Identification of Ubiquitin Ligases Required for Skeletal Muscle Atrophy

See allHide authors and affiliations

Science  23 Nov 2001:
Vol. 294, Issue 5547, pp. 1704-1708
DOI: 10.1126/science.1065874


Skeletal muscle adapts to decreases in activity and load by undergoing atrophy. To identify candidate molecular mediators of muscle atrophy, we performed transcript profiling. Although many genes were up-regulated in a single rat model of atrophy, only a small subset was universal in all atrophy models. Two of these genes encode ubiquitin ligases: Muscle RING Finger 1 (MuRF1), and a gene we designate Muscle Atrophy F-box(MAFbx), the latter being a member of the SCF family of E3 ubiquitin ligases. Overexpression of MAFbx in myotubes produced atrophy, whereas mice deficient in either MAFbx orMuRF1 were found to be resistant to atrophy. These proteins are potential drug targets for the treatment of muscle atrophy.

Muscle atrophy occurs as a consequence of denervation, injury, joint immobilization, bed rest, glucocorticoid treatment, sepsis, cancer, and aging (1). Unfortunately, there are no effective treatments for muscle atrophy. The maintenance of muscle mass is controlled by a balance between protein synthesis and protein degradation pathways, which is thought to shift toward protein degradation during atrophy (1). Recently, a signaling pathway that increases protein synthesis was shown to promote muscle hypertrophy, thereby overcoming muscle atrophy (2, 3). Although protein degradation systems have been extensively studied, specific molecular mediators of atrophy-related degradation have not been defined, nor has it been demonstrated whether blocking such specific mediators can inhibit muscle atrophy. Further, it is unknown whether muscle atrophy caused by disparate perturbations is controlled by a common signaling pathway or whether distinct pathways can lead to muscle wasting.

Denervation, immobilization, and unweighting in rats all result in similar rates of loss in mass of the medial gastrocnemius muscle, a result which is consistent with the idea that there are common mechanisms leading to atrophy (Fig. 1A). To identify potential universal markers of atrophy, we attempted to identify genes regulated in immobilization, and then determined which of these were similarly regulated in multiple other models. We first compared mRNA from rat medial gastrocnemius muscle that had been immobilized for 3 days to mRNA from control muscle, via the Gene- Tag differential display approach (4). We analyzed a relatively early time point (3 days) to identify genes that may function as potential triggers, as well as markers, of the atrophy process. Only genes whose expression changed by threefold or more were accepted as being differentially regulated. Seventy-four transcripts were identified (4) and then assayed for universality by Northern analysis using panels of mRNA prepared from muscle subjected to denervation, immobilization, or unweighting for periods of 1 to 14 days.

Figure 1

(A) Three different models of skeletal muscle atrophy. A time course of mass loss in the rat medial gastrocnemius muscle was examined in three in vivo models: denervation, immobilization, and hindlimb suspension. Female Sprague-Dawley rats weighing 250 to 275 g were used in all models. For the denervation procedure, the right sciatic nerve was cut in the midthigh region, leading to denervation of the lower limb muscles. For the immobilization procedure, the right ankle joint was fixed at 90° of flexion by insertion of a screw (1.2 × 8 mm) through the calcaneous and talis into the shaft of the tibia. For the hindlimb suspension procedure, the hindlimbs were unloaded by suspending the rats by their tails with a tail-traction bandage as described (18). On the indicated days, rats were killed and hindlimb muscles were removed, weighed, and frozen. Weight-matched untreated rats served as controls. Data are means ± SEM, n= 10 rats. (B) Northern blots (4) showing the effect of atrophy on MuRF1 and MAFbx transcripts. Medial gastrocnemius muscle was obtained from rats undergoing a time course (0, 1, 3, and 7 days) of three atrophy models: ankle-joint immobilization, denervation, and hindlimb suspension. (C) Northern blots (4) showing the effect of dexamethasone (DEX) and IL-1 on expression ofMuRF1 and MAFbx. Medial gastrocnemius muscle was obtained from untreated rats (CON) and from rats treated with DEX, delivered orally at a concentration of 6 μg/ml for 9 days, and from rats treated with IL-1, delivered subcutaneously daily at a dose of 0.1 mg per kilogram of body weight for 3 days. (D) Tissue-specific expression of MuRF1 and MAFbx. SK, skeletal. mRNA obtained from rat and human tissues (Clontech) was hybridized with probes for the indicated genes (4).

