Dr Turner serves as a consultant for Eli Lilly and Company, Merck & Co., Inc., and NPS Pharmaceuticals. All other authors have no conflict of interest.
Genetic Effects for Femoral Biomechanics, Structure, and Density in C57BL/6J and C3H/HeJ Inbred Mouse Strains†
Version of Record online: 1 OCT 2003
Copyright © 2003 ASBMR
Journal of Bone and Mineral Research
Volume 18, Issue 10, pages 1758–1765, October 2003
How to Cite
Koller, D. L., Schriefer, J., Sun, Q., Shultz, K. L., Donahue, L. R., Rosen, C. J., Foroud, T., Beamer, W. G. and Turner, C. H. (2003), Genetic Effects for Femoral Biomechanics, Structure, and Density in C57BL/6J and C3H/HeJ Inbred Mouse Strains. J Bone Miner Res, 18: 1758–1765. doi: 10.1359/jbmr.2003.18.10.1758
- Issue online: 2 DEC 2009
- Version of Record online: 1 OCT 2003
- Manuscript Accepted: 19 MAY 2003
- Manuscript Revised: 14 APR 2003
- Manuscript Received: 28 JAN 2003
- quantitative trait loci;
- bone mineral density;
- inbred mice;
Genome-wide QTL analysis for bone density, structure, and biomechanical phenotypes was performed in 999 (B6xC3H)F2 mice. Multivariate phenotypes were also derived to test for pleiotropic QTL effects. Highly significant QTLs were detected with pleiotropic effects on many of these phenotypes, and QTLs with unique effects on specific phenotypes were found as well.
Introduction: The inbred C57BL/6J (B6) and C3H/HeJ (C3H) mouse strains were previously shown to segregate quantitative trait loci (QTLs) for femoral bone density.
Materials and Methods: The 999 s filial (F2) mouse progeny were further phenotyped for measures of femoral biomechanics (load to failure, Fu; work to failure, U; stiffness, S), structure (polar moment of inertia, Ip; moment of inertia ratio, Ir), and more specific femoral midshaft bone density measures (cortical and total vBMD). Two novel multivariate phenotypes were computed using principal component analysis, thus aiding in the exploration of pleiotropic effects of the QTLs detected.
Results and Conclusions: Results of a genome-wide analysis provided strong evidence of pleiotropic QTL effects on chromosome 4, with six of the seven primary phenotypic measures, representing femoral biomechanics, density, and structure, producing LOD scores greater than 8. Chromosomes 1, 8, 13, and 14 were also identified as harboring QTLs that affect phenotypes in two of the three aspects of bone properties. QTLs uniquely contributing to variability in biomechanical measures were identified on chromosomes 10 and 12, whereas a QTL solely affecting structure was found on chromosome 17. Analysis of the evidence for pleiotropic effects using principal component analysis revealed pleiotropic QTLs on chromosomes 4 and 14, influencing nearly all the bone phenotypes measured and revealed QTLs on chromosomes 1, 8, 13, and 17 with pleiotropic effects restricted to either density or the structure and stiffness phenotypes. The use of multivariate phenotypes has allowed us to identify pleiotropic effects of several QTLs previously linked in studies of other mouse strains and in human studies of bone mineral density and femoral structure, which will provide important insight regarding the importance of allelic variation on the entire skeleton.
Osteoporosis is a complex disorder resulting from the influence of both genetic and environmental factors. Twin and family studies have consistently reported a substantial genetic contribution to osteoporosis.(1–3) Important factors contributing to osteoporosis include low bone mineral density (BMD),(4,5) poor bone structure,(6,7) and impaired bone biomechanics.(8) Both BMD(9) and bone structure(10) have been shown to be highly heritable. However, genes contributing to osteoporosis, bone density, and bone structure have not been consistently identified.
