Peak bone mass is a major determinant of osteoporotic fracture risk. Gender differences in peak bone mass acquisition are well recognized in humans and may account for a substantial share of the increased prevalence of fragility fractures in women compared with men. Skeletal development is regulated by both heritable and environmental factors. Experimental animal models provide a means to circumvent complicating environmental factors. In this study we examined the heritability of peak bone mineral density (BMD) in genetically distinct laboratory mouse strains raised under strict environmental control and sought to identify genetic loci that may contribute to gender differences in this skeletal phenotype. Peak whole body BMD of male and female mice from a panel of 18 recombinant inbred (RI) strains derived from a cross between C57BL/6 and DBA/2 progenitors (BXD) was measured by dual-energy X-ray absorptiometry (DXA). A highly significant relationship existed between body weight and BMD in the BXD RI mice (r2 = 0.25; p = 1 × 10−43). To allow for comparison between male and female RI strains, whole body BMD values were corrected for the influence of body weight. The distribution of weight-corrected BMD (WC-BMD) values among the strains indicated the presence of strong genetic influences in both genders, with an estimated narrow sense heritability of 45% and 22% in male and female mice, respectively. Comparison of RI strain results by two-way analysis of variance (ANOVA) revealed a significant strain-by-gender interaction (F1,17,479 = 6.13; p < 0.0001). Quantitative trait locus (QTL) analysis of the BXD RI strain series provisionally identified nine chromosomal sites linked to peak bone mass development in males and seven regions in females. In two cases, the provisional chromosomal loci were shared between genders, but in most cases they were distinct (five female-specific QTLs and six male-specific QTLs). QTL analysis of a genetically heterogeneous F2 population derived from the B6 and D2 progenitor strains provided additional support for the gender specificity of two loci. A significant phenotype-genotype correlation was only observed in male F2 mice at microsatellite marker D7Mit114 on chromosome 7, and a correlation at D2Mit94 on chromosome 2 was only observed in female F2 mice. The present data highlight the important role of gender in the genetic basis of peak bone mass in laboratory mice. Because the male phenotype is associated with considerable fracture risk reduction, an elucidation of the nature of that effect could provide the basis for novel diagnostic, preventative, or therapeutic approaches.
OSTEOPOROSIS IS one of the most common bone and mineral disorders in both men and women. It is characterized by skeletal fragility resulting in fractures from relatively minor trauma. Bone density achieved in early adulthood (peak bone mass) is a major determinant of risk of osteoporotic fracture. Although lifestyle and environmental factors play a role in the achievement of peak bone mass, there now is clear evidence that genetic factors appear to be of great importance.(1) It has been calculated that up to 75% of bone mineral density (BMD) is determined genetically, a contribution that appears to be polygenic in nature. An essential direction in genetic research is to identify the specific chromosomal elements that contribute to bone mass development.
Quantitative trait loci (QTLs) are chromosome sites containing alleles (genes) that influence a normally distributed (quantitative) trait. Recently developed statistical methods make it possible to detect and genetically map the chromosomal locations of such QTLs. Experimental animal models for complex human traits provide a means for circumventing two factors—genetic heterogeneity and environment—that complicate human studies. Because extensive linkage conservation (homology) exists between the mouse and human genomes (about 80%), it is very likely that the mouse can be used to determine human chromosomal sites suitable for further testing in human populations.
There are obvious gender differences in the skeletal phenotype at the time peak bone mass is achieved. Generally, men have larger bone size and greater cortical mass.(2, 3) The factors that influence skeletal development and determine these gender differences have been assumed to be related primarily to sex steroid action. For instance, androgens have been reported to increase bone formation,(4) particularly in cortical areas, and the absence of androgen action has been related to lower bone mass and smaller size.(5) Androgens also may influence weight and muscle mass, thereby indirectly affecting skeletal development. Although it is likely that androgens contribute to gender differences in skeletal character, the molecular mechanisms by which these effects are mediated are unclear.
Gender has been found to exert effects on other phenotypes via genetic mechanisms apparently unrelated to the control of androgen levels. We postulated that the marked gender differences in peak bone size/mass were in part the result of genetic influences. QTL analysis of males in inbred strains derived from a cross between C57BL/6 and DBA/2 progenitors (BXD) revealed evidence of chromosome locations of a number of QTLs affecting peak whole body BMD in male mice. Of particular interest, QTL analysis indicates that the chromosomal loci associated with peak BMD are in some cases shared between the genders, but in other cases are distinct. Additional selective genotyping of an F2 population of male and female mice confirmed some chromosomal loci associated with BMD to be gender specific.
