Endocrinology and Metabolism

AMP-activated protein kinase agonists increase mRNA content of the muscle-specific ubiquitin ligases MAFbx and MuRF1 in C2C12 cells

Brian J. Krawiec, Gerald J. Nystrom, Robert A. Frost, Leonard S. Jefferson, Charles H. Lang

Abstract

The hypothesis of the present study was that exposure of differentiated muscle cells to agonists of the AMP-activated protein kinase (AMPK) would increase the mRNA content of the muscle-specific ubiquitin ligases muscle atrophy F-box (MAFbx) and muscle RING finger 1 (MuRF1). C2C12 cells were incubated with incremental doses of 5-aminoimidazol-4-carboximide ribonucleoside (AICAR) or metformin for 24 h. Both MAFbx and MuRF1 mRNA increased dose dependently in response to these AMPK activators. AICAR, metformin, and 2-deoxy-d-glucose produced time-dependent alterations in ubiquitin ligase expression, typified by a biphasic pattern of expression marked by an acute repression followed by a sustained induction. AMPK-activating treatments in conjunction with dexamethasone produced a pronounced synergistic effect on ligase mRNA expression at later time points. This cooperative response occurred in the absence of a dexamethasone-dependent increase in AMPK expression or activity, as determined by immunoblotting for phosphorylation and expression of AMPKα and its downstream target acetyl-CoA carboxylase (ACC). These responses elicited by AMPK activation singly or in combination with dexamethasone did not extend to the mRNA expression of the UBR box family E3s UBR1/E3αI and UBR2/E3αII. Treatment with the AMPK inhibitor compound C prevented increases in MAFbx and MuRF1 mRNA in response to serum deprivation, as well as AICAR and dexamethasone treatment individually or jointly. Stimulation of AMPK activity in vivo via AICAR injection increased both MAFbx and MuRF1 mRNA in murine skeletal muscle. These data suggest that activation of AMPK in skeletal muscle results in a specific upregulation of MAFbx and MuRF1, responses that are reminiscent of the proposed atrophic transcriptional program executed under various conditions of skeletal muscle wasting. Therefore, AMPK may be a critical component of the intercalated network of signaling pathways governing skeletal muscle atrophy, where its input acts to modify anti- and proatrophic signals to influence gene expression in reaction to catabolic perturbations.

  • atrogin-1
  • proteolysis
  • 2-deoxyglucose

for proper growth and survival, all organisms require means of responding to diverse inputs originating from a discontinuous environmental milieu. Failure to both properly integrate these signals and accordingly adjust metabolism within favorable physiological limits may ultimately result in maladaptation, disease, and death. One critical component of this control system in mammalian systems is the AMP-activated protein kinase (AMPK). AMPK functions as a heterotrimer (consisting of a catalytic α-subunit and regulatory β- and γ-subunits), which senses alterations in the AMP-to-ATP ratio within cells (15, 25, 35). In response to energy-depleting stressors (which consequently increase cellular AMP concentrations), AMPK acts to balance energy consumption with production by suppressing ATP-expensive processes and activating ATP-repleting ones. Regulation of its activity is achieved in an elegant fashion through several AMP-dependent mechanisms (15). Binding of AMP to the γ-subunit allosterically activates the kinase, makes the kinase a better substrate for upstream AMPK kinases, and prevents dephosphorylation of the kinase by phosphatase activity. Conversely, high ATP concentrations inhibit these stimulatory effects of AMP, perhaps via competitive and mutually exclusive binding of AMP vs. ATP to the γ-subunit (46). These mechanisms consort to make AMPK sensitive to a critical threshold AMP concentration (16), above which it endeavors to drive the system back toward energetic homeostasis.

AMPK has proven to be an essential intermediate in the control of fundamental cellular processes such as growth (22, 23), proliferation (24), and survival (7, 23). In addition, AMPK is equally important in orchestrating multiple signaling pathways controlling nutrient uptake and fuel metabolism in many tissue types (15, 25, 35). Furthermore, AMPK plays a crucial underlying role in more complex physiological and behavioral phenomena, such as interorgan communication via various cytokines and adipokines (25, 35), and control of feeding behavior (25, 35), voluntary energy expenditure (38), and cognitive ability (7). AMPK induces these and other effects at a cellular level primarily through two means: through direct phosphorylation of rate-limiting or otherwise strategic components involved in pathways of metabolic control and through a less-well understood control of gene expression. The outcome of AMPK activation on the expression of specific genes is, in general, consistent with the corresponding metabolic responses and tissue adaptations observed after acute and chronic activation of the kinase. The preponderance of data has focused on either the repression of gluconeogenic (1, 27, 34, 47) and lipogenic (26, 47, 56, 59) genes in the liver or the induction of genes involved in mitochondrial biogenesis (2, 39, 55, 60), glucose and lipid metabolism (39, 40, 51), and glucose transport (18, 58) in skeletal muscle. However, despite its established role in altering protein balance by promoting catabolism through its suppression of both mammalian target of rapamycin (mTOR) signaling (5, 22, 23) and translational elongation (20) and its stimulation of macroautophagy (36), little is known regarding the ability of AMPK to influence the representation of specific genes involved in protein breakdown.

Skeletal muscle gene expression patterns have proven to be especially responsive to diverse atrophic conditions. This reaction is typified by increased expression of genes involved in ubiquitin conjugation and proteolysis through the ubiquitin-proteasome pathway (UPP) (3, 13, 30). The strict regulation and specificity that epitomize the UPP are largely dependent on the function of E3 ubiquitin ligases. These enzymes recognize a specific degron (or degradation signal) of a protein, bind it, and catalyze the ultimate step in targeting: the covalent attachment of the ubiquitin moiety to an internal lysine residue or NH2-terminal amino group of the protein (12). Two muscle-specific ubiquitin ligases, muscle atrophy F-box (MAFbx; also known as atrogin-1) and muscle RING finger 1 (MuRF1), are proposed to be archetypal markers for muscle wasting because they are upregulated in a multitude of catabolic perturbations (3, 13, 30). The necessity of these aforementioned E3s in skeletal muscle atrophy has been demonstrated previously where their genetic ablation rendered muscle refractory to a substantial portion of muscle loss after denervation (3). Of particular note are the findings that demonstrate sensitivity of one or both of their gene expressions to conditions of nutrient and energy deprivation in vivo (3, 13, 30) and in vitro (44).

