There is a subset of women who experience particularly rapid bone loss during and after the menopause. However, the factors that lead to this enhanced bone loss remain obscure. We show that patterns of bone loss after ovariectomy vary among inbred strains of mice, providing evidence that there may be genetic regulation of bone loss induced by estrogen deficiency.
Introduction: Both low BMD and increased rate of bone loss are risk factors for fracture. Bone loss during and after the menopause is influenced by multiple hormonal factors. However, specific determinants of the rate of bone loss are poorly understood, although it has been suggested that genetic factors may play a role. We tested whether genetic factors may modulate bone loss subsequent to estrogen deficiency by comparing the skeletal response to ovariectomy in inbred strains of mice.
Materials and Methods: Four-month-old mice from five inbred mouse strains (C3H/HeJ, BALB/cByJ, CAST/EiJ, DBA2/J, and C57BL/6J) underwent ovariectomy (OVX) or sham-OVX surgery (n = 6-9/group). After 1 month, mice were killed, and μCT was used to compare cortical and trabecular bone response to OVX.
Results: The effect of OVX on trabecular bone varied with mouse strain and skeletal site. Vertebral trabecular bone volume (BV/TV) declined after OVX in all strains (−15 to −24%), except for C3H/HeJ. In contrast, at the proximal tibia, C3H/HeJ mice had a greater decline in trabecular BV/TV (−39%) than C57BL/6J (−18%), DBA2/J (−23%), and CAST/EiJ mice (−21%). OVX induced declines in cortical bone properties, but in contrast to trabecular bone, the effect of OVX did not vary by mouse strain. The extent of trabecular bone loss was greatest in those mice with highest trabecular BV/TV at baseline, whereas cortical bone loss was lowest among those with high cortical bone parameters at baseline.
Conclusions: We found that the skeletal response to OVX varies in a site- and compartment-specific fashion among inbred mouse strains, providing support for the hypothesis that bone loss during and after the menopause is partly genetically regulated.
LOW BMD IS among the strongest risk factors for fragility fracture. A given BMD measurement reflects the peak value achieved during growth and the extent of age- and menopause-related bone loss. Intergenerational family and twin studies indicate that genetic factors explain a large part of the population variance in peak BMD.(1) In contrast, the contribution of genetic factors to age- and menopause-related bone loss is unclear.
A better understanding of the factors that govern the rate of bone loss is important, because individuals with high rates of bone loss or high bone turnover indices are at increased risk for fracture, independent of their BMD status.(2,3) Although premenopausal bone loss has been reported, the predominant mechanism underlying bone loss in women is the gradual decline in estrogen production during and after menopause.(4) Reduced estrogen levels lead to increased bone turnover, with resorption exceeding formation, and thus, enhanced rates of bone loss. The rate and extent of bone loss varies tremendously among individuals, as well as by skeletal compartment and site.(5–10) Importantly, evidence suggests that there is a subset of women who experience particularly rapid bone loss during the early menopausal period.(10–14) The factors that lead to enhanced bone loss in this subset of women remain obscure, although it has been suggested that genetic factors may render some women more susceptible to bone loss after estrogen deficiency.(15)
In this regard, although there is strong evidence for heritability of BMD and bone geometry, heritability of the rate of bone turnover and bone loss remains controversial.(16) However, at least one study indicates that cortical bone loss is under strong genetic control,(17) and polymorphisms in genes that mediate the effects of sex steroids on bone are associated with rates of bone turnover, bone loss, and fracture risk.(18–24) However, human studies alone may be insufficient to identify the multiple genetic factors influencing the rate of bone loss. Because the decline in BMD is generally similar to the precision error of BMD measurements, relatively long duration, longitudinal studies with large sample sizes are needed. Moreover, most clinical studies assess bone loss by measuring areal BMD, which does not delineate specific changes in the cortical and trabecular bone compartments. This represents a major limitation as early postmenopausal bone loss is most dramatic in the trabecular bone compartment.(25) For these reasons, animal models may be useful for initial identification of genetic determinants of bone loss that can then be validated in human populations.
Accordingly, inbred mouse strains have recently been established as a valuable tool for studying the genetic regulation of skeletal phenotypes. The heritability of skeletal traits is well established in mice, and several of these traits, including BMD, bone geometry, microarchitecture, and strength, have been shown to vary among inbred mouse strains.(26–33) Furthermore, skeletal responsiveness to altered mechanical loading also varies among inbred mouse strains.(34–39) Thus, the biologic factors that govern both the acquisition and maintenance of skeletal traits show genetic regulation.