MATERIALS AND METHODS
All mice used in these experiments were bred under identical conditions at the Portland Veterans Affairs (VA) Veterinary Medical Unit from stock originally obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Male and female mice from the two progenitor strains C57BL/6 and DBA/2 and 18 of the BXD recombinant inbred (RI) strains were available in sufficient number (n ≥ 10 of each gender within each strain) for testing. B6D2F2 mice (n = 601 female mice and 393 male mice) were derived from an intercross of F1 offspring from a cross between B6 females and D2 males. The B6.D2 congenic strain possessing an introgressed B6 segment from chromosome 2 (20-65 cM) was initiated by mating RI strain BXD-6 females with D2 males. Subsequent generations were produced by backcrossing to the recipient inbred strain (D2).(6) At the time of weaning, all mice of the same strain and gender were housed in groups of four to five animals and maintained with ad libitum water and laboratory rodent chow (Diet 5001: 23% protein, 10% fat, 0.95% calcium, and 0.67% phosphorus; PMI Feeds, Inc., St. Louis, MO, USA). All procedures were approved by the VA Institutional Animal Care and Use Committee and performed in accordance with National Institutes of Health (NIH) guidelines for the care and use of animals in research.
Bone mineral measurements were performed with a pencil beam Hologic QDR 1500 densitometer (Hologic, Waltham, MA, USA) that was calibrated daily with a hydroxyapatite phantom of the human lumbar spine. Analysis was performed using the mouse whole body software (version 3.2), provided by Hologic. Densitometric analysis was performed on freshly killed mice. Food was withheld the night before death (to eliminate confounding effects of undigested rodent chow on BMD assessment), and the mice were killed by CO2 inhalation. The animals were weighed to the nearest 0.1 g and immediately underwent dual-energy X-ray absorptiometry (DXA) scanning. The scans were done with a 1.27-mm-diameter collimator, 0.76-mm line spacing, 0.38-mm-point resolution, and an acquisition time of 9 minutes. The global window was defined as the whole body image minus the calvarium, mandible, and teeth. Data were gathered as bone mineral content (BMC; milligrams of hydroxyapatite) and BMD (mg/cm2). Precision error (expressed as the CV) for BMC was 0.99 ± 0.51% and for BMD was 1.71 ± 0.33%.
Polymerase chain reaction genotyping
Genomic DNA was isolated from individual mouse spleens using a salting-out method.(7) Mice were genotyped using microsatellite markers from the Massachusetts Institute of Technology (MIT) series using a method adapted from Dietrich et al.(8) and Serikawa et al.(9) Markers were chosen with information from the MIT database (http://www.genome.wi.mit.edu/cgi-bin/mouse/index) and the Mouse Genome Database (http://www.informatics.jax.org). All primers were purchased from Research Genetics (Huntsville, AL, USA). Amplification was performed on a Perkin-Elmer 9700 thermocycler (Perkin-Elmer, Branchburg, NJ, USA). Reaction mixture was 25 μl of total volume, consisting of approximately 150 ng of genomic DNA, 264 nM of both forward and reverse primers, 0.2 mM of each deoxynucleoside triphosphate (dNTP), 1 U of Taq polymerase (Perkin-Elmer Cetus, Branchburg, NJ, USA), 2.5 μl of GeneAmp 10× polymerase chain reaction PCR Buffer containing 100 mM Tris-HCl (pH 8.3), 15 mM MgCl2, 500 mM KCl, and 0.01% (wt/vol) gelatin. Thermal cycling included two 5-minute denaturation steps at 95°C and then at 80°C, 40 cycles of 30 s at 94°C, 30 s at 53°C, 30 s at 72°C, and a final extension step for 10 minutes at 72°C. PCR products were separated on 4% agarose gels and visualized with ethidium bromide staining.
To test for markers associated with peak whole body BMD in the RI strains, we used the approach first described by Plomin et al.(10) Over 1500 informative genetic markers have been genotyped in the BXD RI strains, mostly comprising microsatellite simple-sequence repeat polymorphisms determined by PCR so that the location and the B6- or D2-like repeat length of each of these alleles is known for each strain.(11) Because of the derivation of these strains from B6 and D2 progenitors, each RI strain necessarily possesses either two copies of the B6 allele or two copies of the D2 allele. The QTL analysis involved correlating the genetic marker information with the quantitative phenotype. To this end, point biserial correlation coefficients (r) were calculated between the strain means for the phenotype and each marker in the data bank. Analysis of B6D2F2 genotyping data for each gender was accomplished by carrying out a 2 × 2 χ2 test for each marker between the tails of the distribution (low and high) and allele frequencies (B6 or D2). Individual p values for the congenic experiment were determined using linear least squares statistical analyses. For each marker tested; individual mice were assigned a genotypic score of 0, 1, or 2 based on gene dosage (the number of D2 alleles at that locus). Correlation coefficients (Pearson's r) were determined and are equivalent to regressing BMD on gene dosage for each marker.(12) Only a subset (20%) derived from the extreme tails of each of the two B6D2F2 populations was genotyped and therefore analyzed, but all mice from the intercross between heterozygotes of the fourth backcross generation of the B6.D2 congenic strain were genotyped and scored as shown previously. The theoretical rationale underlying the QTL analysis methods have been described previously.(10, 13) The B6D2F2 genotyping data were subjected to analysis using the MapManager QT (beta version 28) computer program(11) to determine the position of the peak logarithm of the likelihood for linkage (log10 of the odds ratio [LOD]) and 1 LOD support intervals. StatView statistical software for the Macintosh (SAS Institute, Inc., Cary, NC, USA) and Systat for the DOS environment(14) (SPSS Science, Chicago, IL, USA) were used to perform all other statistics.
Phenotype values for all strains are presented as the mean ± SEM. Analysis of variance (ANOVA) was used to detect significant strain differences and provide an estimate of heritability. For the RI strain data, multiple regression analyses were performed for whole body BMD to estimate the total amount of genetic variance accounted for by all of the significant gene marker associations. This analysis corrects for intercorrelations among markers and in the process, provides an estimate of the number of effective factors (genes) contributing to the genetic variation for which the entire marker set accounts. In other words, because of chance intercorrelations among markers, some correlations of marker and phenotype may be fortuitous. However, multiple regression analyses do not identify which markers these may be. This approach has been compared with the same analysis in which every marker was included and has shown comparable results.(13) The theoretical rationale underlying the QTL analysis methods have been described previously.(10, 13) Associations attaining p < 0.01 were interpreted to indicate the provisional presence of a gene (QTL) near the correlated marker contributing some proportion of trait variance. Similarly, a gender difference was ascribed when the p values for a marker correlation differed by greater than three orders of magnitude between the two genders, which is equivalent to about p < 0.001.
Shown in Fig. 1 are body weight and BMD values for groups of 12-week-old B6 and D2 mice. As expected, males exhibited heavier body weights than females, but there was no progenitor strain difference in body weight. A significant gender-specific difference in peak whole body BMD was observed in both progenitor inbred strains (p < 0.0001). Body weight, another phenotype with a strong genetic component, is associated closely with bone mass. To account for the observed differences in body weight between the two genders, whole body BMD was corrected for body weight by regression residuals. No strain-specific or gender-specific differences in weight-corrected BMD (WC-BMD were identified (Fig. 1), but two-way ANOVA revealed a significant strain-by-gender interaction (F1,1,132 = 6.4; p < 0.02). The finding of nonsignificant progenitor strain differences is uncommon for a QTL mapping study; usually, such differences are the basis for the mapping attempt. However, it should be noted that progenitor strain differences are not required for successful QTL mapping,(13) because large genetic differences may exist even when phenotypic differences do not. Therefore, we proceeded to examine peak whole body BMD in male and female BXD RI mice. Similar to that observed in the progenitor strains (data not shown), a highly significant relationship existed between body weight and BMD in the BXD RI mice (Fig. 2). Because the goal of these experiments was to compare the genetic determinants of peak whole body BMD in male and female mice, we accordingly corrected the whole body BMD values for weight. In both genders, the distributions of peak WC-BMD were continuous, inferring polygenic control of WC-BMD in both males and females (Fig. 3). Peak whole body WC-BMD varied by 18% across the male BXD RI panel, ranging from 56.6 ± 0.9 mg/cm2 in BXD-27 to 66.8 ± 0.7 mg/cm2 in BXD-6. Peak whole body WC-BMD varied by 17% across the female BXD RI panel, ranging from 56.4 ± 1.1 mg/cm2 in BXD-16 to 66.2 ± 1.5 mg/cm2 in BXD-31. One-way ANOVA, grouped by RI strain, confirmed significant variations in whole body WC-BMD in both the male BXD RI strain set (F17,203 = 11.72; p < 0.0001) and the female BXD RI strain set (F17,276 = 3.07; p < 0.0001). The BXD RI strain distribution for whole body WC-BMD appeared to be gender specific. The gender divergence is illustrated in Fig. 4 in which mean BMD strain values are plotted for each gender. Indeed, when RI strain results from male and female mice were analyzed together, two-way ANOVA revealed a significant strain-by-gender interaction (F1,17,479 = 6.13; p < 0.0001).
Because these genetically distinct RI strains were raised in a controlled environment (i.e., nutritional intake, physical activity, etc.), the differences observed in peak BMD primarily were the result of genetic variation. Heritability (h2) or the proportion of the total variation due to additive genetic sources can be estimated from the r2 from a one-way ANOVA by strain. The heritability reflects the reliability of predicting phenotype from genotype, which is critically important for successful QTL mapping. The estimated narrow sense heritability for whole body WC-BMD in males was 0.45 (p < 0.0001) in the male RI series and 0.22 (p < 0.0001) in the female RI series, thus showing a substantial degree of genetic control of this trait in both genders. These heritability estimates are well above those found to lead to successful QTL mapping of other traits.(1) Interestingly, in male RI mice the estimated narrow sense heritability for whole body BMD (without adjusting for the influence of body weight) essentially was unchanged at 0.46 (p < 0.0001), while in female RI mice the heritability estimate for uncorrected BMD was somewhat higher at 0.31 (p < 0.0001).
To identify and compare the putative location of relevant QTLs in male and female mice, the strain mean values for whole body WC-BMD were correlated with the allele (B6 or D2) at each of the 1522 marker loci in our database. Because of the strain × gender interaction, it was anticipated that QTL analysis of the BXD RI data set would identify chromosomal loci associated with peak BMD that were substantially different in male and female mice. As shown in Table 1, nine discrete regions on chromosomes 2, 7, 9, 11, 13, 17, 18, and X were found to be associated with peak whole body BMD in male BXD RI mice at p < 0.01 and seven discrete regions on six chromosomes 1, 2, 7, 8, 11, and 16 were found in female BXD RI mice. In two cases the provisional chromosomal loci associated with peak BMD were shared between genders (midportion of chromosome 2 and proximal portion of chromosome 7), but in most cases they were distinct (5 female-specific QTLs and 6 male-specific QTLs). We have reported previously the chromosomal locations of QTLs affecting the acquisition of whole body BMD (without weight correction) in female mice.(2) Three of the present WC-BMD QTLs (those found on proximal chromosome 2, chromosome 7, and chromosome 16) also were identified in the previous analysis.
Table Table 1.. Candidate QTL Sites for Whole Body BMD in Male BXD RI Mice
We next examined whole body BMD in an F2 population derived from the progenitor strains (B6 and D2). As was observed in the progenitor strains, males exhibited heavier body weights than females (33.1 ± 0.2 g vs. 26.5 ± 0.1 g; p < 0.0001; Fig. 5). However, in contrast to what was observed in the inbred B6, D2, and RI strains of mice, the distributions of whole body BMD in male and female populations of genetically heterogeneous B6D2F2 mice were not significantly different (65.3 ± 0.1 mg/cm2 vs. 65.1 ± 0.1 mg/cm2; p = 0.26). A total of 20% of each B6D2F2 population by gender (10% from each tail of the frequency distribution) were genotyped. This selective genotyping approach, first suggested by Lander and Botstein,(15) required only 20% of the genotyping expense compared with genotyping the entire F2 population, yet retained >70% of the linkage information needed to detect and map QTLs.