Despite ample evidence implicating MAFbx and MuRF1 as contributing factors to the (patho)physiological response of skeletal muscle to many atrophic stimuli, a complete understanding of the initiating stimuli and cellular signaling underlying their induction is still lacking. Given the pivotal roles AMPK plays in the adaptive responses to energy insufficiency, it is reasonable to suspect its activation may be capable of regulating the expression of one or more of these genes. Therefore, the purpose of these studies was to examine changes in mRNA content of specific ubiquitin ligases complicit in atrophy in response to AMPK activation using in vitro and in vivo experimental approaches.

MATERIALS AND METHODS

Cell culture.

C2C12 myoblasts (American Type Culture Collection, Manassas, VA) were maintained in Eagle's minimum essential medium (EMEM) supplemented with 10% FBS, penicillin (100 IU/ml), streptomycin (100 μg/ml), and amphotericin (250 ng/ml) (all from Mediatech, Herndon, VA) under 5% CO2 at 37°C. For experimental treatments, myoblasts were subcultured into six-well tissue culture plates (Grenier Bio-One, Frickenhausen, Germany). At ∼100% confluence, the cells were switched to medium consisting of EMEM with the above antibiotics-antimycotics and 10% bovine calf serum (Hyclone, Logan, UT) to promote myoblast fusion and differentiation to myotubes. Cells were allowed to differentiate for 4 days before experimental manipulation. Myotubes were provided with fresh differentiation medium for 2 h immediately preceding treatment on the 5th day. All experiments were performed with serum-free EMEM plus antibiotics-antimycotics. 5-Aminoimidazol-4-carboximide ribonucleoside (AICAR; Toronto Research Chemicals, Ontario, Canada), 1,1-dimethylbiguanide hydrochloride (metformin), 2-deoxy-d-glucose (2-DG), d-mannitol, dexamethasone (all from Sigma-Aldrich, St. Louis, MO), and/or compound C (Calbiochem, San Diego, CA) were administered singly or in combination at concentrations and times specified in the figures and text. Dexamethasone and compound C were solubilized in 100% ethanol and 100 mM HCl, respectively. Under no condition did the volume of solvent exceed 0.25% of the volume of the culture medium and have a significant effect on the parameters investigated (data not shown).

Animals.

C57BL/6 mice were obtained from Charles Rivers Laboratories (Wilmington, MA). All mice were housed in a controlled environment and provided water and standard rodent chow (Harlan Teklad, Indianapolis, IN) ad libitum for 1 wk before use. At the time of the study, mice were 8–9 wk of age and weighed 22.8 ± 0.4 g. All experiments were performed in adherence with the National Institutes of Health “Guide for Care and Use of Laboratory Animals” and with the approval of The Pennsylvania State University College of Medicine Institutional Animal Care and Use Committee. On the morning of study, fed mice were injected intraperitoneally with AICAR (1 mg/g body wt; 0.5 ml/mouse) or an equivalent volume of isotonic saline. Six hours after the injection of AICAR or saline, mice were anesthetized with ketamine-xylazine (90 and 9 mg/kg, respectively), and the gastrocnemius-plantaris complex was excised and frozen in liquid nitrogen.

Multiprobe template production for RNase protection assay.

Primer selection for mouse genes of interest was determined with the help of GeneFisher software (11). The lengths of amplified regions were chosen to allow distinct resolution during electrophoretic separation. Primers were synthesized (IDT, Coralville, IA) with restriction sites for EcoRI or KpnI at the 5′ end and with three extra bases at the extreme 5′ end as follows: for MAFbx/atrogin-1, forward = 5′-GCA GAA TTC CAC ATC CTT ATG CAC ACT GGT GCA-3′ and reverse = 5′-GCA GGT ACC GGT ACT GGC AGA GTC TCT TCC ACA-3′; for MuRF1, forward = 5′-GCA GAA TTC AGT GTG TCT TCT CTC TGC TCA GAG A-3′ and reverse = 5′-GCA GGT ACC AGA CCC AGC CCT CCC ACC AA-3′; for UBR1/E3αI, forward = 5′-GCA GAA TTC CCC TAA CCC AGC ACA GAG GGA A-3′ and reverse = 5′-GCA GGT ACC ACT TGC AGA GCG GGC ATA GGT A-3′; for UBR2/E3αII, forward = 5′-GCA GAA TTC CTG AAG TGC ATG CAG GGA ATG GA-3′ and reverse = 5′-GCA GGT ACC CCA CCG AGT GTC CAC AAA TAC TGA-3′; for L32, forward = 5′-GCA GAA TTC CGG CCT CTG GTG AAG CCC AA-3′ and reverse = 5′-GCA GGT ACC CCT TCT CCG CAC CCT GTT GTC A-3′. PCR was conducted with HotStarTaq DNA polymerase (Qiagen, Valencia, CA), and mouse total RNA was reverse transcribed with Superscript first-strand synthesis system for RT-PCR (Invitrogen, Carlsbad, CA). PCR products were phenol-chloroform extracted, ethanol precipitated, and sequentially digested with KpnI and EcoRI (Promega, Madison, WI). Digested products were gel purified, reextracted, and cloned into KpnI/EcoRI-digested pBluescript II SK+ (Stratagene, La Jolla, CA). Plasmid DNA was isolated with both QIAprep spin miniprep and plasmid maxi kits (Qiagen). Plasmids with inserts were verified by sequencing in the Pennsylvania State College of Medicine Molecular Genetics Core Facility. Final constructs were linearized with EcoRI, gel purified, and quantitated spectrophotometrically. The template was prepared so that a 2-μl aliquot contained 10 ng of MAFbx/atrogin-1, 30 ng of MuRF1, 10 ng of UBR1/E3αI, 10 ng of UBR2/E3αII, and 20 ng of L32.

