Contraction and AICAR Stimulate IL-6 Vesicle Depletion From Skeletal Muscle Fibers In Vivo

Protocols for animal use were in accordance with the guidelines of the Institutional Animal Care and Use Committee of the Joslin Diabetes Center and the National Institutes of Health. All animals were housed in a 12:12-h light:dark cycle and fed a standard laboratory chow and water ad libitum. Whole-body IL-6 knockout (B6.129S2-Il6^tm1kopf/J) and corresponding wild-type (C57BL/6J) mice were obtained from The Jackson Laboratory. Wild-type NMRI (Naval Medical Research Institute) mice were obtained from Taconic, Denmark. AMPKα2-inactive transgenic mice on an FVB background and matched controls were also studied (24).

Animals underwent a 12-h fast and were anesthetized with 90 mg/kg pentobarbital. Sciatic nerves were bilaterally isolated, and electrodes were placed around each nerve, which were connected to a Grass S88 stimulation unit. One tibialis anterior (TA) muscle was stimulated for 45 min (1 train/sec, 500-ms train duration, 100 Hz, 0.1-ms pulse duration, 1–13 V), while the muscle of the contralateral leg served as the control experiment. After the conclusion of the contraction protocol, TA muscles were removed and immediately fixed as previously described (25).

From the fixed TA muscles single fibers were isolated. Thirty fibers from each muscle were arbitrarily isolated from the muscle with fine forceps and were subjected to immunostaining as previously described (25,26). Primary antibodies against IL-6 (ab-6672) and Golgi marker 130 (GM130 [ab-40881]) were obtained from Abcam (Cambridge, MA). The secondary Alexa488 antibody (#A11008) was obtained from Molecular Probes (Eugene, OR). Confocal images were obtained blindly on a Zeiss LSM-410 confocal microscope with Argon Krypton 488-nm laser line for excitation of Alexa488. Z stacks were collected using a 63× Apochromat (1.4 NA) Zeiss oil immersion objective and a zoom of 2. Z stacks were collected at every third muscle fiber (n = 8 out of the 30 fibers arbitrarily isolated) at two different intracellular locations. The distance covered 6 µm from the interior (approximately 12 µm from the surface) toward the surface of the fiber using 8× frame image averaging.

The plasmid “IL-6-pEGFP C-fusion” was generated by PCR amplification of the IL-6 coding region from a mouse full-length EST clone (BC132458, IMAGE #40130735; Source BioScience, Nottingham, U.K.) using the following primers: 5′-GCG GAA TTC GTC AAT TCC AGA AAC CGC TA-3′and 5′-CGC CTC GAG GGT TTG CCG AGT AGA TCT CAA AGT-3′. This PCR product was cut with EcoRI and XhoI and inserted in the EcoRI and BamHI sites of pEGFP-N1 (Clontech Laboratories, Inc., Mountain View, CA). To enable ligation between the XhoI site and the BamHI site, a linker was added, which was made by hybridization between the two oligonucleotides 5′-TCGACGGATATCCGCGGGCCCGG-3′and 5′-GATCCCGGGCCCGCGGATATCCG-3′. This three fragment cloning produced a coding sequence fusion of IL-6 to the N-terminus of enhanced green fluorescent protein (EGFP). The inserted sequence was confirmed by sequencing. The Golgi marker pEYFP (enhanced yellow fluorescent protein)-Golgi was used as a nonsecreted control and was obtained from Clontech (#6909–1). DNA was grown in E. coli TOP10 cells and extracted using a Plasmid Mega Kit (Qiagen, Valencia, CA). DNA was dissolved in water.

For intravital imaging experiments, the quadriceps muscles of 7- to 9-week-old male ICR mice (Taconic, Derwood, MD) were anesthetized, the skin covering the quadriceps was opened, and the superficial muscle fibers were transfected with gene gun bombardment with 1 μg IL-6-EGFP or EYFP-Golgi cDNA/0.5 mg gold, as previously described (26,27). After transfection, the skin was closed with sutures and the animals recovered for 5 days.

