Research Article

Interleukin-13 drives metabolic conditioning of muscle to endurance exercise

See allHide authors and affiliations

Science  01 May 2020:
Vol. 368, Issue 6490, eaat3987
DOI: 10.1126/science.aat3987

IL-13 hits the gym

Interleukin-13 (IL-13) is a cytokine secreted by T cells, innate lymphoid cells (ILC2s), and granulocytes. It acts as a central mediator in allergy and antihelminth defense with various effects. Knudsen et al. report a distinct role for IL-13 in exercise and metabolism (see the Perspective by Correia and Ruas). Mice subjected to endurance training showed increases in circulating IL-13, which correlated with ILC2 expansion in the muscles. By contrast, exercise-induced increases in muscle fatty acid utilization and mitochondrial biogenesis were erased when mice lacked IL-13. Activation of signaling pathways downstream of the muscle IL-13 receptor was key to this effect. Intramuscular injection of adenoviral IL-13 could recapitulate exercise-induced metabolic reprogramming. This signaling pathway may have evolved to combat the metabolic stresses of parasite infection.

Science, this issue p. eaat3987; see also p. 470

Structured Abstract

INTRODUCTION

Exercise provides a vast array of health benefits. The increased metabolic activity of contracting skeletal muscle elicits an integrated response involving multiple tissues and signaling pathways to cope with increased energy and oxygen demands. A coordinated effort to promote endurance is mediated by a switch from glycolytic to oxidative metabolism favoring fatty acids as the energy source. This metabolic fueling strategy is met with specialized muscle fibers that exhibit distinct energy substrate preferences and mitochondrial oxidative capacity. These adaptive changes—such as increased cardiorespiratory capacity, enhanced muscle oxidative metabolism, and improved whole-body glucose homeostasis—promote metabolic fitness. However, the mechanisms that mediate these adaptive responses remain unclear.

RATIONALE

In the early 1960s, labile blood and lymph factors were found to mediate some of the metabolic effects of exercise. Recent studies further support the notion that communication between resident immune cells and their host tissues is important for regulating the metabolic setpoint and thereby maintaining tissue function. We found that endurance exercise increased circulating levels of the cytokine interleukin-13 (IL-13) in mice and humans. Endurance exercise also led to the expansion of type 2 innate lymphoid cells (ILC2s), one of the primary IL-13–producing cell types within mouse muscle. This implicated a role for IL-13 in the control of the adaptive responses elicited by exercise. We used several molecular and bioenergetic assays and generated three genetic models to determine the role of IL-13 signaling in the metabolic reprogramming of skeletal muscle in response to endurance exercise training.

RESULTS

Relative to wild-type control animals, Il13-deficient mice showed reduced running capacity on a treadmill. RNA sequencing of skeletal muscle from control and Il13-deficient mice was performed to examine the role of IL-13 in exercise physiology. IL-13 did not have an appreciable effect on metabolic gene expression in resting muscles. However, endurance training increased a network of mitochondrial and fatty acid oxidation genes in muscle of control animals, which was lost in mice lacking Il13. Il13-deficient muscle showed defective fatty acid utilization after a single bout of exercise and failed to increase mitochondrial biogenesis after endurance training. Furthermore, endurance training in control animals led to increased numbers of muscle oxidative fibers and improvements in mitochondrial respiration, endurance capacity, and glucose tolerance. All of these metabolic benefits of exercise training required intact IL-13 signaling.

We found that IL-13 acts directly on skeletal muscle through its receptor IL-13Rα1, leading to the activation of Stat3. Stat3 phosphorylation was elevated in muscle after both a single session and endurance training—an effect lost in Il13-deficient mice. In C2C12 myotubes, IL-13 treatment increased mitochondrial respiration that was dependent on Il13ra1 and Stat3. The IL-13–Stat3 axis controlled the metabolic program elicited by exercise training partly through a coordinated transcriptional regulation with two nuclear receptors and mitochondrial regulators, ERRα and ERRγ. Mice specifically lacking Il13ra1 or Stat3 in skeletal muscle displayed reductions in muscle fatty acid oxidation and endurance capacity. By contrast, increasing levels of IL-13 in skeletal muscle recapitulated the metabolic reprogramming induced by endurance exercise in a Stat3-dependent manner, leading to improvements in systemic glucose homeostasis and running capacity.

CONCLUSION

IL-13 signaling appears to be activated immediately after exercise and stabilized by endurance training, with the effects of modulating substrate utilization and mediating mitochondrial biogenesis, respectively. As such, it fits the criteria of a humoral factor that regulates exercise-induced metabolic effects. IL-13 exerts direct effects on skeletal muscle to increase transcriptional programs encoding fatty acid oxidation and mitochondrial electron transport chain complexes through IL-13Rα1 and the downstream effector Stat3. This adaptive response, an interplay of the immune and metabolic pathways, primes muscle for sustained physical activity. These observations highlight the importance of immune signaling in the maintenance of tissue metabolic fitness.

IL-13 mediates muscle metabolic programming to support endurance exercise via IL-13Rα1 and Stat3.

Image shows gastrocnemius muscle cross section stained for myosin heavy chain (MyHC) isoforms to determine muscle fiber type composition. Blue, green, and red indicate MyHC I-, IIa-, and IIb-positive muscle fibers, respectively. Type I and type IIa muscles contain more oxidative fibers and are characterized by high endurance, whereas type IIb muscle fibers are glycolytic and prone to fatigue.

Abstract

Repeated bouts of exercise condition muscle mitochondria to meet increased energy demand—an adaptive response associated with improved metabolic fitness. We found that the type 2 cytokine interleukin-13 (IL-13) is induced in exercising muscle, where it orchestrates metabolic reprogramming that preserves glycogen in favor of fatty acid oxidation and mitochondrial respiration. Exercise training–mediated mitochondrial biogenesis, running endurance, and beneficial glycemic effects were lost in Il13–/– mice. By contrast, enhanced muscle IL-13 signaling was sufficient to increase running distance, glucose tolerance, and mitochondrial activity similar to the effects of exercise training. In muscle, IL-13 acts through both its receptor IL-13Rα1 and the transcription factor Stat3. The genetic ablation of either of these downstream effectors reduced running capacity in mice. Thus, coordinated immunological and physiological responses mediate exercise-elicited metabolic adaptations that maximize muscle fuel economy.

Exercise reduces the risk of multiple diseases, notably metabolic syndrome. Many beneficial effects of regular exercise occur independently of body weight changes (1). Early studies implicated unidentified humoral factors that mediate insulin-independent, exercise-induced muscle glucose uptake (2). Subsequently, several “myokines” or muscle-produced factors have been reported (3). Interleukin-6 (IL-6) is one of the first myokines described to be acutely induced after exercise. In vitro studies suggested that IL-6 enhances glucose utilization in differentiated myotubes (3). However, resting IL-6 levels are actually reduced by endurance training and increased in obesity (3). Furthermore, Il6 gene deletion in mice does not affect glucose uptake, substrate metabolism, or running capacity, which suggests that additional factors are involved in metabolic adaptations to prolonged physical activity (4, 5).

IL-13 is an exercise-induced regulator of endurance capacity

To better understand the immunological response and identify circulating factors elicited by endurance exercise, we profiled a panel of human type 1 and 2 cytokines in plasma from obese, normal-weight sedentary, and endurance-trained women (performing >1 hour of aerobic exercise at least four times per week). Consistent with prior reports, obese female participants had higher plasma IL-6 levels (Fig. 1A and table S1). Endurance-trained women had significantly higher IL-13 levels but lower IL-6 levels (Fig. 1A and table S1). In a separate cohort of normal-weight sedentary men, college cross-country runners, and American-football players (samples collected postseason), male athletes had significantly higher plasma IL-13 levels but showed no differences in IL-6 levels relative to controls (Fig. 1B). We observed similar (although not statistically significant) trends for higher IL-13 and lower IL-6 circulating concentrations when female or male mice subjected to 4 to 5 weeks of endurance training (40 min of low-intensity treadmill running daily, 5 days per week) were compared to untrained controls (fig. S1, A and B). Exercise is reported to increase myokine production. However, Il13 mRNA was not detectable in mouse primary myotubes but was enriched in Percoll-isolated immune and stromal cells from muscle lysate (Fig. 1C). Exercise training further up-regulated Il13 expression in these cells (Fig. 1D). The myokine Il6 was expressed in both myotubes and muscle immune/stromal cells. Its expression was not affected by exercise training.