RNA extraction and RNase protection assay.

Total RNA was extracted from cells or powdered muscle tissue using Tri reagent (Molecular Research Center, Cincinnati, OH). mRNA expression was determined by RNase protection assay. A 2-μl aliquot of template was prepared with T7 polymerase with buffer (Fermentas, Hanover, MD), NTPs, and tRNA (Sigma-Aldrich), RNasin and DNase (Promega), and [32P]UTP (Amersham Biosciences, Piscatawy, NJ). Unless otherwise noted, the entire RNase protection assay procedure, including labeling conditions, component concentrations, sample preparation, and gel electrophoresis, was as published (BD Pharmingen, San Diego, CA). Hybridization buffer was 80% formamide and 20% stock buffer (200 mM PIPES, pH 6.4, 2 M NaCl, and 5 mM EDTA). Hybridization proceeded overnight at 56°C in a dry bath incubator (Fisher Scientific, Pittsburgh, PA) without the use of mineral oil. Samples were treated with RNase A+T1 (Sigma) in 1× RNase buffer (10 mM Tris·HCl, pH 7.5, 5 mM EDTA, and 300 mM NaCl) followed by proteinase K (Fisher Scientific) in 1× proteinase K buffer (50 mM Tris, pH 8.0, 1 mM EDTA, 1% Tween 20). After ethanol precipitation, samples were resuspended in 5 μl of loading buffer [98% formamide (vol/vol), 0.05% xylene cyanol (wt/vol), 0.05% bromphenol blue (wt/vol), and 10 mM EDTA]. Polyacrylamide gels (34 × 45 cm) were run at 75 W for 70 min in an S3S sequencing system (Owl Separation Systems, Portsmouth, NH), transferred to chromatography paper, and dried for 10 min at 80°C (FB GD 45 gel dryer; Fisher Scientific). Gels were exposed to a PhosphorImager screen (Molecular Dynamics, Sunnyvale, CA). Data were visualized and analyzed by ImageQuant software (version 5.2; Molecular Dynamics). Signal densities for mRNAs were normalized to densities for L32 mRNA.

Immunoblot analysis.

After drug treatment, cells were rinsed with cold Dulbecco's PBS (Invitrogen) and collected on ice in lysis buffer (20 mM HEPES, 50 mM β-glycerophosphate, 1% Triton X-100, 100 mM KCl, 2 mM EDTA, 50 mM NaF, 1 mM DTT, 0.5 mM PMSF, 1 mM benzamidine, 1 mM sodium orthovanadate, and 2 μg/ml leupeptin). Lysates were then passed several times through a 27-gauge needle and centrifuged at 1,500 g for 10 min at 4°C. A portion of the resulting cell supernatant was used to determine protein concentration via a bicinchoninic acid assay kit (Pierce, Rockford, IL). Sample buffer (5×) was added to an aliquot of supernatant.

Samples were loaded according to total protein content (20 μg) on polyacrylamide gels for separation by SDS-PAGE. Proteins were transferred to PVDF membrane (Biotrace; PALL, Pensacola, FL), blocked in nonfat dry milk, and incubated overnight at 4°C with phosphospecific antibodies for AMPKα (Thr172) and acetyl-CoA carboxylase (ACC; Ser79) (both from Cell Signaling Technology, Beverly, MA). Excess primary antibody was removed by washing in 1× TBS + 0.1% Tween 20, and membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse secondary antibody (Sigma-Aldrich) at room temperature. Blots were developed with enhanced chemiluminescence (Amersham Biosciences) in accordance with the manufacturer's instructions and exposed to BioMax XAR X-ray film (Kodak, Rochester, NY) in a cassette equipped with a DuPont Lightning Plus intensifying screen. Developed film was scanned (ScanMaker IV; Microtek USA, Carson, CA).

After development, antibody was removed from membranes by treatment with a solution containing 62.5 mM Tris, pH 6.8, 2% (wt/vol) SDS, and 100 mM β-mercaptoethanol in a 50°C water bath for 15 min. Blots were then blocked with nonfat dry milk and incubated overnight at 4°C with antibodies for AMPKα and ACC (both from Cell Signaling Technology). An antibody against β-tubulin (Santa Cruz Biotechnology, Santa Cruz, CA) served as a control for equal protein loading of samples. Membranes were then processed as above.

Statistics.

Results for individual cell experiments were replicated in at least three independent experiments and (when applicable) are presented as means ± SE calculated from the pooled data. Data were analyzed by unpaired Student's t-test in two-group comparisons and ANOVA with a Student-Neuman-Keuls posttest in multigroup comparisons to determine treatment effect when ANOVA indicated a difference among the means. Differences between groups were considered significant at P < 0.05.

RESULTS

AMPK agonists AICAR and metformin induce muscle-specific ubiquitin ligase mRNA content in a dose-dependent fashion.