Interval mapping of data from a genomewide scan of the two F2 populations combined thus far has confirmed the presence of four QTLs on chromosomes 1, 2, 4, and 11.(16) However, when the male and female F2 populations were considered separately, evidence for gender specificity at certain chromosomal loci became apparent. Shown in Fig. 6 are the average WC-BMD values of male and female animals for the three possible genotypes for the D1Mit291, D2Mit94, D4Mit312, D7Mit114, and D11Mit349 microsatellite markers. Significant associations between genotype and phenotype were found in both genders for the D4Mit312 and D11Mit349 markers. However, only weak genetic effects were present at markers D1Mit291 and D2Mit94 in male F2 mice, neither of which were significant and consequently these QTLs would not have been detected in studies of male mice alone. In contrast, analysis of the male F2 population alone revealed a significant association between genotype and phenotype at D7Mit114 that was not present in female F2 mice. QTL analysis of this region on chromosome 7 in the two combined F2 populations(17) failed to reach the level of statistical significance proposed by Lander and Kruglyak(18) to confirm linkage (LOD = 2.9 vs. 3.1 for significant and 4.3 for confirmed), presumably, because of the gender-related dichotomy at this chromosomal locus.
Guidelines for the determination of gender divergence have yet to be established by the genetics community. However, using our predefined criterion for gender difference (more than three orders of magnitude difference in p values), the chromosome 2 and 7 QTLs clearly exhibited gender specificity. Interval mapping was performed on both male and female B6D2F2 data for the chromosome 2 and 7 QTLs with Map Manager QT. The LOD plots based on least squares regression methods (df = 2) are shown in Fig. 7. A significant phenotype-genotype relationship existed for female mice at the midportion of chromosome 2 and for male mice at the proximal portion of chromosome 7, whereas in the opposite gender these same QTLs did not appear to exist.
Currently, we are generating a chromosome 2 congenic strain to further analyze this BMD-related QTL.(6) To assess the fidelity of this line during development, we have intermated heterozygotes (confirmed by microsatellite markers flanking the donor B6 chromosomal segment) to generate an F2 population. This experiment was performed at a stage in the development of this line (fourth backcross) when we estimate that the recipient (D2) strain contributes ∼95% of the unlinked genome. Shown in Fig. 8 are the mean whole body WC-BMD values of male and female mice for the three possible genotypes for the D2Mit249 marker that resides within the introgressed B6 donor region. In female mice, a large difference between female mice homozygous for the D2 allele versus those homozygous for the B6 allele is apparent, as is the intermediate BMD of heterozygous mice. In contrast, male mice again show no impact of allelic variation at this genomic site on peak whole body BMD.
There is good evidence in human populations that bone mass is highly heritable.(19) In both men and women, bone mass has been noted to have genetic determinants in twin and family studies.20-22) Similarly, the relative risk of fracture is increased by a family history of osteoporosis.(23, 24) In female mice, we(25) and others26-28) have shown a similar phenomenon, and here we show a strong genetic basis for peak bone mass in males as well. Because the mouse and human genomes are quite syntenic and the mouse is extremely useful for genetic studies, there are obvious opportunities for better understanding the genetic basis for skeletal development.
In previous studies we used QTL analyses to identify chromosomal loci associated with peak bone mass in female mice. Several of these murine loci have been reported by other groups or are in regions syntenic with chromosomal loci associated with bone mass in humans.(29) This convergence of data strongly supports the presence of genes of importance for skeletal development in these areas. The specific genes responsible have not been identified with certainty, but methods are available to do so. In fact, several known genes located in chromosomal loci identified by QTL analysis appear to be promising candidates.(25)
Epidemiological studies have shown that body weight is a strong predictor of BMD.30-33) Our studies indicate that a similar relationship exists in the laboratory mouse. The coincidence of increased body weight with greater BMD values could stem from environmental factors such as complementary nutritional effects on body composition and skeletal mass, the impact of mechanical loading on skeletal development, or body weight, and BMD may be modulated by linked genes or perhaps even the same genes. Our current experimental models do not allow us to distinguish among these alternatives. However, the fact that correcting whole body BMD for variations in body weight had no impact on the heritability of BMD in male RI mice but reduced the heritability of BMD in female RI by 30% favors the possibility that body weight and peak BMD may be influenced by linked genes or perhaps by common genes with pleiotropic effects.
The substantial differences observed in the BXD RI strain distributions for WC-BMD between male and female mice argues strongly for gender specificity in the regulation of bone mass. QTL analyses of the male and female RI strain sets provisionally identified two chromosomal loci associated with peak BMD that were shared between genders (midportion of chromosome 2 and proximal portion of chromosome 7), indicating that in some cases common genetic mechanisms influence skeletal development. On the other hand, the majority of the provisional loci appear to be associated with peak BMD in one but not both genders. These studies strongly suggested the presence of gender-specific determinants of BMD.
