It has been suggested that the metabolic benefits of physical exercise could be mediated by myokines. We examined here the effect of exercise training on skeletal muscle expression of a panel of myokines in humans. Pathways regulating myokine expression were investigated in human myotubes.
Eleven obese non-diabetic male subjects were enrolled in an 8-week endurance training program. Insulin sensitivity was assessed by an oral glucose tolerance test. Subcutaneous adipose tissue and Vastus lateralis muscle biopsy samples were collected before and after training. RNAs were prepared from adipose tissue and skeletal muscle. Primary culture of myoblasts was established.
As expected, exercise training improved aerobic capacity and decreased fat mass. No significant change in interleukin 6, fibroblast growth factor 21, myostatin (MSTN) or irisin mRNA level was found in muscle after training. A twofold increase in apelin mRNA level was found in muscle but not in adipose tissue. No change in circulating myokine and adipokine plasma levels was observed in the resting state in response to training. Interestingly, apelin was significantly expressed and secreted in primary human myotubes. Apelin gene expression was upregulated by cyclic AMP and calcium, unlike the other myokines investigated. Importantly, changes in muscle apelin mRNA levels were positively related to whole-body insulin sensitivity improvement.
Collectively, our data show that exercise training upregulates muscle apelin expression in obese subjects. Apelin expression is induced by exercise signaling pathways and secreted in vitro in human primary myotubes, and may behave as a novel exercise-regulated myokine with autocrine/paracrine action.
Regular physical activity protects against numerous chronic diseases such as obesity, type 2 diabetes and cardiovascular diseases.1 Some of the beneficial effects of regular exercise include lower blood pressure, improved glucose homeostasis and lipid profile, higher resting energy expenditure and reduced fat mass. Several mechanisms underlie such benefits. It is now widely accepted that regular exercise increases nutrient metabolism in various tissues by regulating the expression and activity of key metabolic control genes, leading to enhanced insulin sensitivity and metabolic flexibility.2 The skeletal muscle exhibits remarkable metabolic adaptations to exercise, including mitochondrial biogenesis and improved substrate metabolism.3, 4 How exactly the contracting muscle mediates the metabolic and physiological adaptations of exercise is still unclear.5 It has long been hypothesized that the muscle can produce endocrine signals capable of mediating the health benefits of exercise.6 As the skeletal muscle is the largest organ of the body, the discovery of several factors secreted by the contracting muscle has led to a new field of research. These so-called myokines are secreted in response to exercise and can regulate in an autocrine and endocrine fashion the function of muscle and other organs.6 Skeletal muscle has been first considered as an endocrine organ because of its ability to produce interleukin 6 (IL6) as an exercise-released factor.7 IL6 induces lipolysis and improves insulin-stimulated glucose uptake.8 Another well-documented myokine is myostatin (MSTN). MSTN exhibits antihypertrophic effects in the skeletal muscle and MSTN null mice are characterized by an excessive muscle mass.9 Irisin originating from the proteolytic cleavage of fibronectin type III domain-containing protein 5 (FNDC5) and fibroblast growth factor 21 (FGF21) was recently identified as novel myokine.10, 11 Irisin improves glucose homeostasis in mice, and seems to behave as a thermogenic factor involved in the browning of white subcutaneous adipose tissue in mice.10 FGF21 enhances whole-body insulin sensitivity and thermogenesis in brown adipose tissue.12, 13 Although the effect of exercise training on the plasma levels of these myokines has been previously studied, inconsistent findings have been reported.14, 15, 16, 17, 18
In the present study, we investigated skeletal muscle myokines (IL6, MSTN, Irisin, FGF21) gene expression and plasma levels in obese individuals in response to an 8-week aerobic exercise training intervention. To extend the knowledge on the adaptation of muscle to training, we performed a comprehensive gene expression profiling of human skeletal muscle. This led to the identification of apelin (APLN) as a novel factor upregulated by exercise training. Apelin expression, secretion and regulation by exercise-activated signaling pathways were next investigated in vitro in human primary myotubes.
Subjects and methods
Eleven sedentary obese male volunteers who had stable weight during the previous 3 months were recruited at the Toulouse Clinical Investigation Centre (Table 1). The subjects were on their usual diet before the study and ate a weight-maintaining diet consisting of 35% fat, 16% protein and 49% carbohydrates 2 days before the experiment. None were previously enrolled in an endurance activity training. They were asked to maintain their dietary habits during the study and to refrain from vigorous physical activity 48 h before each clinical investigation day. Dietary intake was assessed by a dietician from a 3-day weighed food record, including 2 week days and 1 weekend day, the week before the first investigation day. Dietary records were assessed at baseline and during the last week of the program. Nutrient intake was calculated using PRoFIL software v6.7 (Audit Conseil en Informatique Médicale, St Doulchard, Bourges, France) with the CIQUAL French food composition database for diet composition.
