Impaired exercise tolerance, mitochondrial biogenesis, and muscle fiber maintenance in miR-133a–deficient mice

Abstract

Exercise promotes multiple beneficial effects on muscle function, including induction of mitochondrial biogenesis. miR-133a is a muscle-enriched microRNA that regulates muscle development and function. The role of miR-133a in exercise tolerance has not been fully elucidated. In the current study, mice that were deficient in miR-133a demonstrated low maximal exercise capacity and low resting metabolic rate. Transcription of the mitochondrial biogenesis regulators peroxisome proliferator-activated receptor-γ coactivator 1-α, peroxisome proliferator-activated receptor-γ coactivator 1-β, nuclear respiratory factor-1, and transcription factor A, mitochondrial were lower in miR-133a–deficient muscle, which was consistent with lower mitochondrial mass and impaired exercise capacity. Six weeks of endurance exercise training increased the transcriptional level of miR-133a and stimulated mitochondrial biogenesis in wild-type mice, but failed to improve mitochondrial function in miR-133a–deficient mice. Further mechanistic analysis showed an increase in the miR-133a potential target, IGF-1 receptor, along with hyperactivation of Akt signaling, in miR-133a–deficient mice, which was consistent with lower transcription of the mitochondrial biogenesis regulators. These findings indicate an essential role of miR-133a in skeletal muscle mitochondrial biogenesis, exercise tolerance, and response to exercise training.—Nie, Y., Sato, Y., Wang, C., Yue, F., Kuang, S., Gavin, T. P. Impaired exercise tolerance, mitochondrial biogenesis, and muscle fiber maintenance in miR-133a–deficient mice.

Skeletal muscle is highly plastic in response to environmental and physiologic changes. Endurance exercise training produces multiple beneficial adaptations to improve skeletal muscle function. For example, exercise increases respiratory activity and glycolytic-to-oxidative fiber type switching that increases muscle fatigue resistance (1–3). In addition, exercise also promotes new capillary formation (angiogenesis) in skeletal muscle to facilitate oxygen and nutrient delivery (1, 4). Of note, mitochondrial adaptation is a major phenomenon in skeletal muscle that is exposed to exercise, including mitochondrial biogenesis, mitochondrial dynamics (fusion and fission), and clearance of damaged mitochondria (mitophagy) (5).

MicroRNAs (miRNAs) are a type of small, noncoding RNA that regulate post-transcriptional gene expression by binding to the 3′-untranslated region of target messenger RNAs, which results in mRNA degradation or translation repression. Primary miRNAs (pri-miRNAs) are transcribed and cleaved into 60- to 70-nt precursors in the nucleus and then cut into 19- to 22-nt mature miRNAs in the cytoplasm to exert gene silencing functions (6–8). Several miRNAs have been characterized as skeletal muscle–enriched miRNAs, such as miR-1, -133a, -133b, -206, -486, and -499 (9–13), and are closely related with skeletal muscle development and muscle function. miR-1 inhibits histone deacetylase 4 to activate myocyte enhancer factor-2–mediated myogenesis (14). miR-206 assists myoblast exit from the cell cycle and entry into differentiation via targeting DNA synthesis that is related gene DNA polymerase A (15). In addition, miR-486 modulates Akt signaling through repression of phosphatase and tensin homolog and forkhead box protein O1a (FoxO1a) (12), whereas miR-499 shows an important role in slow fiber determination (13).

miR-133, one of most studied muscle-enriched miRNAs, contains 3 different members, miR- 133a-1, miR-133a-2, and miR-133b. Mature miR-133a-1 and miR 133a-2 have nearly identical sequences, with a 2-nt difference between miR-133b at the 3′ terminus and are generated from 3 bicistronic miRNA clusters that encode miR-133a-1/miR-1-2, miR-133a-2/miR-1-1, and miR-206/miR-133b separately (11, 16–18). miR-133a enhances myoblast proliferation via targeting serum response factor (14) and inhibits smooth muscle gene expression during cardiomyocyte development (16). miR-133a also represses PR domain containing 16 to prevent brown adipose determination from muscle cells (19, 20) and is a key determinant of fast fiber type by inhibiting the slow fiber related gene TEA domain family member 1 (21). Recently, miR-133a-1/miR-133a-2 double-knockout (DKO) mice were generated, and one half of DKO mice die as embryos or neonates from ventricular-septal defects. Mice that survive to adulthood develop dilated cardiomyopathy and some suffer sudden death at 5–6 mo of age (16). Progressive centronuclear myopathy is observed in skeletal muscle and is associated with a fast-to-slow fiber switch and mitochondrial dysfunction (22).

In recent years, miRNA-mediated gene regulation has been implicated in endurance exercise–induced muscle adaptation. miR-23 and miR-696, which target peroxisome proliferator-activated receptor-γ coactivator 1-α (Pgc1α), are down-regulated after endurance exercise to promote mitochondrial biogenesis and improve mitochondrial function (23–25). In addition, miR-16 is decreased to enhance vascularization by increasing Vegf and Vegf receptor-2 in skeletal muscle after exercise training (26). Several human studies show that miR-133 is increased in response to acute endurance exercise (27, 28), but is decreased after long-term exercise training (27, 29). In mice, miR-133 is unchanged after acute endurance exercise (23). Nevertheless, these lines of evidence provide a clue for the potential of miR-133 in the regulation of exercise-induced skeletal muscle adaptations. To test whether miR-133 is important for exercise-induced adaptations, we used miR-133a-1 and -133a-2 DKO mice to assess exercise tolerance in miR-133a–deficient mice and their response to exercise training. We hypothesized that exercise tolerance and exercise-induced training adaptations would be impaired in miR-133a-knockout (KO) mice compared with wild-type (WT) littermates.

