Endocrinology and Metabolism

Muscle growth after postdevelopmental myostatin gene knockout

Stephen Welle, Kirti Bhatt, Carl A. Pinkert, Rabi Tawil, Charles A. Thornton


Constitutive myostatin gene knockout in mice causes excessive muscle growth during development. To examine the effect of knocking out the myostatin gene after muscle has matured, we generated mice in which myostatin exon 3 was flanked by loxP sequences (Mstn[f/f]) and crossed them with mice bearing a tamoxifen-inducible, ubiquitously expressed Cre recombinase transgene. At 4 mo of age, Mstn[f/f]/Cre+ mice that had not received tamoxifen had a 50–90% reduction in myostatin expression due to basal Cre activity but were not hypermuscular relative to Mstn[w/w]/Cre+ mice (homozygous for wild-type myostatin gene). Three months after tamoxifen treatment (initiated at 4 mo of age), muscle mass had not changed from the pretreatment level in Mstn[w/w]/Cre+ control mice. Tamoxifen administration to 4-mo-old Mstn[f/f]/Cre+ mice reduced myostatin mRNA expression to less than 1% of normal, which increased muscle mass ∼25% over the following 3 mo in both male and female mice (P < 0.005 vs. control). Fiber hypertrophy appeared to be sufficient to explain the increase in muscle mass. The pattern of expression of genes encoding the various myosin heavy-chain isoforms was unaffected by postdevelopmental myostatin knockout. We conclude that, even after developmental muscle growth has ceased, knockout of the myostatin gene induces a significant increase in muscle mass.

  • Cre recombinase
  • tamoxifen
  • muscle fiber hypertrophy
  • myosin heavy chains
  • conditional knockout

myostatin [also known as growth differentiating factor 8 (GDF-8)] is a member of the transforming growth factor-β (TGF-β) family. In mice, constitutive knockout of the third exon of the myostatin gene, which encodes the active portion of the peptide, leads to a marked increase (∼2-fold) in skeletal muscle bulk (8, 14). Excessive muscle growth has also been observed in mice with conditional myostatin knockout dependent on myoblast differentiation (4), expression of a dominant negative form of myostatin (mutation that prevented normal cleavage of the propeptide) (18), expression of a dominant negative form of the myostatin receptor (7), constitutive knockout of myostatin receptors (6), or expression of proteins that bind myostatin [follistatin (7) and the portion of the myostatin propeptide on the NH2-terminal side of its cleavage site (7, 17)]. All of these transgenic mice have reduced myostatin activity very early in life, so that these studies of genetically altered mice have not elucidated whether reducing myostatin activity postdevelopmentally can cause muscle growth. A few studies have examined the consequences of reducing myostatin activity after mice were weaned, by systemic administration of anti-myostatin antibodies or other proteins that bind to myostatin (13, 6, 15, 16). Increased muscle mass was observed when myostatin activity was inhibited after weaning, generally in the range of 10–30% depending on the length of treatment and age of the mice. Most of this research involved mice young enough that muscle was still growing in the untreated controls (treatment initiated at 4–10 wk of age). Only one published study (15) reported effects of an anti-myostatin antibody in mice old enough that developmental muscle growth would have ceased before treatment (started at 24 wk of age). In that study, there was an increase of muscle mass of only 12% after antibody administration for 5 wk, which might be too short a period to observe the full effect of myostatin deficiency in mature muscle.

Because administration of antibodies or other proteins that inactivate myostatin requires a large supply of proteins that must be administered regularly, and because the specificity and extent of myostatin inactivation caused by administration of these proteins in vivo is uncertain, alternative approaches would be useful for further research on the physiological and molecular effects postdevelopmental inhibition of myostatin activity. Thus we generated a line of mice in which a critical segment of the myostatin gene is flanked by loxP (floxed) (14). Introduction of an inducible Cre recombinase transgene into this line should permit knockout of the floxed segment after muscles have matured. Potential advantages of this approach are the specificity of the genetic effect, having a measure of the extent of myostatin deficiency (i.e., residual myostatin expression) and the reduced cost and effort required to study the effect of long-term myostatin deficiency. Here, we show that a tamoxifen-inducible Cre recombinase consistently reduces myostatin expression to less than 1% of normal and that postdevelopmental myostatin deficiency causes a significant increase in muscle mass, although a limitation with this particular Cre is a significant loss of myostatin expression prior to tamoxifen administration.