AICAR is a pharmacological agent commonly used to artificially activate AMPK (6). Once taken up by intact cells, AICAR is phosphorylated to form 5-aminoimidazole-4-carboxamide ribonucleoside monophosphate, which is capable of producing stimulatory effects identical to those of AMP on AMPK, but in the absence of detectable changes in adenine-nucleotide levels. To assess the sensitivity of ubiquitin ligase mRNA to AMPK signaling, C2C12 cells were exposed to increasing concentrations of AICAR for 24 h (Fig. 1). Both MAFbx (Fig. 1A) and MuRF1 (Fig. 1B) increased dose dependently in response to increasing concentrations of AICAR up to 1 mM; exposure of cells to concentrations >2 mM resulted in substantial loss of cell viability (unpublished observations). Effective doses of AICAR concomitantly induced phosphorylation of the α-subunit of AMPK at the Thr172 residue (Fig. 1C), a signaling event shown to be requisite for nearly all AMPK activity (49). Activation of AMPK in response to AICAR was further corroborated by a slight gel-shift in an immunoblot for total AMPKα protein (Fig. 1C). Furthermore, the phosphorylation of ACC, a well-established target of AMPK, is frequently used as an indirect measure of AMPK activity. Phosphorylation of ACC was increased at all doses capable of inducing ligase expression (Fig. 1C) in a manner parallel to AMPKα phosphorylation. In both cases, maximal expression was achieved with a concentration of 1 mM, which was subsequently used for all further studies involving AICAR.

Fig. 1.

5-Aminoimidazol-4-carboximide ribonucleoside (AICAR) increases muscle atrophy F-box (MAFbx) and muscle RING finger 1 (MuRF1) mRNA in a dose-dependent manner. A, top: quantitation of MAFbx mRNA. The value for AICAR-free control is set at 1.0 arbitrary units (AU). A, bottom: representative autoradiographs of MAFbx and L32. B, top: quantitation of MuRF1 mRNA. The value for AICAR-free control is set at 1.0 AU. B, bottom: representative autoradiographs of MuRF1 and L32. Results in bar graphs are means ± SE for n = 6/group. Means with different letters are statistically different from one another (P < 0.05). C: representative immunoblots of phosphorylation (P) at the Thr172 site of AMPKα, total AMPKα, phosphorylation at the Ser79 site of acetyl-CoA carboxylase (ACC), total ACC, and β-tubulin.

The conclusiveness of data obtained regarding AMPK-dependent responses to AICAR are limited by the potential non-AMPK-specific “side effects” of the treatment (6). To confirm the role of AMPK in modulating ligase expression, cells were also treated with metformin (Fig. 2). Metformin is an antidiabetic biguanide that activates AMPK in a mechanistically distinct manner from that of AICAR (see discussion below). Similarly to AICAR treatment, exposure of myocytes to increasing concentrations of metformin up to 2 mM dose dependently increased mRNA for MAFbx (Fig. 2A). Doses above 2 mM compromised cell survival (unpublished observations). MuRF1 expression, which was only induced with the highest tolerated concentration of metformin (Fig. 2B), appeared to be the notable exception to this recurring pattern of dose dependence observed in AICAR- and metformin-treated cells. Effective doses again correlated with Thr172 phosphorylation of AMPKα, gel shifting of native AMPK, and Ser79 phosphorylation of ACC (Fig. 2C). Maximal expression for both MAFbx and MuRF1 was achieved with a concentration of 2 mM; consequently, all further studies utilizing metformin were performed with this dose.

Fig. 2.

Metformin (Met) increases MAFbx and MuRF1 mRNA in a dose-dependent manner. A, top: quantitation of MAFbx mRNA. The value for metformin-free control is set at 1.0 AU. A, bottom: representative autoradiographs of MAFbx and L32. B, top: quantitation of MuRF1 mRNA. The value for metformin-free control is set at 1.0 AU. B, bottom: representative autoradiographs of MuRF1 and L32. Results in bar graphs are means ± SE for n = 8 or 9 per group. Means with different letters are statistically different from one another (P < 0.05). C: representative immunoblots of phosphorylation at the Thr172 site of AMPKα, total AMPKα, phosphorylation at the Ser79 site of ACC, total ACC, and β-tubulin.

AICAR increases MAFbx and MuRF1 mRNA content time dependently in a biphasic pattern.

Given the substantial induction of MAFbx and MuRF1 after prolonged exposure to activators of AMPK, the acute and intermediate kinetics of these agents on ligase expression were examined. Cells were stimulated with AICAR (1 mM) for incremental periods of time up to and including 24 h, and results were compared with those from time-matched control cells (Figs. 3 and 4). Treatment beyond 24 h resulted in significant cell death (unpublished observations). Serum deprivation alone of control cells increased MAFbx mRNA (Fig. 3A), which peaked at 2 h (P < 0.05) and remained elevated above 1-h baseline values thereafter. In addition, this deprivation appeared to slightly induce AMPK and ACC phosphorylation rapidly (within 1–2 h) and persistently (through 24 h; Fig. 5). The presence of AICAR initially repressed MAFbx (at 1 and 2 h; Fig. 3B), whereas more protracted exposure led to an eventual increase in expression at later time points, with an apparent maxima at 24 h. Signaling to (α-subunit phosphorylation) and from (ACC phosphorylation) AMPK in response to AICAR appeared to be competent at all times examined (Fig. 5).

Fig. 3.

Time dependence of AICAR-induced MAFbx mRNA expression. A: inter-time point quantitation of MAFbx mRNA. The value for 1-h AICAR-free control is set at 1.0 AU. All other control and treatment times are compared with the 1-h AICAR-free control. Data in line graph are means ± SE for n = 8 or 9 per time point per treatment. The presence of solvent in control treatments did not have a statistically significant effect on the indexes measured (see materials and methods); therefore, control groups from all individual experiments were pooled into a single group for this and all subsequent analyses. Here, control data in line graph are means ± SE for n = 25–27 per time point. B: intra-time point quantitation of MAFbx mRNA. The value for each AICAR-free time-matched control is set at 0.0 AU. Treatments at each time are compared with their corresponding time-matched control. Data in bar graph are means ± SE for n = 25–27 for control and n = 8 or 9 per group. Means with different letters are statistically different from one another (P < 0.05). Statistics comparing control treatment over time are not graphically represented but cited in the text. C: representative autoradiographs of MAFbx and L32 under indicated treatment conditions. Dex, dexamethasone.

Fig. 4.