Five days after transfection, the overnight fasted mice were anesthetized with phenobarbital sodium (90 mg/kg, intraperitoneally), and the skin covering the quadriceps muscle was opened to expose the quadriceps. The mice were mounted on their side in dental cement, as previously described (27,28). In the animals used for contraction-stimulation experiments, microeletrodes were placed in the groin and knee region of the quadriceps muscle, just before applying a coverglass. In the animals used for AICAR- or caffeine-stimulation experiments, a catheter was placed in a tail vein. Ten minutes after mounting, a confocal image was recorded just beneath the surface (∼3 µm below the sarcolemma) of the transfected muscle fiber. Another image 6 µm deeper inside the fiber was also recorded. There was an additional 30-min temperature and movement stabilization period and another basal pair of images were obtained and used as t = 0 (t denotes accumulated contraction time) throughout the study. The images obtained just after mounting and 30 min later did not differ significantly. The 512 × 512 pixel confocal images were collected with a 63×, 1.2 NA Zeiss C-Apochromat water immersion objective on a Zeiss LSM-410 confocal microscope with ArKr 488-nm laser line for excitation of EGFP. Green emission light was collected between 500 and 530 nm using the standard fluorescein isothiocyanate filter. Immediately after basal image pair (t = 0), direct electrical muscle stimulation was initiated (frequency 2 Hz, duration 60 ms, voltage 1.1–3 V for 3 × 5 min + 1 × 10 min eliciting repeated single contractions). The contractions of the imaged fibers were continually monitored through the oculars of the microscope. Voltage was adjusted accordingly to maintain force. Contraction periods were separated by 90 s of rest. Confocal images were collected at the end of each contraction period. AICAR was given as an intravenous bolus (1 g/kg), and images were collected every 10 min after injection. In the animals given caffeine stimulation, a continuous infusion was given intravenously for 20 min (85 mg/kg) followed by monitoring until 70 min after infusion. The caffeine concentration was modeled after a caffeine concentration (510 mg/kg) previously shown to increase glucose transport in incubated muscle without contractions (29). However, infusing this higher concentration in vivo would result in whole-body contractions, instability, or death of the mouse, thereby interfering with the experiment and imaging.

Images obtained in immunostained muscle fibers were obtained as TIFF images with the Zeiss confocal software and were imported into Metamorph Software (V. 6.1; Universal Imaging Corp., West Chester, PA). Image stacks were created, and maximal and average projections were created. The total number, average gray value intensity, and average area of IL-6 and GM130 vesicular staining were quantified throughout the field of view in the muscle fiber. For the time-lapse image stacks of IL-6-EGFP or EYFP-Golgi, the threshold and the classifying settings of the Metamorph software were adjusted to count the IL-6 vesicles and avoid nonspecific background. Because of variation in the level of IL-6-EGFP or EYFP-Golgi expression between individual transfected fibers, the actual vesicle count divided by the vesicle count at t = 0 (T/T₀) is shown.

Vesicle counts obtained from fibers in Metamorph were imported into Sigma plot 10.0 and subjected to paired t test or one way ANOVA against ROI fluorescence values at t = 0 (T/T₀).

To determine if mouse TA muscle fibers contain IL-6 protein, we analyzed the intracellular localization of endogenous IL-6 by immuno-staining of intact fixed fibers (Fig. 1). Briefly, mice were anesthetized and the sciatic nerve was attached to electrodes, and the TA muscle in one leg was contracted in situ for 45 min. The contralateral leg was used as a resting, basal control. After contractions, both muscles were excised and fixed. Single muscle fibers were teased from both muscles and immunostained for IL-6. In the TA muscle fibers from the rested leg, IL-6 protein was observed in discrete, dot-like vesicular structures near the sarcolemma (Fig. 1A, vertical arrow) and inside the fibers in the area of the T-tubule membranes (Fig. 1A, diagonal arrow). The IL-6–positive vesicles were not present in muscle fibers from whole-body IL-6 knockout mice (Fig. 1C), demonstrating the specificity of staining. In the TA muscles from the contracted leg, the number of IL-6–positive vesicles within the muscle fibers was significantly reduced (Fig. 1B, arrows), which corresponded to a 62% reduction in IL-6 vesicles throughout the muscle fibers (Fig. 1D). In order to rule out the possibility that the decrease in IL-6 vesicle number was a result of vesicle fusion, average vesicle intensity (Fig. 1E) and average vesicle size (Fig. 1F) were quantified. There was no significant difference in IL-6 vesicle intensity or average area. These data demonstrate that IL-6 protein is abundant in vesicular structures inside mouse muscle fibers and that the number of intramyofibril IL-6 vesicles is reduced following muscle contractions.