Generation of mice in which exon 3 of the myostatin gene is floxed was described previously (14). In the present study, we used mice with floxed (f) alleles from which the neomycin resistance gene had been deleted. Thus the f allele was the same as the wild-type allele (w) except for replacement of 37 bases in the second intron with a loxP sequence (34 bases) and replacement of ∼150 bases 0.8 kb downstream of exon 3 with ∼100 bases (including a loxP sequence) from the gene targeting vector (pKI, constructed and donated by Dr. Lin Gan). The 129S6/SvEvTac genome from the embryonic stem cells used for homologous recombination had been crossed into the C57BL/6 background at least four generations before the CAGGCre-ER transgene was introduced into these lines. Both the Mstn[f/f] and Mstn[w/w] lines were descended from Mstn[f/w] founders, so that 129S6/SvEvTac genes other than myostatin (and neighboring genes in linkage disequilibrium) had an equal chance of being introduced into both lines.

A mouse (congenic with the C57BL/6 strain) with the CAGGCre-estrogen receptor (ER) transgene (5) was purchased from The Jackson Laboratory [B6.Cg-Tg(cre/Esr1)5Amc/J] to start the Mstn[f/f]/Cre+ and Mstn[w/w]/Cre+ lines. Mice homozygous for the CAGGCre-ER transgene are not viable, so the lines have been maintained by mating Cre+ mice with Cre− mice. The CAGGCre-ER transgene is expressed ubiquitously (5). Transcription is driven by a chimeric promoter of the cytomegalovirus immediate-early enhancer and the chicken β-actin promoter/enhancer (CAGG). The Cre recombinase is fused with a mutant form of the ligand binding domain of the estrogen receptor with a low affinity for estradiol. The Cre-ER protein stays in the cytoplasm, sequestered by HSP90, in the absence of ligand binding to the ER domain. Upon association of 4-OH-tamoxifen (produced in vivo from injected tamoxifen) with the ligand binding domain, the fusion protein dissociates from HSP90 and is free to enter the nucleus and excise DNA flanked by loxP sequences.

For genotyping, tail biopsies were obtained at weaning (3 wk of age) and digested overnight at 55°C with proteinase K (0.3 mg/ml in 10 mM Tris, pH 9, 50 mM KCl, 0.1% Triton X-100, total volume 200 μl per tail) and then heated at 95°C for 25 min to inactivate proteinase K. Insoluble material was removed by centrifugation, and 1 μl of supernatant was used as the template in 40-μl PCR reactions. Primers are listed in Table 1. The altered DNA sequence in the downstream loxP region was the basis for distinguishing f and w alleles (∼550 bp amplicon for f allele, ∼600 bp for w allele). Excision of exon 3 was detected with three-primer PCR: a forward primer upstream of the 5′ loxP sequence, reverse primer A in the floxed DNA, and reverse primer B downstream of the 3′ loxP sequence. Primer A generated an amplicon of ∼300 bp if exon 3 was present. If there was no excision of exon 3, there was no amplicon from primer B because of its long distance from the forward primer. If exon 3 was excised, primer B produced an amplicon of ∼400 bp.

View this table:
Table 1.

PCR primers and probes

At 4 mo of age, male and nulliparous female Mstn[f/f]/Cre+ and Mstn[w/w]/Cre+ mice either were killed for examination of muscle mass and myostatin expression (n = 8 mice per genotype) or were treated with tamoxifen (n = 11 mice per genotype). Tamoxifen was dissolved by sonication in corn oil at a concentration of 20 mg/ml, and the solution was stored in a refrigerator for no longer than 1 wk. Intraperitoneal injections of 0.25 ml of this solution (5 mg tamoxifen) were administered at 2- to 3-day intervals until eight injections had been given. Tamoxifen-treated mice were examined 3 mo after the final injection.