During the past decade, mouse models of osteoporosis have been actively investigated as a complementary means to identify genes contributing to osteoporosis susceptibility. A series of studies in recombinant inbred lines and a variety of inbred F2 crosses have yielded several chromosomal regions consistently linked to quantitative trait loci (QTLs) for BMD, including segments of mouse chromosomes 1, 4, 11, 13, and 18.(8) Importantly, the use of different inbred strains has resulted in the identification of unique BMD QTLs.(11–13) Unlike human studies, animal models are ideal for the study of biomechanical measures. Human studies of BMD(9) and femoral bone structure(10) have identified unique QTLs for each phenotype. Similarly, QTLs, which contribute uniquely to BMD and femoral structure(13–15) and to biomechanical properties,(16,17) have also been identified in mice. Thus, the study of different complementary phenotypes of bone biomechanics, BMD, and bone structure is critical for the identification of genes influencing osteoporosis.
A series of unique studies has been ongoing to thoroughly characterize the bone related phenotypes of the C3H/HeJ (C3H) and C57BL/6J (B6) mice. Previous linkage studies identified QTLs on mouse chromosomes 1, 4, 6, 13, and 18 for femoral volumetric BMD (vBMD) in a sample of (B6 × C3H) F2 mice.(12) Additional femoral phenotypes have now been examined in these F2 mice, including measures of bone structure and bone biomechanics. In addition, a more refined measure of femoral vBMD was also been developed.
We have obtained data on multiple correlated, heritable phenotypes representing bone density, structure, and biomechanics. One possible explanation for the observed correlation structure is a single gene influencing several of the phenotypes, or pleiotropy. To informally test for pleiotropy, given genetic data on multiple quantitative phenotypes, we have used the statistical technique of principal component analysis. These components, or multivariate phenotypes, can be used as additional phenotypes for quantitative genetic linkage analysis. The presence of a QTL linkage peak for a principal component phenotype provides support for the hypothesis of pleiotropic effects of a gene at that chromosomal location on the multiple phenotypes contributing to the principal component measure. In this way, we have now been able to test whether some genes affect multiple aspects of bone (pleiotropy) and also to identify those genes that uniquely affect a limited number of bone phenotypes.
MATERIALS AND METHODS
This study used two inbred strains of mice, B6 and C3H, previously shown to differ greatly in total femur vBMD.(18) All mice were produced and maintained at The Jackson Laboratory as previously described.(12) A total of 999 female mice from the second filial (F2) generation from B6 and C3H progenitors were raised, and femurs were collected for analysis. The mice used for experimental analyses were necropsied at 16 weeks of age when acquisition of adult BMD was achieved. Femurs were isolated for experimental measurements and stored in 95% ethanol. Kidneys and spleens from each mouse were frozen in liquid nitrogen and stored at −60°C for later extraction of genomic DNA. Use of mice in this research project was reviewed and approved by the Institutional Animal Care and Use Committee of The Jackson Laboratory. Femur length and weight were measured for each animal for consideration as possible covariates for the bone structure, density, and biomechanical measures.
Femora were tested at the mid-shaft by three-point bending at room temperature. Load was applied in the anteroposterior direction midway between two supports that were 5 mm apart. Load-displacement curves were recorded at a crosshead speed of 0.5 mm/s using a microforce materials testing machine (Vitrodyne 2000; Liveco, Burlington, VT, USA). From the load displacement curve, the load to failure (Fu) was calculated, reflecting the strength of the bone.(19) Stiffness (S), which reflects bone rigidity, and work to failure (U), which is the energy necessary to cause a fracture, were also computed.(19)
Measurement of bone structure:
Femurs from the mice were analyzed for bone structure using peripheral quantitative computed tomography (pQCT; QCT Research SA Plus; Norland Medical Systems, White Plains, NY, USA). Each pQCT measurement included information from one 110-μm slice from the midpoint of the femoral shaft. A voxel size of 70 μm was used for analysis. Structure of the femoral shaft was measured from pQCT scans using BonAlyse software (BonAlyse Ltd., Jyväskylä, Finland). A subset of the bones (n = 48) were analyzed using microcomputed tomography (μCT 20; Scanco Medical AG, Bassersdorf, Switzerland), with structure measurements taken at the femur midshaft at 17-μm resolution. Scion Image 4.0.2 (Scion Corp., Frederick, MD, USA) was used to calculate femoral structure from μCT images.