In this study, we explored the potential role of genetic factors in modulating bone loss subsequent to estrogen deficiency by comparing the skeletal response to ovariectomy (OVX) in five inbred strains of mice. Specifically, we asked whether trabecular and cortical bone changes after OVX varied with mouse strain. We also sought to extend previous observations regarding differences in skeletal morphology among inbred mouse strains and to determine whether the baseline bone morphology influenced the response to OVX.
MATERIALS AND METHODS
We performed OVX or sham-OVX (SHAM) surgery in five inbred strains of mice at 4 months of age (n = 6-9/group): C57BL/6J (B6), BALB/cByJ (BALBc), C3H/HeJ (C3H), CAST/EiJ (CAST), and DBA/2J (DBA/2). Mice had free access to food (NIH 31 diet; Purina Mills International, Richmond, IN, USA) and water and were killed 1 month after surgery. Bone weight and uterine weight were measured at necropsy. The right femur, right tibia and fifth lumbar vertebra were harvested for analysis of trabecular and cortical bone properties. The protocol and procedures were approved by the Institutional Animal Care and Use Committee of the Jackson Laboratory.
We assessed trabecular and cortical bone architecture using μCT (μCT40, Scanco Medical AG, Basserdorf Switzerland). Specifically, trabecular bone architecture was evaluated at the fifth lumbar vertebra and proximal tibia, whereas cortical bone morphology was evaluated at the femoral midshaft. CT images were reconstructed in 1024 × 1024-pixel matrices and stored in 3-D arrays. The resulting grayscale images were segmented using a constrained Gaussian filter (sigma = 0.8, support = 1) to remove noise, and a fixed threshold (22% of the maximal grayscale value for vertebrae and tibia and 30% for midfemoral cortical bone) was used to extract the structure of the mineralized tissue.
For the vertebrae, 250-300 transversely oriented CT slices (12-μm isotropic voxel size) were obtained such that the entire vertebral body was included in the scanned region. The trabecular bone region within the vertebral body, excluding the superior and inferior endplates, was manually identified (10-90% of vertebral height). For the tibia, 100 coronal CT slices were acquired, and 70 slices centered in the tibia were evaluated. The region of interest was manually outlined on each CT slice, excluding the primary spongiosa, and extending 1.8 mm distally from the growth plate. Morphometric parameters, including bone volume fraction (BV/TV, %), trabecular number (Tb.N, mm−1), trabecular thickness (μm), trabecular separation (Tb.Sp, μm), structure model index (SMI), and connectivity density (ConnDens, mm−3) were computed without assumptions regarding the underlying bone architecture.(40,41)
At the femoral midshaft, 50 transverse CT slices were obtained (12-μm isotropic voxel size) and used to compute the total cross-sectional area (TA, mm2), cortical bone area (BA, mm2), medullary area (MA, mm2), cortical thickness (CortTh, μm), and bone area fraction (BA/TA, %). We also used the CT images to measure the antero-posterior (AP) and medio-lateral (ML) diameters, the area moments of inertia about the AP and ML axes (Iap and Iml, mm4), and the polar moment of inertia (J, mm4). Finally, we scanned a hydroxyapatite phantom (100, 200, 400, and 800 mg HA/cm3) to estimate the mineralization density of the cortical bone in the femoral midshaft in mg/cm3 hydroxyapatite.
Variables were evaluated using standard descriptive statistics. To test whether bone loss after OVX was influenced by genetic factors, we conducted a two-factor ANOVA, with mouse strain and treatment (SHAM versus OVX) as grouping variables. To test whether OVX-induced bone loss was significant within any given strain, values for the variable of interest in the SHAM and OVX groups were compared using an unpaired Student's t-test. To test whether skeletal parameters differed among these inbred strains, we compared the values in SHAM-operated animals using ANOVA, followed by posthoc testing with Fisher's protected least squares difference test. All tests were two-tailed, and differences were considered statistically significant at p < 0.05.
Strain-related differences in response to OVX
Effectiveness of the OVX procedure was confirmed by a dramatic reduction in uterine weight per body weight (SHAM = 0.456 ± 0.028% versus OVX = 0.109 ± 0.006%, p < 0.0001). Body weight did not differ between OVX and SHAM, except for C3H, in which OVX mice were heavier than SHAM (29.3 ± 1.3 versus 24.8 ± 1.1 g, p = 0.03).
The patterns of skeletal deterioration in response to OVX varied among the inbred strains in a site- and compartment-specific fashion (Table 1; Fig. 1). For trabecular BV/TV, thickness, and connectivity density, the response to OVX depended on the strain, as judged by a statistically significant interaction between strain and treatment (Table 1). Overall, strain-related differences in the response to OVX were more pronounced at trabecular than at cortical sites.
Table Table 1.. Effect of OVX on Trabecular and Cortical Bone Parameters in Five Inbred Mouse Strains Expressed as the Mean Percent Difference Between SHAM and OVX* (95% CI)
Vertebral trabecular BV/TV was lower in OVX than SHAM mice in four of five strains (Table 1). Only C3H appeared resistant to OVX-induced trabecular bone deterioration at the vertebrae. The effect of OVX on vertebral BV/TV in BALB/c mice (−24%, p = 0.0002 versus SHAM) differed significantly from the effect in B6 (−14.7%, p = 0.01 versus SHAM; p = 0.02 versus BALB/c), C3H (−5.4%, not significant versus SHAM; p = 0.002 versus BALB/c), and CAST (−19.3%, p = 0.03 versus SHAM; p = 0.006 versus BALB/c). Vertebral trabecular number declined significantly in BALB/c mice only (p = 0.03 versus SHAM), whereas trabecular thickness declined in B6 (p = 0.0004 versus SHAM) and BALB/c (p < 0.0001 versus SHAM), and tended to decline in DBA (p = 0.08 versus SHAM). The effect of OVX on trabecular thickness differed significantly between BALB/c and all other strains (0.01 < p < 0.0001) and between B6 and CAST (p = 0.004).
At the proximal tibia, the effect of OVX on trabecular BV/TV also depended significantly on mouse strain (p = 0.0002; Table 1). However, in contrast to the vertebral body, where C3H mice were resistant to OVX-induced bone loss, at the proximal tibia, C3H mice had the greatest decline in trabecular BV/TV (−39.2%, p = 0.004 versus SHAM). The other strains also exhibited a decline in tibial trabecular BV/TV, ranging from −18% to −27%, but this decline was statistically significant only in BALB/c (p = 0.005 versus SHAM). Trabecular number declined in DBA/2 only (−14.3%, p = 0.005 versus SHAM). Trabecular thickness decreased in C3H (−9.5%, p = 0.05 versus SHAM) and BALB/c (−13.1%, p = 0.0002 versus SHAM), and increased in DBA/2 (+14%, p = 0.015 versus SHAM). The change in trabecular thickness after OVX differed significantly between B6 and BALB/c (p = 0.04), DBA/2 and B6 (p = 0.006), BALB/c (p < 0.0001), and C3H (p = 0.002), and between BALB/c and CAST (p = 0.025).
Overall, the effect of OVX on midfemoral cortical bone did not depend on mouse strain (Table 1). However, cortical bone from B6 mice was more consistently responsive to OVX than the other strains, showing significant changes in all parameters except for total cross-sectional area. Cortical bone area fraction declined significantly (−3.4% to −8.6%; 0.0002 < p < 0.04; Fig. 1), and cortical thickness showed a trend to decline (−3.0 to -9.5%; 0.0002 < p < 0.09) in all strains, except for C3H. The mineralization density showed a similar trend. Interestingly, the polar moment of inertia tended to respond differently in B6 versus CAST (p = 0.06), with B6 showing a statistically significant decline (−10.6%, p = 0.05) and CAST exhibiting a nonsignificant increase.
Strain-related differences in normal body weight, uterine weight, and skeletal morphology
We confirmed and extended previous observations regarding strain-related differences in normal mice by comparing those mice that underwent SHAM surgery. Body weight was highest in BALB/c, intermediate in B6, C3H, and DBA/2, and lowest in CAST (Table 2). Uterine weight (per body weight) was highest in BALB/c and CAST, intermediate in C3 and DBA, and lowest in B6 mice (Table 2).
Table Table 2.. Body Weight, Uterine Weight, and Skeletal Morphology in SHAM-Operated Mice From Five Inbred Strains (mean ± SE)
Consistent with other studies, we found differences in skeletal morphology among inbred strains that depended on skeletal site (Table 2; Fig. 2). In the fifth lumbar vertebrae, trabecular BV/TV was highest in BALB/c, intermediate in B6 and DBA/2, and lowest in C3H and CAST mice (p < 0.0001; Table 2). Vertebral trabecular bone in BALB/c was characterized by increased trabecular number and thickness and reduced trabecular separation compared with other strains. In comparison, vertebral trabecular bone in C3H mice was characterized by reduced trabecular number, but increased trabecular thickness, whereas the low BV/TV in CAST was caused by both reduced trabecular number and thickness (Table 2).
At the proximal tibia, trabecular BV/TV was highest in C3H and BALB/c, intermediate in B6 and DBA/2, and lowest in CAST (Table 2). Both trabecular thickness and number were statistically significantly higher in C3H and BALB/c compared with other strains. CAST had the fewest trabeculae and consequently the greatest trabecular separation.
At the femoral midshaft, total cross-sectional area was highest in B6 and BALB/c, intermediate in C3H, and lowest in DBA/2 and CAST (Table 2). Cortical thickness, mineral density, and the ratio of cortical bone to total area (BA/TA) were highest in C3H, intermediate in BALB/c, DBA/2 and CAST, and lowest in B6.
Effect of baseline bone status on response to OVX
To test whether the morphology of the bone at baseline influenced the subsequent response to OVX, we tested the correlation between the mean value of vertebral trabecular BV/TV, proximal tibia trabecular BV/TV, and midfemoral BA/TA in SHAM-operated mice from each strain and the respective percent or absolute difference between sham and OVX for each parameter. Both vertebral and proximal tibia trabecular bone loss (expressed as percent difference versus sham) were inversely correlated with BV/TV values in SHAM-operated mice (r = −0.60 and r = −0.76, respectively). Thus, mouse strains with higher values of trabecular BV/TV at baseline tended to have a relatively greater decline in trabecular BV/TV than strains that had lower BV/TV to start with. This pattern was similar when the data were analyzed using the absolute change in BV/TV rather than percentage change. In cortical bone, the pattern was the opposite, with a strong, positive linear association between midfemoral BA/TA in sham-operated mice and the percentage decline in BA/TA after OVX (r = 0.92).
In this study, we examined whether genetic factors influence the bone loss caused by estrogen deficiency by studying the skeletal response to OVX in five inbred strains of mice. We found that OVX-induced bone loss varies among inbred mouse strains in a compartment- and site-specific fashion. We also confirmed and extended our own and others' previous observations regarding differences in trabecular and cortical bone morphology among inbred mouse strains and noted that, among these inbred strains, the response to OVX seems to depend partly on the baseline bone morphology.
With regard to skeletal compartment (i.e., trabecular versus cortical), the effect of mouse genetic background on OVX-induced bone loss was particularly notable at trabecular bone sites, with a statistically significant interaction between strain and treatment for several key trabecular bone traits at both the vertebral body and proximal tibia (Table 1). In comparison, the effects of genetic background on cortical bone changes after OVX were generally less prominent (Table 2). It may be that a study of longer duration is required to see more marked strain-related differences in the cortical bone response to estrogen deficiency.
With regard to skeletal site-specificity, evidence for site-specific responses to OVX within a single bone compartment is seen by examining the vertebral body and proximal tibia. For vertebral trabecular BV/TV, differences between SHAM and OVX depended on the strain, ranging from not significant (in C3H) to −24% (in BALB/c). OVX-induced changes in trabecular BV/TV also varied by strain at the proximal tibia. However, in contrast to the vertebral body, the decline in trabecular BV/TV at the proximal tibia was greatest in C3H mice (−39%) and was twice that seen in B6 mice (−18%).
Previous studies have also shown that genetic factors modulate bone loss in mice. For example, female B6 mice undergo significantly greater endosteal resorption after sciatic neurectomy than do C3H mice.(35) Furthermore, the skeletal response to hindlimb suspension varies among the B6, C3H, and BALB/c inbred strains in a site- and compartment-specific fashion.(39) Taken together with results from this study, these observations confirm that the skeletal response to catabolic stimuli is governed by complex relationships among genetic background, skeletal site, and bone compartment.
Another important finding from our study is that the skeletal response to OVX in mice seems to be influenced by their baseline bone mass and morphology. Although this observation is limited by the inclusion of only five strains, the data suggest that mice with higher trabecular BV/TV at baseline lose more bone after OVX, both in absolute and relative terms. One may speculate that mice with low BV/TV to start with cannot afford to lose more and that there is some mechano-stat driven feedback that allows them to maintain trabecular bone volume at a minimal level even in the face of estrogen deficiency. Additional studies are required to confirm this hypothesis. Another important implication of this observation is that it infers that the genes that influence attainment of adult bone mass and architecture ultimately also influence, perhaps indirectly, bone loss after estrogen deficiency. Thus, the underlying bone architecture, particularly the amount of bone surface and rate of bone turnover may be major determinants of the skeletal response to estrogen deficiency. To sort out independent genetic effects on the rate of bone loss after estrogen deficiency, future studies could focus on measuring the response to OVX in mouse strains with similar trabecular BV/TV at baseline.
