Gene expression profiling of skeletal muscles treated with a soluble activin type IIB receptor

Fedik Rahimov, Oliver D. King, Leigh C. Warsing, Rachel E. Powell, Charles P. Emerson Jr., Louis M. Kunkel, Kathryn R. Wagner


Inhibition of the myostatin signaling pathway is emerging as a promising therapeutic means to treat muscle wasting and degenerative disorders. Activin type IIB receptor (ActRIIB) is the putative myostatin receptor, and a soluble activin receptor (ActRIIB-Fc) has been demonstrated to potently inhibit a subset of transforming growth factor (TGF)-β family members including myostatin. To determine reliable and valid biomarkers for ActRIIB-Fc treatment, we assessed gene expression profiles for quadriceps muscles from mice treated with ActRIIB-Fc compared with mice genetically lacking myostatin and control mice. Expression of 134 genes was significantly altered in mice treated with ActRIIB-Fc over a 2-wk period relative to control mice (fold change > 1.5, P < 0.001), whereas the number of significantly altered genes in mice treated for 2 days was 38, demonstrating a time-dependent response to ActRIIB-Fc in overall muscle gene expression. The number of significantly altered genes in Mstn−/− mice relative to control mice was substantially higher (360), but for most of these genes the expression levels in the 2-wk treated mice were closer to the levels in the Mstn−/− mice than in control mice (P < 10−30). Expression levels of 30 selected genes were further validated with quantitative real-time polymerase chain reaction (qPCR), and a correlation of ≥0.89 was observed between the fold changes from the microarray analysis and the qPCR analysis. These data suggest that treatment with ActRIIB-Fc results in overlapping but distinct gene expression signatures compared with myostatin genetic mutation. Differentially expressed genes identified in this study can be used as potential biomarkers for ActRIIB-Fc treatment, which is currently in clinical trials as a therapeutic agent for muscle wasting and degenerative disorders.

  • myostatin
  • ActRIIB-Fc

myostatin, also known as growth/differentiation factor 8 (GDF8), is a member of the transforming growth factor (TGF)-β superfamily of signaling peptides that plays an important role in regulation of skeletal muscle growth. Mice deficient for the gene encoding myostatin (Mstn) display significantly increased muscle mass (16). An over twofold enlargement of skeletal muscles observed in Mstn−/− mice has been shown to result from a combination of muscle fiber hypertrophy and hyperplasia. Furthermore, hypermuscular phenotypes in cattle (8, 17), sheep (5), and dogs (21) with natural myostatin mutations demonstrate strong evolutionary conservation of the myostatin signaling pathway in muscle growth regulation. Inactivating mutations in the human MSTN gene also result in hypermuscularity with no overt detrimental effects (27). Myostatin is expressed in both developing and adult skeletal muscle, and systemic administration of myostatin in adult mice induced severe muscle and fat loss, indicative of its postnatal role in the maintenance of fully differentiated muscle (36).

In vivo and in vitro studies have suggested that myostatin signals through the activin type IIB receptor (ActRIIB) (13). ActRIIB is a transmembrane serine-threonine kinase receptor for select TGF-β superfamily members and results in activation of Smad transcription factors. Transgenic mice expressing a truncated form of ActRIIB show significant increases in skeletal muscle mass (13). Based on this evidence, a soluble form of ActRIIB was recently generated by fusing the extracellular domain of the receptor with the immunoglobulin Fc region, referred to as ActRIIB-Fc, which presumably acts as a decoy receptor for myostatin. Treatment of normal mice with ActRIIB-Fc for 2 wk resulted in up to 60% muscle mass increase (14), improved hypoxia-induced muscle dysfunction in normal mice (23), and increased muscle mass and strength in a mouse model for amyotrophic lateral sclerosis (20). Furthermore, treatment of several mouse models of cancer-induced cachexia with ActRIIB-Fc prevented muscle degeneration, reversed prior loss of skeletal muscle, and prolonged survival time (35).

ActRIIB-Fc is not a selective myostatin inhibitor. ActRIIB-Fc has been shown to also bind activin and GDF11, whose role in postnatal muscle growth is unclear (20, 25). Mstn−/− mice treated with ActRIIB-Fc have increased muscle mass compared with untreated Mstn−/− mice; however, myostatin is believed to play a substantial role in ActRIIB-Fc-mediated muscle growth compared with other yet unknown muscle growth regulatory factors inhibited by ActRIIB-Fc (14). ActRIIB-Fc is currently in clinical trials (2), and there is a need for biomarkers to correlate with outcome measures.