To assess structural rigidity, the maximum (Imax) and minimum (Imin) moments of inertia were calculated from the pQCT scans. Polar moment of inertia Ip was computed as the sum of Imax and Imin, and moment of inertia ratio Ir as the ratio of Imax to Imin. Moments of inertia represent the resistance of a bone to torsional and bending loads. Imax and Imin were highly correlated with Ip (r > 0.97; p < 0.0001); therefore, only Ip and Ir data are presented in this paper. Calculations based on measurements of the same bones by pQCT and μCT were highly correlated (pQCT = 0.01 + 1.01 × μCT; r = 0.96; n = 48).
Measurement of BMD:
Previously, Beamer et al.(12) used a composite femur vBMD phenotype that was an average computed at 2-mm intervals over the entire femoral shaft. Thus, the measure was affected by femoral size, mineralization, and the relative proportions of cortical and cancellous bone. Current studies reported herein used a simplified vBMD measure that only included cortical bone at the femoral midshaft. Total and cortical vBMD were assessed using pQCT. A threshold of 900 mg/ml was used to separate cortical from trabecular bone. Total vBMD represents the bone mineral at the femoral midshaft divided by the total bone volume (including bone marrow space). Cortical vBMD is the bone mineral divided by only the cortical bone volume. Because pQCT measurement of BMD in mice is affected by partial volume averaging, we verified the accuracy of the cortical cross-sectional area measurement (this measurement is typically altered most by partial volume averaging) using high resolution μCT, which has minimal partial volume averaging error. Cortical areas measured using the two techniques were highly correlated (pQCT = 0.04 + 0.99 × μCT; r = 0.96; n = 48), suggesting that pQCT, as it was used in this study, has sufficient accuracy to measure cortical area, which is required for accurate estimation of cortical vBMD.
The preparation of genomic DNA and the genotyping of microsatellite markers were performed as previously described.(12) A total of 107 markers were genotyped in 999 of the F2 progeny. Marker maps were generated using MAPMAKER/EXP(20) with data from all available genotyped F2 animals. Marker order and distances were compared with those previously published (The Jackson Laboratory, Bar Harbor, ME, USA). The average intermarker distance was 18.5 cM.
Stepwise regression analysis was used to identify significant (p < 0.05) covariates with the two density (total and cortical vBMD), two structure (Ip and Ir,), and three biomechanical variables (Fu, S, and U). Body weight was a significant predictor of all seven phenotypes; therefore, regression residuals, representing weight-adjusted measures, were computed and used in all subsequent analyses. Pairwise correlation analysis was performed among the seven measures to examine the underlying phenotypic relationships among the variables.
The seven adjusted phenotypic variables were then subjected to principal component analysis (PCA) (SAS PRINCOMP; SAS v. 6.12; SAS Institute, Cary, NC, USA) to identify novel multivariate phenotypes for pleiotropic genetic effects. Examination of the correlation structure of the multiple phenotypes resulted in a derived measure, or “first principal component.” This measure is a simple linear combination of the original measured phenotypic values and explains the maximum possible amount of the common phenotypic variation that can be achieved with a single summary measure. Subsequent independent components can also be generated, each explaining the maximum possible amount of the remaining variation among the phenotypes. These components can then interpreted based on the estimated coefficient for each variable. Those variables with larger coefficient estimates, both positive and negative, have a greater linear contribution to the final component estimate. The eigenvalue corresponding to each principal component was examined to identify those components explaining a substantial proportion of the underlying phenotypic variation (>5%). These components were then used in subsequent genome wide linkage analyses.
Linkage analyses were then performed to identify chromosomal regions linked to the seven primary bone measures as well as the two principal component phenotypes. Interval mapping, as implemented in the program QTL/Cartographer,(21) was used to evaluate evidence for linkage. ANOVA was performed for the most significant marker in each QTL region to further characterize significant genotypic group differences.