To ensure that the decrease in IL-6 vesicles was not caused by a general contraction-induced protein leak from the muscle fibers, we stained for an endogenous nonsecreted protein, GM130. GM130 is a widely used marker that is bound to the Golgi membranes helping to maintain the cis-Golgi structure (30). Under resting conditions, GM130-positive structures were located in dot-like vesicles throughout the muscle fiber at both the sarcolemma and T-tubule regions (Fig. 2A, arrows). The localization, number, intensity, and average area of GM130-positive vesicle structures did not change during 45 min of in situ contractions (Fig. 2B, arrows; Fig. 2C). Thus, in contrast to IL-6 (Fig. 1), the GM130-positive structures are not reduced during fiber contraction. These data indicate that the contraction-induced reduction in IL-6–positive vesicles is not due to a general protein leak from the muscle fibers (Fig. 2).

To determine the kinetics of contraction-mediated IL-6 vesicle reduction, we analyzed IL-6 vesicle localization and number in muscle fibers before, during, and after a bout of muscle contractions in situ. We expressed IL-6-EGFP in the superficial portion of the quadriceps muscle in living mice using gene gun transfection (26–28,31–33). Five days later, mice were anesthetized and the transfected fibers were subjected to intravital imaging (34). An image was collected just beneath the sarcolemma surface (fiber surface position) along with another image 6 µm inside the fiber (fiber interior position). During basal conditions, IL-6-EGFP was localized in vesicular structures at the surface position of the transfected muscle fiber (Fig. 3A, t = 0) and throughout the interior location (Fig. 3B, t = 0). Thus, IL-6-EGFP–positive vesicles were located similarly to the endogenous IL-6–positive vesicles described in Fig. 1. Immediately after recording the basal (t = 0) images, the muscle fibers were subjected to in situ contractions for 3 × 5 min and 1 × 10 min, with each contraction bout separated by 90 s of rest. Images were obtained between each contraction bout at both surface (Fig. 3A) and interior positions (Fig. 3B). The number of IL-6-EGFP vesicles in both positions was reduced by approximately 50% during the first 15 min of contractions (Fig. 3C), and the last 10 min of contraction further reduced IL-6-EGFP vesicles to 40% of basal (Fig. 3C). The quantitative data are similar to the reduction of endogenous IL-6 found by immunostaining (Fig. 1). To confirm the imaging results of contraction-mediated IL-6 depletion from single muscle fibers on a whole-muscle level, IL-6-EGFP was expressed in gastrocnemius muscles of NMRI mice using in vivo electroporation. In situ contractions resulted in a 68% reduction in IL-6-EGFP protein (Fig. 3D, P = 0.02, n = 3), comparable to the reduction in vesicle number observed using imaging.

Taken together, these intravital images show that IL-6 is localized to vesicular structures throughout the muscle fibers and that contractions induce a continuous reduction of IL-6 within the muscle fibers over time (Fig. 3).

Similar to the experiments of GM130 immunostaining, we performed in vivo imaging of a fluorescent tagged trans-Golgi marker 1,4-galactosyltransferase (35) fused N-terminally to EYFP (EYFP-Golgi). This was done to ensure that the disappearance of IL-6-EGFP was not due to a contraction-induced protein leak from muscle fibers during in vivo contractions. In the basal state, EYFP-Golgi was localized in a dotted vesicular like pattern (Fig. 4A and B; t = 0 in B) similar to GM130 (Fig. 2). There was no reduction in EYFP-Golgi after 25 min of in situ contractions (Fig. 4A–C; t = 5–25 in B).