Fig. 1 IL-13 is an exercise-inducible factor regulating endurance capacity.

(A) Resting plasma concentrations of IL-13 and IL-6 in sedentary (n = 36 obese; n = 23 normal weight) and exercise-trained (n = 25) women. Linear regressions were used to examine associations adjusting for age and obesity status. Detailed data are in table S1. (B) Resting plasma concentrations of IL-13 and IL-6 in normal-weight sedentary controls (n = 48) and exercise-trained men (n = 25 cross-country; n = 24 football). Linear regressions were used to examine associations adjusting for age and training type. Detailed data are in table S2. (C) mRNA expression of Il13 and Il6 in primary myotubes (differentiated from primary myoblasts or satellite cells) or muscle immune/stromal cells (Percoll gradient–isolated) measured by qPCR. n = 4 to 6 per group, experiment performed twice. (D) mRNA expression of Il13 and Il6 in muscle immune/stromal cells from untrained or 4-week endurance–trained mice measured by qPCR. n = 6 per group, experiment performed twice. (E) Quantification of percentage of ILC2 (Gata3+), ILC3 (Rorγt+), and T cells (CD3+) among CD45+ cells in skeletal muscle of untrained or 4-week endurance–trained mice. n = 6 per group, 10-week-old male mice, experiment performed twice. (F) Quantification of percentage of IL-13+ ILC2 (Gata3+) and T cells (CD3+) among CD45+ cells in skeletal muscle of untrained or 4-week endurance–trained mice. n = 6 per group, 10-week-old male mice, experiment performed twice. (G) Endurance capacity test performed by treadmill running in wild-type and Il13–/– mice. n = 9 per group, 16-week-old male mice, experiment performed four times. Additional mouse cohort data are in table S3. Error bars indicate SEM. *P < 0.05, **P < 0.01, ***P < 0.001 [unpaired Student t test in (C) to (G)].

A primary source of IL-13 is tissue-resident type 2 innate lymphoid cells (ILC2s) (6). After 4 weeks of endurance training, the percentage of ILC2s in the CD45+ cell population was significantly increased, whereas the percentage of ILC3s or T cells did not change (Fig. 1E and fig. S1C). In addition, among CD45+ cells, ILC2s constituted a greater percentage of IL-13–producing cells than did T cells (Fig. 1F). A well-characterized activator of ILC2s is IL-33, which is produced by multipotent stromal cells (MSCs) in skeletal muscle and adipose tissue (710). In mice, serum IL-33 concentrations increased after one bout of exercise and remained elevated for at least 2 hours (fig. S1D). Treatment with IL-33 or phorbol myristate acetate plus ionomycin up-regulated Il13 expression in muscle immune/stromal cells (fig. S1E). Thus, ILC2s in skeletal muscle may serve as a source of IL-13–producing cells activated by endurance exercise.

To examine the relevance of IL-13 in exercise physiology, we performed treadmill-running tests comparing wild-type and Il13 whole-body–knockout (Il13–/–) mice. Il13–/– mice displayed a significant reduction in running time and distance (Fig. 1G). However, there were no differences in their muscle strength relative to wild-type controls (fig. S1F). Similar results were observed for male and female mice (the data presented are primarily from male mice unless stated otherwise). Thus, IL-13 is a candidate humoral factor induced by endurance training that modulates running capacity.

IL-13 regulates energy substrate utilization during endurance exercise

To determine the molecular basis of IL-13 signaling in regulating exercise physiology, we performed mRNA expression profiling by RNA sequencing (RNA-seq) using gastrocnemius muscle samples from wild-type control and Il13–/– mice with or without exercise training, as described earlier. Exercise training significantly up-regulated the expression of 155 genes and down-regulated the expression of 81 genes in skeletal muscle of wild-type mice [false discovery rate (FDR) < 0.05]. However, only two of these genes were similarly regulated by exercise training in muscle of Il13–/– mice (fig. S2A). A similar discordance was observed among genes nominally (P < 0.05) regulated by exercise when comparing wild-type with Il13–/– mice (fig. S2A). We therefore included differentially expressed genes with P < 0.05 in Gene Ontology analysis to ensure sufficient power for identifying enriched biological functions. Gene Ontology enrichment analyses identified fatty acid/lipid metabolism (FDR-enrichment < 0.05) and tricarboxylic acid (TCA) cycle (FDR-enrichment = 0.056) among the most up-regulated biological processes, comparing exercise-trained wild-type mice with untrained wild-type controls (fig. S2B and data S1). Differential expression of selected key genes in these biological processes could be validated by real-time quantitative polymerase chain reaction (qPCR; fig. S2C). Among the genes up-regulated by exercise training in muscle of wild-type mice were those involved in fatty acid uptake [e.g., lipoprotein lipase (Lpl) and acyl–coenzyme A (CoA) synthetase long-chain family member 1 (Acsl1)], fatty acid β-oxidation [e.g., carnitine palmitoyltransferase 1b (Cpt1b) and acyl-CoA dehydrogenase very long chain (Acadvl)], and TCA cycle enzymes [e.g., pyruvate dehydrogenase α1 (Pdha1) and isocitrate dehydrogenase 2 (Idh2)] (Fig. 2A). In addition, glycogen branching enzyme (Gbe1; required for glycogen synthesis) and lactate dehydrogenase b (Ldhb; an enzyme converting lactate to pyruvate for entrance into the TCA cycle) were also increased in exercised muscle. By contrast, glycogenolysis [e.g., phosphorylase kinase subunit γ1 (Phkg1)] was among the major biological processes down-regulated by endurance exercise in wild-type muscle (Fig. 2A and fig. S2, B and C). This agrees with the notion that endurance running preserves muscle glycogen stores and promotes a metabolic switch from glycolysis to fatty acid oxidation (11). However, this metabolic reprogramming was lost in Il13–/– mice.

Fig. 2 IL-13 regulates metabolic substrate utilization in exercising muscle.

(A) Illustration of key metabolic genes in muscle metabolism regulated by endurance training in gastrocnemius of wild-type and Il13–/– mice identified by RNA-seq. Data are presented as a heat map (log2-fold change, trained versus untrained of the same genotype) with wild-type animals at left and Il13–/– at right. Red indicates higher expression in exercised muscle; blue indicates lower expression. n = 4 per group, 20-week-old male mice. (B and C) Fatty acid uptake and oxidation (B) and glucose uptake (C) in C2C12 myotubes treated with rIL-13 (10 ng/ml) overnight. n = 5 or 6 biological replicates per group per experiment, experiment performed four times, statistical analysis performed using unpaired Student t test. (D) Triglyceride (TG) and glycogen levels in the quadriceps of nonexercised (control) and single session–exercised, wild-type and Il13–/– mice. n = 4 or 5 per group, 24-week-old male mice, experiment performed twice, statistical analysis (nonexercise versus single session of the same genotype) performed using unpaired Student t test. (E) Serum concentrations of free fatty acids (FFA), TG, and glycerol in control and single session–exercised wild-type and Il13–/– mice. n = 4 or 5 per group, 24-week-old male mice, experiment performed twice, statistical analysis (nonexercise versus single session of the same genotype) performed using unpaired Student t test. (F) Left: Respiration of exercising wild-type and Il13–/– mice at increasing running speeds. Right: Maximum oxygen uptake (VO2 max) under the experimental setting was assessed. n = 4 per group, 24-week-old female mice, experiment performed twice, statistical analysis performed using two-way ANOVA with Bonferroni post hoc test with animal group and time points modeled as variables (respiration) and unpaired Student t test (VO2 max). Error bars indicate SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Consistent with the RNA-seq data, recombinant IL-13 (rIL-13) increased fatty acid uptake, β-oxidation, and glucose uptake in differentiated C2C12 myotubes (Fig. 2, B and C); similar results were observed in primary myotubes (fig. S2D). Single bouts of acute exercise without exhaustion (30 min, 12 m/min) were used to examine the role of IL-13 in muscle substrate utilization in vivo. Wild-type mice had significantly lower intramuscular triglyceride (TG, measured in quadriceps) and maintained glycogen levels after exercise (Fig. 2D). Mice lacking Il13 had reduced muscle glycogen content, whereas TG content was unchanged. Muscle TG and glycogen levels did not differ between wild-type and Il13–/– mice at resting state. Serum free fatty acid (FFA), TG, and glycerol levels were similarly elevated by exercise in both genotypes (Fig. 2E). This suggests that the exercise-elicited defect in Il13–/– mice was muscle-intrinsic and not at the substrate level. Using an enclosed metabolic treadmill chamber in a running test, we determined that Il13–/– mice had lower maximal oxygen uptake (VO2 max) (Fig. 2F), consistent with their reduced running capacity (Fig. 1G). Muscle structure and mass were similar between wild-type and Il13–/– mice (fig. S2, E and F). There was no significant difference in basal oxygen consumption rate between untrained and trained wild-type and Il13–/– mice (5 months old; fig. S2G). Thus, IL-13 plays a role in regulating the metabolic substrates that fuel exercising muscle.