Time dependence of AICAR-induced MuRF1 mRNA expression. A: inter-time point quantitation of MuRF1 mRNA. The value for 1-h AICAR-free control is set at 1.0 AU. All other control and treatment times are compared with the 1-h AICAR-free control. Data in line graph are means ± SE for n = 25–27 for control and n = 8 or 9 per time point per treatment. B: intra-time point quantitation of MuRF1 mRNA. The value for each AICAR-free time-matched control is set at 0.0 AU. Treatments at each time are compared with their corresponding time-matched control. Data in bar graph are means ± SE for n = 25–27 for control and n = 8 or 9 per group. Means with different letters are statistically different from one another (P < 0.05). Statistics comparing control treatment over time are not graphically represented but cited in the text. C: representative autoradiographs of MuRF1 and L32 under indicated treatment conditions.

Fig. 5.

Phosphorylation of AMPK and ACC in response to AICAR treatment. Shown is a time course of phosphorylation events in response to treatment with AICAR singly (1 mM) or in combination with dexamethasone (10 μM). Representative immunoblots of phosphorylation at the Thr172 site of AMPKα, total AMPKα, phosphorylation at the Ser79 site of ACC, total ACC, and β-tubulin are presented.

Serum deprivation also increased MuRF1 mRNA content (Fig. 4A), which climbed steadily until reaching a peak at 16 h (P < 0.05). Similarly to MAFbx, MuRF1 mRNA also exhibited an acute repression (at 2 h), followed by a robust increase in response to AICAR (Fig. 4B). However, for MuRF1, this induction occurred earlier; by 8 h, a maximal response had already been achieved, which was sustained throughout the remainder of the time course.

AICAR and dexamethasone act synergistically to induce MAFbx and MuRF1 mRNA content.

Dexamethasone induces MAFbx both in culture (29, 43, 44, 50) and in animals (3). Here, dexamethasone reduced expression of the ligase acutely (at 1 and 2 h; Fig. 3B), a result that has not been reported in previous investigations. By 4 h, dexamethasone had reciprocally increased MAFbx mRNA, an effect that persisted throughout the remainder of the time course (Fig. 3B). This induction occurred in the absence of alterations in phosphorylation of AMPKα and ACC compared with time-matched controls (data not shown), suggesting dexamethasone treatment alone does not activate the AMPK signaling cascade.

To examine the potential interactive effects between AMPK activity and glucocorticoid action on MAFbx and MuRF1 mRNA content, myocytes were incubated with both AICAR and dexamethasone together (Figs. 3 and 4). Acutely, this drug combination behaved similarly to AICAR alone (at 1 and 2 h; Fig. 3B) with respect to MAFbx mRNA. Interestingly, this repression appeared to be extended through 4 h, a response distinct from that for either AICAR or dexamethasone treatment alone. By 8 h, the dual treatment had begun to induce expression of the ligase at an intermediate level between that of AICAR and dexamethasone administered singly. At later time points (16 and 24 h; Fig. 3B), the combination dramatically stimulated MAFbx expression in a synergistic manner. These effects could not be attributed to dexamethasone-dependent alterations in AMPK signaling, as AMPKα and ACC were phosphorylated to a similar extent between AICAR alone and the combination treatment at all time points (Fig. 5).

Although dexamethasone has been shown to stimulate MuRF1 expression in vitro (29, 43, 50), such an effect was not observed in the present study. Although dexamethasone appeared to slightly induce the ligase, this increase did not reach statistical significance (Fig. 4B). The pattern of MuRF1 mRNA expression in response to a combination of AICAR and dexamethasone was reminiscent of that for MAFbx. Acutely, MuRF1 expression in response to the dual treatment mirrored that for AICAR alone (2 h), whereas the two treatments produced a synergistic response at all time points where cells had been previously shown to be responsive to AICAR alone (8, 16, and 24 h; Fig. 4B).

Metformin-induced alterations in MAFbx and MuRF1 mRNA content.

Time course studies using metformin were conducted as a second independent means for exploring a causal role of AMPK activation in increasing ubiquitin ligase mRNA expression (Fig. 6). Metformin treatment at later time points produced alterations in mRNA content of MAFbx (Fig. 6A) and MuRF1 (Fig. 6B) that were qualitatively similar to those achieved with AICAR administration. Additionally, as with AICAR, the combination of metformin and dexamethasone synergistically induced both ligases. The above results correlated with increases in phosphorylation of AMPKα and ACC caused by metformin, independent of dexamethasone cotreatment (data not shown).

Fig. 6.

Time dependence of metformin-induced MAFbx and MuRF1 mRNA expression. A: inter-time point quantitation of MAFbx mRNA. The value for 1-h metformin-free control is set at 1.0 AU. All other control and treatment times are compared with the 1-h metformin-free control. Data in line graph are means ± SE for n = 23–27 for control and n = 8 or 9 per time point per treatment. B: inter-time point quantitation of MuRF1 mRNA. The value for 1-h metformin-free control is set at 1.0 AU. All other control and treatment times are compared with the 1-h metformin-free control. Data in line graph are means ± SE for n = 24–26 for control and n = 8 or 9 per time point per treatment.

Disruption of cellular energy homeostasis reproduces the biphasic pattern of MAFbx and MuRF1 mRNA content.

Although both AICAR and metformin are accepted pharmacological activators of AMPK, neither activates the kinase by directly disturbing the energy status of the cell. In an effort to determine whether muscle-specific ubiquitin ligase mRNA expression was sensitive to cellular energy starvation, C2C12 cells were incubated with 2-DG (Fig. 7). 2-DG is a d-glucose analog that is phosphorylated but not further metabolized postuptake. As a result, import of 2-DG inhibits hexokinase action through a negative feedback mechanism and ultimately restricts cellular glucose utilization (10). Cells were incubated with 25 mM 2-DG for time periods up to 16 h (beyond which considerable lethality resulted; unpublished observations). To account for any potential effect(s) of osmotic stress [an environmental stimulus itself capable of stimulating AMPK (8)] independent of energy depletion, equimolar amounts of d-mannitol were added to controls. The presence of 25 mM mannitol did not appreciably shift the expression profile of MAFbx or MuRF1 during serum deprivation (data not shown). Furthermore, there was no notable increase in phosphorylation of ACC in mannitol-treated controls (Fig. 7C), which argues against significant activation of AMPK in response to the presence of the osmolite.