To obtain appropriate genome-wide significance thresholds for the linkage results, permutation tests(22) were performed. Specifically, the seven adjusted traits and two principal components measured on each animal were randomly reassigned as a group across the 999 animals, resulting in a permuted data set. By keeping all phenotypic data together, the underlying phenotypic correlations were preserved. For 5000 such permuted data sets, interval mapping was then performed across the whole genome, and the resulting maximum logarithm of the odds (LOD) scores for linkage for each phenotype were recorded. In this manner, the LOD significance thresholds for the 95th and 99th percentiles of the maximum genome-wide LOD scores across all phenotypes were found to be 3.5 and 4.4, respectively. The above procedure was repeated for the PCA, where the first two principal components for each animal were randomly reassigned as a pair across individuals. For this case, the LOD thresholds for the 95th and 99th percentiles were found to be 3.5 and 4.2, respectively.
Novel phenotypic measures
Seven femoral phenotypes were measured in 999 female (B6 × C3H) F2 animals. Greatest pairwise correlations were observed among the phenotypes measuring similar aspects of bone properties (biomechanics, structure, and BMD; Table 1). The seven phenotypes were then subjected to a principal component analysis, from which two novel multivariate phenotypes explaining 65.1% of the common phenotypic variability (Table 2) were retained for subsequent linkage analysis. Only these two components had associated eigenvalues greater than the traditional threshold of one. The weightings of these two components, representing the degree and direction of each of the measured variable's contribution to each component, are shown in Table 2 as well.
The first principal component was positively weighted for all seven femoral measures and therefore may represent common influences on all three measured aspects of bone properties. The second principal component explained a lower proportion of the common phenotypic variation. It was most strongly positively weighted for cortical and total vBMD density. This principal component was negatively weighted for the femoral structure variables, Ip and Ir, and the biomechanical phenotype, stiffness (S). Thus, the second principal component implied independent factors influencing femoral structure (Ip and Ir) and stiffness or vBMD.
Loci with pleiotropic effects on multiple aspects of bone
Evidence of linkage (p < 0.01) of at least two phenotypes measuring each of the three aspects of bone properties was observed in mouse chromosome 4 (Fig. 1A). The QTL on chromosome 4 attained maximum LOD scores of ≥16 with phenotypes representing biomechanics (load to failure, Fu), BMD (total vBMD), and femoral structure (polar moment of inertia, Ip). Three additional phenotypes, U, cortical vBMD, and S, all had LOD scores ≥8. Analysis of the first principal component resulted in a maximum LOD score of 25 in the same chromosomal region (Fig. 2).
Significant linkage (p < 0.01) was also detected in markers on chromosome 14 at 37 cM, with multiple phenotypes representing biomechanics and density measures (Fig. 1E). A maximum LOD score of 10 was attained with the biomechanical phenotype S and a LOD score of 8 with Fu. For BMD (total vBMD), the maximum LOD score was 6 in the same chromosomal region. The first principal component also attained a highly significant LOD score (LOD = 7) in the same region of chromosome 14 (Fig. 2).
On the telomeric portion of chromosome 13, LOD scores of 10 or greater were obtained for bone phenotypes representing biomechanics and structure (Fig. 1D). A QTL for structure (Ir) and biomechanics (Fu) was detected near the telomere (53–59 cM) of chromosome 13, with LOD scores of 10 and 14, respectively. Two additional biomechanical measures (S and U) both had LOD scores of 8 and also mapped to the telomeric region of chromosome 13. The second principal component had its greatest evidence of linkage in the centromeric region of chromosome 13 (Fig. 2).
Several unique QTL regions were found that contributed to variability in femoral structure and/or biomechanical properties without having major effects on BMD (LOD <5). These non-BMD linkages were on chromosomes 8, 10, 12, and 17 (Table 3). On chromosome 8, evidence of linkage (LOD = 14; 60 cM) to femoral structure was found using the Ip phenotype. Linkage to femoral biomechanics was also detected in the same chromosomal region, with significant LOD scores (p < 0.01) for both Fu and S (Fig. 1B). The second principal component, contrasting density with the structure and biomechanical phenotypes, had a LOD score of 5 in this same region of chromosome 8 (Fig. 2). The chromosome 8 QTL finding was unique among the significant QTLs for bone biomechanical, density, and structure phenotypes, in that mice homozygous for the B6 allele had the highest mean phenotypic value. For the other QTLs, the opposite trend was observed (Table 4).