Exercise training–induced muscle mitochondrial biogenesis and glucose tolerance require IL-13

Endurance training increases mitochondrial oxidative metabolism by increasing both mitochondrial biogenesis and respiratory capacity. For genes differentially expressed in muscle of wild-type and Il13–/– mice, Gene Ontology analysis further identified oxidation reduction as a key biological process dysregulated in Il13–/– muscle (FDR-enrichment < 0.05), in addition to fatty acid metabolism, especially under trained conditions (fig. S3A and data S2). Most genes in the oxidation reduction category were also up-regulated by endurance exercise in wild-type mice (Fig. 3A and fig. S3B). These included genes involved in the electron transfer flavoprotein, mitochondrial ribosome, mitochondrial protein import, and components of the electron transport chain (ETC) complexes. The expression of these genes was either unchanged or instead down-regulated by exercise training in Il13–/– mice.

Fig. 3 Mitochondrial biogenesis in endurance-trained muscle requires IL-13 signaling.

(A) Representative mitochondrial genes differentially regulated in the gastrocnemius of endurance-trained wild-type and Il13–/– mice identified by RNA-seq. Data are presented as a heat map (log2-fold change, trained versus untrained of the same genotype) with wild-type animals at left and Il13–/– at right. Red indicates higher expression in exercised muscle; blue indicates lower expression. n = 4 per group, 20-week-old male mice. (B) Left: Mitochondrial respiration of C2C12 myotubes treated with rIL-13 (10 ng/ml) overnight. Oligomycin (1) was added to block ATP-coupled respiration, FCCP (2) to induce maximal respiration, and antimycin A/rotenone (3) to block mitochondrial electron transport. n = 5 biological replicates per group per experiment, experiments performed six times, statistical analysis performed using two-way ANOVA with Bonferroni post hoc test with treatment and time points modeled as variables. Right: mtDNA content of C2C12 myotubes treated with rIL-13 (10 ng/ml) overnight. Relative mtDNA content was measured by qPCR normalized to nuclear DNA (nDNA). n = 3 biological replicates per group per experiment, experiments performed six times, statistical analysis performed using unpaired Student t test. (C) Immunoblot analyses of gastrocnemius mitochondrial protein content from untrained and endurance-trained wild-type and Il13–/– mice. n = 4 per group, experiment performed twice, 20-week-old male mice. (D) mRNA expression of oxidative and glycolytic muscle fiber type markers measured by qPCR in gastrocnemius muscle from untrained and endurance-trained wild-type and Il13–/– mice. n = 6 or 7 per group, 20-week-old male mice, experiment performed three times, statistical analysis (untrained versus trained of the same genotype) performed using unpaired Student t test. (E) Quantification of succinate dehydrogenase–positive (SDH+) muscle fibers in cross sections of gastrocnemius from untrained and endurance-trained wild-type and Il13–/– mice. n = 8 per group, 20-week-old male mice, experiment performed twice, statistical analysis (untrained versus trained of the same genotype) performed using unpaired Student t test. Error bars indicate SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

In C2C12 myotubes, rIL-13 treatment led to higher basal and maximal mitochondrial oxygen consumption rates. This was accompanied by an increase in mitochondrial biogenesis, as determined by the ratio of mitochondrial DNA (mtDNA) to nuclear DNA (Fig. 3B). Similar results were observed in primary myotubes (fig. S3C). By contrast, rIL-6 failed to promote respiration (fig. S3D). Consistent with the RNA-seq results, endurance training increased mitochondrial ETC complex proteins in the gastrocnemius muscle of wild-type but not Il13–/– mice (Fig. 3C). Electron microscopy further showed that endurance exercise increased the average muscle mitochondrial area in wild-type mice (fig. S3E). Muscle mitochondrial area was not altered by exercise in Il13–/– mice. Endurance training also promoted a shift in muscle fiber type composition toward increased numbers of oxidative fibers (types I and IIa; succinate dehydrogenase activity–positive) and reduced numbers of glycolytic fibers (types IIb and IIx) in wild-type but not Il13–/– mice (Fig. 3, D and E, and fig. S3, F and G). These changes were associated with IL-13–dependent increases in muscle mitochondrial respiration when the complex II and complex IV substrates—succinate and ascorbate/tetramethyl-p-phenylenediamine, respectively—were used (Fig. 4A). Similar results were obtained for complex IV activity (Fig. 4B). In addition, endurance training increased the running capacity and glucose tolerance of wild-type male (Fig. 4, C and D) and female mice (fig. S4, A and B) relative to untrained animals. This training-induced increase in metabolic fitness was lost in Il13–/– mice. Thus, although IL-13 signaling is dispensable for basal mitochondrial respiration, it is a prerequisite for increasing muscle mitochondrial biogenesis and oxidative capacity and for enhancing systemic glucose homeostasis in response to exercise training.

Fig. 4 Endurance training increases running capacity and glucose tolerance through IL-13 signaling.

(A) Electron flow assay using Seahorse bioanalyzer with mitochondria isolated from gastrocnemius of untrained and endurance-trained wild-type and Il13–/– mice. Complex I (C-I) respiration was measured using pyruvate and malate as substrates and blocked with rotenone. Complex II (C-II) was measured using succinate as substrate and blocked with antimycin A. Complex IV (C-IV) respiration was measured by injecting tetramethyl-p-phenylenediamine/ascorbate. n = 15 to 25 replicates per group, data combined from three experiments, statistical analysis (untrained versus trained of the same genotype) performed using unpaired Student t test. OCR, oxygen consumption rate. (B) Complex IV activity assay performed with gastrocnemius mitochondria from untrained and endurance-trained wild-type and Il13–/– mice. The activity was assessed by the rate of cytochrome c oxidation measured by the decline in reduced cytochrome c (absorbance at 550 nm). n = 3 to 5 per group, 20-week-old female mice, experiment performed twice, statistical analysis performed using two-way ANOVA with Bonferroni post hoc test with animal group and time points modeled as variables. (C) Endurance capacity test performed by treadmill running in untrained and endurance-trained wild-type and Il13–/– mice. n = 6 or 7 per group, 16-week-old male mice, experiment performed twice, statistical analysis (untrained versus trained of the same genotype) performed using unpaired Student t test. (D) Left: Glucose tolerance test (GTT) of untrained and endurance-trained wild-type and Il13–/– mice. Center: Area under the curve (AUC) of the GTT. Right: Fasting serum insulin levels of untrained and endurance-trained wild-type and Il13–/– mice. n = 5 per group for GTT/AUC and n = 4 per group for serum insulin, 20-week-old male mice, experiment performed three times, statistical analysis (untrained versus trained of the same genotype) performed using two-way ANOVA with Bonferroni post hoc test with animal group and time points modeled as variables (GTT) and unpaired Student t test (AUC and serum insulin). Error bars indicate SEM. *P < 0.05, **P < 0.01; ns, not significant.