To begin to develop in vitro assays that can be used as biomarkers of ActRIIB-Fc response, we assessed gene expression signatures in murine skeletal muscles treated with ActRIIB-Fc and compared changes observed in mRNA expression levels with muscles from Mstn−/− and control mice. In this study we report a range of potential mRNA biomarkers that respond to ActRIIB-Fc treatment as well as candidate genes playing a role downstream of the myostatin signaling pathway.



ActRIIB-Fc (RAP-031) was a kind gift of Acceleron Pharma, Inc. (3, 25). Three-month-old C57BL/6 female mice were injected intraperitoneally with 10 mg of soluble ActRIIB-Fc per 1 kg of body weight based on previous dose finding studies (12). Mice were treated twice weekly for 2 wk (4 doses) before muscle harvest or treated once (1 dose) and killed after 2 days. Control mice were injected with phosphate-buffered saline (PBS) and killed at 2 days along with the one-dose ActRIIB-Fc-treated mice. Mstn−/− mice originated from the colony of Dr. Se-Jin Lee (Dept. of Molecular Biology and Genetics, Johns Hopkins University) (16) and were killed along with the four-dose ActRIIB-Fc-treated mice. Our initial study used n = 3 mice from each of these four groups. An independent replication study used n = 8 Mstn−/− mice and n = 5 mice from each of the other three groups. Quadriceps muscles were dissected and immediately snap frozen in liquid nitrogen, followed by RNA and protein isolation. Quadriceps muscles were chosen because of convenience and previous demonstration of robust response to ActRIIB-Fc (14). All animals were housed in the same colony according to regulations approved by the Institutional Animal Care and Use Committee at Johns Hopkins University.

RNA extraction and cDNA synthesis.

Total RNA was isolated from the frozen muscle tissues with TRIzol reagent (Invitrogen). RNA was quantified with UV absorption at 260 nm with a NanoDrop ND-1000 Spectrophotometer, and its integrity was assessed with the RNA 6000 Nano chip on the Agilent 2100 Bioanalyzer (Agilent). cDNA for quantitative real-time polymerase chain reaction (qPCR) analysis was synthesized from 0.5 μg of total RNA with the RNA-to-cDNA Master Mix (Applied Biosystems) and a combination of oligo(dT) and random hexamer primers in 20-μl reactions.

Microarray analysis.

Total RNA was labeled with biotin with the GeneChip Whole Transcript (WT) Sense Target Labeling Assay according to the manufacturer's instructions. Affymetrix GeneChip Mouse Gene 1.0 ST arrays were hybridized with the labeled cDNA overnight. Arrays were processed and scanned with Fluidics Station 450 and GeneChip Scanner 3000 7G (Affymetrix). Array data were preprocessed with RMA (10) and analyzed with the limma package (version 3.4.4) (28) from the Bioconductor project in the statistical computing environment R (version 2.11.1). When multiple probe sets mapped to the same gene, the probe set with the highest interquantile range across all arrays was retained (20,405 probe sets in total; see Supplemental Table S1).1 Genes were flagged as significantly differentially expressed if the fold change between classes was >1.5 and if the nominal P value was <0.001. The microarray CEL files and normalized probe set intensity values are deposited in the GEO database ( and are available under accession number GSE25306. Biological pathway analysis was carried out with the Ingenuity Pathway Analysis tool (IPA 8.6; Ingenuity Systems). We also used the Kyoto Encyclopedia of Genes and Genomes (KEGG) ( and Gene Ontology ( databases to obtain gene lists to use in heat maps and cluster analysis.

Quantitative real-time PCR.

High-throughput qPCR was performed on the BioMark 48.48 Dynamic Array (Fluidigm) with inventoried TaqMan Gene Expression Assays (Applied Biosystems) according to the manufacturer's instructions. To increase the target transcript levels to detectable concentrations in nanoliter reaction chambers of the BioMark array without altering their ratios in the transcriptome, the target cDNAs were preamplified with a diluted pool of TaqMan assays for 14 cycles. To compensate for cDNA concentration differences between samples and introduced by pipetting errors, expression of the target genes was normalized relative to the expression levels of two housekeeping genes, Gapdh and B2m, and fold change values were computed with the 2−ΔΔCt method (where Ct is threshold cycle) (26). All qPCR reactions were run in triplicate, and mice from the same treatment group were analyzed as biological replicates.

