modulation of some inflammatory signals (42), our results in men with type 2 diabetes and cultured myotubes are compelling. miR-20b-5p targets the STAT3 gene in MCF-7 breast cancer cells (34), and we confirm this association in our microarray data, as well as at the protein level. Collectively, these results suggest that miR-20b-5p directly targets the STAT3 gene in human skeletal muscle cells.

Lifestyle intervention programs to increase physical activity and promote weight loss in adults at risk for type 2 diabetes are associated with changes in circulating miRNAs, including reductions in miR-20b-5p (43). In mouse models, miR-20a-5p promotes adipocyte differentiation (44). Nevertheless, a direct link between changes in miRNA-20b-5p abundance and improvements in glucose homeostasis is unknown. Here, we report that overexpression of miR-20b-5p in human skeletal muscle cells impacted mRNA expression of genes involved in several metabolic pathways, including cholesterol homeostasis, fatty acid metabolism, and glycolysis. While genes involved in fatty acid oxidation or glucose uptake were unaltered in miR-20b-5p–transfected myotubes (data not shown), glycogen synthesis was affected. We found that basal glucose incorporation into glycogen was increased in myotubes expressing miR-20b-5p, and insulin action was blunted. mRNA content of several genes directly relevant to glycogen synthesis, as well as insulin signaling, was altered in miR-20b-5p–transfected myotubes, which could partly explain the alterations in glucose metabolism. Although there was no change in mRNA, myotubes expressing miR-20b-5p had reduced total glycogen synthase protein, likely reflecting posttranslational downregulation. Phosphorylation of glycogen synthase renders it inactive (45). We found that basal glycogen synthase phosphorylation was reduced in myotubes overexpressing miR-20b-5p, which is consistent with the increased basal rate of glycogen synthase noted in these cells. Upon insulin stimulation, the level of phosphorylated glycogen synthase was significantly reduced in control cells as expected (coincident with increased glycogen synthesis), while in myotubes expressing miR-20b-5p, insulin did not reduce glycogen synthase phosphorylation, and in miR-20b-5p–expressing cells there was an impaired insulin-stimulated increase in glycogen synthesis. The effects of miR-20b-5p on skeletal muscle metabolism are likely to reflect changes in protein content from miR-20b-5p targets as well as secondary consequences of these changes in protein expression, which are evident at the level of insulin signal transduction.

We identified AKTIP as a direct miR-20b-5p target. AKTIP, also known as mouse Ft1 orthologous (46), is reported to interact directly with AKT and modulate threonine kinase AKT phosphorylation and activation by PDK1 (47). AKTIP facilitates telomeric DNA replication (48), but the functions of AKTIP in other AKT signaling pathways are ambiguous. For example, RNA interference–mediated reduction of AKTIP in primary human fibroblasts leads to a profound proliferation defect arrested in late S phase (48). We provide evidence that AKTIP is a direct target of miR-20b-5p. The reduced AKTIP mRNA in miR-20b-5p–overexpressing myotubes was coincident with a reduced insulin-stimulated AKT phosphorylation. The glycogen synthase gene has a putative miRNA-20b target sequence in the 3′UTR region, and although we did not observe changes of glycogen synthase mRNA after miRNA-20b transfection, the total protein content of glycogen synthase was reduced. miRNAs exert effects both by reducing message stability and by preventing translation. A schematic overview of putative functional effects of miR-20b-5p in skeletal muscle is presented in Fig. 4C and D, which are likely to reflect modification of both primary (direct miRNA-20b-5p targets) and indirect effects.

Exosomes are released from cells into the circulation and transported to target cells to deliver cargo, including proteins and nucleic acids, such as various RNA species. Functional miRNA are delivered to target cells (11,49); however, little is known of the molecular machinery that regulates this process, including the tissue of origin of the exosomes, the specific target tissues of exosomes, and the manner in which the cargo is delivered. Thus, we are unable to ascertain the cellular mechanism by which the exosomes derived from serum of patients with type 2 diabetes are released or targeted and the specific cell or tissue involved, and this remains a limitation of the current study. Downregulation of miR-20b-5p targets, including STAT3 and AKTIP, are likely to have tissue-specific effects. Further studies are required to understand the physiological role of increased serum miR-20b-5p in individuals with type 2 diabetes.

Here, we provide evidence that miR-20b-5p is highly expressed in serum exosome-enriched extracellular vesicles isolated from patients with type 2 diabetes. Moreover, when introduced into skeletal muscle cells, miR-20b-5p alters cellular glucose metabolisms and the STAT3 and AKT signaling pathway. Since serum exosomal miR-20b-5p is correlated with low 2-h glucose values as well as downregulation of inflammatory pathways, it is possible that miR-20b-5p may not be an effector of impaired insulin signaling and glycogen synthesis but, instead, may be part of a homeostatic mechanism attempting to improve metabolic control. Taken together, our results highlight potential interactions between exosome-enriched extracellular vesicles and metabolic regulation.

Acknowledgments. The authors thank Dr. Jorge Ruas and Dr. Lars Ketscher, both at Karolinska Institutet, for helpful discussion and input. The authors also acknowledge the core facility for Bioinformatics and Expression Analysis at Novum, where the gene arrays were performed. This facility is supported by the board of research at the Karolinska Institute and the research committee at the Karolinska University Hospital.

Funding. This work was supported by grants from the Strategic Diabetes Program at Karolinska Institute, European Research Council Ideas Program (ICEBERG, ERC-2008-AdG23285), Vetenskapsrådet (Swedish Research Council) (2011-3550, 2012-1760, 2015-165), Swedish Diabetes Foundation (DIA2015-032, DIA2015-052), Stiftelsen för Strategisk Forskning (Swedish Foundation for Strategic Research) (SRL10-0027), Diabetes Wellness Sweden, Novo Nordisk Foundation (NNF14OC0009941), Swedish Research Council for Sport Science (FO2016-0005), the Swedish Heart Lung Foundation (20150423), and Stockholm Läns Landsting (Stockholm County Council).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.K. conceived and performed experiments, analyzed data, and wrote the manuscript. O.P.B.W. and S.E.-A. provided expertise and feedback. T.F. and K.C. performed clinical analysis and provided feedback. J.R.Z. and A.K. conceived and planned the study, secured funding, and wrote the manuscript. M.K., O.P.B.W., T.F., K.C., S.E.-A., J.R.Z., and A.K. read and approved the final manuscript. M.K. and A.K. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.