MATERIALS AND METHODS

Mice and animal care

miR-133a-1 and miR-133a-2 double-null (miR-133a1^−/−a2^−/−) C57BL/6 mice were obtained as previously described (16), and heterozygous miR-133a1^+/−a2^+/− mice were generated via crossing miR-133a1^−/−a2^−/− mice with WT C57BL/6 mice. The same genetic background for WT mice and 133a1^−/−a2^−/− KO mice were then obtained from the intercross of miR-133a1^+/−a2^+/− offspring. Gender, age, and genetic background matched WT mice, and miR-133a1^−/−a2^−/− KO mice were used throughout the study. Mice were housed with free access to water and standard rodent chow. Mouse maintenance and experimental use were performed according to protocols approved by the Purdue University Animal Care and Use Committee.

Muscle cryosectioning and Immunofluorescence

Fresh extensor digitorum longus (EDL) muscle was embedded in optimal cutting temperature compound and frozen in isopentane that was chilled on dry ice. Frozen muscles were then cut into 10-μm-thick cross-sections by using a Leica CM1850 cryostat (Leica, Wetzlar, Germany). For capillary staining, frozen section samples were fixed in 4% paraformaldehyde for 5 min and then permeabilized and blocked in PBS that contained 5% goat serum, 2% bovine serum albumin, 0.2% Triton X-100, and 0.1% sodium azide for 1 h. Sections were incubated with an anti-rat CD31 antibody (1:200; BD Pharmingen, Brea, CA, USA) for 2 h. Cell membranes were identified by costaining with dystrophin (1:500; Abcam, Cambridge, MA, USA). For fiber type staining, samples were directly permeabilized and blocked with blocking buffer, and MyHC2a (type 2A) and MyHC2b (type 2B) antibodies from mAb clones HB277 and HB283 were used (American Type Culture Collection, Manassas, VA, USA). After each primary antibody, sections were incubated with respective fluorescent-labeled anti-IgG secondary antibodies, mouse anti-rat 594 for CD31, goat anti-mouse IgG1 594 for MyhC2a, goat anti-mouse IgM 486 for MyhC2b, and goat anti-rabbit 647 for dystrophin (1:1000; Thermo Fisher Scientific, Waltham, MA, USA) for 1 h. Fluorescent images were captured with a CoolSnap HQ charge-coupled-device camera (Photometrics, Tucson, AZ, USA) by using a Leica DM6000 microscope. For 1 whole EDL muscle, 4–5 images were taken under ×100 magnification to merge into an intact EDL muscle image for analyses in Adobe Photoshop CC (Adobe, San Jose, CA, USA). Capillary and fiber number counting were quantified by count tool manually (Adobe). Fiber types were traced and fiber type was recorded by immunostaining images (red for type 2A, green for type 2B, nonstained for type 2X or 1). For fiber size determination, a single fiber was selected by magic wand tool (Adobe). Area measurement was performed after setting the measurement scale (1 pixel = 0.57 μm; ×100 magnification). Cross-sectional area of 100 fibers from the central region of EDL muscle was recorded for each fiber type.

Indirect calorimetry

Mouse resting energy metabolism was monitored with an Oxymax Open Circuit indirect calorimeter (Oxymax; Columbus Instruments, Columbus, OH, USA) as previously described (30). In brief, mice were placed in metabolic chambers with free access to food and water. Oxygen consumption (Vo₂) and carbon dioxide production (Vco₂) were recorded individually for 24 h under a constant environmental temperature (22°C) and 12-h light (6 am–6 Pm)/12-h dark cycle (6 Pm–6 am) after 1 d of acclimation. Vo₂ and Vco₂ were normalized to body mass. Energy expenditure (EE) in calories was calculated:

where RER (respiratory exchange ratio) = Vco₂/Vo₂.

Measurement of exercise capacity

Exercise capacity was determined by using a graded exercise treadmill (Columbus Instruments), and mice were run to exhaustion. Mice were acclimated on the treadmill at a speed of 10 m/min on a 10% grade for 5 min on 3 constitutive days before the capacity test. For determination of maximal exercise capacity, mice started by running at 10 m/min on a 10% grade for 5 min, with subsequent increases in speed by 2 m/min every 1 min until mice were unable to maintain running speed. Running speed, time, and distance were recorded. Work and power were calculated as follows:

Exercise protocol

Acute exercise was performed at 14 m/min on a 10% grade for 60 min. Long-term exercise training of mice was performed as previous described (31). In brief, 3-mo-old mice were trained on the treadmill at 14 m/min on a 10% grade for 60 min, 5 d/wk, for 6 wk. Mice were euthanized by cervical dislocation 3 h after the acute exercise bout and 1 d after completion of exercise training. Tibialis anterior (TA), EDL, and gastrocnemius muscles were collected and frozen at −80°C for further analysis.