The study was performed according to the latest version of the Declaration of Helsinki and the Current International Conference on Harmonization (ICH) guidelines. It was approved by the Ethics Committee of Toulouse University Hospitals and all subjects gave written informed consent. The study is registered in Clinical Trials NCT01083329 and EudraCT 2009-012124-85.
Anthropometric and clinical parameters
Anthropometric parameters, blood samples, adipose tissue and skeletal muscle biopsy samples were analyzed during a 2-day investigation, 1 week apart, before and after the training program as follows. On day 1, after an overnight fast, maximal oxygen consumption (VO2max) was measured on a braked bicycle ergometer as described in de Glisezinski et al.19 On day 2, after an overnight fast, blood samples were drawn and percutaneous biopsy samples of the Vastus lateralis muscle and of the abdominal subcutaneous adipose tissue were obtained as previously described.20, 21 Ninety minutes after the end of biopsy sampling, an oral glucose tolerance test (OGTT) with 75 g glucose load was performed at 30-min intervals (times 30, 60, 90, 120). Body composition was assessed by dual-energy X-ray absorptiometry performed with a total body scanner (DPX, Software 3.6, Lunar Radiation Corp., Madison, WI, USA). Blood glucose was assayed using the glucose oxidase technique (Biomérieux, Paris, France). Plasma nonesterified fatty acids were assayed with an enzymatic method (Wako kit, Unipath, Dardilly, France). Serum insulin was measured by using a bi-insulin IRMA kit (Bertin Pharma, Montigny le Bretonneux, France). Plasma FGF21 and apelin were quantified with the Human FGF-21 Quantikine ELISA Kit (R&D Systems Europe, Lille, France) and the human EIA apelin-12 kit (Phoenix Pharmaceuticals, Belmont, CA, USA), respectively. Retinol binding protein 4 (RBP4) was analyzed by immuno-nephelemetry on a BN ProSpec (Siemens HealthCare Diagnostics, Cergy-Pontoise, France). Other parameters were determined using standard clinical biochemistry methods. As it is a good reflection of the insulin sensitivity measured by an euglycemic insulin clamp, the Matsuda insulin sensitivity index (ISI-M) derived from oral glucose tolerance test is widely used in clinical and epidemiological research.22 ISI-M was calculated as: 10 000 per square root of [(fasting glucose × fasting insulin) × (mean glucose × mean insulin during oral glucose tolerance test)].
Exercise training program
The exercise training program was performed at the Centre de Ressources d'Expertise et de Performance Sportives (CREPS) of Toulouse. The 45–60-min exercise sessions consisted mainly of cycling and running, 5 times a week, for 8 weeks. Subjects exercised 3 times per week under supervision during the first 4 weeks and 2 times per week during the last 4 weeks. They exercised on their own during other sessions. All daily sessions consisted of at least a 20-min warm-up at 35% VO2max followed by progressively increasing exercise intensity (up to 85% VO2max) and duration (up to 1 h) throughout the training program. The subjects exercised at a target heart rate corresponding to 35–85% of their VO2max. Heart rate was monitored with a Suunto T3 Cardiometer (MSE, Strasbourg, France). Compliance with training was good and the percentage of sessions completed was greater than 85% at the end of the study. Subjects were instructed to keep their usual dietary habits during the study. Adherence to the training program was self-reported. At the end of the 8-week training program they were investigated 48–72 h after the last exercise bout.
For determination of VO2max, gas exchanges were measured as previously described.23 Breath-by-breath measurements were taken at rest and throughout the exercise to assess air flow, and O2 and CO2 concentrations in expired gases using a computerized ergospirometer (Ultima PFX, Medical Graphics, St. Paul, MN, USA). Oxygen concentration was analyzed by a zirconium cell and CO2 concentration by an infrared analyzer. Certified calibration gases were used to calibrate the analyzers every day before the beginning of the assay. The VO2max exercise trial occurred in a ventilated room to ensure a constant room temperature and hygrometry from the calibration just before the trial.