The IL-13Rα1–Stat3 axis mediates IL-13 activity in muscle

IL-13 signals by means of a heterodimeric receptor comprising IL-13Rα1 and IL-4R. Knockdown of Il13ra1 in C2C12 myotubes blocked rIL-13–mediated increases in mitochondrial respiration (fig. S5A). Similar to Il13–/– mice, whole-body Il13ra1–knockout (Il13ra1–/–) mice had reduced endurance running capacity (Fig. 5A). Subsequently, skeletal muscle–specific Il13ra1-knockout (skmIl13ra1–/–) mice were generated by crossing Il13ra1f/f allele with ACTA1-cre mice (fig. S5B). skmIl13ra1–/– mice retained normal muscle weight and morphology (fig. S5, C and D) but had a significant reduction in endurance running capacity and muscle fatty acid oxidation (Fig. 5B). Stat6 is the canonical transcription factor mediating transcription downstream of IL-13 receptors in immune cells. However, it is not expressed in muscle. De novo motif analysis of the promoters of mitochondrial genes differentially regulated in muscle of exercise trained wild-type and Il13–/– mice identified several enriched binding sites for transcription factors, such as the Err nuclear receptors and Stat3 (fig. S5E and data S3). Activation of Stat3 through Jak signaling downstream of IL-13Rα1 has been described in other cell types, including hepatocytes (12, 13). Stat3 phosphorylation was elevated in muscle after both a single session and endurance training (fig. S5F and Fig. 5C)—an effect lost upon Il13 deletion. In concert, muscle IL-13 protein levels were increased by one exercise session and, to a greater extent, by endurance training (fig. S5G; IL-13 protein increase in single-session exercise was not significant). rIL-13 treatment in C2C12 myotubes directly increased Stat3 phosphorylation, whereas Stat3 knockdown blocked rIL-13–induced mitochondrial respiration (fig. S5, H and I). Skeletal muscle–specific Stat3-knockout (skmStat3–/–) mice, generated by crossing Stat3f/f (control) to ACTA1-cre mice (fig. S5, J and K), also ran less and had reduced soleus muscle fatty acid β-oxidation (Fig. 5D), similar to Il13–/–, Il13ra1–/–, and skmIl13ra1–/– mice.

Fig. 5 IL-13Rα1 and Stat3 are downstream effectors of IL-13 in muscle.

(A) Endurance capacity test of wild-type and Il13ra1–/– mice performed by treadmill running. n = 6 per group, 20-week-old male mice, experiment performed twice, statistical analysis performed using unpaired Student t test. (B) Endurance capacity test performed by treadmill running and ex vivo fatty acid oxidation of isolated soleus muscle from Il13ra1f/f (control) and skmIl13ra1–/– mice. n = 6 to 8 per group (endurance capacity) and n = 5 to 8 per group (fatty acid oxidation), 24-week-old male mice, experiment performed twice, statistical analysis performed using unpaired Student t test. (C) Immunoblot analyses and quantification of Stat3 phosphorylation (Tyr705) in quadriceps muscle of untrained and endurance-trained wild-type and Il13–/– mice. n = 6 per group, 16-week-old female mice, experiment performed three times, statistical analysis (untrained versus trained of the same genotype) performed using unpaired Student t test. p-Stat3, phospho-Stat3; t-Stat3, total Stat3. (D) Endurance capacity test performed by treadmill running and ex vivo fatty acid oxidation of isolated soleus muscle from Stat3f/f (control) and skmStat3–/– mice. n = 12 per group for endurance capacity and n = 5 per group for fatty acid oxidation, 24-week-old male mice, experiment performed twice, statistical analysis performed using unpaired Student t test. Error bars indicate SEM. *P < 0.05, **P < 0.01.

Elevated IL-13 signaling in muscle confers an endurance training–like effect

To examine the dependence of IL-13 signaling on muscle Stat3, we used intramuscular injection of adenoviral IL-13 (adIL-13) to increase IL-13 levels in skeletal muscle; adenoviral green fluorescent protein (adGFP) was used as a control. Injection of adIL-13 into gastrocnemius muscle (fig. S6, A and B) induced Stat3 phosphorylation (as well as total Stat3 protein) in uninjected quadriceps muscle (fig. S6C), with concomitant increases in quadriceps and serum IL-13 levels (fig. S6D) and without altering muscle mass (fig. S6E). Elevated IL-13 signaling in muscle through gastrocnemius adIL-13 injection was sufficient to regulate the expression of exercise-controlled genes identified by RNA-seq (fig. S6F), increase respiration of isolated mitochondria given substrates for complexes II and IV (fig. S6G), enhance treadmill running time and distance as well as soleus muscle fatty acid oxidation (Fig. 6A), and improve glucose tolerance in wild-type mice (Fig. 6B). These effects were completely abolished in skmStat3–/– mice. Thus, IL-13 activates muscle Stat3 to regulate mitochondrial oxidative metabolism during exercise. In addition, increasing IL-13 signaling locally in the muscle drives an exercise-like effect on mitochondrial activity that is accompanied by increased running capacity and glucose tolerance.

Fig. 6 Increased IL-13 signaling in muscle drives a Stat3-dependent, exercise-like effect.

(A) Left and center: Endurance capacity test performed by treadmill running of Stat3f/f and skmStat3–/– mice injected with adGFP or adIL-13 into gastrocnemius muscle. n = 6 to 8 per group, 24-week-old male mice, experiment performed twice, statistical analysis (adGFP versus adIL-13 of the same genotype) performed using unpaired Student t test. Right: Ex vivo fatty acid oxidation of isolated soleus muscle from Stat3f/f and skmStat3–/– mice injected with adGFP or adIL-13 into gastrocnemius muscle. n = 5 or 6 per group, 24-week-old male mice, experiment performed once, statistical analysis (adGFP versus adIL-13 of the same genotype) performed using unpaired Student t test. (B) Glucose tolerance test of mice injected with adGFP or adIL-13 into gastrocnemius muscle. n = 6 to 8 per group, 24-week-old male mice, experiment performed twice, statistical analysis performed using two-way ANOVA with Bonferroni post hoc test with animal group and time points modeled as variables. (C) mRNA expression of Esrra and Esrrg in gastrocnemius muscle of untrained and endurance-trained wild-type and Il13–/– mice measured by qPCR. n = 6 or 7 per group, 20-week-old male mice, experiment performed twice, statistical analysis (untrained versus trained of the same genotype) performed using unpaired Student t test. (D) mRNA expression of Esrra and Esrrg in quadriceps muscle of Stat3f/f and skmStat3–/– mice after gastrocnemius injection of adGFP or adIL-13 measured by qPCR. n = 6 to 8 per group, 24-week-old male mice, experiment performed once, statistical analysis (adGFP versus adIL-13 of the same genotype) performed using unpaired Student t test. (E) Stat3 regulation of Esrrg gene promoter in reporter transient transfection assays, shown as relative luciferase activity of reporters driven by Esrrg promoters in AD293 cells cotransfected with increasing amounts of Stat3 expression vector. “1-kb,” “2-kb WT,” and “2-kb mutant” refer to reporter constructs containing 1-kb, 2-kb, and 2-kb with the mutated Stat3 binding site of Esrrg promoter, respectively. n = 6 biological replicates per group per experiment, experiment performed four times, statistical analysis performed using unpaired Student t test. (F) Relative luciferase activity of 2-kb Esrrg promoter in wild-type and Stat3–/– C2C12 myoblasts. Stat3–/– cells were cotransfected with a control or Stat3 expression vector ± rIL-13 overnight. n = 12 biological replicates per group per experiment, experiment performed three times, statistical analysis performed using unpaired Student t test. Error bars indicate SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