Quantitative real-time PCR

TA muscle was used for gene expression analysis. Total RNA was extracted by using Trizol reagent (Thermo Fisher Scientific). For mRNA reverse transcription, first-strand cDNA was generated by random hexamer primers with MMLV Reverse Transcriptase (Thermo Fisher Scientific). For mature miRNA reverse transcription, multiple adenosine nucleotides were first added to the 3′ end of total RNA with Escherichia coli DNA polymerase (New England BioLabs, Ipswich, MA, USA), and cDNA was then synthesized with a Poly T primer including an adaptor sequence (Supplemental Table 1) using MMLV Reverse Transcriptase. Real-time PCR reactions were performed with a SYBR green PCR kit in a Roche LightCycler 480 System (Roche, Basel, Switzerland). Primers for mRNA are listed in Supplemental Table 1. For the determination of mature miRNA, the specific miRNA sequence was used as the forward primer and the adaptor sequence from the Poly T primer was used as the universal reverse primer for all miRNAs. Gene expression was determined with the 2^−ΔΔCt relative quantification method and was normalized to 18 s for mRNA or sno202 for miRNA.

Mitochondrial content

White and red parts of the gastrocnemius were first separated by anatomic location and color. Total DNA was then isolated by using a standard phenol-chloroform protocol. Quantitative real-time PCR was used to determine relative mitochondrial copy number by calculating the ratio of amplification between mitochondrial DNA and nuclear DNA. Primers for mitochondrial DNA and nuclear DNA amplification were previously described (32) and are listed in Supplemental Table 1.

Citrate synthase activity

Citrate synthase (CS) activity was determined spectrophotometrically in TA muscle tissue. In brief, muscle samples were homogenized in sucrose buffer (20 mM tris, 40 mM KCl, 2 mM EGTA, 250 mM sucrose) that contained protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA) and phosphatase inhibitors (NaF and Na₃VO₄). Of muscle homogenate, 20 μg was incubated with 100 mM Tris, 0.1% Triton X-100, 100 μM DTNB, and 0.3 mM acetyl-CoA. Reaction was initiated by adding 0.5 mM oxaloacetic acid. Absorbance was measured on a spectrophotometer (Thermo Fisher Scientific) at 412 nm for 3 min. Enzyme activity was reported as micromolar per minute per milligram protein.

Immunoblot analysis

Muscle samples from TA muscle were homogenized with RIPA buffer that contained a protease inhibitor cocktail (Sigma-Aldrich) and phosphatase inhibitors (NaF and Na₃VO₄). Protein concentration was measured by using a bicinchoninic acid protein quantification kit (Pierce, Rockford, IL, USA). Equal amounts of protein were loaded for electrophoresis (Bio-Rad, Hercules, CA, USA). Proteins were transferred to PVDF membranes (Bio-Rad) and incubated with primary antibodies, followed by anti-rabbit or anti-mouse IgG-horseradish peroxidase (Cell Signaling Technology, Danvers, MA, USA)– or fluorescence (Li-Cor Biosciences, Lincoln, NE, USA)-conjugated secondary antibody. Signals were detected by using fluorescence or chemiluminescence Western blot detection reagent (Santa Cruz Biotechnology, Santa Cruz, CA, USA) on a FluorChem E system (ProteinSimple, San Jose, CA, USA). Mitochondrial oxidative complex cocktail was purchased from Abcam; FoxO1, pAkt, Akt, pp70S6k, p70S6k, pS6, S6, LC3 were from Cell Signaling Technology; and IGF-1 receptor (Igf1R), P62, glyceraldehyde 3-phosphate dehydrogenase were from Santa Cruz Biotechnology.

Statistical analysis

All analyses were conducted with Student's t test with a 2-tail distribution or a 2-way ANOVA, followed by post hoc comparisons. All experimental data were presented as means ± SEM. Comparisons with values of P ≤ 0.05 were considered statistically significant.

RESULTS

miR-133a is induced after aerobic exercise in skeletal muscle

To comprehensively assess the role of miR-133 in regulation of muscle function in response to aerobic exercise, we first examined the expression level of miR-133 after acute exercise in mice with or without 6 wk of exercise training. Total RNA from TA muscle was harvested 3 h after a single bout of exercise. The basal level of miR-133a had a 20-fold abundance compared with miR-133b (Fig. 1A), which indicated that miR-133a is the primary isoform of miR-133 in TA muscle. Acute exercise did not alter the transcription of miR-133a-1, miR-133a-2, and miR-133b, but long-term training increased the transcriptional level of primary miR-133s (Fig. 1B–D), which suggested a higher requirement of miR-133 in training muscle. Mature miR-133s were then examined. Consistent with previous human studies (27, 28), acute exercise increased miR-133a in both untrained and trained mice, which suggested enhanced processing of miR-133a precursors after acute exercise (Fig. 1E). In addition, exercise training increased mature miR-133a (Fig. 1E); however, acute and trained exercise had no effect on the expression level of miR-133b (Fig. 1F), which further demonstrated that miR-133a is the main isoform of miR-133 involved in exercise regulation.

Figure 1.

Aerobic exercise induces the expression of miR-133a. TA muscles from 3 h after 1 h of acute aerobic exercise in mice with or without 6 wk of aerobic exercise training were used for gene analysis. A) Relative expression levels of miR-133a and miR-133b (n = 7). B–D) Expression levels of miR-133 primary transcripts miR-133a-1 (B), a-2 (C), and b (D) (n = 4) in response to acute exercise (A-Exe). E, F) Expression levels of mature miR-133a (E) and miR-133b (F) in response to A-Exe (n = 4). Data are presented as means ± se. **P < 0.01.

Impaired exercise tolerance in miR-133a–deficient mice

Because miR-133a was increased in response to acute exercise and with exercise training, miR-133a-1 and miR-133a-2 double-null mice (16) were used to investigate the effect of miR-133a deletion on exercise performance. Primary miRNA of 133a-1 and 133a-2 were completely abolished in TA muscle of KO mice (Fig. 2A) and mature miR-133a expression was decreased by >90% (Fig. 2B), which indicated successful deletion of miR-133a in muscle tissue. We first assessed the exercise capacity of the miR-133a–deficient mice. In response to graded, maximal exercise, exercise capacity was 33% lower (22 m/min vs. 33 m/min) in miR-133a-KO compared with WT mice (Fig. 2C). Accordingly, the running time was 40% and running distance 50% lower in KO vs. WT mice (Fig. 2D, E). Work and power were 50 and 20% lower in KO vs. WT mice, respectively (Fig. 2F, G). These results from the maximal treadmill test demonstrate profound exercise intolerance in KO vs. WT mice, which implicates potential muscle dysfunction in miR-133a–deficient mice.

Figure 2.

Impaired exercise capacity in miR-133a–deficient mice. Total mRNA was extracted in TA from WT and KO mice. A) Expression levels of primary miR-133a transcripts, miR-133a-1 and miR-133a-2. B) Expression level of mature miR-133a (n = 7). Maximal graded treadmill exercise was performed in WT and KO mice after 3 d acclimatization. C–E) Speed (C), distance (D), and time (E) at maximum were recorded. F, G) Work (F) and power (G) were calculated normalized to individual body mass (WT, n = 6; KO, n = 8). n.d., not detectable. Data are presented as means ± se. *P < 0.05; **P < 0.01.

Skeletal muscle is a key tissue for energy metabolism regulation. We questioned whether energy expenditure was altered in miR-133a–deficient mice. Resting metabolic rate was evaluated by indirect calorimetry. Vo₂ was lower in both light (11% decrease) and dark phases (10% decrease), along with a decrease in Vco₂ production (Fig. 3A, B). An increased trend for respiratory exchange ratio was found in miR-133a-KO mice, which suggested fuel use switching from free fatty acid to glucose (Fig. 3C). Glucose oxidation uses less oxygen than fat, which supports the lower oxygen consumption in miR-133a-KO mice. Accordingly, calculated energy expenditure was significantly lower in KO compared with WT mice (Fig. 3D).

Figure 3.

Reduced energy expenditure in miR-133a–deficient mice. Indirect calorimetry was performed in 3-mo-old mice. O₂ consumption (Vo₂) (A), CO₂ production (Vco₂) (B), RER (C), and energy expenditure (D) were recorded under a 12-h light/dark cycle (WT, n = 9; KO, n = 10). RER, respiratory exchange ratio. Data are presented as means ± se.

Impaired mitochondrial biogenesis in miR-133a–deficient mice

Decreased aerobic exercise capacity and energy expenditure implicate a deficiency in oxidative phosphorylation in miR-133a–deficient mice, which reflects mitochondrial dysfunction. In fact, abnormal mitochondria have been observed in miR-133a–deficient mice (22). Here, we further investigated the mitochondrial changes at the molecular level. The transcription levels of Pgc1α, peroxisome proliferator-activated receptor-γ coactivator 1-β (Pgc1β), nuclear respiratory factor-1 (Nrf1), and transcription factor A, mitochondrial (Tfam)—the central transcriptional factors that regulate mitochondrial biogenesis—were all decreased in KO muscle (Fig. 4A) and the protein level of Pgc1α was also decreased (Fig. 4E). Mitochondrial content was determined by the ratio of mitochondrial DNA to nuclear DNA. Significant loss of mitochondria was found in both glycolytic (white) and oxidative (red) fibers of gastrocnemius muscle (Fig. 4B). This conclusion was supported by decreases of mitochondrially encoded transcription of NADH-ubiquinone oxidoreductase chain 1 (Nd1) and cytochrome c oxidase subunit I (Cox1) (Fig. 4C). Oxidative phosphorylation in mitochondria is mediated by the complex family (I–V) of proteins in the electron transport chain (ETC). We analyzed complexes I–V by Western blot. Complexes I and III were dramatically lower in miR-133a-KO muscle, whereas complexes II, IV, and V were unchanged (Fig. 4E), which suggested an impairment in the respiratory chain in KO mice. Finally, mitochondrial function was assessed by CS activity, which was lower by 20% in KO mice (Fig. 4D), suggesting that the mitochondrial impairment induced by miR-133 KO impacts both the tricarboxylic acid cycle and ETC. Collectively, these findings revealed an important role for miR-133a in the regulation of mitochondrial biogenesis.

Figure 4.

Deletion of miR-133a decreases mitochondria biogenesis in skeletal muscle. A) Expression levels of mitochondrial biogenesis regulators, Pgc1α, Pgc1β, Nrf1, and Tfam. B) Genomic DNA was extracted from the white and red part of the gastrocnemius muscle separately. Mitochondrial content was determined by the ratio of mitochondrial DNA and nuclear DNA. C) Expression levels of mitochondrially encoded Nd1 and Cox1. D) CS activity was measured spectrophotometrically. E) Protein levels of Pgc1α and representative proteins in mitochondrial oxidative complexes I–V (CI–V). Gapdh, glyceraldehyde 3-phosphate dehydrogenase. Data are presented as means ± se (n = 4). *P < 0.05; **P < 0.01.

Deletion of miR-133a abolished exercise training–induced mitochondrial biogenesis

Long-term exercise training exerts multiple beneficial effects on skeletal muscle mitochondrial physiology, including mitochondrial dynamics (fusion and fission), clearance of damaged mitochondria (mitophagy), and mitochondrial biogenesis (5). In light of the up-regulation of miR-133a during exercise (Fig. 1) and impaired mitochondrial biogenesis in miR-133a–deficient mice (Fig. 4), we next addressed whether miR-133a is also involved in exercise training–induced mitochondrial biogenesis. As expected, exercise training–induced miR-133a expression was blocked in miR-133a–deficient mice (Supplemental Fig. 1A, B, D). In WT mice, 6 wk of exercise training increased the transcriptional expression of the mitochondrial biogenesis regulators, Pgc1α, Pgc1β, Nrf1, and Tfam (Fig. 5A–D), but failed to increase the expression of those regulators in miR-133a–deficient mice (Fig. 5A–D). However, training only showed an increase of complex III protein, but no effect on protein levels of Pgc1α and complexes I, II, IV, and V (Supplemental Fig. 2A–G), which suggested that post-transcriptional regulation for mitochondrial proteins might require more intensive exercise. Although mitochondrial content was elevated after exercise training, training did not restore mitochondrial mass in miR-133a–deficient mice (Fig. 5E). Similar effects were found in the expression of mitochondrially encoded genes Nd1 and Cox1 (Fig. 5F, G). Moreover, mitochondrial function was assessed by CS activity. In agreement with decreased mitochondrial mass in miR-133a–deficient mice, CS activity was still lower in miR-133a-KO compared with WT mice after exercise training (Fig. 5H). These data indicate a limited effect of exercise training on mitochondrial physiology of miR-133a–deficient mice, which further demonstrates that miR-133a is essential for exercise-induced mitochondrial biogenesis in skeletal muscle.

Figure 5.

Exercise training fails to rescue mitochondrial biogenesis in miR-133a–deficient mice. Mice were treadmill trained 5 d/wk, 60 min/d at 14 m/min speed for 6 wk. A–D) Expression levels of mitochondrial biogenesis regulators Pgc1α (A), Pgc1β (B), Nrf1 (C), and Tfam (D). E) Mitochondrial content was determined by the ratio of mitochondrial DNA (mtDNA) and nuclear DNA (nDNA). F, G) Expression levels of mitochondria-encoded Nd1 (F) and Cox1 (G). H) CS activity was measured spectrophotometrically. Data are presented as means ± se (n = 7). *P < 0.05 vs. all other groups; ^#P < 0.05 vs. all other groups.

Deletion of miR-133a compromised exercise training–induced fiber type switch

Muscle fiber shifting from glycolytic to oxidative fiber and increased capillarization classically characterize the skeletal muscle adaptation to aerobic exercise, which is closely associated with increased mitochondrial biogenesis. We first assessed exercise training–induced fiber type distribution in miR-133a–deficient mice. Consistent with increased mitochondrial biogenesis, muscle fibers significantly switched from glycolytic type 2B to more oxidative type 2A after exercise training (Fig. 6A, B). Specifically, percentage of type 2A fiber was increased by 70% in WT mice vs. only 15% in miR-133-KO mice (Fig. 6B), whereas type 2B fiber was decreased by 16 and 19% in WT and KO groups, respectively (Fig. 6C). No significant change was found in type 2X/1 fiber (Fig. 6D). These data showed a compromised training effect on fiber type switch in miR-133a–deficient mice. In addition, type 2B fibers were decreased without the relative increase of type 2X/1 and type 2A fibers in miR-133a-KO mice (Fig. 6B–D), which implicated a significant loss of type 2B fibers in miR-133a–deficient mice and which might be a secondary effect of impaired mitochondrial function. Subsequently, we measured the effects of exercise training on skeletal muscle capillarization. The capillary-to-fiber ratio was not altered after training in both WT and KO mice, as expected (Fig. 6E, F), which suggested that greater intensity of exercise might be required for increased skeletal muscle capillarization.

Figure 6.

Exercise training has a compromised effect on fiber type distribution in miR-133a–deficient mice. EDL muscle from WT and miR-133a–deficient was harvested for immunostaining after 6 wk training. A) Muscle fiber typing was determined by immunostaining with specific antibodies to type 2A (red), type 2B (green), and type 2X and I (nonstained) fibers, with dystrophin (blue) stain to identify cell membrane. B–D) Total fibers in whole EDL muscle were counted and fiber distribution was presented as the percentage of total fibers. E) Capillaries in EDL muscle were identified with CD31 staining. F) Capillaries in whole EDL muscle were counted and expressed as capillary-to-fiber ratio (n = 5/group). Data are presented as means ± se. Scale bars: 100 μm (A), 50 μm (E).

Abnormal Akt activation is involved in the impairment of mitochondrial biogenesis in miR-133a–deficient mice

We next investigated the underlying mechanism for the impairment of mitochondrial biogenesis in miR-133a–deficient mice. As Igf1r is a predicted target of miR-133a (33), Igf1r was examined in miR-133a–deficient mice. mRNA level of Igf1R was not changed (Fig. 7A), but Igf1R protein was increased in KO muscle, demonstrating a post-transcriptional regulation of miR-133a on Igf1R (Fig. 7B). In skeletal muscle, Igf1 induces muscle hypertrophy via Igf1R-activated Akt signaling. We found not only greater Akt phosphorylation, but also greater total Akt protein level in KO mice (Fig. 7B). mRNA analysis demonstrated greater Akt1, but not Akt2, in miR-133a-KO muscle (Fig. 7A). Activation of Akt has been reported to inhibit FoxO1 translocation and transcription, which contributes to the suppression of Pgc1α transcription (34–36). Consistent with this premise, FoxO1 was much lower in miR-133a–deficient mice, which was likely a result of hyperactivated Akt (Fig. 7B). Because exercise training failed to restore mitochondrial biogenesis in miR-133a–deficient mice, we asked if Akt is constitutively activated during exercise training. mRNA level and protein level of Igf1R, Akt1, and Akt2 were not altered after exercise training in WT muscle (Fig. 7C, D). In contrast, Igf1R and Akt were increased in miR-133a-KO mice, whereas FoxO1 was accordingly lower after exercise training (Fig. 7C, D).

Figure 7.

Akt is activated in miR-133a–deficient mice. A) Expression levels of Igf1R, Akt1, and Akt2. B) Protein levels of Igf1R, phosphorylation of Akt, total Akt, and FoxO1. C) Expression levels of Igf1R, Akt, and Akt2 in mice that received 6 wk of exercise training. D) Protein levels of Igf1R, phosphorylation of Akt, total Akt, and FoxO1 in trained muscle. Gapdh, glyceraldehyde 3-phosphate dehydrogenase. Data are presented as means ± se (n = 4). *P < 0.05; **P < 0.01.

Hyperactivation of Akt contributes to muscle hypertrophy in type 2A fibers, but loss of type 2B fibers

Mitochondrial oxidative capacity has an inverse relationship with fiber size in skeletal muscle (37). In light of decreased mitochondrial function and hyperactivation of Akt in miR-133a–deficient mice, muscle fiber size was determined in miR-133a-KO mice. Akt is well known as the key hypertrophic pathway through decreased protein degradation and increased protein synthesis in skeletal muscle. Both phosphorylation and protein levels of p70S6K and p70S6, which are important for protein synthesis, were greater in miR-133a–deficient muscle (Fig. 8A). In addition, the atrophic genes, Atrogin, Murf1, FoxO3a, were all dramatically lower (Fig. 8B). Subsequently, average fiber cross-sectional area of EDL muscle was greater in KO compared with WT mice (Fig. 8C). When analyzed by fiber type, fiber cross-sectional area of only type 2A fibers were larger in miR-133a–deficient EDL muscle (Fig. 8E), whereas there were no differences in type 2B fiber size. Surprisingly, although Akt-FoxO3a signaling is a classic antiatrophic pathway, there was a large loss of type 2B fibers not as a result of a switch to type 2A and 2X, but rather because of a reduction of total fibers in EDL of miR-133a-KO mice (Fig. 8D). Autophagy markers, P62 and LC3-I/II, were then analyzed to examine whether the loss of type 2B fibers might be caused by increased autophagy. Protein levels of P62 and LC3-II were significantly increased in miR-133a-KO muscle (Fig. 8F). Subsequent immunofluorescence of LC3 in EDL muscle cross-section further demonstrated that autophagy was present in type 2B fibers, not type 2A fibers (Fig. 8G). These results indicate a diverse response of glycolytic and oxidative fibers to the impaired energy homeostasis and mitochondrial dysfunction in miR-133a deficient mice.

Figure 8.

Hyperactivation of Akt contributes to hypertrophy in type 2A fibers, but loss of type 2B fibers. A) Protein levels of p70S6K (S6k) and p70S6, total p70S6K and p70S6. B) Expression levels of atrophic gene atro1, murf1, and fox3a (n = 4). C) Fiber cross-sectional area of EDL muscle (n = 4). D) Fiber numbers were counted after fiber type immunostaining in EDL muscle (n = 4). E) Representative images of type 2A (red) and type 2B (green) fibers in EDL muscle. Cross-sectional areas of myofibers were determined by using Photoshop software, and 100 fibers were counted for each EDL (n = 3). F) Protein levels of autophagy markers, P62 and LC3 (n = 4). G) Representative image of LC3 immunostaining in EDL muscle (n = 3). Arrows indicate fibers with autophagy. Gapdh, glyceraldehyde 3-phosphate dehydrogenase. Data are presented as means ± se. Scale bars: 100 μm (E), 50 μm (G). *P < 0.05 (type 2B); **P < 0.01 (total fiber).