Although most genes perturbed during immobilization were similarly regulated during denervation, most of these genes were unaltered in the unweighting model, despite the fact that similar rates of atrophy were seen in these models (Fig. 1A). Two genes, however, were identified that were up-regulated in all three models of atrophy:MuRF1 (for Muscle RING Finger 1) and MAFbx (for Muscle Atrophy F-box) (Fig. 1B).

MuRF1 and MAFbx expression were analyzed in two additional models of skeletal muscle atrophy: treatment with the cachectic cytokine interleukin-1 (IL-1) (5) and treatment with the glucocorticoid dexamethasone (6). Both cachectic agents caused an up-regulation of MuRF1 andMAFbx, with dexamethasone resulting in a greater than 10-fold increase in expression of both genes (Fig. 1C). Analysis of rat and human tissues revealed that MuRF1 and MAFbxmRNAs were expressed selectively in cardiac and skeletal muscle (Fig. 1D), which is consistent with their serving specific roles in these tissues.

Because MAFbx had not been previously identified, we cloned full-length rat and human cDNAs for this gene. Open reading frames of rat and human MAFbx cDNA sequence predict proteins that are 90% identical (4). MAFbx contains an F-box domain, which is characteristic of proteins that are components of a particular type of E3 ubiquitin ligase called an SCF ubiquitin-ligase complex (7, 8). The SCF complex is thus named because it involves stable interactions between Skp1, Cullin1, and one of many F-box–containing proteins (Fbps). More than 38 different Fbps have been identified in humans (9, 10), with the closest relative to MAFbx being Fbx25 (10). Two lines of evidence indicate that MAFbx is in fact an SCF-type E3 ubiquitin ligase. First, yeast two-hybrid cloning using full-length MAFbx as a “bait” resulted in 94 independent clones of Skp1, out of a total of 94 clones obtained in the interaction experiment (11). Second, affinity purification of MAFbx from mammalian cells transfected with MAFbx resulted in the copurification of both endogenous Skp1 and Cullin1 (Fig. 2A). This copurification was dependent on the presence of the F-box domain in MAFbx (Fig. 2A, compare lanes 3 and 4). The F-box motif has been shown to be necessary for interaction between Fbps and Skp1 (12).

Figure 2

(A) MAFbx interacts with the SCF complex components Cullin1 and Skp1. Vectors encoding glutathione S-transferase (GST) (GST/CON), GST-MAFbx, or GST-MAFbxΔFb (an F-box deletion of MAFbx, amino acids 216 through 263) were transiently transfected into Cos7 cells (4). Both Cullin1 and Skp1 could be copurified, with glutathione-agarose beads, from lysates of cells transfected with GST-MAFbx (lane 3). Deletion of the F box markedly reduced the amount of Cullin1 and Skp1, which coprecipitated (lane 4). I.B., immunoblot. (B) Overexpression of MAFbx causes atrophy. C2C12 myotubes were either uninfected (CON) or infected with an adenovirus expressingEGFP, or with an adenovirus expressing both a Myc-epitope–tagged rat MAFbx gene and EGFP(MAFbx-EGFP). At day 4 after differentiation, fluorescent myotubes were photographed, and myotube diameters were measured (right). The adenoviruses were generated as described (13). Calibration = 50 μm. (C) Determination of infection levels. Because the EGFP and MAFbx-EGFP viruses contained the EGFP gene, an anti-EGFP immunoblot allowed for a relative determination of infection levels (top). An immunoblot of lysates confirmed the presence of Myc-epitope–tagged MAFbx protein in the myotubes infected with the MAFbx virus (bottom). (D) MuRF1 protein has ubiquitin ligase activity. Purified glutathione-Sepharose–bound MuRF1 protein (GST-MuRF1) was added to a ubiquitin ligase reaction as described (19). Briefly, recombinant GST-MuRF1 (100 ng) was incubated with 32P-labeled ubiquitin (3 μg) in the presence of ATP, E1, and recombinant Ubc5c (lane 5). In lanes 1 through 4, indicated components were omitted. Aliquots of the reaction were analyzed by 12.5% SDS-polyacrylamide gel electrophoresis to detect 32P-labeled high-molecular-weight ubiquitin conjugates. The “ubiquitin polymer” may include ubiquitinated Ubc5c and MuRF1.