Mice were euthanized by CO2 inhalation until cessation of breathing followed by cervical dislocation. Gastrocnemius and quadriceps muscles were rapidly dissected from both legs, weighed, and immediately immersed in liquid nitrogen. Quadriceps of six tamoxifen-treated mice (2 male and 1 female f/f, 2 male and 1 female w/w) were immersed in melting isopentane (prechilled in liquid nitrogen) rather than liquid nitrogen, so that they could be used for histology. Muscles were stored at −75°C.

RNA was extracted from gastrocnemius muscles and quantified by UV absorbance, as described previously (14). The amount of RNA recovered from the muscles was not significantly different between genotypes either before (μg/mg tissue was 0.56 ± 0.02 in w/w, 0.64 ± 0.03 in f/f) or after (0.75 ± 0.04 in w/w, 0.67 ± 0.05 in f/f) tamoxifen administration. cDNA was generated from 1.5 μg of total RNA with MMLV reverse transcriptase and oligo(dT) primer. Quantitative real-time PCR (qRT-PCR) was done with the ABI Prism 7900HT Sequence Detection System and SDS 2.2 software. The cDNAs for myostatin, GAPDH, and various isoforms of myosin heavy chain (MyHC 1, 2a, 2b, 2x) were quantified with the primers and probes listed in Table 1. Except for the MyHC 1 forward primer, and use of 3′ Iowa Black quencher instead of 3′ TAMRA, these were the same as the MyHC primers and probes used by Girgenrath et al. (3). With equal amounts of cDNA in the PCR reactions, mean cycle thresholds for GAPDH mRNA were similar (within 0.3 cycles) in Mstn[f/f]/Cre+ and Mstn[w/w]/Cre+ mice both before and after tamoxifen treatment; so GAPDH mRNA was used as the reference in the relative quantification of the other transcripts.

Cryostat sections of (10 μm thick) were cut perpendicular to the long axis of the quadriceps muscles, near the midsection. Hematoxylin and eosin (H&E) staining was used to examine general morphological features. For quantitative analysis of fiber cross-sectional areas, sarcolemmal membranes were stained with an anti-caveolin-3 rabbit polyclonal antibody (Abcam) and Cy-3-labeled donkey anti-rabbit secondary antibody (Jackson Immunoresearch). Nuclei were stained with DAPI (Molecular Probes). Images of 200 fibers from each muscle sample were obtained with a Nikon Eclipse E600 fluorescence microscope equipped with a Photometrics Coolsnap HQ camera. Fiber areas were computed from the caveolin images with Metaview v. 6.1r5 (Universal Imaging) software. Myonuclei were counted after the caveolin-3 and DAPI images were merged. Myonuclei were defined as DAPI-stained nuclei judged by the technician to be located inside the sarcolemmal boundaries defined by caveolin-3 staining. All analyses of fiber size and myonuclei counts were performed by the same technician, who did not know the genotype or sex of the mice.

Statistical significance of differences between groups was assessed by t-tests. Values are presented as the mean ± 1 SE.


Mstn[f/f], Mstn[f/f]/Cre+, and Mstn[w/w]/Cre+ mice are healthy and breed normally. When Mstn[f/f]/Cre+ mice were weaned at 3 wk of age, typically there was no detectable Cre-mediated excision of the floxed exon in the tail, confirming that there was no germ-line knockout in the Cre+ parent. However, PCR analysis of tail DNA occasionally revealed Cre-mediated recombination, which could be due either to germ-line Cre activity in the parent (heterozygous knockout) or Cre activity at embryonic, fetal, or postnatal stages (chimeric mice). Mice with Cre-mediated recombination in the tail were not included in the study.