Western blotting.

Frozen quadriceps muscle samples were homogenized in Tissue Extraction Reagent I (Invitrogen), followed by quantification with the BCA Protein Assay Kit (Thermo Scientific). Twenty micrograms of tissue lysates from two mice from each group were subjected to polyacrylamide gel electrophoresis on NuPAGE Novex 4–12% Bis-Tris resolving gels under reducing conditions, transferred to polyvinylidene difluoride (PVDF) membrane, and visualized with the Western Breeze Kit (Invitrogen) according to step-by-step kit instructions. Primary antibodies were purchased from Santa Cruz Biotechnology (α-T-catenin, sc-98616; R-cadherin, sc-7941; Igfbp5, sc-13093; Fhl-1, sc-133580; Myozenin 2, sc-66990; Crp3, sc-98827; Troponin I, sc-15368; Myh7, sc-53089; Myl3, sc-58804) and Abcam (Myl3, ab680).


Treatment of mice with ActRIIB-Fc significantly increases skeletal muscle mass.

To evaluate the effects of ActRIIB-Fc on murine skeletal musculature and to obtain muscle tissues for biomarker identification, 3-mo-old female wild-type mice were treated acutely (1 dose) or chronically (4 doses over 2 wk) with ActRIIB-Fc. Quadriceps muscle from mice chronically treated with ActRIIB-Fc had significantly increased muscle weight (63%) compared with vehicle-treated control mice (P < 0.001). Previous studies with ActRIIB-Fc have demonstrated that this is the maximal amount of muscle growth achieved with postnatal administration of ActRIIB-Fc (14, 25). Myostatin-null mice had an approximate doubling of muscle mass as has been previously shown (16), while animals acutely treated with ActRIIB-Fc had no significant muscle growth compared with control animals (data not shown).

Genomewide expression profiling of mice treated with ActRIIB-Fc and Mstn−/− mice.

Transcriptional profiles of the quadriceps muscles from mice treated acutely and chronically with ActRIIB-Fc, Mstn-null mice, and age-matched control mice (n = 3/group) were determined with the Affymetrix expression GeneChip Mouse Gene 1.0 ST arrays. In contrast to the oligonucleotide arrays that were used to assess gene expression in myostatin-null mice in previous studies (4, 29, 33), the Gene 1.0 ST arrays interrogate the whole transcript with multiple probes distributed along the entire transcribed regions of 28,853 genes in the mouse genome, thus eliminating the 3′-end biased representation of transcripts and providing a more complete and accurate expression measure.

Quadriceps muscles from mice deficient for the Mstn gene and mice treated with ActRIIB-Fc showed both up- and downregulation of genes relative to control mice, with downregulation more common in both cases (Fig. 1). Moreover, we observed a time-dependent effect of ActRIIB-Fc treatment. With fold change > 1.5 and nominal P < 0.001 as cutoffs, 134 genes were differentially expressed in mice chronically treated with ActRIIB-Fc compared with control mice, whereas 38 genes were differentially expressed in the acutely treated mice compared with control mice. It should be noted, however, that the false discovery rate (FDR) at nominal P = 0.001 was 11% and 40% for these two comparisons, respectively. On the other hand, Mstn−/− mice had 360 genes altered compared with control mice (with FDR 4% at nominal P = 0.001), demonstrating that complete loss of functional myostatin protein has a larger impact on global muscle gene expression (these genes are listed in Supplemental Table S1). Among these 360 genes were 88 of the 134 genes differentially expressed in the chronic ActRIIB-Fc-treated mice relative to control mice, indicating that most of the significant expression changes seen in chronically treated mice are also seen in the Mstn−/− mice. Moreover, among the 360 genes for which expression levels were significantly different between Mstn−/− and control mice, in 293 cases the expression levels in the chronically treated mice were more similar to the levels in the Mstn−/− mice than to the levels in the control mice. This indicates a significant overall similarity between the Mstn−/− and chronically ActRIIB-Fc-treated mice (P < 10−30, binomial test).

Fig. 1.