Fig. 7.

Time dependence of 2-deoxyglucose-induced MAFbx and MuRF1 mRNA expression. A: intra-time point quantitation of MAFbx mRNA. The value for each mannitol-treated time-matched control is set at 0.0 AU. Treatments at each time are compared with the corresponding time-matched control. Data in bar graph are means ± SE for n = 17 or 18 for control and n = 8 or 9 per group. B: intra-time point quantitation of MuRF1 mRNA. The value for each mannitol-treated time-matched control is set at 0.0 AU. Treatments at each time are compared with the corresponding time-matched control. Data in bar graph are means ± SE for n = 17 or 18 for control and n = 8 or 9 per group. Note that, in both A and B, data for the dexamethasone-only treatment were replotted from Figs. 3 (for MAFbx) and 4 (for MuRF1) for comparison purposes. Means with different letters are statistically different from one another (P < 0.05). C: time course of AMPK and ACC phosphorylation events in response to treatment with 2-deoxyglucose singly (25 mM) or in combination with dexamethasone (10 μM). Representative immunoblots of phosphorylation at the Thr172 site of AMPKα, total AMPKα, phosphorylation at the Ser79 site of ACC, total ACC, and β-tubulin are presented.

Exposure to 2-DG rapidly decreased MAFbx mRNA at early time points (1 and 2 h; Fig. 7A). This effect had reversed completely by more intermediate times, where a small but significant increase in mRNA content had occurred (4 and 8 h; Fig. 7A). MuRF1 mRNA expression demonstrated a comparable pattern of regulation, where 2-DG acted as a potent inducer at later times (particularly at 8 h; Fig. 7B). AMPK signaling appeared to be active at all time points examined (Fig. 7C).

Dual treatment with 2-DG and dexamethasone augmented expression of MAFbx (Fig. 7A) and MuRF1 (Fig. 7B) synergistically, albeit not as strikingly as observed earlier with AICAR. However, for MAFbx, this interaction was transient and was not present at 16 h, where 2-DG had a slightly antagonistic effect on the ability of dexamethasone to increase ligase expression (Fig. 7A). Identical to results with AICAR and metformin outlined above, dexamethasone had no effect on 2-DG-induced increases in AMPK signaling events (Fig. 7C).

AMPK activation does not increase the mRNA content of the UBR box family ligases UBR1/E3αI and UBR2/E3αII.

UBR box family ubiquitin ligases (52) serve as the E3 recognition factors for the NH2-end rule pathway, a ubiquitous process found in all cells whereby specific proteins bearing destabilizing NH2-terminal residues are marked for degradation by the proteasome. Several members of this E3 family have been implicated in the pathogenesis of skeletal muscle atrophy observed during cachexia, sepsis, and diabetes (28, 31, 48). To determine whether activation of AMPK results in an accumulation of mRNA content for all ubiquitin ligases implicated in skeletal muscle proteolysis, the expression of UBR1/E3αI and UBR2/E3αII in response to AICAR was determined (Fig. 8). In contradistinction to the responses of the muscle-specific ligases detailed above, both UBR1/E3αI and UBR2/E3αII were depressed in response to AICAR in a dose- (data not shown) and time-dependent manner (Fig. 8, A and B, respectively). Furthermore, at no time was this effect synergistic or additive with that of dexamethasone. Although there was a similar effect on both, the mechanism by which AMPK suppressed expression appears distinct for each ligase. For UBR1/E3αI, the time course suggested that AMPK activation decreased mRNA content of the E3, whereas for UBR2/E3αII a serum deprivation-induced increase in expression was prevented (data not shown). Similar results were obtained in cells treated with metformin and 2-DG (data not shown).

Fig. 8.

AICAR decreases mRNA expression of the UBR box family ligases UBR1/E3αI and UBR2/E3αII. A: intra-time point quantitation of UBR1/E3αI mRNA. The value for each AICAR-free time-matched control is set at 0.0 AU. Treatments at each time point are compared with their corresponding time-matched control. Data in bar graph are means ± SE for n = 26 or 27 for control and n = 8 or 9 per group. B: intra-time point quantitation of UBR2/E3αII mRNA. The value for each AICAR-free time-matched control is set at 0.0 AU. Treatments at each time point are compared with their corresponding time-matched control. Data in bar graph are means ± SE for n = 24–27 for control and n = 8 or 9 per group. Means with different letters are statistically different from one another (P < 0.05).

AMPK inhibition prevents serum deprivation-, dexamethasone-, and AICAR-induced increases in MAFbx and MuRF1 mRNA content.

To address the specificity of the treatments used previously to activate AMPK, cells were cotreated with AMPK agonists and compound C (Fig. 9), a small molecule reversible inhibitor of AMPK (59). Initial dose-response experiments determined that a concentration of 20 μM was optimal on the basis that it achieved maximal inhibition of ligase mRNA content, although not significantly impacting cell viability during extended culture times of up to 16 h, after which the inhibitor proved lethal (data not shown). Addition of compound C effectively prevented the increase in mRNA of both MAFbx and MuRF1 in response to treatment with AICAR (Fig. 9). Interestingly, the inhibitor also blocked the induction of the ligases by dexamethasone (whether added individually or in combination with AICAR). In all cases, including control cells treated with compound C, MAFbx and MuRF1 mRNA were reduced below vehicle-treated control levels, demonstrating that the serum deprivation-induced increase for each ligase noted previously during time course studies (Figs. 3 and 4) was antagonized by AMPK inhibition. Similar results were obtained in cells exposed to 2-DG and compound C for 8 h (data not shown).