DISCUSSION

Endurance exercise improves muscle function through multiple adaptations in skeletal muscle (1). To investigate the potential role of miRNA in endurance exercise–mediated muscle adaptations, miR-133a–deficient mice were used to determine the potential role of miR-133a in exercise tolerance. In the current study, we reveal that exercise capacity is reduced in miR-133a–deficient mice as a consequence of impaired mitochondrial biogenesis, which supports the previous finding that miR-133a–deficient muscle has an impairment in mitochondrial function (22). Long-term aerobic exercise training failed to rescue abnormal mitochondrial biogenesis in miR-133a–deficient mice, which may result from hyperactivation of Akt-mediated down-regulation of FoxO1 in miR-133a–deficient muscle. Finally, miR-133a deficiency led to loss of type 2B muscle fibers. Our data suggest an indispensable role of miR-133a in mitochondrial biogenesis and muscle function.

In skeletal muscle, mitochondrial biogenesis is mediated by several central transcriptional factors, including Pgc1α, Pgc1β, Nrf1, and Tfam responses to exercise (38). We observed that primary miR-133s are up-regulated after exercise training in concert with the increase of mitochondrial regulators, implicating the requirement of miR-133a during mitochondrial biogenesis. However, mature miR-133a is increased 3 h postexercise without increasing the transcription of primary miR-133s, which suggests that miR-133a is acutely and transiently regulated by exercise. Exercise training increased muscle Pgc1α, Pgc1β, Nrf1, and Tfam expression in WT, but not in miR-133a-KO mice, which is consistent with a lack of mitochondrial biogenesis in miR-133a-KO mice. Considering the inability of exercise training to restore mitochondrial biogenesis in miR-133a–deficient mice, miR-133a plays a key role in the regulation of exercise signals via transcription of these mitochondrial regulators.

Several signal pathways are activated to regulate the expression of mitochondrial transcription after endurance exercise, such as AMPK, calcium, and p38 MAPK (38); however, little evidence points to those possible targets through which miR-133a facilitates mitochondrial biogenesis. One possibility is miR-133a regulation of Igf1 that mediates differentiation via direct targeting of Igf1R in muscle cells (33), which suggests that miR-133a might be involved in Igf1-Akt signaling. Our findings support this hypothesis in that Akt signaling was enhanced in miR-133a–deficient muscle. Indeed, insulin and Igf1 diminish AMPK activity through Akt phosphorylation of AMPK Ser485/Ser491 in skeletal muscle (39), implying that Akt is a negative regulator for AMPK-mediated mitochondrial biogenesis. Furthermore, hyperactivation of Akt in muscle seems to be a common phenotype in some severe myopathy models with impaired muscle oxidative capacity and mitochondrial function. Dystrophin-deficient mdx mice have decreased mitochondrial oxidative phosphorylation (40), but increased Akt activation at both the transcriptional and phosphorylation level (41). Mice with muscle-specific deletion of mammalian target of rapamycin (mTOR) also exhibit decreased mitochondrial function and hyperactivation of Akt signaling (42). These lines of evidence reveal an underlying connection of uncontrolled Akt signaling and mitochondrial dysfunction in skeletal muscle. The mechanism through which Akt regulates mitochondrial biogenesis requires further investigation; however, suppression of FoxO1 activity by Akt may provide some clues for this connection (43–45). In hepatocytes, FoxO1 directly promotes the transcription of Pgc1α (34), whereas insulin inhibits Pgc1α-mediated gluconeogenesis via suppression of FoxO1 by Akt (35). In addition, insulin suppresses nuclear import of FoxO1, thereby decreasing Pgc1α and its downstream gene expression in human muscle (36). Consistent with these findings, FoxO1 total protein was greatly reduced coincident with Akt hyperactivation and decreased Pgc1α transcription in the current study (Fig. 6A).

Igf1-Akt signaling is a critical pathway for myofibrillar protein synthesis through the tuberous sclerosis complex (TSC)-mTOR-p70S6K cascade, eventually contributing to muscle hypertrophy; however, resistance exercise training, which results in muscle hypertrophy, has no effect on CS activity (46) or mRNA expression of Pgc1α (47). Conversely, aerobic exercise training enhances mitochondrial biogenesis and mitochondrial function (48), but has minimal impacts on muscle hypertrophy compared with resistance exercise (37). Of interest, miR-133a expression level is down-regulated in resistance exercise after functional muscle overload (49), whereas our findings demonstrate that miR-133a is up-regulated by aerobic exercise. Further studies are required to further elucidate the role that miR-133a–mediated Igf1-Akt plays in the initiation of myofibrillar protein synthesis and mitochondrial biogenesis.

In addition to a lack of changes in oxidative capacity, exercise training has a comprising fiber type shift in miR-133a–deficient muscle. A possible explanation is that fiber type determination shares common signaling pathways with mitochondrial biogenesis, such as Pgc1α. On the basis of specific Pgc1α gain and loss of function studies, Pgc1α is a key factor that regulates slow fiber determination based in muscle (50, 51). Exercise training failed to increase the expression of Pgc1α; therefore, it is not surprising that exercise training failed to induce the same degree of fiber type shift in miR-133a–deficient mice as it does in WT mice. However, decreases in Pgc1α could not explain the change of fiber type composition in untrained miR-133a–deficient mice, which have a higher percentage of oxidative fibers. This observation is consistent with previous research that describes a fast-to-slow fiber type shift in untrained miR-133a–deficient mice (22). Although inhibition of miR-133a increases slow gene expression by directly targeting TEA domain family member in muscle (21), supporting the oxidative fiber determination after the deletion of miR-133a, this is discrepant with the phenotype of impaired mitochondrial function in miR-133a–deficient muscle, as slow fibers are well known to have higher mitochondrial activity. Our subsequent analysis demonstrated a significant loss of type 2B fibers in EDL muscle from miR-133a KO mice, which accounts for the increase of type 2A fiber percentage. Consistent with miR-133a-KO mice, loss of fast fibers also occur in mice with muscle-specific deletion of mTOR that also have hyperactivation of Akt and impaired mitochondrial dysfunction (43).