There was some Cre activity in muscle in the absence of tamoxifen, as reflected by recombination of DNA (not shown) and by a 50–90% reduction in myostatin mRNA expression at 4 mo of age in untreated Mstn[f/f]/Cre+ mice relative to Mstn[w/w]/Cre+ mice (Fig. 1). The pre-tamoxifen decline in myostatin expression was greater in female mice, probably because of higher endogenous estrogen levels. Premature recombination is most likely a very gradual stochastic process, since initial studies of DNA recombination in muscles of Mstn[f/w]/Cre+ mice at ∼6 wk of age indicated that the abundance of the f allele was generally within 10% of abundance of the w allele. Myostatin mRNA expression was normal in Mstn[f/f] mice with no Cre transgene (Fig. 1), indicating that the reduced myostatin mRNA expression was caused by loss of the floxed DNA rather than by impaired expression from the floxed allele.

Fig. 1.

Mean ± SE myostatin mRNA/GAPDH mRNA ratios in gastrocnemius by real-time quantitative RT-PCR. Myostatin and Cre recombinase genotypes are shown on horizontal axis. Top: percentage of average value in normal males (Mstn[ww]/Cre−). Bottom: log10 relative myostatin expression (normal = 0). ○, Females; •, males; tmx, tamoxifen.

The reduction in myostatin gene expression prior to tamoxifen administration was not associated with a hypermuscular phenotype. There was no significant difference between untreated Mstn[f/f]/Cre+ and Mstn[w/w]/Cre+ female mice in body weight, muscle weights, or muscle mass per gram body weight (Table 2). The mean body and muscle mass of the untreated Mstn[f/f]/Cre+ males was 15% greater than that of the Mstn[w/w]/Cre+ males (Table 2), but muscle mass per gram body weight was not different (Fig. 2).

Fig. 2.

Mean ± SE muscle mass/body mass ratios at 4 mo of age without tamoxifen treatment (pre-tmx) and at 7.5 mo of age, 3 mo after tamoxifen treatment (post-tmx). P, post-tmx comparison between Mstn[w/w]/Cre+ and Mstn[f/f]/Cre+ mice by t-test. P > 0.60 for all pre-tmx f/f vs. w/w comparisons.

View this table:
Table 2.

Body and muscle mass

Three months after the final tamoxifen injection, myostatin mRNA levels consistently were reduced by more than 99% (P < 0.001) in Mstn[f/f]/Cre+ mice relative to mean mRNA levels in Mstn[w/w]/Cre+ mice (Fig. 1). This effect also was observed 5 days after the final tamoxifen injection (not shown), indicating that tamoxifen, rather than longer exposure to basal Cre activity, was responsible for the knockout.

In Mstn[w/w]/Cre+ mice, muscle weights 3 mo after tamoxifen treatment (mice ∼7.5 mo of age) were not significantly different from those of untreated 4-mo-old Mstn[w/w]/Cre+ mice (Table 2). Total body weight continued to increase past 4 mo, so that muscle mass per gram body weight was ∼5–10% lower 3 mo after tamoxifen treatment in these control mice compared with muscle/body weight ratios prior to tamoxifen (Fig. 2).

Loss of myostatin in Mstn[f/f]/Cre+ mice was associated with significant muscle growth. Muscle mass was ∼25% greater 3 mo after tamoxifen treatment whether expressed in absolute terms compared with muscle mass in untreated Mstn[f/f]/Cre+ mice at 4 mo of age (Table 2) or in terms of muscle mass per gram body weight compared with tamoxifen-treated Mstn[w/w]/Cre+ controls (Fig. 2). The effect was similar in males and females, and in both gastrocnemius and quadriceps muscles.

H&E-stained muscle sections of mice with postdevelopmental myostatin knockout were unremarkable except for the larger fiber size. Quantitative analysis was done with sections stained with anti-caveolin-3 antibody to define fiber perimeters and DAPI to label nuclei. Individual fiber size distributions are illustrated in Fig. 3. The mean cross-sectional area of individual quadriceps fibers in tamoxifen-treated Mstn[f/f]/Cre+ mice was 42% greater (P < 0.001; Table 3) than that of same-sex tamoxifen-treated Mstn[w/w]/Cre+ controls (this was not inconsistent with the difference in quadriceps mass shown in Table 2, because the difference in quadriceps mass between the 3 knockout and 3 control muscles examined histologically was greater than the mean difference based on all 11 mice in each group). Myostatin knockout increased the average fiber area per myonucleus more than it increased the number of myonuclei per fiber (Table 3).