Venn diagrams indicating the overlap between sets of differentially expressed genes in ActRIIB-Fc-treated and Mstn−/− mice. Left: numbers of genes upregulated relative to controls. Right: numbers of genes downregulated relative to controls. There were no genes that were upregulated in 1 of the 3 comparisons and downregulated in another.

A replication cohort of 23 mice (n = 8 Mstn−/− mice, n = 5 for each of the other 3 groups) was subsequently tested to verify the findings from the first set of 12 mice. Although we did not observe a complete concordance of differentially expressed genes between the replication cohort and the initial cohort, a significant fraction of the genes that were altered in the first cohort were also significantly altered in the replication cohort (P < 10−10 by Fisher's exact test for both up- and downregulation for Mstn−/− vs. control and for chronic ActRIIB-Fc treated vs. control).

Validation of array results with quantitative real-time PCR.

To validate our array results, we selected 30 genes from the up- and downregulated gene lists and analyzed their expression levels with qPCR using TaqMan gene expression assays on the BioMark 48.48 Dynamic Array. In addition to genes whose expression levels were dramatically altered, we also picked a few genes with modest changes as controls. Expression levels of these 30 genes were normalized with two different housekeeping genes (Gapdh and B2m) presumed to be stably expressed in the wild-type, ActRIIB-Fc-treated, and Mstn−/− mice. We observed significant similarities between fold change differences determined by the arrays and qPCR. Genes that were determined to be significantly downregulated on arrays were also significantly downregulated by qPCR analysis, except for the Dysfip1 gene, which showed slightly opposing changes in Mstn−/− muscles, with qPCR typically showing larger fold change differences relative to arrays for the majority of tested genes (Table 1). The overall correlation coefficient between array and qPCR data was at least 0.89 in all three groups of mice (Fig. 2). Moreover, the correlation between the qPCR data from the first cohort of mice and the microarray data from the second independent cohort of mice was at least 0.75 in all three comparisons (not shown).

View this table:
Table 1.

Comparison of fold change values detected by arrays versus qPCR

Fig. 2.

Correlation between log2 fold changes from microarrays and from quantitative real-time polymerase chain reaction (qPCR) for acute ActRIIB-Fc treated vs. control (left); chronic ActRIIB-Fc treated vs. control (center); and Mstn−/− vs. control (right). qPCR values are normalized relative to B2m.

Protein analysis.

To determine whether changes at the mRNA level were also reflected at the protein level, we quantified with Western blotting assays the expression levels of nine proteins (α-T-catenin, R-cadherin, Igfbp5, Fhl-1, Myozenin 2, Crp3, Troponin I, Myh7, and Myl3) encoded by genes selected arbitrarily from the lists of up- and downregulated genes. Only for myosin light chain 3 (Myl3) did protein levels mirror the changes in transcript levels in some but not all tested mice in each ActRIIB-Fc-treated and Mstn−/− group, respectively (data not shown). The poor correlation between the mRNA and protein levels for the remaining eight proteins might be explained by the limited elapsed time since the injection of ActRIIB-Fc relative to an estimated myofibrillar protein half-life of 3 wk in mature mice, which is apparently not affected by myostatin deficiency (31). In addition, the higher sensitivity of mRNA analysis compared with protein quantification assays may also have been a factor.

Biological processes affected by ActRIIB-Fc treatment.

To determine biological pathways influenced by myostatin gene deficiency and inhibition with ActRIIB-Fc, we analyzed the sets of genes that showed at least 1.25 fold change differences between groups with Ingenuity Pathway Analysis software. The most significantly affected 10 canonical pathways in chronic ActRIIB-Fc-treated and Mstn−/− quadriceps are listed in Tables 2 and 3, respectively.

View this table:
Table 2.

Top 10 canonical pathways affected by chronic ActRIIB-Fc treatment in quadriceps muscle identified by Ingenuity Pathway Analysis

View this table:
Table 3.