Muscle mass maintenance is regulated by alterations in both muscle fiber number and muscle fiber volume. We identified autophagy as a potential mechanism for the loss of type 2B fibers as autophagy was present exclusively in type 2B fibers, whereas we speculate that increased Akt activity contributes to type 2A fiber hypertrophy. Oxidative stress–triggered autophagy might explain the opposite response of type 2B and type 2A fibers in miR-133a–deficient mice. Decreased mitochondrial function is often associated with increased reactive oxygen species (ROS). Although ROS was not measured in miR-133a KO muscle, complexes I and III in the ETC are main sites of ROS production (52) and these are significantly decreased in miR-133a-KO muscle. Complex I deficiency has been linked to increased ROS production (53). As a result of the high H₂O₂ scavenging capacity (54) and activities of antioxidant enzymes, superoxide dismutase, glutathione peroxide, and catalase (55–57) in slow fibers, type2A fibers are more resistance to oxidative stress. Oxidative stress–regulated autophagy is not fully understood, and mitochondrial ROS production in miR-133a–deficient mice is a potential mechanism (22), perhaps via mitophagy-triggered Bcl-2/adenovirus E1B 19-kDa-interacting protein 3 or phosphatase and tensin homolog-induced putative kinase 1 (58) autophagy pathways.

In this study, exercise training fails to increase protein levels of Pgc1α and most mitochondrial complexes, as well as CS activity and skeletal muscle capillarization, as expected, which suggests that more intensive exercise might be required to induce such changes. Tanner et al. (59) found mitochondrial complex proteins are increased after 3 wk of training initiated with 80% maximal tolerated speed, but not Pgc1α protein. In Rossiter et al. (60), young rats trained with maximum tolerated speed increased capillary density, but failed to alter the capillarization if running with the maximal speed of old rats, which is 30% lower than that of young rats in the beginning of exercise training. Here, 40 and 65% maximal speed of training were applied in WT and KO mice separately; therefore, the failure of Pgc1α, mitochondrial complexes, CS activity, and capillarization to respond to exercise training is not surprised. In fact, the absolute work of KO mice is much higher than that of WT mice; however, worse exercise training response in miR-133a–deficient mice further demonstrates the importance of miR-133a in exercise regulation.

Of note, primary miR-133b transcript is compensatively increased in miR-133a-KO mice (Supplemental Fig. 1C); however, mature miR-133b is significantly decreased rather than up-regulated (Supplemental Fig. 1E). miR-206, which shares the transcript with miR-133b, shows a dramatically compensatory increase in miR-133a-KO mice (Supplemental Fig. 1F), and the interesting finding reveals that the different ways of processing mature miR-133b and miR-206 may exist in skeletal muscle. In light of the impaired muscle function in miR-133a–deficient mice, the compensatory effects of miR-206 are limited in miR-133a-KO mice, which indicates that miR-206 might not be involved in the miR-133a–regulated mitochondrial biogenesis and exercise-induced muscle adaption. In addition, impaired cardiac contractility at systole was found in miR-133a-KO heart (16), but whether abnormal heart function impaired exercise performance in miR-133a-KO mice is yet to be fully understood.

In summary, we have demonstrated that loss of miR-133a leads to reduced mitochondrial biogenesis, exercise intolerance, and an impaired response to exercise training. The loss of miR-133a leads to hyperactivation of Akt that, in turn, might impair mitochondrial biogenesis and promote loss of type2B muscle fibers. Thus, miR-133a plays an essential role in muscle function and muscle plasticity, providing a potential therapeutic intervention for muscle disease.

Glossary

Cox1

cytochrome c oxidase subunit I

CS

citrate synthase

DKO

double knockout

EDL

extensor digitorum longus

ETC

electron transport chain

FoxO1

forkhead box protein O1

Igf1R

IGF-1 receptor

KO

knockout

miRNA

microRNA

Nd1

NADH-ubiquinone oxidoreductase chain 1

Nrf1

nuclear respiratory factor-1

Pgc1α

peroxisome proliferator-activated receptor-γ coactivator 1-α

Pgc1β

peroxisome proliferator-activated receptor-γ coactivator 1-β

pri-miRNA

primary microRNA

ROS

reactive oxygen species

TA

tibialis anterior

Tfam

transcription factor A, mitochondrial

WT

wild-type

AUTHOR CONTRIBUTIONS

Y. Nie, S. Kuang, and T. P. Gavin designed the research and wrote the paper; Y. Nie and Y. Sato performed the experiments; and Y. Nie, C. Wang, F. Yue, S. Kuang, and T. P. Gavin analyzed the data.