Fig. 3.

Cross-sectional fiber areas for quadriceps muscles of tamoxifen-treated Mstn[f/f]/Cre+ (n = 2 male, 1 female) and Mstn[w/w]/Cre+ mice (n = 2 male, 1 female) 3 mo after final tamoxifen injection. Fibers (200) from each muscle are ordered from smallest to largest. Solid lines, f/f males (myostatin-deficient); dashed lines, w/w males (normal myostatin expression); ○, f/f female; ▿, w/w female.

View this table:
Table 3.

Quadriceps muscle fiber size and myonuclear number

Expression of MyHC genes in gastrocnemius muscles was analyzed by qRT-PCR. MyHC 2b mRNA expression was much greater than MyHC 1, MyHC 2a, or MyHC 2x expression, accounting for ∼85% of all MyHC mRNA (Fig. 4). There was no significant difference between genotypes, either before or after tamoxifen treatment, in the expression profile of MyHC isoforms. MyHC 2a mRNA (not shown) accounted for less than 0.1% of total MyHC mRNA in the gastrocnemius in all mice.

Fig. 4.

Mean ± SE expression of types 1, 2x, and 2b MyHC mRNAs in gastrocnemius as %total MyHC mRNA. Type 2a MyHC mRNA < 0.1% of total in all mice (not shown). Open symbols, age 4 mo, no tamoxifen (pre-tmx); closed symbols, age 7.5 mo, 3 mo after tamoxifen treatment (post-tmx); circles, Mstn[f/f]/Cre+ (myostatin-deficient); triangles, Mstn[w/w]/Cre+ (normal myostatin expression).


Postweaning administration of antibodies or other proteins that inactivate myostatin induces muscle growth in mice (13, 6, 15, 16). However, in mice old enough that normal muscle growth has ceased, there was an increase in muscle mass of only ∼12% after 5 wk of treatment with an anti-myostatin antibody (15). In the present study, muscle mass increased ∼25% during 3 mo of >99% myostatin deficiency induced after developmental muscle growth had plateaued. The larger effect with gene knockout than with the anti-myostatin antibody might be explained by the longer period of myostatin deficiency, more effective inhibition of myostatin activity, or both.

Constitutive myostatin deficiency in mice causes both hyperplasia and hypertrophy of muscle fibers. However, when myostatin activity is inhibited after weaning, it appears that fiber hypertrophy is sufficient to explain the increase in muscle mass (1, 2, 6, 15). Our results are consistent with these observations. In principle, the fiber hypertrophy could be caused by expansion of myonuclear domain volumes or by incorporation of new myonuclei from myoblasts. Our results suggest that expansion of myonuclear domain volume contributes more to the fiber hypertrophy than does addition of more myonuclei to the fibers. This observation is consistent with studies of both constitutive myostatin knockout and postweaning inhibition of myostatin activity, in which excessive muscle growth was associated with increased protein/DNA ratios (6, 8, 14) or fiber hypertrophy without an increase in the number of myonuclei (1, 18). Most studies of mechanisms of action of myostatin have focused on myoblast proliferation, but to fully understand the effects of myostatin deficiency we need to learn the mechanism whereby myostatin regulates myonuclear domain volume.

There was reduced myostatin gene expression in 4-mo-old Mstn[f/f]/Cre+ mice prior to tamoxifen treatment, but this was insufficient to cause hypermuscularity. The fact that the myostatin expression prior to tamoxifen was lower in females than in males suggests that endogenous estrogens might be responsible for this Cre activity. Different inducible Cre recombinase transgenes are available and might have lower basal Cre activity with adequate inducibility. We did not detect Cre activity (exon 3 excision) or expression (RT-PCR) in muscle with a doxycycline-inducible Cre transgene (13). We are currently evaluating another tamoxifen-inducible transgene that has additional mutations of the ER-binding domain and a different promoter than the transgene used in the present study (11). Preliminarily, it appears that basal Cre activity in muscle is much lower with this transgene (95% of normal myostatin expression in males, 88% in females, at 4 mo of age), but that more than eight tamoxifen injections might be required to achieve >99% knockout.