Because QTL mapping with only 18 genotypes (RI strains) will not yield definitive results, we extended our preliminary studies to the more convincing analysis of a genetically heterogeneous F2 population derived from the same B6 and D2 progenitor strains. Strong associations between peak bone mass and three chromosomal locations (chromosomes 1, 4, and 11) were identified in QTL analyses of both male and female B6D2F2 mice, further documenting the presence of shared genetic determinants in chromosomal regions identified in earlier studies in female mice.(17) However, there was no relationship with bone mass in male mice at a fourth locus strongly associated with bone mass in female mice (chromosome 2) and the reverse situation (no relationship with bone mass in female mice at a fifth locus strongly associated with bone mass in males) was found at chromosome 7. The experimental design we used (the analysis of a large F2 population) has substantial power, and our results strongly support the existence of sex-specific pathways engaged in the genetic regulation of peak bone mass. A recent study of opposite-sex twins also hints at the existence of sex-linked genes influencing the variance of BMD in humans.(34)
It is informative to contrast the results obtained in these analyses to those obtained when gender is not considered in the study design (see Klein in accompanying article(16)). When male and female animals were studied together, the QTL on chromosome 7 was not detected using standard criteria. Unless specifically sought, gender effects can lead to a lessened ability to detect potentially important genetic determinants.
Although these results point to genes within regions on chromosomes as being primarily involved in gender differences in peak bone mass, the effects we have observed also are compatible with the effects of other genes acting at these sites (epistasis).(35, 36) That possibility must be considered in designing future studies to identify specific genes and mechanisms of their actions. It should be recognized that our selective genotyping approach markedly reduced our likelihood of detecting epistatic interactions, especially those involving dominance. Our intention is to explore more fully epistasis with complete genotyping of the entire B6D2F2 population in the future. If epistasis is an important element of the effects we observed, similar findings may not be observed in other strains of mice (or other species) that do not share the same complement of genes.
Gender exerts a profound effect on the skeleton. Our studies are the first to identify gender divergence at specific chromosomal loci that influence skeletal development. Sex-specific quantitative trait loci have not been described commonly, but there are several examples. For instance, stress-induced analgesia(37) and alcohol preference(38) in mice, hypertension in rats,(39) and longevity in Drosophila(40) are all affected by sex-specific loci. Because the phenotype of bone density is complex and gender differences are known to exist in other important skeletal characteristics (e.g., bone size), it might be expected that the genetic determinants of these phenotypes also would be influenced by gender.
The major differences in bone size and mass appear at adolescence.(41, 42) This divergence in skeletal phenotypes has long been assumed to exist based on sex steroid action. For instance, male/female differences may in part result from gender-specific effects of sex steroids on periosteal bone formation.(43) Gender also effects the response to growth hormone therapy(44) and the differentiation of adipocytes from mesenchymal precursors.(45) Certainly, the chromosomal loci that reveal gender divergence may be related to the control of or responsiveness to sex steroid action. On the other hand, our results cannot support or refute the hypothesis that gender differences are sex steroid dependent. In fact, the presence of a clear gender divergence in the genetic basis of peak bone mass raises the intriguing possibility that there also may be other mechanisms by which gender influences gene activation. Other mechanisms that may be involved are uncertain but include genes on X or Y chromosomes with complementary activities at autosomal sites. Another possibility is that genomic imprinting results in silencing of either maternal or paternal expression.(46) The influence of imprinting can be examined,(47) but it requires extensive characterization of the genotypes of all progenitors, information not available in the animals studied here. Even if sex steroid action is the root of gender differences, a better understanding of the genes mediating those effects is important.
In summary, the identification of the specific genes responsible for the chromosomal associations with bone mass noted here should yield important insights into the control of skeletal homeostasis. Gender strongly affects bone development and the adult phenotype, and an understanding of the genetic basis for male female differences in growth would be quite informative. Similarly, because the male phenotype is associated with considerable fracture risk reduction, an elucidation of the nature of that effect could provide the basis for novel diagnostic, preventative, or therapeutic approaches. Finally, these results demand that gender differences be considered when the genetic foundations of bone biology are examined.
The authors thank Virginia Chambers, Renn Turner, Kristina Vartanian, and Amy Carlos for skilled technical assistance and Tamara Phillips and John Crabbe for encouragement and advice. This work was supported by funds from the NIH (AA 10760 and AR 44659) and the Medical Research Service of the Department of VA.