Incubation of intact soleus and extensor digitorum longus muscles in vitro with the AMPK activator AICAR for 2 h has been shown to inhibit IL-6 production (23), whereas AICAR incubation of cultured human myotubes for 24 h increases IL-6 mRNA (21,36). Here, we determined if short term AICAR infusion results in the release of IL-6 from muscle fibers. Quadriceps muscles were transfected with IL-6-EGFP, and 5 days later mice were injected intravenously with a bolus of AICAR (1g/kg). Changes in IL-6-EGFP were measured using time-lapse imaging of IL-6-EGFP in situ. AICAR injection resulted in a gradual decrease in IL-6-EGFP vesicles at both surface (Fig. 5A) and interior (Fig. 5B) positions of the muscle fibers. The reduction of IL-6-EGFP was specific to AICAR injection, since saline did not result in a significant reduction of IL-6-EGFP (Fig. 5C and D). The decrease in IL-6-EGFP reached a plateau 70 min after injection, with an approximately 50% reduction in both positions (Fig. 5C and D). To determine if AMPK activity mediated the effects of AICAR on IL-6, we used muscle-specific transgenic mice that express inactive AMPKα2, the major AMPK catalytic isoform expressed in skeletal muscle. AICAR did not decrease IL-6-EGFP vesicle number in muscle-specific AMPKα2-inactive mice (Fig. 5E and F). AICAR had no effect on the Golgi marker EYFP-Golgi (Fig. 6A–D). These results indicate that AICAR decreases IL-6-EGFP vesicle number in muscle fibers to a degree comparable to that of in situ contractions (Fig. 4) and that AMPKα2 mediates AICAR-stimulated IL-6-EGFP vesicle reduction.

Calcium is essential for excitation-contraction coupling in muscle and is a well-established trigger signal for vesicle reduction in many secretory cell types (37). To determine if calcium mediates IL-6 vesicle reduction, the calcium releasing agent caffeine was infused intravenously from 0–20 min at the highest dose (85 mg/kg) possible without inducing whole-body muscle contractions and larger movements due to the systemic delivery in the living mice. Caffeine had no effect on IL-6-EGFP or Golgi marker EYFP-Golgi vesicles during the 20 min of caffeine infusion or during the 50-min period after infusion (Fig. 7A–C). In contrast, the caffeine infusion led to significant increases in phosphorylation of Ca²+/calmodulin-dependent protein kinase II both during and after infusion, indicating that the caffeine infusion protocol had an effect at the cellular level (Fig. 7E).

Our results with AICAR and caffeine led us to hypothesize that AMPK, but not calcium signaling, is important for contraction-mediated IL-6 reduction. Therefore, we used the muscle-specific AMPKα2 inactive transgenic mice to test this hypothesis. Under basal conditions, IL-6-EGFP–positive vesicles were similarly distributed throughout the muscle fibers when comparing control mice with AMPKα2-inactive transgenic mice (Fig. 8A and B). Muscle contractions similarly reduced IL-6-EGFP vesicles at the surface and interior positions in both wild-type and transgenic mice, demonstrating that the contraction-induced reduction of IL-6 in AMPKα2 inactive transgenic was fully intact (Fig. 8A and B).

Physical exercise is an important modality to improve whole-body metabolic state and insulin sensitivity. It has been postulated that IL-6 secretion from exercising skeletal muscle contributes to the exercise-induced changes in whole-body metabolism, making IL-6 essential for the beneficial effects of exercise on glucose homeostasis (38). However, data establishing the presence and/or secretion of IL-6 in fully differentiated skeletal muscle has been inconclusive (11–13). Previous studies using lower resolution imaging of sectioned skeletal muscle suggested that under basal conditions there is very little IL-6 protein expression within the muscle fibers (11–13). Here, we imaged single intact fixed muscle fibers or living muscle fibers at high resolution and detected significant amounts of intracellular IL-6–positive vesicles under resting conditions at both the sarcolemma and T-tubule regions. The vesicles were specific for IL-6, as these vesicles were not present in the muscle fibers of IL-6 knockout mice. Thus, these data clearly establish that IL-6 is present within skeletal muscle fibers.