Fig. 9.

Compound C antagonizes the stimulatory effects of AICAR and dexamethasone on MAFbx and MuRF1 mRNA expression. Cells were pretreated with compound C (20 μM) for 30 min preceding the switch to treatment medium containing the reagents specified in the figure with readdition of the inhibitor. Representative autoradiographs of MAFbx, MuRF1, and L32 in response to the indicated treatments singly or in combination with compound C are presented.

AMPK activation in vivo stimulates expression of MAFbx and MuRF1 in murine skeletal muscle.

In light of the mounting body of evidence suggesting that AMPK activation increases muscle-specific ubiquitin ligase mRNA in vitro, it was desirable to determine whether similar effects could be reproduced in intact skeletal muscle. To establish whether activation of AMPK in vivo would increase MAFbx and MuRF1 mRNA, mice were injected intraperitoneally with AICAR (Fig. 10). The mRNA content of both ligases increased in the gastrocnemius of AICAR-treated mice relative to saline-treated controls 6 h postinjection. Correspondingly, AICAR administration significantly increased Thr172 phosphorylation of AMPKα in skeletal muscle (∼70%; 4,484 ± 286 vs. 2,618 ± 206 arbitrary units, AICAR vs. control muscle; P < 0.05).

Fig. 10.

AICAR injection increases MAFbx and MuRF1 mRNA in vivo. Top: quantitation of MAFbx and MuRF1 mRNA isolated from the gastrocnemius-plantaris complex. The value for control mice given saline (ip) is set at 1.0 AU. Data in bar graph are means ± SE for n = 10 for control and n = 5 for treatment animals per group. *P < 0.05 compared with the respective control value. Bottom: representative autoradiographs of MAFbx, MuRF1, and L32.

DISCUSSION

A wealth of investigation has highlighted the importance of AMPK as a master regulator of cellular metabolism during times of energetic famine induced by a variety of physiological and pathological stimuli. In several of these contexts (particularly nutritional starvation), there is an established acceleration of proteolysis that occurs selectively in skeletal muscle, which, if left unchecked, leads to debilitating wasting (37). Because AMPK has proven to be an obligate effector of many of the responses of an organism to nutrient limitation, it was reasonable to suspect its activation may contribute to the development and progression of skeletal muscle atrophy. To address this possibility, we analyzed the genetic signature of an in vitro model of murine skeletal muscle after exposure to agonists of AMPK for changes in mRNA expression of E3 ubiquitin ligases previously identified as high-fidelity markers of the atrophic process (3, 13, 30). Treatment of C2C12 cells with several AMPK activators resulted in a dose- and time-dependent modulation of mRNA content of the muscle-specific ubiquitin ligases MAFbx and MuRF1, characterized by an acute repression preceding a sustained induction. The stimulatory effect was potently synergistic with the synthetic glucocorticoid dexamethasone, which was not a result of amplified signaling through the AMPK cascade. This response was selective and did not extend to the ubiquitin ligases UBR1/E3αI and UBR2/E3αII, both of which were decreased. Inhibition of AMPK signaling prevented increases in MAFbx and MuRF1 mRNA content in response to AMPK agonists and dexamethasone. Finally, in vivo activation of AMPK via AICAR injection recapitulated the stimulation of MAFbx and MuRF1 expression in skeletal muscle. Therefore, activation of AMPK results in discriminatory gene regulation of E3s in skeletal muscle, whereby muscle-specific ligases are strongly induced by persistent AMPK activity, whereas those with more ubiquitous tissue expression patterns are suppressed. These observations begin to delineate a heretofore unknown relationship between AMPK and UPP-dependent protein degradation in skeletal muscle and represent the first implication of the AMPK cascade as an intracellular signaling pathway that may contribute to the etiology of the proteolysis-associated aspects of skeletal muscle atrophy.

Here the role of AMPK in provoking E3 expression was demonstrated using three independent agents capable of activating AMPK in distinctive manners: AICAR, metformin, and 2-DG. Although AICAR and 2-DG activate the kinase by acting as an AMP mimetic or placing cells directly under energetic duress (respectively), metformin is one of a few select stimuli that appears to activate the kinase without detectably perturbing levels of AMP or ATP (8, 17), perhaps involving an alternative pathway of activation involving mitochondrial-derived reactive nitrogen species and the function of PKC-ζ as an AMPK kinase kinase toward LKB1 (57). Although none of these reagents is entirely specific in stimulating AMPK, that similar results were achieved with each treatment (albeit with varying time courses and intensities) suggests that changes in gene expression for these ubiquitin ligases are indeed AMPK dependent and not a function of some alternative kinase or signaling pathway. Furthermore, the stimulatory effects of these treatments were wholly antagonized by the addition of compound C, a previously characterized inhibitor of AMPK. The specificity of compound C itself has been demonstrated against select structurally related kinases (59), although an alternate independent analysis has revealed that the application of the inhibitor concomitantly with AICAR may interfere with AICAR uptake itself (and therefore may not antagonize AMPK directly) (9). However, that virtually identical results were achieved with inhibition during 2-DG treatment reiterates that data generated with compound C demonstrate a causal role for AMPK itself in controlling muscle-specific E3 expression. Collectively, this evidence supports the notion that AMPK is indeed a bona fide modulator of both MAFbx and MuRF1 mRNA content in skeletal muscle.