Canonical pathways affected by Mstn gene deletion in quadriceps muscle identified by Ingenuity Pathway Analysis

Genes playing a role in oxidative phosphorylation and mitochondrial function were among the top two downregulated pathways in both ActRIIB-Fc-treated and Mstn−/− mice (Supplemental Table S2 and Supplemental Fig. S1). A heat map of genes involved in oxidative phosphorylation is shown in Fig. 3. Similar changes in oxidative phosphorylation in Mstn−/− mice have also been reported by Chelh et al. (4), using custom oligonucleotide array and two-dimensional gel electrophoresis. A shift in fiber type composition from slow, oxidative fibers to fast, glycolytic fibers has been described in Mstn−/− mice and may account for gene expression changes in these animals (7). However, no fiber type switching has been observed in postnatal myostatin inhibition or with ActRIIB-Fc treatment (3). Similarly, impairment in mitochondrial function and energy metabolism was observed in Mstn−/− muscle fibers (22). Expression of genes known to be important for mitochondrial function is reduced in both ActRIIB-Fc and Mstn−/− muscle as displayed in Fig. 4.

Fig. 3.

Heat map of 118 genes involved in oxidative phosphorylation pathway, in (from left to right) control, acute ActRIIB-Fc-treated, chronic ActRIIB-Fc-treated, and Mstn−/− mice. Color bar at top shows range of expression values from decreased expression (red) to increased expression (green) relative to average values in controls. rep, Replicate.

Fig. 4.

Heat map of 1,269 genes with Gene Ontology (GO) annotation “Mitochondria” in (from left to right) control, acute ActRIIB-Fc-treated, chronic ActRIIB-Fc-treated, and Mstn−/− mice. Color bar at top shows range of expression values from decreased expression (red) to increased expression (green) relative to average values in controls.

Another class of genes that was altered in both ActRIIB-Fc-treated and Mstn−/− mice involves the ubiquitin-proteasome proteolytic pathway. Ubiquitin is a small protein tag that directs unneeded proteins to the proteasome for degradation and recycling. Reduced protein degradation has been suggested to be one of the mechanisms of ActRIIB-Fc-induced muscle growth (35). Using Northern assays, Zhou et al. (35) demonstrated sharp reduction in ubiquitin and muscle-specific ubiquitin ligases atrogin-1/MAFbx and MuRF1 mRNAs in cachexic mice on treatment with ActRIIB-Fc. The mRNA expression levels of atrogin-1/MAFbx and MuRF1 were also reduced 1.5-fold in acutely treated mice, with P values of 0.0004 and 0.01, respectively, although their expression levels were not significantly altered in chronically treated and myostatin-null mice. Thus protein ubiquitination, an important precursor step for protein degradation, appears to be a transcriptional target of myostatin and perhaps of other ActRIIB ligands. Differentially expressed genes from these three pathways are listed in Supplemental Table S2.


Muscle degenerative and wasting disorders, including muscular dystrophy, sarcopenia, and cachexia, reduce the quality of life and survival of affected individuals and currently lack effective therapies. Pharmacological inhibition of the myostatin signaling pathway is considered a promising therapeutic approach to improve muscle mass and function in patients with these conditions. ActRIIB-Fc is a potent inhibitor of myostatin as well as other members of the TGF-β superfamily and has recently entered clinical trials. In an attempt to identify mRNA-based biomarkers of the skeletal muscle in response to ActRIIB-Fc, we assessed gene expression signatures in mice treated with ActRIIB-Fc and compared the changes in gene expression of control and Mstn−/− mice.

Previous studies using microarray and proteomic approaches in mice with targeted disruption of the myostatin gene in the developing mouse embryo have yielded interesting results elucidating some of myostatin's downstream target genes and pathways (4, 29). In contrast to these studies, we used a new generation of Affymetrix expression arrays to identify differentially expressed genes in mouse skeletal muscles as a consequence of postnatal ActRIIB-Fc treatment. We anticipate that the effects of postnatal and time-dependent treatment with ActRIIB-Fc will be more relevant to clinical use than the effect of the complete genetic lack of the myostatin.

In our study, among the most strongly differentially expressed genes in ActRIIB-Fc-treated and Mstn−/− mice were, predictably, several genes that encode for muscle-specific proteins. Genes encoding muscle structural and contractile proteins, in particular those encoding a number of type I or slow isoforms, showed at least threefold reduction in both chronically ActRIIB-Fc-treated and Mstn−/− mice quadriceps muscles (Table 4). Similar to the study by Steelman et al. (29), there was no concomitant increase in fast-twitch fiber protein gene expression. Several sarcomeric protein genes whose expression was highly reduced in our study in quadriceps muscle (Myh7, Myl3, Myoz2, Tnnc1, Csrp3, Tcap; reviewed in Ref. 15) and Nexn (9) have mutations in human homologs associated with hypertrophic cardiomyopathy. However, no cardiac phenotype has been observed in either ActRIIB-Fc-treated or Mstn−/− mice (6, 16, 19).