Muscle contraction and AICAR stimulation reduced IL-6–positive vesicles from both surface and interior positions within living muscle fibers. Because muscle fibers have a T-tubule network that plays a central role in transmembrane substrate and hormone exchange with serum (27,31,39), the localization of IL-6 inside the muscle fiber in the vicinity of the T-tubule membranes is logical, since this would increase the surface area available for efficient release of IL-6. Consistent with this hypothesis, contraction-induced IL-6 vesicle reduction was rapid, with a significant decrease in IL-6 vesicles occurring after only 5 min of muscle contractions. Our findings are compatible with a secretory function of muscle fibers and demonstrate that skeletal muscle fibers function as classic endocrine cells containing depots of IL-6–positive intracellular vesicles that are reduced upon contraction stimulation. The concept of skeletal muscle fibers as a secretory cell is also supported by data showing that muscles express vesicle-associated membrane protein family members localized to GLUT4 vesicles, and these vesicles are involved in regulated secretion (40). Although the current study presents many lines of evidence that IL-6 vesicles are undergoing stimuli depletion from muscle fibers, consistent with a release of IL-6 into serum, our imaging method cannot visually show a release of IL-6. However, all of the data support that this is the case. In addition to the IL-6 imaging data, we have also shown that the Golgi marker is not degraded with any of the stimulation protocols, making it unlikely that IL-6 undergoes nonspecific degradation or leakage. Furthermore, our findings that certain stimuli don’t decrease IL-6 vesicle number (AICAR in AMPKα2-inactive transgenic mice, caffeine) also suggest that IL-6 vesicle reduction is specific and not the result of a general, stimuli-induced degradation.

We found that the maximal degree of reduction in number of IL-6 vesicles in the muscle fibers was ∼60%, regardless of the intensity and length of the contraction protocol. In contrast, the Golgi marker did not change with any of the protocols. IL-6 vesicle reduction was found without a change in IL-6 vesicle intensity or average area and was confirmed by a similar reduction in IL-6-EGFP protein, indicating that vesicle fusion or redistribution is not the cause of a reduced IL-6 signal. The maximal decrease with AICAR stimulation was similar, with a 50% reduction of IL-6 vesicle content. Interestingly, we have previously reported that muscle contraction results in a maximal reduction in GLUT4 vesicle depots of approximately 70% (34), comparable to maximal IL-6 reduction. These findings suggest there is a limit to the degree of decrease in protein-containing vesicles in skeletal muscle fibers in the order of approximately 60–70%. A maximal limit for secretory vesicle reduction has been described in other cell types such as insulin-secreting pancreatic cells (5–30%) (41), neuroendocrine cells (25%) (42), growth hormone–secreting cells (5–25%) (43), mast cells (30–40%) (44), and chromaffin cells (30%) (45). Lower maximal values for other cell types suggest that the secretory function of muscle may be higher. The higher capacity in muscle may enable protein secretion even under conditions where a low number of muscle fibers or mass of muscles are contracting. This would allow for secretory functions such as myokine secretion or GLUT4 vesicle movement to occur even under conditions such as low-intensity exercise.

Even though our data showing a contraction-mediated IL-6 vesicle reduction is in line with the classical concept of secretory cell kinetics, our findings are in contrast to a study analyzing muscle IL-6 before and after exercise (11). Using immunohistochemistry of muscle sections from human biopsies, it was reported that IL-6, localized primarily in the sarcolemma region, was increased 2 h after bicycle exercise (11). The discrepancy with our data, which revealed a decrease in IL-6 signal, could be due to differences in methods because the previous study was based on muscle biopsies and not intact muscle in vivo (11). Interestingly, mouse plantaris muscles subjected to surgical injury showed significant IL-6 protein in the sarcolemma region (10), and it has also been reported that the main source of IL-6 within the muscle in response to muscle biopsy or muscle isolation may be from circulating monocytes such as macrophages (14). Thus, the increase in IL-6 protein observed in the human muscle biopsies after exercise could result from the surgical procedure combined with increased macrophage migration during the exercise bout. Furthermore, an increase in IL-6 protein in the muscle fibers simultaneous with an increase in the circulation is not consistent with classical protein secretion consisting of a net reduction of secretory content from the endocrine cell.