Our observation that mouse strains with the highest trabecular bone volume lose the most bone after OVX contradicts one study in women where forearm pQCT measurements showed that those with the lowest trabecular vBMD at baseline had the greatest bone loss.(10) On possible explanation for the discrepancy between the mouse and human data may relate to comparing weight-bearing sites (in the mouse) to a non-weight-bearing site (in humans). Additionally, factors other than baseline bone morphology (such as baseline turnover rate, levels of sex steroids) may differ between mice and humans and may influence the rate of bone loss subsequent to estrogen deficiency.
Regarding cortical bone, we noted an opposite pattern, such that mice with high cortical bone properties at baseline seem to be more protected from cortical bone loss after estrogen deficiency compared with mice with lower cortical bone properties at baseline. We note that additional studies are required to confirm these observations.
To interpret these data, it is important to consider the skeletal maturity of the mice. We performed our study in 4-month-old mice. Although mice, like other rodents, never achieve epiphyseal closure, by 4 months of age, bone elongation is minimal and cortical bone mass is maximal.(26) We therefore consider that our study evaluates the effects of estrogen deficiency in skeletally mature, rather than growing, animals. In rats, in whom the majority of studies regarding OVX-induced bone loss have been conducted, bone elongation in minimal by 6 months of age and cortical bone mass peaks at 8-9 months of age. Thus, in terms of skeletal maturity, 4-month-old mice are approximately equivalent to 8- to 9-month-old rats.
Overall, the factors that govern skeletal sensitivity to estrogen deficiency are poorly understood.(13) One consistent finding in human studies is that peri- and postmenopausal bone loss is attenuated in those with higher body weight or body mass index.(11,42,43) The reduced rates of bone loss seen in heavier individuals may be explained by a greater mechanical stimulus to the skeleton and/or by increased estrogen levels caused by aromatization of androgen into estrogen in fat tissue. In this study there was no consistent association between body weight or body composition and the variation in bone loss after OVX (data not shown). One possible explanation for this discrepancy between humans and mice may be that there are differences in the extent of extragonadal steroid production in mice compared with humans.(44) Another explanation may be that the variability in body weight and body composition in these inbred mice is small relative to that seen in human populations.
There were several limitations associated with our study. First, our sample size was modest, and we may have been able to detect additional strain- or OVX-related differences with larger groups. We chose not to correct our analysis for multiple comparisons to allow us to have the greatest chance of detecting subtle, but important biologic differences between the strains given our modest sample size. Second, we assessed only a single time point (4 weeks) after OVX and therefore we cannot fully assess the rate and extent of bone loss. As mentioned previously, longer-duration studies may be needed to detect changes in cortical bone. Third, no biomechanical testing was performed, and therefore the effects of OVX on functional capacity can only be inferred from the changes in morphology. Fourth, this study did not incorporate biochemical markers of bone turnover or histomorphometry, and thus it is impossible to attribute the observed changes in bone morphology to any underlying biologic processes. Finally, it should be pointed out that, although OVX is currently the accepted model for bone loss associated with estrogen deficiency, surgical removal of the ovaries does not mimic the gradual cessation of ovarian function that occurs in women at the menopause. Furthermore, ovarian production of androgens may attenuate menopausal bone loss, a phenomenon that cannot be studied using the OVX model.
Despite these limitations, our findings raise some interesting and provocative issues. First and most importantly, it seems that genetic factors are important in defining the extent of bone loss in mice after OVX. Second, the rate of bone loss with estrogen deprivation in mice may depend strongly on peak bone mass and morphology, traits which are also under strong genetic regulation. Third, there is clear skeletal site-specificity in the response to estrogen-induced bone loss, a phenomena often noted clinically. Finally, our data strongly suggest that studies of OVX-induced bone loss in genetically manipulated mice, such as transgenics or knockouts, might be compromised if the background strain is heterogeneous.
In summary, we found that the skeletal response to OVX varies in a site- and compartment-specific fashion among 4-month-old inbred mouse strains, providing support for the hypothesis that bone loss during and after the menopause may be partly genetically regulated. These data provide strong rationale for additional studies designed to delineate the genetic factors and associated biologic pathways that modulate the skeletal response to estrogen deficiency. Ultimately these studies may afford identification and early treatment of those women at high risk for enhanced postmenopausal bone loss.
This study was supported by NIH Grant AR43618. We are grateful to Robert McClean for assistance with statistical analyses.