The motif enrichment analysis indicated that the Err nuclear receptor family, which is up-regulated by exercise (14) and controls oxidative metabolism in both skeletal and cardiac muscle (15, 16), may be downstream of muscle IL-13–Stat3 signaling. RNA-seq data (validated by real-time PCR) demonstrated that exercise-induced muscle expression of Esrra and Esrrg was IL-13–dependent (Fig. 6C). Furthermore, adIL-13 increased Esrra and Esrrg in the quadriceps of Stat3f/f but not skmStat3–/– mice (Fig. 6D). Putative Stat3-binding sites were identified ~2 kb upstream of the transcriptional initiation site for both Esrra and Esrrg genes (fig. S6H). The overexpression of Stat3 in AD293 cells was sufficient to increase the 2-kb reporter activity of either Esrra or Esrrg in a dose-dependent manner (Fig. 6E and fig. S6I), but not the 1-kb promoter that lacked Stat3-binding sites. Mutation of the only Stat3-binding site in the Esrrg promoter abolished Stat3 regulation (Fig. 6E). To examine the regulation in a relevant cell type, we generated Stat3–/– C2C12 myoblasts by CRISPR-Cas9 genome editing. rIL-13 increased the luciferase activity of the 2-kb promoter in control but not Stat3–/– myoblasts (Fig. 6F and fig. S6J). The introduction of Stat3 in Stat3–/– myoblasts increased basal promoter luciferase activities and restored the sensitivity to rIL-13 treatment. Thus, IL-13 appears to act as a regulator of the adaptive metabolic response of muscle to exercise training. This occurs, in part, through a coordinated transcriptional program mediated by Stat3 and Errα/Errγ.

Discussion

Metabolic conditioning of muscle to endurance exercise enhances fatty acid utilization and mitochondrial respiration, while reserving glycogen utilization, as a strategy to sustain energy supply for prolonged physical activity (17) (fig. S6K). This coordinated effort extends metabolic flexibility and efficiency, which contributes to the beneficial effects of exercise on health. Our results demonstrate the interplay between immune and metabolic pathways through IL-13 signaling in the control of energy substrate utilization in endurance exercise. IL-13 acts directly on muscle cells to increase fatty acid oxidation and mitochondrial ETC complex activity that is dependent on the downstream effector Stat3. Among the targets of the IL-13–Stat3 signaling are Esrra and Esrrg, two nuclear receptors also known to control fat catabolism and mitochondrial respiration in muscle (15, 18). A previous study examining Errα/Errγ cistromes in mitochondrial oxidative metabolism using chromatin immunoprecipitation sequencing has predicted Stat3 binding sites colocalized with those occupied by Errs (19), implicating a feedforward mechanism for robust transcriptional outputs. Errα and Errγ are “orphan” nuclear receptors without well-defined endogenous ligands (20). How these receptors are activated remains unclear. Our results suggest that IL-13 may function as an upstream signal of the Stat3-Err transcriptional network.

Several lines of evidence indicate that IL-13 fits the criteria of a humoral factor mediating the metabolic benefits of exercise training. Endurance exercise increases circulating IL-13 levels in both women and men. rIL-13 treatment increases the uptake of glucose and fatty acid in myotubes. In mice, exercise-induced mitochondrial biogenesis and glucose homeostasis require IL-13 signaling. Furthermore, elevated IL-13 signaling in muscle drives exercise-like metabolic effects. IL-13 is not a classical myokine because it is not produced by the muscle. One potential source of IL-13 is ILC2s. In concert, exercise leads to a higher percentage of ILC2s in the CD45+ cell population in mouse muscle. Although IL-13–mediated Stat3 phosphorylation could be observed after a single exercise session, a significant increase in muscle IL-13 levels is achieved after training. This indicates that IL-13 signaling is induced immediately after exercise to modulate substrate utilization and is stabilized by endurance training to mediate mitochondrial biogenesis.

Both ILC2s and IL-13 play important functions in the immune response to parasitic worms. Interestingly, prior infection with helminths in humans is associated with a higher VO2 max (21), which is indicative of mitochondrial oxidative capacity and a predictor of running fitness. Thus, the coevolution of helminths and hosts may have resulted in increased metabolic efficiency in the latter, perhaps in response to the increased metabolic burden from parasite infection. Thus, IL-13 signaling may have evolved to regulate metabolic adaptation to energy stress during various states including parasitic infection, endurance exercise, and exposure to cold temperatures (22).

Materials and methods

Animal studies

All animal studies were approved by the Harvard Medical Area Standing Committee on Animal Research. Mouse strains used in all experiments were in the C57Bl/6J background. Mouse genetic models were validated by both DNA genotyping and mRNA expression. Littermate controls were randomized to treatment groups based on body weight and genotype. Animals were housed at 22°C in a barrier facility and kept on a 12-hour light, 12-hour dark cycle with free access to food and water. Experiments were performed in both sexes when possible. Detailed descriptions of all animal cohorts used for this study are included in table S2.

Il13–/– mice in the Balb/c background were obtained from A. McKenzie. Generation of Il13–/– animals in the C57Bl/6J background was described in (13). These animals have been backcrossed >9 generations to C57Bl/6J mice from The Jackson Laboratory. Stat3f/f (23) animals were crossed to ACTA1-Cre mice (JAX#006149) from The Jackson Laboratory to generate skeletal muscle specific deletion of Stat3.

Il13ra1–/– and Il13ra1-floxed animals were generated from targeted Il13ra1tm1a(EUCOMM)Hmgu ES cells (KO first allele) derived from the European Conditional Mouse Mutagenesis Program (EUCOMM). These ES cells were of C57Bl/6N background. ES cells were injected at the Boston Nutrition and Obesity Research Center Transgenic Core at the Beth Israel Transgenic Mouse Facility. Chimeras were bred to C57Bl/6J mice from The Jackson Laboratory. Il13ra1–/– mice were further backcrossed to C57Bl/6J animals for >8 generations and SNP genotyping was used to selectively breed animals displaying the highest inheritance of C57Bl/6J alleles (24). Il13ra1-floxed animals were generated by crossing Il13ra1–/– animals with transgenic mice expressing Flippase (JAX#005703) from The Jackson Laboratory to remove the cassette containing the artificial splice acceptor and produce animals with loxP sites flanking exon 4 of Il13ra1, which contains the cytokine binding domain. Il13ra1-floxed animals were crossed to ACTA1-Cre mice (JAX#006149) to generate skeletal muscle–specific deletion of Il13ra1.

Human samples

For the cohort of women, plasma samples were collected from obese participants, normal-weight sedentary controls, and endurance athletes (25). For the cohort of men, plasma samples were collected from normal-weight sedentary controls, cross-country runners, and American-football players postseason (26). Demographic characteristics are included in tables S1 and S2. Human studies were approved by the institutional review boards at Texas Tech University and Emory University and informed consent was obtained prior to beginning any procedures.

C2C12 myoblast culture and cell line generation

C2C12 myoblasts were obtained from ATCC (CRL-1722). Myoblasts were maintained in culture at <80% confluence in DMEM (high glucose), 10% fetal bovine serum (FBS). For differentiation, myoblasts were plated on collagen-coated plates. At confluence, media was changed from 10% FBS to 2% horse serum to induce differentiation. Myotube formation was monitored visually for 7 to 10 days prior to experimental use. For overnight treatment with rIL-13, myotubes were treated with rIL-13 (10 ng/ml; R&D Systems, #413-ML) or vehicle control (0.1% BSA in PBS) in DMEM (high glucose) containing 2% horse serum for 18 to 24 hours. Acute signaling experiments were performed in myotubes after 1 hour of incubation in serum-free DMEM. rIL-13 was added at a final concentration of 10 ng/ml.

For the generation of stable shRNA-expressing cell lines, small hairpin target sequences were cloned into the pSIREN-RetroQ vector. The sequence for shRNA oligonucleotides can be found in table S4. Retrovirus was produced by transfecting retroviral vectors into Phoenix packaging cells, followed by conditioned media collection. C2C12 myoblasts were incubated with retrovirus-containing conditioned media with polybrene (4 μg/ml) and infected cells were selected with puromycin (2 μg/ml).