To determine whether MAFbx expression was sufficient to cause muscle atrophy, we generated an adenovirus encoding ratMAFbx and Enhanced Green Fluorescent Protein(EGFP) (13). We then infected differentiated, postmitotic, multinucleated C2C12 skeletal myotubes (14) with either a control adenovirus expressingEGFP alone or a MAFbx-EGFP expressing adenovirus (in which myc-tagged MAFbx was coexpressed withEGFP, via an internal ribosomal entry sequence). After 2 days, the myotubes expressing the MAFbx gene were significantly thinner than the EGFP-infected cells, as determined by measuring myotube diameter (Fig. 2B). An immunoblot of protein lysates with an antibody to EGFP allowed for a relative determination of infection levels in EGFP andMAFbx-EGFP cultures (Fig. 2C). Immunoblotting with an antibody to the MAFbx myc-epitope tag confirmed the production of MAFbx protein in the MAFbx-infected myotubes (Fig. 2C).

The RING finger protein MuRF1 was previously identified by virtue of its interaction in a yeast two-hybrid experiment with a domain of the sarcomeric protein titin (15). MuRF1 contains all the canonical structural features of RING-domain–containing monomeric ubiquitin ligases (16, 17). After confirming that MuRF1 protein levels increased during atrophy (4), we tested recombinant MuRF1 protein for ubiquitin ligase activity in an in vitro assay. MuRF1 was required for the formation of a high-molecular-weight ubiquitin polymer, indicating that it functions as a ubiquitin ligase (Fig. 2D).

To investigate the in vivo function of MAFbx andMuRF1, we genetically engineered mice with null alleles for these genes. For MAFbx, a specific gene deletion was generated by replacing the genomic DNA spanning the ATG through the exon encoding the F-box region with a LacZ/neomycin cassette (4). For MuRF1, genomic DNA spanning the ATG through the fifth exon was replaced with a LacZ/neomycincassette (4). The addition of the LacZcassette in each case allowed us to simultaneously disrupt gene function and perform β-galactosidase (β-Gal) staining to determine gene expression patterns. Analysis of the genetic loci demonstrated the expected perturbations in MAFbx (Fig. 3A) and MuRF1 (Fig. 3C) +/– and –/– mice. Further, MAFbx –/– mice were null for MAFbx mRNA (Fig. 3B), andMuRF1 –/– mice were null for MuRF1mRNA (Fig. 3D). Both MAFbx –/– andMuRF1 –/– mice were viable and fertile and appeared normal. Knockout mice had normal growth curves relative to those of wild-type litter mates, and skeletal muscles and the heart had normal weights and morphology.

Figure 3

(A) The targeting of theMAFbx gene was confirmed in embroyonic stem (ES) cells and in both heterozygous and homozygous MAFbx mutant mice by digestion of genomic tail DNA with Eco RI and probing with a 5′ 1.1-kb Sac II fragment to detect the endogenous (end. allele) 3.1-kb and targeted (mut. allele) 4.9-kb Eco RI fragments (4). (B) The targeted mutation in the MAFbx gene was verified by probing mRNA from both tibialis anterior (TA) and gastrocnemius (GA) muscle prepared from MAFbx +/+, +/–, and –/– mice with a MAFbx probe, spanning base pairs (bp) 660 through 840 of coding sequence (MAFbx, upper panel), as well as with a probe of the inserted LacZ gene (lower panel). (C) The targeting of the MuRF1 gene was confirmed in ES cells and in both heterozygous and homozygous MuRF1mutant mice by digestion of genomic tail DNA with Eco RI and probing with a 5′ 0.5-kb Bgl II fragment to detect the endogenous (end. allele) 15-kb and targeted (mut. allele) 10-kb Eco RI fragments (4). (D) The targeted mutation in the MuRF1 gene was verified by probing mRNA from both TA and GA muscle prepared fromMuRF1 +/+, +/–, and –/– mice with a probe spanning bp 1 through 500 of rat MuRF1 coding sequence (MuRF1, upper panel), as well as with a probe of the inserted LacZgene (lower panel).