View this table:
Table 4.

Genes encoding slow isoforms of muscle proteins are downregulated in ActRIIB-Fc-treated and Mstn−/− mice

In addition to constitutive myostatin-null mice, expression profiling in mice with conditional postdevelopmental myostatin deletion has also been recently explored by Welle and colleagues (33). In contrast to embryonic deletion of the myostatin gene, postdevelopmental deletion of Mstn affects the expression of fewer genes, does not significantly affect contractile protein gene expression, and does not significantly affect expression of mitochondrial proteins. In the present study with postnatal pharmacological treatment with ActRIIB-Fc, we similarly find many fewer expression changes in the chronically treated animals (134) versus the myostatin-null animals (360). We do, however, see a larger downregulation of contractile protein genes in chronically ActRIIB-Fc-treated mice than were reported for the postdevelopmental myostatin deletion mice, and likewise for mitochondrial proteins including those involved in oxidative phosphorylation. Some of these differences may be related to specifics of individual muscles, as quadriceps were used in our study as an example of muscle that responds robustly to ActRIIB-Fc while gastrocnemius muscle was used by Welle et al. (33).

Suppression of mitochondrial and oxidative phosphorylation genes with ActRIIB-Fc treatment or in Mstn−/− muscles is noteworthy, as impaired energy metabolism can have undesirable clinical consequences. Reduction in the expression levels of these genes (Figs. 3 and 4) may be associated with decreased mitochondria number and lower oxygen consumption, as was previously reported in myostatin-deficient muscle (1, 22). Furthermore, the expression level of a key transcription factor, estrogen-related receptor α (Errα), involved in mitochondrial biogenesis that induces hundreds of oxidative phosphorylation pathway genes in skeletal muscle (18) is reduced 1.6 (P = 0.008)- and 1.8 (P = 0.002)-fold in mice chronically treated with ActRIIB-Fc-treated and Mstn−/− muscles, respectively. The expression level of transcriptional coactivator peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α) gene, a major factor that regulates mitochondrial biogenesis together with Errα (18), is also downregulated by 1.7- and 1.5-fold in treated and Mstn−/− mice, although the differences were not statistically significant.

Expression changes of two particular genes, Nos1 and Ddah1, involved in nitric oxide synthesis that may influence muscle function were highlighted by Welle et al. (33). Both of these genes were significantly downregulated in postdevelopmental myostatin gene deletion. We also detected statistically significant downregulation of the Nos1 gene in quadriceps from all three groups of mice (fold change values in acutely treated mice −2.2, P = 8 × 10−8; chronic −1.6, P = 2 × 10−5; Mstn−/− −1.4, P = 0.0003). Ddah1 was downregulated in each comparison (acute −1.6, chronic −1.4, and Mstn−/− −1.5), at nominal P < 0.001. Postnatal treatment with a neutralizing myostatin antibody was also shown to reduce the expression levels of these two genes (32), potentially making the Nos1 and Ddah1 genes good candidate biomarkers for ActRIIB-Fc treatment and postnatal myostatin inhibition. Interestingly, Zmynd17 was among the few genes that showed more than twofold reduction in postdevelopmental myostatin deletion and was also downregulated in our study in mice treated both acutely (−4.3-fold, P = 1.7 × 10−8) and chronically (−3.3-fold, P = 2.1 × 10−7) with ActRIIB-Fc. Furthermore, the expression of this gene was markedly reduced in mice treated with a neutralizing myostatin antibody (32). The function of Zmynd17 is currently unknown. We propose that this gene could also serve as an excellent biomarker for ActRIIB-Fc treatment.

Two cell-cell adhesion protein genes, Ctnna3, the gene that encodes for α-T-catenin, and Cdh4 encoding cadherin 4, were upregulated in mice chronically treated with ActRIIB-Fc as well as in Mstn−/− mice (Table 1). The catenins are ubiquitously expressed cytoplasmic proteins that associate with E-cadherin at cellular junctions. Catenin/cadherin complexes play an important role in mediating cellular adhesion. α-T-catenin binds to α-E-catenin as well as to β-catenin, and it functions to inhibit Wnt signaling. The upregulation of these genes is in agreement with a previous study that demonstrated that lack of myostatin inhibits the Wnt signaling pathway (29).

