Increased bone mineral density in the femora of GDF8 knockout mice


  • Mark W. Hamrick

    Corresponding author
    1. Department of Cellular Biology and Anatomy, Medical College of Georgia, Augusta, Georgia
    • Department of Cellular Biology and Anatomy, Laney Walker Blvd., CB2915, Medical College of Georgia, Augusta, GA 30912
    Search for more papers by this author
    • Fax: (706) 721-6120


GDF8 (myostatin), a member of the transforming growth factor (TGF)-β superfamily of secreted growth and differentiation factors, is a negative regulator of skeletal muscle growth. GDF8 knockout mice have approximately twice the skeletal muscle mass of normal mice. The effects of increased muscle mass on bone modeling were investigated by examining bone mineral content (BMC) and bone mineral density (BMD) in the femora of female GDF8 knockout mice. Dual-energy X-ray absorptiometry (DEXA) densitometry was used to measure whole-femur BMC and BMD, and pQCT densitometry was used to calculate BMC and BMD from cross-sections taken at two different locations: the midshaft and the distal metaphysis. The DEXA results show that the knockout mice have significantly greater femoral BMD than normal mice. The peripheral quantitative computed tomography (pQCT) data indicate that the GDF8 knockout mice have approximately 10% greater cortical BMC (P = .01) at the midshaft and over 20% greater cortical BMC at the metaphysis (P < .001). Likewise, knockouts show approximately 10% greater cortical thickness (P < .001) and significantly greater cortical BMD (P < .001) at both locations. These results suggest that inhibitors of GDF8 function may be useful pharmacological agents for increasing both muscle mass and BMD. Anat Rec Part A 272A:388–391, 2003. © 2003 Wiley-Liss, Inc.

The forces imposed upon bones by muscles are significantly larger than those gravitational forces associated with body mass, which suggests that muscle strength should be a primary determinant of peak bone strain and therefore a primary factor in determining bone strength (Burr, 1997). Schoenau and Frost (2002) argued that the largest mechanical loads on bones dominate the postnatal development of bone strength, and the largest loads on bones are in turn proportional to muscle mass (Frost, 1997, 1998). However, it is also clear that load frequency and strain rate are just as important as (if not more important than) load magnitude in stimulating bone modeling. Indeed, low-magnitude, high-frequency vibration has been found to increase trabecular bone density in the adult sheep femur by >30% (Rubin et al., 2001, 2002). Turner et al. (1994) have shown that if load frequency is increased, but strain magnitude is held constant, bone formation rate increases significantly. As Burr et al. (2002, p. 785) noted, “high strain magnitudes are not required to stimulate bone adaptation if the strain rate is sufficiently high.”

We attempted to further clarify the complex relationship between muscle mass and bone strength using a new mouse model: mice that lack GDF8 (myostatin). GDF8 is a negative regulator of skeletal muscle growth, and knockout mice that lack GDF8 have approximately twice the skeletal muscle mass of normal mice at both 2 and 10 months of age (McPherron et al., 1997; McPherron and Lee, 1997). The effect of GDF8 appears be dose-dependent, as mice heterozygous for the disrupted GDF8 sequence have muscle weights that are intermediate between those of normal mice and mice homozygous for the GDF8 mutation (McPherron and Lee, 2002). In situ hybridization data show that GDF8 is first expressed in mouse embryos in the myotome compartment of somites, and GDF8 transcripts can still be detected in adults (McPherron et al., 1997; Ji et al., 1998). GDF8 expression during normal growth and development has so far only been detected in skeletal muscle, although it was recently observed in bone immediately following fracture (Cho et al., 2002). GDF8 knockout mice do not differ from normal mice (relative to body weight) in metabolic rate, food consumption, or body temperature (McPherron and Lee, 2002).

In a preliminary study (Hamrick et al., 2000) we examined cross-sectional geometry of the femur at two sites—the third trochanter and the femoral midshaft—in a mixed-sex sample of adult normal and GDF8 knockout mice. We found that the knockout mice differed from normal mice primarily in expansion of the third trochanter. We did not, however, collect any data on bone mass or density from these specimens. The objective of the present report is to provide new data on bone mineral content (BMC), bone mineral density (BMD), and femoral morphology in GDF8 knockout mice to increase our understanding of the effects of increased muscle mass on bone modeling in the limb skeleton.



The sample used in this study included 15 female wild-type CD-1 mice (4 months old) and 15 female CD-1 mice (4 months old), homozygous for the disrupted GDF8 sequence. The myostatin gene was originally disrupted in hybrid 129/SvJ/C57BL/6J mice by homologous targeting in embryonic stem cells, as described previously (McPherron et al., 1997). The entire mature C-terminal region was deleted and replaced by a neo-cassette to ensure that the experimental mice would be null for myostatin function. The myostatin-null mutation was later crossed into a line of CD-1 outbred mice. Myostatin knockout mice on the CD-1 background were studied here because these animals tend to be better breeders than the hybrid mice.

Animals were collected at the age of 4 months because studies on inbred mouse strains indicate that mice reach peak bone density at this time (Beamer et al., 1996). Animals were killed by CO2 overdose, as approved by the Medical College of Georgia, and body weights were recorded immediately afterward. The quadriceps muscles (vastus lateralis, vastus intermedius, vastus medialis, and rectus femoris) were dissected from the right femur immediately after the mice were euthanized, cleaned of excess connective tissue, and weighed to the nearest .001 g. The quadriceps femoris was chosen for analysis because it is the largest muscle group in the mouse hindlimb, and is therefore likely to account for a large portion of the muscle-induced stresses imposed upon the femur. Right femora were fixed in 10% buffered formalin and stored in 70% ethyl alcohol for densitometry. The femur length, midshaft mediolateral and anteroposterior diameter, and third trochanter diameter were measured by digital calipers in each specimen prior to densitometry.

Densitometry and Geometry

DEXA densitometry (PIXImus system) was used to measure whole BMC and BMD from the right femora. The PIXImus dual-energy x-ray absorptiometry system allows accurate measurement of BMC and BMD from small laboratory animals using a relatively low x-ray energy (80/35 kVp) and ultra-high resolution (.18 × .18 mm pixel size) to achieve contrast in low-density mouse bone. Replicability data indicate an excellent correlation (.99) between PIXImus (GE Medical Systems, Milwaukee, WI) BMC and total ashed weight. pQCT densitometry was then used to calculate BMC, BMD, and cortical thickness from individual cross-sections at two different locations: the midshaft and the distal metaphysis. The midshaft is primarily cortical bone, whereas the distal metaphysis has a high proportion of trabecular bone. The pQCT technique is known to be a very useful and reliable approach for measuring trabecular and cortical bone area, mineral content and density, and cross-sectional geometry in small rodent bones (Ferretti, 1995; Gasser, 1995; Beamer et al., 1996; Hamrick et al., 2002). Cross-sections (1 mm thick) were scanned at 4 mm/sec, with a voxel size of .070 mm and a threshold value of 524.0 mg/cm3 used to distinguish trabecular from (sub)cortical bone.

Means and standard deviations were determined for all measured and calculated parameters using SYSTAT™ (Richmond, VA) software. All variables were compared between experimental and control animals using a single-factor analysis of variance (ANOVA) with genotype as the factor.


The control and knockout mice included in the sample did not differ significantly from one another in body mass, but the quadriceps muscles of GDF8 deficient mice were approximately 60% larger than those of the controls (Table 1). Linear dimensions indicate that the femora of the knockout mice are about 3% shorter than those of normal mice, and their third trochanters are expanded mediolaterally compared with those of normal mice (Table 1). Likewise, the femoral midshafts of the knockout mice are broader mediolaterally than those of normal mice, but the midshafts of the knockout mice are reduced anteroposteriorly (Table 1). The DEXA results indicated that female mice lacking GDF8 show significantly increased BMD in their femora compared to normal mice (Table 2). pQCT data from the midshaft and distal metaphysis are consistent with the DEXA results. Cross-sectional slices from the midshaft show significant increases in cortical BMD, BMC, and thickness in the knockout mice (Table 2). Cross-sectional slices from the distal metaphysis reveal even greater differences between the two groups: the increase in cortical BMC is >20% in the knockouts, and cortical BMD and thickness are also significantly greater in the experimental animals (Table 2, Fig. 1). The greater difference between normal and experimental animals observed in the distal femur is consistent with previous studies (e.g., Hsieh et al., 2001) showing greater bone strains and bone formation rates in the more distal regions of limb bones. The slices from the distal metaphysis also show an increase in trabecular BMD (Fig. 1), but this difference is not statistically significant (Table 2).

Table 1. Summary statistics [mean (S.D.)] for body mass, quadriceps mass, and femoral dimensions of control (+/+) and GDF-8 deficient (−/−) mice
ParameterControl (+/+) (n = 15)GDF-8 deficient (−/−) (n = 15)P
Body mass (g)37.4 (3.9)37.6 (4.6).93
Quadriceps mass (g).25 (.03).42 (.05)<.001
Femur length (mm)17.4 (.6)16.84 (.6)<.05
Mediolateral shaft diameter (mm)1.81 (.09)1.90 (.12).05
Anteroposterior shaft diameter (mm)1.49 (.10)1.41 (.07)<.05
Third trochanter diameter (mm)2.21 (.18)2.51 (.18)<.001
Table 2. Summary statistics [mean (S.D.)] for densitometric properties of control (+/+) and GDF-8 deficient (−/−) mice
ParameterControl (+/+) (n = 15)GDF-8 deficient (−/−) (n = 15)P
 BMD (g/cm2).062 (.004).070 (.004)<.001
 BMC (g).032 (.004).034 (.003).13
 Cortical thickness (mm).28 (.01).31 (.02)<.001
 Cortical BMD (mg/cm3)1144.9 (37.0)1185.8 (24.1)<.01
 Cortical BMC (mg/mm)1.29 (.12)1.41 (.12)<.05
Distal metaphysis   
 Cortical thickness (mm).26 (.01).29 (.03)<.001
 Cortical BMD (mg/cm3)898.3 (35.5)995.5 (48.4)<.001
 Cortical BMC (mg/mm)1.2 (.1)1.5 (.1)<.001
 Trabecular BMD (mg/cm3)125.5 (32.1)146.0 (30.2).08
 Trabecular BMC (mg/mm).34 (.08).30 (.08).22
Figure 1.

pQCT sections from the distal femur metaphysis of normal (left column) and GDF8 knockout (right column) mice. Note the increased trabecular bone (red, orange) and increased cortical bone (white) in the knockout animals.

The increased cortical BMD and BMC observed in the present study were unexpected, especially since the animals included did not differ significantly in total body mass (primarily because the knockouts do not gain fat (McPherron and Lee, 2002)). It is well known that body mass is one of the best predictors of bone mass in both men and women (Felson et al., 1993). If the control and experimental animals are weight-matched they are not expected to differ in BMC and density, since bone mass should be proportional to body mass. However, body composition can play a major role in determining bone mass. Most studies on this subject have demonstrated a stronger relationship between fat mass and BMD than between lean (muscle) mass and BMD, which may be due in part to the fact that fat cells (adipocytes) produce hormones (e.g., leptin and estrogen (Thomas and Burguera, 2002)) that are anabolic to bone. The results presented here suggest the opposite; that is, increased muscle mass is associated with increased BMD. This result is consistent with Young et al.'s (2001) finding that lean mass is most strongly correlated with bone mass in young girls until about age 18, after which fat mass is more strongly correlated with BMD.

We are currently investigating bone strength, densitometry, and geometry in other regions of the skeleton, such as the lumbar vertebrae and the femoral neck, in order to investigate the potential use of GDF8 inhibitors as pharmacological agents for increasing BMD and bone strength in the skeleton. Recent experiments using mouse models have shown that inhibitors of GDF8 function have the potential to slow and/or prevent the development of obesity, type 2 diabetes, and muscle-wasting disorders such as cachexia (McPherron and Lee, 2002; Zimmers et al., 2002). The results presented here suggest that some of the benefits associated with GDF8 deficiency include localized increases in BMC and BMD in addition to decreased body fat and increased muscle mass.


We are grateful to Dr. Gary Schneider, NEOUCOM, for providing us access to the pQCT laboratory facility under his supervision, and to K. Grecco for helpful assistance in collecting the pQCT data. The PIXImus data were collected in the lab of Dr. Carlos Isales, and we appreciate the use of the PIXImus facility. Drs. Alexandra McPherron and Se-Jin Lee provided helpful advice and assistance with the knockout mice.