We then challenged the mice in an atrophy model to determine the effect of MAFbx and MuRF1 deficiency on skeletal muscle loss. Muscle atrophy was induced by cutting the sciatic nerve, resulting in denervation and disuse of the tibialis anterior and gastrocnemius muscles. Denervation resulted in the up-regulation of LacZexpression from the MAFbx and MuRF1 gene loci in all muscle fibers, as demonstrated by β-Gal staining in the tibialis anterior muscle of heterozygous MAFbx and MuRF1mice (Fig. 4A). Wild-type mice had significant muscle atrophy in the gastrocnemius muscles at 7 and 14 days after denervation (Fig. 4B). In contrast, theMAFbx –/– mice had significant muscle sparing relative to MAFbx +/+ mice at both 7 and 14 days (Fig. 4B). At 14 days, the MAFbx –/– mice exhibited a 56% sparing of muscle as compared with the wild-type mice.MAFbx –/– mice exhibited no additional muscle loss between 7 and 14 days, whereas MAFbx +/+mice continued to lose mass. The sparing of muscle mass at 14 days was also reflected in a preservation of mean fiber size and fiber size variability; MAFbx –/– mice had significantly larger fibers than did the MAFbx +/+ mice, and muscle in the denervated limb had the same fiber size variability (standard deviation) as that in the control limb (wild type versus knockout: control, 2146 ± 375 μm2 versus 2053 ± 420 μm2; denervated, 1068 ± 122 μm2 versus 1508 ± 284 μm2) (Fig. 4C).

Figure 4

(A) TheMAFbx and MuRF1 genes are up-regulated in muscle after denervation. The regulation of the MAFbx andMuRF1 genes was examined with β-Gal staining inMAFbx +/– and MuRF1 +/−mice. The right sciatic nerve was cut in heterozygous mice, resulting in denervation of the TA muscle. Seven days later, the right and left TA muscles were sectioned and stained for β-Gal activity, in the same media, for equivalent times. In control muscle, there is a low level ofMAFbx expression in some (primarily deep region), but not all, muscle fibers of the TA muscle. In comparison,MuRF1 is expressed in all fibers at a slightly higher level than MAFbx. After denervation, both MAFbx andMuRF1 expression are up-regulated in all muscle fibers.(B) Muscle from MAFbx- and MuRF1- deficient mice maintains muscle mass after denervation, as compared to that of wild-type (+/+) mice. The right hindlimb muscles of adult mice (MAFbx +/+ and –/–) were denervated by cutting the right sciatic nerve. The left hindlimb of each animal served as its own control. At 7 and 14 days after denervation, the right and left GA muscle complex was removed and weighed. Muscle weights (GA) are plotted as a percent of control, calculated as the right/left muscle weights. Data are means ± SEM, n = 5 to 10 mice.(C) Muscle fiber size and variability were maintained in muscles from MAFbx-deficient mice after denervation. Cross sections taken from the TA muscle were stained with an antibody against laminin (Sigma). Representative cross sections are shown from the TA muscle: wild type (+/+), control left side (upper left); wild type (+/+), 14-day-denervated (DEN) right side (lower left); homozygous (–/–), control left side (upper right); homozygous, 14-day-denervated right side (lower right).

The MuRF1 –/– mice also had significant muscle sparing at 14 days of denervation but not at 7 days (Fig. 4B). At 14 days, the MuRF1 –/– exhibited a 36% sparing of muscle as compared with the wild-type (+/+) mice. These data provide strong evidence that both MAFbx and MuRF1 are critical regulators of muscle atrophy, most likely through the regulation of the degradation of crucial muscle proteins.

The discovery of two ubiquitin ligases as markers for multiple models of skeletal muscle atrophy suggests that highly disparate perturbations, ranging from denervation to glucocorticoid treatment, activate common atrophy-inducing pathways. Further, the particular function of ubiquitin ligases—to target discrete substrates for proteolysis by the adenosine triphosphate (ATP)–dependent proteasome—suggests that either a single protein degradation pathway is up-regulated during atrophy, which requires both MAFbx and MuRF1; or that parallel pathways in which these genes play required roles are up-regulated.

Because both MuRF1 and MAFbx are also specifically expressed in cardiac muscle, it will be important to examine the roles of these ubiquitin ligases in heart remodeling and disease and to determine whether loss of either MuRF1 orMAFbx affects cardiac function. The current studies identifyMuRF1 and MAFbx as markers of skeletal muscle atrophy and potential targets for therapeutic intervention to prevent the loss of skeletal muscle in clinical settings of atrophy.

  • * To whom correspondence should be addressed. E-mail:{at}


View Abstract

Navigate This Article