Insulin-like growth factor (IGF) II and the IGF binding protein 5 (Igf2 and Igfbp5) genes were upregulated in ActRIIB-Fc-treated mice and in Mstn−/− mice (Table 1). The IGF-activated signaling pathway is one of the best-understood pathways regulating skeletal muscle differentiation and growth. Increased IGF2 expression due to a regulatory polymorphism is associated with large muscle mass in pigs (30), and IGFBP5 was shown to mediate myostatin-induced inhibition of porcine embryonic myogenic cell proliferation (11). Furthermore, these two genes are believed to coregulate muscle growth through an autoregulatory loop (24). A recent study revealed significant differences in the expression levels of several IGF and IGF components, including IGF2 and IGFBP5, in muscle, heart, liver, and serum of Mstn−/− mice compared with control mice (34).

In addition to myostatin, other members of the TGF-β superfamily have been shown to bind ActRIIB-Fc (14, 20). Moreover, treatment of Mstn−/− mice with ActRIIB-Fc resulted in further enlargement of muscle size, demonstrating that ActRIIB-Fc modulates a factor regulating muscle growth in addition to myostatin (14). One of the aims of our study was to compare differentially expressed genes between ActRIIB-Fc-treated and Mstn−/− mice. Genes that are affected by ActRIIB-Fc treatment but not by myostatin gene knockout could provide clues to potential downstream targets of other TGF-β family members implicated in muscle growth and suggest novel therapeutic targets for muscle disorders. We identified 36 genes differentially affected by either ActRIIB-Fc treatment or myostatin gene deficiency (Table 5). Of these 36 genes, six (Zmynd17, Igf2, Cdkn1c, Igfn1, Edn1, H19) were significantly (P < 0.001) altered only in chronically treated mice but not in Mstn−/− mice. The overall similarity of gene expression in chronic ActRIIB-Fc-treated and myostatin gene-deleted animals is striking: of the 134 genes significantly altered in chronic ActRIIB-Fc-treated animals compared with control animals, 66% (88 of 134) were also significantly altered in Mstn−/− mice compared with control mice, and for 87% (117 of 134) the expression level in Mstn−/− mice was more similar to the level in chronic ActRIIB-Fc-treated animals than in control mice. This suggests that the predominant biological effect of ActRIIB-Fc on muscle occurs via inhibition of a common TGF-β pathway shared by myostatin and other ActRIIB-Fc binding ligands.

View this table:
Table 5.

Genes differentially affected by ActRIIB-Fc treatment and myostatin gene deletion

Clinical trials with ActRIIB-Fc have been initiated in disease populations. Although their ultimate success will be judged by improvement in function and quality of life, analysis of the trials will be aided by determining whether a biological response was observed in various cohorts. Biomarker identification in ActRIIB-Fc-treated muscles is one approach to this clinical trial challenge. Differentially expressed genes determined in this study provide candidate biomarkers to correlate with gene expression signatures obtained in human clinical trials of ActRIIB-Fc. Finally, analysis of the changes in expression following ActRIIB-Fc should aid in our efforts to understand the molecular mechanisms by which TGF-β family members regulate skeletal muscle growth.


Microarray expression experiments were performed by the Microarray Core Facility of the Molecular Genetics Core Facility at Children's Hospital Boston supported by National Institutes of Health (NIH) Grants P50-NS-40828 and NIH-P30-HD-18655. This study was supported by NIH Grant 5U54-HD-060848 to the Senator Paul D. Wellstone Muscular Dystrophy Cooperative Research Center for Facioscapulohumeral Muscular Dystrophy (FSHD) at the Boston Biomedical Research Institute.


No conflicts of interest, financial or otherwise, are declared by the author(s).


We thank the staff of the Molecular Genetics Core Facility at Children's Hospital Boston. We thank Acceleron Pharma, Inc. for generously providing ActRIIB-Fc. We also thank Genri Kawahara and Anu Subramanian for their technical assistance with Western blotting experiment setup and for their insightful comments.


  • 1 Supplemental Material for this article is available online at the Journal website.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
View Abstract