We cannot explain why the basal Cre activity in Mstn[f/f]/Cre+ mice, which caused 50–90% loss of myostatin gene expression by the time they were 4 mo old, did not induce hypermuscularity. With in vitro myostatin activity assays, a 90% drop in the myostatin concentration from 200 to 20 ng/ml has hardly any effect, whereas a further decline to 1 ng/ml almost eliminates activity (Ref. 19 and product information from R&D Systems). Thus, if the normal myostatin concentration in mature muscle is sufficiently high, it might be necessary to reduce it by more than 90% to have a significant effect on muscle mass in the absence of any other stimulus for muscle growth. The pre-tamoxifen decline in myostatin expression probably was very gradual, and loss of myostatin protein must lag behind loss of mRNA. Thus it is possible that not enough time with low myostatin had passed for muscle hypertrophy to be detectable in untreated mice at 4 mo of age. We were not able to assess relative myostatin protein levels with typical immunoblotting procedures, a problem also noted recently by another laboratory (9). Several laboratories have published papers reporting muscle myostatin levels based on immunoblotting, but the anti-myostatin antibodies might have detected proteins other than myostatin. After disulfide bonds are reduced, processed myostatin is a monomer of ∼12.5 kDa. This 12.5-kDa monomer was the only protein that yielded a specific myostatin signal (i.e., not present in myostatin knockout mice) among the several serum proteins that reacted with a polyclonal anti-myostatin antibody (19). When myostatin was overexpressed in C2C12 cells, immunoblotting detected the 12.5-kDa monomer (10). We did not detect the monomer in reduced muscle extracts with three different anti-myostatin antibodies (Chemicon AB3239, Santa Cruz SC-6884, Novus Biologicals NB 100-281). These antibodies reacted with higher-molecular-weight proteins, but these same proteins also were present in muscle extracts from mice with constitutive or postdevelopmental myostatin knockout. Thus myostatin either is unstable during tissue storage (−75°C) or extraction (even though protease inhibitors were used in the homogenization buffer) or is below the limit of detection in normal gastrocnemius and quadriceps muscles. This limit was ∼0.5 ng when we used the NB 100-281 antibody with typical immunoblotting methods, as determined by spiking recombinant mouse myostatin (R&D Systems) into muscle extracts. Considering that maximal myostatin activity might be achieved at concentrations of ∼200 pg/μl (19), it would not be surprising if there were less than 0.5 ng of myostatin in ∼1 mg muscle. An attempt to concentrate nonreduced myostatin by immunoprecipitation with the NB 100-281 antibody was unsuccessful (recombinant myostatin spiked into a muscle sample was not recovered).

Constitutive myostatin knockout causes a shift in gene expression from slow to fast myosin isoforms in pectoral, soleus, and extensor digitorum longus (EDL) muscles (3, 12). The gastrocnemius muscle, examined in the present study, has a MyHC expression profile similar to that of EDL. Postdevelopmental myostatin knockout did not significantly affect this profile. This result is consistent with the observation that 12 wk of treatment with an anti-myostatin antibody, started at 10 wk of age, did not alter the MyHC expression profile of EDL muscle in SCID mice (3).

The method for postdevelopmental myostatin knockout was very robust. All Mstn[f/f]/Cre+ mice treated with tamoxifen had myostatin mRNA levels less than 1% of the average of control mice. This line of mice should be useful to investigators who want to examine effects of myostatin deficiency in mice that do not have the severe double-muscled phenotype or type 1 fiber deficiency characteristic of mice with constitutive myostatin knockout or to those who want to differentiate developmental from postdevelopmental effects of myostatin knockout.


This work was supported by National Institutes of Health Grants AG-19853, ES-45533, RR-16286, and NS-48843 and by institutional funds.


We thank Bharati Shah, Don Henderson, and Andrew Cardillo for technical assistance and Drs. Se-Jin Lee and Lin Gan for donating plasmids used to create the gene targeting construct.


  • 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.


View Abstract