The interaction between dexamethasone and AMPK function on E3 gene expression is also noteworthy. To our knowledge, the synergy between these two stimuli has not been previously reported in skeletal muscle and is in fact contrary to the known effects of AMPK activity on dexamethasone-induced gene expression of two key gluconeogenic genes, glucose-6-phosphatase and phosphoenolpyruvate carboxykinase, in rat hepatoma cells (1, 34). One possible explanation for this interaction evidenced here could be a positive regulation of AMPK expression or activity in response to glucocorticoid administration. Such an outcome has been reported in hepatic (53) and cardiac (41) tissue, although the authors of the former study cite unpublished data that this effect was not observed in skeletal muscle. Our results examining native AMPKα expression and activity (Figs. 5 and 7C) are in accord with that latter finding, and thus the synergistic effect of the two agents on muscle-specific ligase expression cannot be attributed to an increased abundance of AMPK or signal transduction through AMPK. Importantly, inhibition of AMPK ablated the stimulatory effects of dexamethasone, indicating that AMPK signaling not only cooperates with the glucocorticoid in upregulating MAFbx and MuRF1 but is required for dexamethasone-dependent increases in both E3s. These results suggest AMPK exerts a dichotomous influence on glucocorticoid-induced gene expression, which is likely tissue and gene dependent, and raises the possibility that AMPK may be an important determinant of glucocorticoid sensitivity in skeletal muscle.

There are several potential mechanisms through which AMPK activity could be acting individually and in cooperation with dexamethasone to produce these effects on ubiquitin ligase expression. Numerous studies have highlighted the impact of AMPK signaling events on proteins, particularly transcription factors and transcriptional coactivators and repressors, which control various aspects of gene expression (32). Several well-established examples of this regulation include the ability of the kinase to disrupt the gluconeogenic gene program by controlling the subcellular localization of the CREB coactivator TORC2 (27) and the stability of the transcription factor hepatocyte nuclear factor-4α (19) through phosphorylation-mediated mechanisms. Likewise, AMPK is able to restrain lipogenic gene expression by repressing mRNA expression of the transcription factor SREBP-1 (59) and by interfering with the DNA binding activity of the coactivator ChREBP (26). Suppression of transcription and/or cofactor activity is not the only consequence of AMPK activation. Expression of the GLUT4 gene by the kinase appears to be driven by positive regulation and redistribution of GEF and MEF2A to the nucleus (18), whereas AMPK-induced mitochondrial biogenesis coincides with increased DNA binding of NRF-1 (2) and the upregulation of the proliferator-activated receptor-γ coactivator-1α and calmodulin-dependent protein kinase IV genes (60). As well as these transcriptional mechanisms, AMPK is also able to regulate gene expression via posttranscriptional means involving control of mRNA stability through the nuclear import of mRNA binding protein HuR (54). Additionally, SNF1 (the yeast equivalent of AMPKα) is capable of influencing chromatin configuration via histone phosphorylation, which results in acetylation of particular promoters by the acetyltransferase GNC5 (33) (although such a role has not been confirmed for mammalian AMPK). Therefore, AMPK is capable of influencing mRNA abundance by regulating the intracellular partitioning, stability, mRNA expression, and DNA binding capacity of specific coactivators or transcription factors, the stability of transcriptional end-products themselves, and perhaps through epigenetic gene regulation. Further studies will be needed to determine which mechanisms are requisite for the control of MAFbx and MuRF1 gene expression.

The understanding of the intracellular signaling and molecular mechanisms underpinning skeletal muscle atrophy has traditionally lagged behind those regulating hypertrophy. It has only recently become appreciated that many of the canonical growth-controlling pathways that promote protein accretion are also intimately involved in suppressing the presentation of classic markers of wasting muscle and/or the atrophic process itself. Both receptor-tyrosine kinase-Akt- and mTOR complex 1-mediated signaling events have been shown to limit MAFbx/atrogin-1 (29, 43, 44, 50) and MuRF1 (43, 50) expression and prevent atrophy (4, 43, 50) in addition to their well-established roles in promoting increases in skeletal muscle cell and tissue size (4, 42). This functional quality adds dramatically to the complex paradigm of skeletal muscle size control, as these pathways (and presumably a number of their ultimate targets) are capable of influencing both aspects of cellular size adaptation. Within this contemporary model of atrophic regulation, our finding that AMPK activation is a positive regulator of ubiquitin ligase expression makes sound contextual and teleological sense. AMPK has an acknowledged role in antagonizing mTOR complex 1 signaling by stimulating the GAP activity of tuberin (21) toward Rheb, an important activator of mTOR (45). Furthermore, emerging evidence suggests that Akt may directly regulate the activity of AMPK (14), thereby linking growth factor and energetic signaling upstream of tuberin. Thus, in the context of atrophic signaling, AMPK could serve to monitor the energetic state of skeletal muscle and provide this information as an input to modify antiatrophic signals derived from mitogenic (receptor-tyrosine kinase-Akt-dependent) and nutritional (mTOR complex 1-dependent) sensors while amplifying proatrophic signals originating from circulating glucocorticoids, all to better control mass in response to environmental cues.

In conclusion, we describe here the role of AMPK in controlling the mRNA content of select muscle-specific ubiquitin ligases using both in vitro and in vivo experimental approaches. Exposure of differentiated C2C12 cells to agonists of AMPK causes dose- and time-dependent increases in mRNA of MAFbx and MuRF1. This stimulatory effect synergizes with dexamethasone, an occurrence that cannot be accounted for by either altered signaling through or protein expression of AMPK itself. Accordingly, signal transduction through the AMPK pathway appears necessary for the induction of these muscle-specific ligases in response to several stimuli, including glucocorticoid treatment and energetic starvation. Lastly, pharmacological activation of AMPK in vivo reproduces the stimulation of MAFbx and MuRF1 mRNA expression demonstrated in vitro. The studies herein elucidate a key piece of the regulatory puzzle that governs elements of the atrophic gene expression response and will likely control other phenotypes associated with skeletal muscle catabolism.

GRANTS

This work was supported in part by National Institutes of Health Grant GM-38032 (C. H. Lang). B. J. Krawiec was supported by National Institutes of Health Predoctoral Training Grant T32 GM-08619.

Acknowledgments

We are grateful to Danuta Huber for assistance with the bicinchoninic acid assays for protein concentration.

Footnotes

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract