The authors state that they have no conflicts of interest.
Introduction: Body weight is positively correlated with bone mass and density, and both muscle mass and body fat are thought to play a role in regulating bone metabolism. We examined bone metabolism in calorically restricted mice to determine how alterations in soft tissue mass affect bone mass, density, and strength.
Materials and Methods: Caloric restriction (CR) was initiated in male mice at 14 wk of age at 10% restriction, increased to 25% restriction at 15 wk, and then increased to 40% restriction at 16 wk, where it was maintained until 24 wk of age when the study was terminated. Control mice were fed ad libitum (AL). Body composition, BMD, and BMC were measured by DXA, BMD and BMC in the femoral metaphysis were measured by pQCT, femora were tested in three-point bending, serum leptin and IGF-1 were measured using immunoassay, and osteoblast and osteoclast numbers were determined using histomorphometry.
Results: Body weight, lean mass, fat mass, percent body fat, serum leptin, and serum IGF-1 were all significantly lower in CR mice than AL mice. Whole body BMC and BMD did not differ significantly between the two groups. Femur BMC, BMD, cortical thickness, and fracture strength decreased significantly in CR mice, but trabecular bone volume fraction in the femur did not change with food restriction. Vertebral cortical thickness also decreased with caloric restriction, whereas spine BMC, BMD, and trabecular bone volume fraction were significantly increased with caloric restriction.
Conclusions: Caloric restriction and its related weight reduction are associated with marked decreases in lean mass, fat mass, serum leptin and IGF-1, and cortical bone mass. Consistent with the opposite effects of leptin on cortical and cancellous bone, trabecular bone mass is spared during food restriction.
Body weight is highly correlated with bone mass and BMD, but the role of body composition (e.g., muscle mass and fat mass) in regulating bone formation and resorption is less clear. A number of studies suggest that body fat has protective effects on bone mass and BMD, particularly in postmenopausal women.(1–3) The positive effects of fat on bone are particularly obvious in those conditions where body fat is decreased significantly. For example, low body weight with conditions such as anorexia nervosa or exercise-induced hypothalamic amenorrhea are associated with reduced bone mass.(4–7) Even less severe, voluntary weight loss is associated with increased rates of bone loss in adults.(8,9) The bone loss that accompanies weight loss is, however, not simply attributable to a decrease in weight-bearing loads on the skeleton. Ballet dancers and girls with anorexia both have lower BMD in non–weight-bearing sites than girls with normal body weight, but the differences in BMD at these sites disappear after adjusting for fat mass, indicating that peripheral body fat is a primary determinant of BMD even in non–weight-bearing regions.(10) Fogelholm et al.(11) also showed that, over a 3-yr period of weight loss and regain in women, BMD at the radius (a non–weight-bearing bone) was more strongly correlated with body weight than was BMD in weight-bearing regions such as the hip and spine. These findings suggest that peripheral body fat can regulate bone metabolism through systemic, circulating factors, in addition to its effects on bone mediated through load-bearing.
It is likely that there are a number of different molecular mechanisms that link peripheral body fat with bone metabolism. Adipose tissue produces estrogens through aromatization of androgens, and fat may positively influence bone mass by serving as a source of estrogen in estrogen-deficient postmenopausal women.(12) High fat mass is associated with hyperinsulinemia, and insulin increases indices of bone formation when administered in vivo.(13) Adipocytes also produce a number of factors including leptin and adiponectin that may directly regulate bone formation by osteoblasts and bone resorption by osteoclasts.(14–16) A number of studies have shown that serum leptin levels are robust predictors of BMD in adult men and women,(17) and the correlation between serum leptin and BMD is even stronger among women than men.(18–20) Leptin treatment can increase serum osteocalcin, testosterone, and growth hormone in food-restricted rodents,(21,22) and leptin treatment restores hormone levels of gonadal steroids and IGF-1 with a concomitant increase in osteocalcin in women with exercise-induced hypothalamic amenorrhea.(23)
One of the most well-documented observations in aging research is the positive effect of caloric restriction (CR) on life expectancy.(24,25) Although caloric restriction extends lifespan in a number of model organisms, it is also associated with bone loss in humans.(26) Likewise, food restriction in rodents leads to reduced skeletal growth and bone loss.(27–29) The purpose of this study is to determine how caloric restriction alters bone mass and bone metabolism. We and others have speculated that trabecular and cortical bone compartments respond differently to weight loss and leptin deficiency, with leptin deficiency preserving trabecular bone volume but decreasing cortical bone mass.(16,30,31) In this study, we use 3 mo of caloric restriction in mice to test this hypothesis.
MATERIALS AND METHODS
Twenty male mice were obtained from the caloric restriction colony at the National Institute on Aging, NIH. Caloric restriction (CR) was initiated in 10 mice at 14 wk of age at 10% restriction, increased to 25% restriction at 15 wk, and then increased to 40% restriction at 16 wk, where it was maintained until 24 wk of age when mice were killed by CO2 overdose following IACUC-approved procedures. Ten control mice were fed ad libitum (AL) and also were killed at 24 wk. CR mice were fed a fortified diet (NIH-31/NIA) that contains the same ingredients as the AL diet (NIH-31), except that the fortified diet contains extra vitamin supplementation to provide CR mice with the same level of micronutrients as mice fed AL (Table 1).
Table Table 1.. Diet Formulations for Mice Fed Ad Libitum (AL; NIH-31 Diet) and Mice on Caloric Restriction (CR; NIH-31/NIA Diet)
DXA analysis, pQCT, and radiography
Body composition and bone densitometry data were collected for each mouse after they were killed using DXA densitometry (GE Lunar PIXImus system). Whole body BMC, BMD, total fat mass (TFM), and fat-free soft tissue mass (FFM) were measured. BMC and BMD of the femur and spine were measured using the PIXImus region of interest option. After mice were killed, the left femur was dissected free and fixed in buffered formalin, and pQCT (Norland Stratec XCT-Research pQCT system) was used to collect densitometric parameters from the femur distal metaphysis. A single cross-section 1 mm thick was scanned at 4 mm/s with a voxel size of 70 μm. pQCT measurements collected include total BMC and total BMD. Trabecular and cortical parameters were collected from histological sections (see below).
Blood was collected by cardiac puncture, allowed to clot in chilled tubes on ice for ∼30 minutes, and centrifuged, and the serum was removed and stored frozen at −80°C. Serum assays were performed as previously described(32) following manufacturer specifications using enzyme immunoassay and radioimmunoassay kits for leptin (Crystal Chem, Downers Grove, IL, USA), IGF-1 (R&D Systems, Minneapolis, MI, USA), PYD (Quidel Corp., San Diego, CA, USA), and osteocalcin (Biomedical Technologies, Stoughton, MA, USA). Any serum that remained (∼3–4 samples per treatment group) was also analyzed for intact PTH levels (Immunotopics, San Clemente, CA, USA).
Right femora of mice were thawed in cold water at room temperature for 1 h and prepared for mechanical testing in three-point anteroposterior bending using a Vitrodyne V1000 Materials Testing system as described previously.(33,34) Femora were mounted on stainless steel fixtures spaced 5 mm apart, 2.5 mm either side of center. Testing was linear displacement control with a displacement rate of 0.10 mm/s using a Transducer Techniques 5-kg load cell. Femora were loaded to failure with data points recorded every 0.01 s. Structural, or extrinsic, properties including ultimate force (Fu; height of curve) and stiffness (S; slope of curve) were calculated from load-displacement curves.
The L4 or L5 vertebra along with the left distal femur were decalcified in EDTA, embedded in paraffin, and sectioned at 4–6 μm for histomorphometry. Alternate sections were stained with H&E for calculation of trabecular BV/TV, trabecular number, trabecular thickness, cortical thickness, and marrow adipocyte density, TRACP (Sigma 386-A) to label osteoclasts, or with an osteocalcin antibody (rabbit anti-mouse polyclonal; Santa Cruz FL-95) to label osteoblasts.(35) Static parameters include osteoclast number as a fraction of tissue area (N.Oc/T.Ar) and osteoblast number as a fraction of tissue area (N.Ob/T.Ar).
Dependent variables were compared between mice fed AL and CR using single-factor ANOVA with genotype as the factor. Because the AL and CR mice were found to differ in body weight, data were adjusted for body weight differences using a general linear model of covariance (ANCOVA) with body mass as a covariate. Coefficients of determination (r2) were calculated for dependent variables using linear regression with body composition and serum leptin and IGF-1 measurements as independent variables.
Effects of caloric restriction on bone and body composition parameters
Caloric restriction reduced body weight by >30% in the mice, and the reduction in body weight was associated with significant reductions in lean mass, fat mass, and quadriceps mass (Table 2). The reductions in muscle mass and body fat are significant not only in absolute terms but also when normalized by body weight (Table 2). Whole body BMC and BMD did not differ significantly between groups (Table 2). Although total BMC and BMD did not differ between groups, calorically restricted animals did show a significant reduction in femur BMC and BMD compared with mice fed ad libitum (Table 3). Serum osteocalcin was significantly decreased in the CR mice, even when normalized by total body BMC, and serum PYD was increased in the CR mice when adjusted for total BMC (Table 2). Serum PTH levels were similar between groups, ranging from 140–190 ng/ml in the AL mice to 105–190 ng/ml in the CR mice. Caloric restriction was associated with a significant decrease in femur midshaft diameter, cross-sectional area, cortical thickness, and second area moment of inertia (Figs. 1 and 2; Table 3). The reduction in bone cross-sectional size was also associated with a significant decrease in bone strength and stiffness (Fig. 2). Analyses of covariance indicate that the two groups of mice do not differ significantly from one another in femur BMC, cross-sectional dimensions, or strength when normalized for body weight (Table 3). Trabecular bone volume fraction (BV/TV) in the metaphysis was slightly, but not significantly, increased in the food-restricted mice. The CR mice showed a greater number of trabeculae than mice fed ad libitum, but these trabeculae were thinner than those of the ad libitum group (Table 3). Marrow adipocytes were observed in the femora of AL mice but were completely absent from food-restricted mice (Table 3). The two groups did not differ from one another in femur length or joint dimensions (e.g., femur head diameter or bicondylar diameter), indicating that longitudinal growth was completed and/or unaffected by the food restriction treatment.
Table Table 2.. Body Composition Parameters and Serum Measurements in Mice Fed Ad Libitum (AL) and Mice on Caloric Restriction (CR)
Table Table 3.. Morphometry and Densitometry Data for the Femur in Mice Fed Ad Libitum (AL) and Mice on Caloric Restriction (CR)
Data for the spine differed somewhat from those of the femur (Fig. 1). DXA data showed that BMC and BMD in the spine were actually higher in the calorically restricted mice (Fig. 1; Table 4), and histomorphometry data showed a similar increase in trabecular BV/TV among the calorically restricted animals associated with a slight increase in trabecular number and no significant change in trabecular thickness (Fig. 1; Table 4). Osteoblast number was decreased (Fig. 3), and osteoclast number was increased (Fig. 4) in the spine and femur of CR mice, and the bone marrow of CR mice showed an abundance of venous sinuses adjacent to bone trabeculae (Fig. 4). It is important to note, however, that vertebral cortical thickness was actually lower in CR mice compared with mice fed ad libitum (Table 4)
Table Table 4.. Morphometry and Densitometry Data for the Spine and L4/5 Vertebra in Mice Fed Ad Libitum (AL) and Mice on Caloric Restriction (CR)
Associations among body composition and bone parameters
Regression analyses showed that percent muscle and fat are better determinants of total BMC than lean and fat mass. Fat mass predicts total BMC with r2 = 0.27, whereas percent fat improves the prediction of total BMC to r2 = 0.34. Quadriceps mass predicts total BMC with r2 = 0.26, whereas relative quadriceps mass (normalized by body weight) improves the prediction of total BMC to r2 = 0.36. Perhaps more importantly, multiple regressions including both relative quadriceps mass and percent body fat increase the prediction of total BMC to r2 = 0.46. In the femur, both fat mass and lean mass are significant predictors of femur BMC (r2 =0.79 and 0.69, respectively), and including both in the regression equation only increases the r2 slightly to 0.80. BMC in the spine was weakly (r = −0.25 to −0.42), and inversely, correlated with the body composition variables, but cortical thickness in the spine showed a strong correlation with body weight (r2 = 0.56) and body fat (r2 = 0.50), more so than with lean mass (r2 = 0.47).
Associations among serum leptin, IGF-1, and bone parameters
Serum leptin and IGF-1 levels both decrease significantly with caloric restriction (Table 2), and both are strongly associated with one another (r = 0.90). IGF-1 was strongly correlated with lean mass (r = 0.84), fat mass, (r = 0.85), and total body BMC (r = 0.30), whereas leptin showed a weaker association with all three of these variables (r = 0.06, 0.68, and 0.14, respectively). In the femur, regression analyses indicated that serum IGF-1 is a better predictor (r2 = 0.75) of femur BMC than serum leptin (r2 = 0.57), but adding fat to the regression with IGF-1 increased the r2 to 0.83. Serum leptin and IGF-1 were weakly (r = −0.27 to −0.31) and inversely associated with BMC in the spine, but spine BV/TV showed a much stronger, and inverse, association with serum IGF-1 (r = −0.66) and serum leptin (r = −0.54).
Previous studies(36) have emphasized the role of overall body weight in regulating bone size, mass, and strength. Our findings indicate that body weight per se is a relatively poor predictor of total body BMC, whereas body composition, expressed either as percent fat or relative muscle mass, is a more robust determinant of whole body bone mass. These two body composition variables together account for more variance in whole body bone mass than either parameter alone. This is most likely because these soft tissues seem to affect trabecular and cortical bone compartments differently, with caloric restriction decreasing cortical bone mass but sparing trabecular bone. These results are consistent with previous studies showing a decrease in body weight and cortical bone thickness with food restriction in rodents.(36–39) Likewise in humans, when weight is lost through caloric restriction, significant loss of BMD is also observed at the hip and spine, particularly in cortical bone.(40)
The increased trabecular number and lower bone cross-sectional area of 6-mo-old mice on caloric restriction is similar to the phenotype observed in young (∼3 mo) C57BL/6 mice. Trabecular bone volume fraction at the distal femur and spine normally decreases with age in C57BL/6J mice, whereas body weight and cortical cross-sectional area increase with age,(41,42) so that younger mice tend to have numerous, thin trabeculae and small cortical cross-sections resembling adult CR mice in this respect. Caloric restriction therefore seems to produce a pedomorphic, or “juvenilized,” skeletal morphology in which trabecular and cortical bone structure characteristic of younger mice is retained in adults (Fig. 5). This may also explain why the mice on caloric restriction have greater osteoclast number but at the same time have greater trabecular BV/TV in the spine. Osteoclast number in trabecular bone of male C57BL/6 mice usually peaks at 5 mo and then declines.(42) Delaying the rise and subsequent fall of osteoclast number in mice with caloric restriction will result in older mice having greater numbers of osteoclasts, and increased trabecular volume, characteristic of younger animals (Fig. 5).
The lower femur BMC and BMD, greater spine BMC and BMD, decreased cortical thickness, and increased trabecular bone volume in the food restricted mice resembles the skeletal phenotype observed in leptin-deficient ob/ob mice and calorically restricted rats.(43,44) We and others(16,30,31) have speculated that the preservation of trabecular bone in the skeleton of leptin-deficient rodents may be an adaptive mechanism for storing and preserving mineral reserves during periods of nutritional stress and/or to maintain some biomechanical competence of cancellous bone in the face of decreasing cortical strength. The fact that trabecular bone volume fraction is preserved in the femur of food-restricted mice, and even increased in the spine, indeed suggests that trabecular bone mass in mice is regulated in a fundamentally different manner than cortical bone. Equally likely is that trabecular and cortical bone mass are coordinated by a common signaling pathway, such as leptin, that exerts opposite effects on the different bone compartments.(45) It is important to note that in leptin-deficient ob/ob mice, as well as in the calorically restricted mice described here, cortical bone thickness is significantly reduced in both the spine and femur.(44) Likewise, femur BMC and cortical thickness decrease with age in mice as serum leptin also declines.(46) The decreased cortical bone size and thickness with CR is also consistent with the effects of NPY, which is elevated with leptin deficiency and suppresses cortical bone formation.(47) The increased trabecular bone volume in the CR mice is also consistent with the effects of lower sympathetic tone and decreased sympathetic outflows mediated by the β2-adrenergic receptor.(48)
Altered leptin signaling is likely to be a major factor underlying the decreased cortical bone mass and increased trabecular bone volume of both leptin-deficient ob/ob mice and calorically restricted rodents, but it is equally likely that alterations in other fat-derived cytokines and nutrition-related peptides are involved.(49) For example, the very low IGF-1 levels measured in the CR mice are likely to contribute to the cortical thinning observed in these mice, which is also observed in liver IGF-1–deficient lid/lid mice.(50) Furthermore, decreased sympathetic tone and absence of functional β-adrenergic receptors is also associated with decreased IGF-1, which is implicated in the decreased cortical bone mass characteristic of leptin-deficient mice and mice lacking β-adrenergic receptors.(45,51) Also, leptin-deficient ob/ob mice show decreased numbers of osteoclasts in their vertebrae, and osteoclast number is increased with leptin treatment, whereas calorically restricted mice show large numbers of vertebral osteoclasts despite being leptin deficient. To explain these seemingly paradoxical results one may speculate that, whereas ob/ob mice have decreased β2 adrenergic receptor–mediated RANKL production as a results of their lower sympathetic tone,(52) the latter is maintained in CR mice as a result of the stress reaction generated by the food restriction and/or that fat loss in these mice results in lower estradiol levels to prompt osteoclastogenesis. Finally, the stress of food restriction may also increase cortisol levels and decrease levels of T3 and insulin, which will together also contribute to bone loss.
It is unclear if the preservation of trabecular bone mass with leptin deficiency and caloric restriction is specific to rodents. Certain studies have shown that severe caloric restriction and weight loss in humans are associated with loss of BMD at both the hip and spine,(26,40) but these studies did not clearly separate data from cortical and trabecular bone. In another study, 31 patients were followed for a year after gastric banding, and weight loss with gastric banding was associated with a decrease in cortical bone mass at the femoral neck and an increase in trabecular bone mass at the greater trochanter and lumbar spine.(53) Thus, whereas caloric restriction may increase lifespan in rodents and perhaps people it is also associated with significant loss of muscle mass and cortical bone, which together are likely to significantly increase fracture risk. It is possible that pharmacological manipulation of the signaling pathways regulating cortical bone formation, such as neuropeptide Y and potentially leptin itself, that are suppressed with caloric restriction may hold potential for preventing significant loss of cortical mass with weight loss.
We are grateful to Dr Nancy Nadon at the National Institute on Aging for making the mice available and to Dr David Pashley and Kelli Agee for use of the Vitrodyne. Two anonymous reviewers provided useful comments that improved the manuscript. Funding for this work was provided by the Medical College of Georgia.