Stat3 knockout myoblasts were generated using LentiCRISPR v2 (27) containing a single gRNA for Stat3 from the mouse GeCKO v2 library (28). The sequence for LentiCRISPR v2 Stat3 gRNA can be found in table S4. Lentivirus was packaged in AD293 cells by transient transfection. Conditioned media containing lentivirus was used to infect C2C12 myoblasts. Cells were selected with puromycin and individual clones were characterized. Stat3 deletion was verified by immunoblot.

Primary myoblast isolation, culture, and differentiation

Primary myoblasts were isolated from mouse gastrocnemius, quadriceps, tibialis anterior, and biceps femoris muscles. Muscle was excised and washed with cold PBS and placed in DMEM on ice. Samples were minced with scissors and transferred to 10 ml of collagenase (type II; 2 mg/ml) and dispase (0.5 mg/ml) in DMEM at 37°C with shaking for 1 hour. The cell suspension was dispersed 20 times with a 10-ml pipette and centrifuged at 50g to spin down large muscle pieces. The supernatant was passed through a 70-μm cell strainer and centrifuged at 500g. The cell pellet was washed twice, resuspended in Ham’s F10, 20% FBS with bFGF (10 ng/ml), and plated on collagen-coated culture dishes. Fresh media and bFGF were added every other day while cells were expanded for 8 days. Cells were then plated for differentiation in collagen-coated culture plates. Media was changed to 4% horse serum in Ham’s F10 without bFGF to induce differentiation. Cells were differentiated for 8 to 10 days prior to experiments. For overnight treatment with rIL-13, primary myotubes were treated with rIL-13 (10 ng/ml) or vehicle control (0.1% BSA in PBS) in Ham’s F10 media containing 4% horse serum for 18 to 24 hours.

AD293 culture and transfection

AD293 cells were used for transcriptional reporter assays. Cells were maintained in 10% FBS in DMEM (high glucose).

Endurance running capacity test

Mice were acclimatized to the treadmill for three consecutive days at low speed (5 m/min) with no incline for 5 min. Running tests were performed with 5° incline. Animals ran at 10 m/min for the first 10 min followed by increases of 2 m/min every 5 min until exhaustion, which was defined by inability to remain on the treadmill for >5 s. Animals were removed immediately after exhaustion. An Exer-3/6 treadmill from Columbus Instruments was used for all treadmill running tests.

Endurance exercise training

Mice were acclimatized to treadmill running, as outlined above, before endurance exercise training. Endurance training was performed as described (29) with minor changes detailed below. Animals performed a single bout of running 5 days per week for the duration of exercise training (see detailed list of animal cohorts). All bouts of exercise were performed at the start of the dark cycle (ZT10-ZT14) to minimize disruption to the circadian rhythm. Animals ran 12 m/min for 30 min per day with 5° incline for the first week of exercise training, followed by an increase to 40 min per day in the second and subsequent weeks. Animals from both genotypes were able to perform all bouts of exercise. We did not observe changes in body weight or food intake in animals performing up to 6 weeks of exercise training.

Tissues and serum from endurance exercise–trained mice were collected 18 to 24 hours after the final bout of exercise to limit acute effects of exercise. Gastrocnemius muscle was used for mRNA and protein expression measurements, fiber type staining, electron microscopy, and isolating mitochondria. Soleus muscle was used for ex vivo functional assays of glucose uptake and fatty acid oxidation. Quadriceps muscle was used to measure mRNA and protein expression.

Acute exercise

Mice were acclimatized to treadmill running as outlined above. Animals were fasted for 4 hours prior to single bouts of acute exercise set for 30 min at 12 m/min with 5° incline. Tissues and serum were harvested immediately after the single bout of exercise.

Muscle strength test

Forelimb grip strength was measured using a Chatillon force meter (30). For the inverted screen test, mice were placed upright on a wire grid that was subsequently turned 180°. Time spent hanging from the screen was recorded with a maximum hanging time of 10 min (30).

Respirometry during exercise

Respiration during exercise was measured using a Metabolic Modular Treadmill connected to an Oxy-Max Comprehensive Lab Animal Monitoring System from Columbus Instruments. Mice were acclimatized to treadmill running and fasted for 4 hours prior to measurements. A single animal was placed in the chamber followed by a baseline respiratory equilibration for 5 min. The animal then performed an endurance running test starting at 10 m/min for 10 min with subsequent increases of 2 m/min every 5 min until exhaustion. Exhaustion was determined by inability to remain on the treadmill for >5 s.

Metabolic cage experiments

Mice were housed for 72 hours in a Comprehensive Laboratory Animal Monitoring System (CLAMS) from Columbus Instruments. VO2 and VCO2 were measured along with food intake and activity. Data collected from day 3 of the experiment were used for analysis.

Intramuscular injection of adenovirus

Virus was administered by direct injection into gastrocnemius as described (31). Animals were anesthetized with isoflurane inhalation. Each muscle received 50 μl of 1010 PFU/ml adGFP (control) or adIL-13 virus. Animals were allowed to recover for 72 hours before any experimental procedures were performed. Mitochondria were isolated from gastrocnemius muscle. Protein and RNA were isolated for immunoblot and gene expression studies from quadriceps.

Glucose tolerance

Mice were fasted overnight prior to glucose tolerance test. Fasting blood glucose was measured and animals were injected with glucose at 1.5 g/kg body weight into the peritoneum. Blood glucose was measured at 20, 40, 60, 90, and 120 min post-injection. Animals performed the previous bout of exercise at least 12 hours prior to start of glucose tolerance test to avoid acute effects of exercise on glucose metabolism.

Cell isolation and flow cytometry

Single cells were isolated from gastrocnemius, quadriceps, tibialis anterior, and biceps femoris. Muscles were minced and digested with Liberase TM (1 unit/ml), DNase I (0.1 mg/ml), 10% FBS in DMEM for 45 min. The cell suspension was centrifuged at 500g for 5 min and the pellet was incubated for another 45 min in fresh digestion buffer. The cell suspension was passed 20 times through a 10-ml pipette to release cells, centrifuged at 50g for 5 min to remove large debris, and passed through a 70-μm cell strainer. Cells were washed with 10% FBS in DMEM and overlaid onto 40% Percoll for centrifugation at 500g for 20 min. Pelleted cells were washed with PBS and red blood cells were lysed before counting. For each animal, 106 cells were stimulated for 3 hours at 37°C with eBioscience Cell Stimulation Cocktail (Thermo-Fisher, 00-4975-03) in 10% FBS in DMEM. Harvested cells were blocked with TruStain FcX (Biolegend, #101320) and stained with Fixable Viability Dye eFluor 455UV (eBioscience, #65086818). Extracellular markers were stained with BV650-conjugated rat α-mouse CD45 (30-F11, 0.2 μg), FITC-conjugated Armenian hamster α-mouse CD3 (145-2C11, 0.5 μg), FITC-conjugated rat α-mouse Ly-6G/Ly-6C (RB6-8C5, 0.5 μg), FITC-conjugated rat α-mouse CD11b (M1/70, 0.2 μg), FITC-conjugated rat α-mouse CD45R (RA3-6B2, 0.5 μg), FITC-conjugated rat α-mouse TER-119 (TER-119, 0.2 μg), PerCP/Cy5.5-conjugated rat α-mouse CD127 (IL-7Rα) (SB/199, 0.5 μg), PE-Cy7-conjugated rat α-mouse CD127 (IL-7Rα) (A7R34, 0.125 μg), PE-Cy7-cojugated Armenian hamster α-mouse CD3 (145-2C11, 0.2 μg), APC-conjugated rat α-mouse IL-33Rα (DIH9, 0.6 μg), APC/Cy7-conjugated rat α-mouse CD3 (17A2, 0.1 μg), APC/Cy7-conjugated mouse α-mouse NK1.1 (PK136, 0.4 μg), Biotin-conjugated rat α-mouse Ly-6G/Ly-6C (Gr-1, 0.25 μg), Biotin-conjugated rat α-mouse TER-119 (TER-119, 0.25 μg), Biotin-conjugated rat α-mouse CD11b (M1/70, 0.25 μg), Biotin-conjugated rat α-mouse CD5 (53-7.3, 0.25 μg), Biotin-conjugated rat α-mouse CD45R (RA3-6B2, 0.25 μg), Biotin-conjugated rat α-mouse CD45R (RA3-6B2, 0.25 μg), and APC-conjugated Streptavidin (Biolegend, 0.04 μg). Cells were then fixed and permeabilized with the Intracellular Fixation and Permeabilization Buffer Set (eBioscience, #88882400). Intracellular staining was performed with PE-conjugated rat α-mouse IL-13 (eBio13A, 1 μg), FITC-conjugated rat α-mouse Gata3 (TWAJ, 0.1 μg), and PerCP-eFluor710-conjugated rat α-mouse Rorγt (B2D, 0.1 μg). Stained cells were analyzed on a BD LSRFortessa Cell Analyzer.

Serum cytokine measurements

Assays of mouse IL-6, IL-13, and IL-33 levels in serum and muscle lysate were performed using assay kits from Meso Scale Discovery (#K152UBK, #K152QXD, #K152WFK) according to manufacturer instructions.

For human plasma samples, cytokines were analyzed by the V-Plex Proinflammatory Panel 1 (#K15049D) and custom U-Plex assays from Meso Scale Discovery according to manufacturer instructions. Cytokines with right-skewed distributions were natural log (ln)–transformed before analyses. For each cytokine, we excluded outliers (out of mean ± 4×SD) and replaced values below detection rate with ½ minimum values.

Metabolite measurements

Serum lipids were measured using NEFA Reagent (Wako Diagnostics, #999-34691) and Infinity TG reagent (ThermoFisher, #TR22421). For measurement of muscle glycogen and triglyceride, tissues were homogenized in buffer containing 50 mM Tris, 100 mM NaCl, and 0.1% NP40. Tissue was dried using a speed-vac centrifuge and dry tissue weight was used for normalization. For glycogen content, dried tissue samples were resuspended in KOH/Na2SO4 and boiled for 10 min. Large macromolecules were precipitated with ethanol and collected by centrifugation. Glycogen was hydrolyzed by boiling the precipitated macromolecules in H2SO4 for 10 min and neutralized with NaOH. Hydrolyzed glucose was quantified using the Glucose Oxidase Assay Kit (Sigma-Aldrich, #GAGO-20). Muscle lipids were extracted with chloroform and dried in a fume hood. Extracted lipids were analyzed using Infinity TG reagent.

Ex vivo and in vitro metabolic assays

Ex vivo fatty acid oxidation assays were performed using freshly isolated soleus muscle. Soleus muscle was incubated in Krebs-Ringer HEPES Buffer containing 5 mM glucose, 2% BSA, and 2 μCi of [3H]palmitate per muscle. Production of 3H2O in conditioned media was measured to determine fatty acid oxidation and normalized to muscle weight. Fatty acid oxidation in differentiated myotubes was performed similarly. Fatty acid uptake was determined by measuring [3H]palmitate uptake into differentiated myotubes in 5 min. Fatty acid oxidation and uptake were normalized to protein content. Glucose uptake was measured by uptake of [3H]2-deoxy-d-glucose in Krebs-Ringer HEPES Buffer in C2C12 myotubes.

Muscle fiber type staining

For SDH staining, frozen muscle sections (10 μm) were incubated in 0.2 M sodium phosphate buffer, pH 7.6, containing 167 mM sodium succinate and 1.2 mM nitro blue tetrazolium for 30 min at 37°C. Slides were washed with diH2O followed by 30%, 60% and 90% acetone and mounted with aqueous mounting media. For staining of myosin heavy chain isoforms, frozen muscle sections (10 μm) were fixed in 10% formalin, permeabilized with 0.3% Triton X-100 in PBS, and blocked with 5% FBS. Sections were stained overnight with mouse α-MhcI (BA-D5, 5 μg/ml), mouse α-MhcIIa (SC-71, 5 μg/ml), and mouse α-MhcIIb (BF-F3, 5 μg/ml) primary antibodies from the Developmental Studies Hybridoma Bank at the University of Iowa. After washing, the secondary antibodies Alexa Fluor 350–conjugated goat α-mouse IgG2b (γ2b, 10 μg/ml), Alexa Fluor 488–conjugated goat α-mouse IgG1 (γ1, 10 μg/ml), and Alexa Fluor 594–conjugated goat α-mouse IgM (μ chain, 10 μg/ml) were incubated with sections for 1 hour. Slides were mounted with Anti-fade mounting solution and sealed with nail polish for visualization. Images were collected using a Life Technologies EVOS FL Auto Imaging system using EVOS 10× and 20× objectives (AMEP4682, AMEP4625). Image analysis was performed using ImageJ.

Electron microscopy

Muscle slices were fixed in 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4. Samples were processed by the Electron Microscopy Core at Harvard Medical School and images were collected using a JEOL 1200EX transmission electron microscope equipped with an AMT 2k CCD camera. Mitochondrial analysis was performed on 10,000× cross section (2 nm per pixel) micrographs from gastrocnemius muscle. Eighteen to 20 micrographs from three mice were analyzed for each of the four groups: wild-type and Il13–/– mice with and without endurance training. Image analysis was performed blinded to study groups. ImageJ was used to calculate mitochondrial area.

RNA-seq and analysis

RNA-seq was performed on four animals per genotype per group. Sequencing and raw data processing were conducted at the IMB Genomics Core and IMB Bioinformatics Service Core at Academia Sinica (Taipei, Taiwan). Briefly, samples were quantified with Ribogreen (Life Technologies, CA) and RNA integrity was checked with a Bioanalyzer 2100 (Agilent, CA) (RIN > 8; OD 260/280 and OD 260/230 > 1.8). RNA libraries were prepared with the TruSeq Stranded mRNA Library Preparation Kit (Illumina) according to manufacturer instructions using 2 μg of RNA per sample. Sequencing was performed using an Illumina NextSeq 500 High Output Kit (75 cycles, 400 million total reads) on an Illumina NextSeq 500 instrument. Raw data were processed using Qiagen CLC Genomics Workbench (v.10.1.1). Raw sequencing reads were trimmed by removing adapter sequences, low-quality sequences (Phred quality score of < 20), and sequences with lengths > 25 bp. Individual samples were examined for quality control. Sequencing reads were mapped to the mouse genome assembly (mm10) from the University of California, Santa Cruz with the following parameters: mismatches = 2, minimum fraction length = 0.9, minimum fraction similarity = 0.9, and maximum hits per read = 5. Normalization and calculation of expression values were performed using the Differential Gene Expression for RNA-seq Tool in the Qiagen CLC Genomics Workbench v10. Normalization of gene expression was based on transcripts per kilobase million (RPKM). Statistical analysis of differential gene expression was calculated by generalized linear model implemented in the EdgeR package in R, accounting for differences in library size between samples (32). A significance cutoff for differentially expressed genes of FDR < 0.05 was used to determine differentially expressed genes. To compare differential gene expression patterns between wild-type and Il13 whole-body knockout (Il13–/–) mice, we also examined genes with P < 0.05. Gene Ontology (GO) analyses were performed with DAVID (https://david.ncifcrf.gov). Because a similar discrepancy pattern of differential gene expression between wild-type and Il13–/– mice was observed, we included genes with P < 0.05 for enrichment testing to increase the statistical power of identifying enrichened GO terms. GO terms with an enrichment FDR < 0.05 were considered as statistically significant, and terms with FDR < 0.1 were considered as suggestively significant. De Novo Motif Enrichment was performed with HOMER (http://homer.ucsd.edu/homer) searching 200-bp upstream and downstream of transcription start sites with default settings. Data have been deposited in GEO under accession number GSE126001.

qPCR

Relative gene expression was determined by real-time qPCR with SYBR Green. Expression was normalized to 36B4 (Rplp0) as the internal standard. All primer sequences are listed in table S5. Relative mtDNA was quantified by real time PCR using primers for mitochondrially-encoded Nd1 normalized to nuclear-encoded 36B4 (Rplp0) DNA. Relative expression levels were quantified using standard curve and normalized to the expression of a housekeeping gene 36b4. Student t tests were used to compare differential gene expression between the two groups. P < 0.05 was considered as nominally significant.

Immunoblotting

Standard Tris-Glycine SDS-PAGE was performed and transferred to PVDF membrane by wet transfer. Primary antibodies were incubated overnight in 1% BSA in TBST buffer. ECL signal was imaged using a BioRad ChemiDoc XRS+ imaging system. Antibodies for immunoblotting were mouse α-rat/mouse OxPhos Cocktail (Invitrogen 458099, 20E9DH10C12, 21A11AE7, 13G12AF12BB11, 1D6E1A8, 15H4C4, 1:5000), rabbit α-mouse β-Tubulin (CST2146, Polyclonal, 1:2000), rabbit α-mouse Phospho-Stat3 Tyr705 (CST9145, D3A7, 1:2000), and rabbit α-mouse Stat3 (CST12640, D3Z2G, 1:2000).

Isolation of mitochondria

Mitochondria from skeletal muscle were isolated by differential centrifugation (33). Briefly, gastrocnemius was collected, placed on ice, minced with scissors, and washed three times with 1 ml of PBS with 10 mM EDTA. Muscle was resuspended in isolation buffer 1 containing 70 mM sucrose, 50 mM Tris, 50 mM KCl, 10 mM EDTA, and 0.2% fatty acid–free BSA (pH 7.4), followed by homogenization with a polytron homogenizer four times for 5 s each. Nuclei and large debris were removed by centrifugation at 700g for 10 min at 4°C. The supernatant was collected and centrifuged at 10,000g for 10 min at 4°C to pellet mitochondria. Mitochondria were resuspended in isolation buffer 2 containing 70 mM sucrose, 200 mM mannitol, 5 mM EGTA, and 10 mM Tris (pH 7.4) and passed through a 70-μm cell strainer. Mitochondria were centrifuged again at 10,000g for 10 min at 4°C to remove debris and resuspended in isolation buffer 2 or specific buffers for downstream applications.

Mitochondrial respiration and activity assays

For measuring respiration of C2C12 myotubes, myoblasts were plated (5 × 105 cells per well) and differentiated for 7 days in Seahorse XF24 microplates. An hour prior to start of the assay, myotubes were switched to DMEM (no bicarbonate) with 5 mM glucose and 1 mM pyruvate. Respiration was measured 3 times followed by injection of oligomycin (2 μM final), carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP; 1 μM), and rotenone/antimycin A (1 μM) with three measurements after each injection. Data were normalized to protein content.

Electron flow assays with isolated mitochondria were performed as described (34). Isolated mitochondria (5 μg per well) were plated in XF24 microplates in buffer containing 70 mM sucrose, 220 mM mannitol, 10 mM KH2PO4, 5 mM MgCl2, 2 mM HEPES, 1 mM EGTA and 0.2% BSA. Initial assay buffer additionally contained 10 mM pyruvate, 2 mM malate, and 6 μM FCCP for complex I–driven respiration. Sequential injections of 2 μM rotenone, 10 mM succinate, 4 μM antimycin a, and 100 μM TMPD/10 mM ascorbate were used to measure complex II and IV respiration.

Complex IV activity was measured using the Complex IV Rodent Enzyme Activity Microplate Assay Kit (Abcam, #AB109911) according to manufacturer instructions using 10 μg of isolated mitochondria per well. Activity was measured by the oxidation of cytochrome c and loss of absorbance at 550 nm by reduced cytochrome c.

Luciferase reporter assays

AD293 cells were plated in 96-well plates (104 cells per well) in DMEM, 10% FBS. The next day, wells were cotransfected with 50 ng of firefly luciferase reporter, 10 ng CMV-β-galactosidase expression vector (internal control) and 10 to 30 ng of control or Stat3 expression vector using TransIT-LT1 transfection reagent (Mirrus, #MIR2300). The media was changed 24 hours later and cells were harvested 48 hours after transfection. Luciferase activity was measured using the Luciferase Activity Assay (Promega, #E1501) and normalized to β-galactosidase activity.

For luciferase assays performed in C2C12 myoblasts, cells were plated in 48-well plates. Each well was transfected with 0.5 μg of firefly luciferase reporter and 30 ng Stat3 expression vector using Lipofectamine 2000 transfection reagent (Thermo Fisher, #11668019). Media was changed 24 hours later to fresh media ± rIL-13 for 24 hours and luciferase activity was measured using the Promega Luciferase Activity Assay.

Stat3 with an N-terminal Flag tag was cloned into the pCMX-PL1 vector containing a CMV promoter using EcoRI and XhoI restriction sites. Primer sequences can be found in table S4. One- and two-kilobase promoter regions of Esrra and Esrrg were cloned into the pGL3-Basic vector using primer sequences provided in table S4. Esrra promoters were cloned starting from bases –24 to –1039 and –24 to –2027 for the 1-kb and 2-kb promoters, respectively, with +1 being the transcriptional start site. Esrrg promoters were cloned starting from bases –108 to –1119 and –108 to –2094 for the 1-kb and 2-kb promoters, respectively, with +1 being the transcriptional start site. For the mutation of the Stat3 binding site in the Esrrg 2-kb promoter, site-directed mutagenesis was performed using oligonucleotides provided in table S4. All plasmids were sequenced to confirm sequence integrity/identity.

Statistical analysis

All data are presented as means ± SEM. For human data, associations between training and plasma cytokines were analyzed using linear regression adjusted for age and other data characteristics. Detailed analyses of RNA-seq were as described above. Other statistical analyses were performed using GraphPad Prism 7. For cell-based studies, comparison of two parameters was performed using two-tailed unpaired Student t test. Multiparameter analyses of cell-based studies were performed with two-way analysis of variance (ANOVA) followed by Bonferroni post hoc tests. Cell-based experiments were performed with >3 biological replicates and repeated at least three times. Two-parameter comparisons of samples from in vivo studies were performed using two-tailed unpaired Student t test. Analysis of in vivo studies with multiple parameters was performed using two-way ANOVA with Bonferroni post hoc tests. Statistical significance was defined as *P < 0.05, **P < 0.01, ***P < 0.001 for all figures.

Supplementary Materials

science.sciencemag.org/content/368/6490/eaat3987/suppl/DC1

Figs. S1 to S6

Tables S1 to S5

Data S1 to S3

References and Notes

Acknowledgments: We thank G. Hotamisligil, R. V. Farese, and K. Inouye for help with treadmill/metabolic cage studies; A. J. Wagers, J. R. Mitchell, and X. Yang for critical comments; U. Unluturk and A. E. McQueen for technical assistance; and the IMB Genomics Core and IMB Bioinformatics Service Core at Academia Sinica (Taipei, Taiwan) for sequencing and RNA-seq data analysis. Funding: This work was supported by grants from NIH (F31DK107256 to N.H.K.; F31GM117854 to R.K.A.; 5R01HL129191 to V.A.N.; R01DK113791 and R21AI131659 to C.-H.L.) and American Heart Association (16GRNT31460005 to C.-H.L.). Y.H.L. was supported by funds from Ministry of Science and Technology, Taiwan. Author contributions: N.H.K., K.J.S., A.L.H., M.M.C., R.K.A., Y.H.L., K.A.S., M.R.G., D.J., and S.L. performed the experiments. E.A.O., J.H.K., C.M.P., and J.A.C. provided human plasma samples. D.H.S. and V.A.N. assisted in electron microscopy analysis and method development. D.F.P. and W.S.G. assisted in flow cytometry analysis of ILCs. J.L. and F.B.H. were consulted for analyses related to RNA-seq and human plasma data and provided recommendations for statistical methods. N.H.K. and C.-H.L. conceptualized the study, designed experiments, interpreted data, and wrote the manuscript. C.-H.L. supervised the study. Competing interests: The authors declare no competing interests. Data and materials availability: RNA sequencing data from this study have been deposited in GEO under accession number GSE126001.

Stay Connected to Science

Navigate This Article