The similarities between our findings with muscle fibers and classical endocrine cells led us to investigate a potential role for calcium in IL-6 vesicle reduction. Although intracellular calcium release is a well-established signal for release of vesicles in secretory cells, we found that caffeine infusion, which resulted in a significant increase in intracellular Ca²+/calmodulin-dependent protein kinase II phosphorylation, had no effect on IL-6 vesicle distribution. Our results are in line with a recent study in incubated mouse muscles showing that calcium release had no effect on IL-6 mRNA production (23). The lack of regulation by calcium is plausible given that even single twitches result in the release of calcium within the myofibers, making discriminate regulation of IL-6 release impossible. Thus, it is likely that calcium signaling does not function in the regulation of IL-6 vesicle reduction in skeletal muscle.

AMPK activation has been hypothesized to regulate, in whole or in part, many of the metabolic and transcriptional responses to exercise, including GLUT4 translocation and glucose transport (46), fatty acid oxidation (47,48), mitochondrial biogenesis (49–51), expression of peroxisome proliferator–activated receptor γ coactivator-1α (51), and fiber type transformation (52). AMPK has also been proposed to be involved in contraction-stimulated IL-6 expression and secretion (19). Bicycle exercise for 60 min was shown to increase both AMPK activity in muscle and IL-6 in serum (19); however, whether AMPK mediates IL-6 vesicle reduction has not been directly tested. Our current finding, that AICAR injection decreases IL-6 vesicles in an AMPKα2-dependent manner, along with our previous work showing that this same AICAR treatment also causes AMPK activation (53), is consistent with the hypothesis that AMPK signaling induces IL-6 vesicle reduction in skeletal muscle. However, the normal contraction-stimulated IL-6 vesicle reduction is intact in AMPKα2-inactive mice. Thus, these results clearly show that AMPKα2 activity is not essential for contraction-stimulated IL-6 reduction. This result is similar to studies in which contraction-mediated glucose transport (24,54,55), GLUT4 translocation (34), and fatty acid oxidation (56) are still intact in animal models of reduced AMPK activity (18). Thus, redundant signaling pathways may also exist for the regulation of IL-6 vesicle reduction with muscle contractions.

In conclusion, we have developed a novel image-based system to analyze the kinetics of vesicular-containing proteins in living skeletal muscle. As new myokines are proposed and their function in metabolic and tissue homeostasis emerges, this methodology will provide a valuable tool to determine the intracellular source and depletion kinetics of these putative myokines. This study provides the first data that clearly demonstrate that intact muscle fibers act as endocrine cells in response to muscle contractions in vivo by rapid, stimuli-dependent decreases in the secretory vesicle content.

This project was supported by the National Institutes of Health (NIH) grants R01AR45670 and R01DK68626 (L.J.G.) and the DERC P30DK036836 at the Joslin Diabetes Center. H.P.M.M.L. was supported by the Weimann Foundation, the Beckett Foundation, and the Danish National Research Foundation. J.B. was supported by an NIH Training grant (T32-DK-07260-29) and a Gettysburg College Professional Development grant. J.T.T. was supported by the Novo Nordisk Foundation Center for Basic Metabolic Research. No other potential conflicts of interest relevant to this article were reported.

H.P.M.M.L. designed and performed experiments and wrote the manuscript. J.B. designed and performed experiments and edited the manuscript. P.S. constructed the IL-6-EGFP plasmid and reviewed and edited the manuscript. H.-J.K. performed experiments. H.G. contributed to discussion and reviewed and edited the manuscript. M.F.H. and J.T.T. contributed to the experiments. L.J.G. designed experiments and wrote and edited the manuscript. L.J.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.