DNA microarray and reverse transcription-quantitative PCR
Total RNA from frozen biopsies was prepared as previously described.20, 24 Based on the concentration (Nanodro ND-1000 Spectrophotometer, Labtech, Jebel Jeloud, Tunisia) and quality (Experion, Bio-Rad, Marnes-la-Coquette, France) check, 9 subjects had enough high quality total RNA available for gene expression study. For reverse transcription-quantitative PCR, 500 ng of total RNA was used for first-strand cDNA synthesis using random hexamers and poly(dT) according to the Multiscribe reverse transcriptase kit (High Capacity cDNA Reverse Transcription Kit, Applied Biosystems, Foster City, CA, USA). TaqMan Assays (Applied Biosystems) were used with 18S RNA (Taqman Control Assays) for gene expression normalization. Microarray experiments were performed using Agilent 4 × 44k oligonucleotide arrays as described in Viguerie et al.25 Hybridization quality check resulted in the analysis of microarray data from eight subjects. Microarray data have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE40551.
Satellite cells were isolated from fresh V lateralis biopsy samples obtained before training and cultured as previously described.26 On day 4 of differentiation, the cells were treated as indicated with ionomycin, insulin, GW7647, GW0742 (Sigma-Aldrich, Courtaboeuf, France) or forskolin (Calbiochem Corp., Darmstadt, Germany). After 24 h, the medium was collected to measure secreted factors and cells were harvested for mRNA extraction. Samples were stored at −80 °C.
Gaussian distribution and homoscedasticity of data were tested with corrected Kolmogorov–Smirnov and Levene tests, respectively. Microarray data analyses were performed as described in Viguerie et al.25 Mann–Whitney or Wilcoxon tests were used in non-parametric simple comparisons. Kruskal–Wallis and Dunn’s post tests were performed in multiple comparisons and non-parametric data analyses. Spearman correlation analysis was used to assess the correlation between variables in non-parametric univariate analysis and P-values adjusted for multiple comparisons using the Benjamini–Hochberg procedure. To test for an independent association between insulin resistance and myokines, multiple linear regression models were computed using hierarchical regression in which changes (Δ) in fat mass were entered in a first block, Δ VO2max, Δ plasma adipokines and myokines, and Δ myokine mRNA levels were introduced in the model using a stepwise procedure in a second block and Δ APLN mRNA level was introduced in the subsequent block. Statistical analyses were performed with GraphPad Prism software (GraphPad Software, La Jolla, CA, USA) and SPSS Statistics 17.0 software (SPSS, Chicago, IL, USA). Threshold for statistical significance was P<0.05.
Anthropometric and clinical characteristics before and after endurance training
The 11 obese male volunteers were aged 35.4±1.5 years. As expected, exercise training increased whole-body aerobic capacity (VO2max) by about 7%, slightly reduced fat mass (mean fat mass loss 0.8±0.4 kg) and tended to increase fat-free mass (mean fat-free mass gain 1.2±0.6 kg) (Table 1a). Changes in body composition occurred despite no change in food intake. No significant changes in ISI-M and fasting plasma glucose were observed, whereas fasting plasma insulin tended to decrease in response to training (Table 1a).
Effect of endurance training on skeletal muscle myokine gene expression
We investigated the mRNA levels of IL6, FGF21, MSTN and FNDC5 in the human skeletal muscle of obese individuals pre- and post-exercise intervention. Endurance training did not significantly change gene expression of the four candidate myokines (Figure 1).
Apelin mRNA level is increased in the skeletal muscle from obese individuals after endurance training
Analysis of the human skeletal muscle transcriptome before and after endurance training led to the identification of APLN as the most upregulated transcript encoding a known protein. APLN encodes apelin, a peptide that was so far known as an adipokine.27 Reverse transcription-quantitative PCR confirmed a twofold increase in APLN after endurance training and displayed no change in its receptor APJ (Figure 2). A positive correlation between changes in muscle APLN mRNA levels and changes in ISI-M was found (r=0.81, P=0.008) in response to exercise training (Figure 3). A significant negative correlation with changes in fasting plasma insulin was also found (r=–0.70, P=0.036). In a stepwise regression analysis, the best predictive model of change in ISI-M included changes in plasma RBP4 (β=−0.723), skeletal muscle APLN mRNA (β=0.241) and fat mass (β=–0.203). This model explained 89% of the variability in ISI-M (P=0.008).
Regulation of myokine mRNA level and secretion in human primary myotubes
We next investigated the regulation of APLN and other candidate myokines in vitro in human primary myotubes established from V lateralis muscle. Myotubes were treated with drugs, mimicking the activation of exercise signaling pathways and enhancing calcium and cyclic adenosine monophosphate (cAMP) intracellular levels, ionomycin (a calcium ionophore) and forskolin (an adenylyl cyclase activator), respectively (Figure 4). Ionomycin treatment induced a 1.6-fold increase in APLN and 2-fold decrease in FNDC5 expression. Ionomycin also slightly increased MSTN by 30% and decreased IL6 by 30%. Forskolin treatment decreased IL6, FGF21, FNDC5 and MSTN by 90%, 50%, 50% and 35%, respectively, whereas APLN increased by 3.3-fold. We also showed that activation of peroxisome proliferator-activated receptors (PPAR) -α and -δ signaling as well as insulin treatment had no effect on APLN gene expression (Supplementary Figure 1). Of interest, FGF21 and apelin were detected in the culture medium at low levels when compared with plasma values (Figure 5 and Table 1B). Forskolin and ionomycin, respectively, decreased by 33% and increased by 2.4-fold the FGF21 concentration in the culture medium (Figure 5a). Apelin concentration in the culture medium was very low—about 50 times less than in plasma—and no significant change in apelin concentration was observed compared with control (Figure 5b).
Effect of endurance training on circulating levels of adipokines and myokines in obese individuals
In agreement with the observed fat mass loss, there was a trend for reduced plasma leptin concentrations, but no significant change in adiponectin or RBP4 (Table 1B). Exercise training did not change significantly the resting plasma levels of IL6, FGF21 and apelin (Table 1B).
Endurance training is known to improve whole-body glucose homeostasis and to reduce the risk of developing type 2 diabetes.2 In this study, we identify apelin as a novel myokine that might contribute to exercise training-mediated improvement of whole-body insulin sensitivity in obese individuals. Interestingly, skeletal muscle gene expression of other myokines with a role evoked in the regulation of insulin sensitivity remained unchanged in response to 8 weeks of exercise training in middle-aged obese men.
Accumulating data suggest that, during and following exercise, the skeletal muscle synthesizes and releases factors that may act either systemically or locally within the muscle tissue to mediate some of the metabolic and physiological adaptations of exercise.6 These secreted factors have been termed as myokines. Little is known on the regulation by acute and chronic exercise of the few myokines identified so far. In the present study, skeletal muscle whole-transcriptome profiling led us to identify APLN as a novel skeletal muscle transcript upregulated by exercise training. Apelin was so far known as an adipocyte-secreted peptide that modulates skeletal muscle glucose and lipid metabolism and increases insulin sensitivity via its receptor, APJ.27 No changes in APJ mRNA levels were noted in the muscle in response to training. In addition, mRNA levels of other known myokines such as IL6, FGF21, MSTN and FNDC5 remained unaffected by the training intervention in muscle (P-values>0.1). Consistently, exercise training did not significantly change resting plasma concentrations of IL6 and FGF21. A decrease in MSTN in skeletal muscle appears to be a hallmark of exercise training. MSTN was previously found decreased in the muscle of old women after 12 weeks of aerobic training.28 Here, despite non-significant change, a tendency to decrease appears for MSTN. FNDC5 encodes a novel myokine, irisin, with thermogenic potential in white adipose tissue.29 There has been a recent controversy on the regulation of skeletal muscle FNDC5 by training in humans. In agreement with our data, unchanged expression of muscle FNDC5 was recently reported in response to 6 weeks of endurance cycling.30 In addition, a 12-week endurance exercise training induced no change in serum IL6 despite a twofold decrease in its skeletal muscle gene expression.31 Conversely, it was recently shown that a 2-week exercise program increased serum FGF21 in young healthy women.15 However, no report was made on skeletal muscle gene or protein expression. The contribution from each tissue to plasma level is unknown. In summary, part of the discrepancies between our study and other studies may be explained by the age of the subjects, their obesity status and possibly the difference in exercise training volume and duration.
Despite non-significant change in food intake, the 8-week training program induced a slight but significant fat-mass loss, indicating that a slight energy deficit was achieved with exercise training. As apelin was primarily described as an adipokine, a decrease in circulating apelin was expected, as recently described.32 In type 2 diabetic individuals, 12–24-month aerobic training programs increased plasma apelin while fat mass was unaffected.33, 34 Here, the absence of changes in plasma apelin level may be the result of compensation due to both, the increase in muscle mass and the enhanced gene expression of skeletal muscle APLN, while no change in APLN expression was found in adipose tissue (data not shown). Also, mice overexpressing apelin had no increase in plasma apelin content, whereas enhanced metabolic function was seen in skeletal muscle.35 Alternatively, the upregulation of muscle APLN expression in the face of no apparent changes in plasma apelin levels may suggest that apelin is produced to act locally on skeletal muscle fibers in a similar fashion to interleukin 8.36 In addition, a positive correlation between change in skeletal muscle APLN and increase in ISI-M was observed. Besides a change in plasma RBP4 level, the increase in skeletal muscle APLN appeared as a weak independent contributor of insulin sensitivity during exercise training. This is in agreement with at least another study.34 RBP4 is an adipokine with a controversial role in insulin resistance,37 whereas apelin has been previously shown to promote glucose uptake in skeletal muscle.38 Collectively, our data indicate that skeletal muscle apelin might have a role in the exercise training-induced improvement of insulin sensitivity through autocrine/paracrine effects within the skeletal muscle.
To confirm that apelin is a novel myokine, we next investigated its expression, secretion and regulation in vitro in human primary myotubes, besides other known myokines (IL6, MSTN, FGF21 and FNDC5). Thus, little is known on the regulation of myokines expression in vitro. Exercise induces muscle contraction through a rise in intracellular cyclic adenosine monophosphate and calcium along with other pathways.5 It was previously shown that activation of cyclic adenosine monophosphate/protein kinase A and calcium signaling pathways in human primary myotubes induces PGC-1α and peroxisome proliferator-activated receptors -δ gene expression, and promotes mitochondrial biogenesis.39 These pathways also favor lipid oxidation and glycogen storage. In this study, activation of cyclic adenosine monophosphate signaling by forskolin consistently downregulated IL6, MSTN, FGF21 and FNDC5 expression in vitro, while it strongly induced APLN. In the same line, activation of calcium signaling by ionomycin significantly induced APLN expression while downregulating IL6 and FNDC5. As previous studies have shown that both exercise and muscle contraction induce IL6 mRNA,40 this suggests that the transcriptional regulation of myokines is complex and involves multiple synergistic pathways. Interestingly, we could show that both apelin and FGF21 are secreted in the culture medium of human primary myotubes. Of note, FGF21 secretion paralleled its gene transcription pattern in response to forskolin treatment. In contrast, apelin concentration in the culture medium was very low, about 50 times less than in plasma, and apelin secretion remained unchanged in response to both forskolin and ionomycin treatments. This suggests that apelin may be secreted by skeletal muscle cells, but because of its very short half-life (<5 min)41 the peptide may be quickly degraded, thus preventing a significant accumulation in the medium overtime. A possible interaction of apelin with circulating proteins35 and instability in human plasma42 was also reported. Additionally, skeletal muscle may not contribute significantly to circulating concentrations of plasma apelin. Furthermore, as various apelin isoforms exist in plasma,42 the possibility of a specific skeletal muscle isoform cannot be excluded. It can be hypothesized that apelin is acutely released by skeletal muscle during exercise to act locally on its own receptor as an autocrine/paracrine regulator. In humans, APLN and APJ mRNA levels are known to be regulated by insulin in adipocytes.27 We found here no direct effect of insulin, PPAR-α and/or PPAR-δ agonists on APLN expression in human myotubes (Supplementary Figure 1). These data also indicate a differential regulation of APLN expression between fat and muscle cells. The upregulation of apelin expression in skeletal muscle is in agreement with apelin transgenic and knockout mice data demonstrating a positive role of apelin in skeletal muscle vascular mass and mitochondrial biogenesis,35 and in the maintenance of insulin sensitivity.43 Considering the potential role of apelin in the regulation of lipid metabolism and insulin sensitivity,27 future studies should investigate the functional and metabolic role of apelin in the human skeletal muscle.
In summary, these data highlight apelin as a novel exercise-regulated myokine in humans. Apelin is expressed, secreted and responsive to exercise-activated signaling pathways in cultured human primary myotubes. Skeletal muscle APLN expression is upregulated by 8 weeks of endurance exercise training in obese male subjects and might contribute to exercise training-mediated improvement of whole-body insulin sensitivity. Collectively, these data suggest that apelin may be locally produced by skeletal muscle fibers in response to exercise and acts locally to improve muscle metabolism and function. Future studies should investigate the influence of acute exercise on skeletal muscle apelin expression, as well as its metabolic role in the human skeletal muscle.
Gene Expression Omnibus
We are very grateful to the staff of Toulouse Clinical Investigation Centre and to the study participants. This study was supported by grants from the National Research Agency ANR-12-JSV1-0010-01 (CM) and ANR-09-GENO-0018-01 (DL), European Federation for the Study of Diabetes/Novo Nordisk and Société Francophone du Diabète (CM), Inserm DHOS Recherche Translationnelle and AOL Hôpitaux de Toulouse (DL), Fondation pour la Recherche Médicale (DL), Inserm DHOS Recherche Translationnelle 2009 (CT, DL), AOL 08 163 02 Hôpitaux de Toulouse (CT, DL) and Glaxo Smith Kline (DL).
About this article
Supplementary Information accompanies this paper on International Journal of Obesity website (http://www.nature.com/ijo)
Nature Reviews Cardiology (2018)