Strong evidence of a QTL (LOD = 21) was found for the new total vBMD phenotype on chromosome 1 at 71 cM. Linkage to this chromosomal region was supported by a LOD score of 10 at 72 cM for the cortical vBMD phenotype and a LOD score of 6 at the same position for the biomechanical phenotype Fu. A LOD score of 9 in this same region (at 69 cM) was obtained for the second principal component, which is a contrast of density with the other bone phenotypes measured.
Loci affecting a single aspect of bone
Significant evidence of a QTL on chromosome 10 was observed uniquely with the bone biomechanical phenotypes. A LOD score of 13 was achieved with the work to failure phenotype (U) at the 18-cM position (Fig. 1C), and a LOD of 9 was attained in the same chromosomal region for the load to failure (Fu) phenotype (22 cM). On chromosome 12, a LOD score of 8 was detected with the U phenotype at position 7 cM, and a LOD score of 6 with the Fu phenotype was detected at the same chromosomal position. Evidence of linkage (LOD = 8; p < 0.01) to the phenotypes of cortical vBMD and total vBMD was detected at the centromeric portion of chromosome 13 (15 cM). This centromeric chromosome 13 QTL was distinct from the QTL detected for biomechanics and structure on the telomeric portion of chromosome 13. On chromosome 17, a LOD score of 8 was observed with the second principal component phenotype. Polar moment of inertia (Ip) achieved a LOD score of 5 on chromosome 17.
On chromosome 4, six of the seven measured femoral phenotypes had LOD scores of 8 or greater. The strong evidence for pleiotropy in this chromosomal region is further supported by a LOD score of 25 with the first principal component. This region of mouse chromosome 4 has consistent synteny or preservation between the species of a chromosomal segment containing multiple genes in the same order, with human chromosome 1p, to which linkage with BMD has been reported previously.(23,24) Linkage to this region in the mouse was also reported with total femur BMD measured in this same F2 sample by Beamer et al.(12) Previous studies of femoral structure in humans(10) and B6 × D2 F2 mice(15) did not detect linkage to this chromosomal region. Femoral structure phenotypes measured in both of these studies were substantially different from the femur midshaft structural properties considered in this study.
In similar fashion, evidence of linkage of three phenotypes, representing bone density and biomechanical properties, was found in mouse chromosome 14. Pleiotropic effects of the QTL on chromosome 14 were further indicated by a LOD of 7 for the first principal component. Klein et al.(13) reported evidence of linkage to chromosome 14 at position 2 cM in a sample of B6 × D2 F2 mice measured for total body BMD by DXA. This region is 30–40 cM from the QTLs identified in this report and may represent a distinct linkage finding. Studies of BMD(9,23,24) and femoral structure(10) in various human samples have not reported linkage to chromosome 13, the most likely syntenic region to mouse chromosome 14.
Linkage to two of the three measured aspects of bone was identified on mouse chromosomes 1, 8, and 13. Density (total vBMD), biomechanical measures (Fu), and the second principal component demonstrated linkage to chromosome 1 at position 71 cM. This region of chromosome 1 was also linked to total body BMD in a sample of B6 × D2 F2 mice.(13) Linkage of femoral BMD and Fu to chromosome 1 was also detected in MRL × SJL F2 mice by Li et al.(16) at position 104 cM. In addition, studies in a sample of premenopausal sister pairs have also detected linkage to the syntenic region on human chromosome 1q.(9)
On chromosome 8, both measures of biomechanics (Fu and S), as well as structure (Ip), yielded significant LOD scores (p < 0.01), with a maximum LOD score of 14 for the Ip phenotype. Klein et al.(15) reported linkage to a similar region of chromosome 8 (30–40 cM) when analyzing femoral cross-sectional area in a sample of B6 × D2 F2 mice. Linkage of biomechanical measures Fu and U to chromosome 8 has been reported in MRL × SJL F2 mice,(16) albeit at a somewhat different position (16 cM). Our observation that the chromosome 8 QTL acts in a different direction from the other significant QTLs (with animals homozygous for the B6 allele having a higher phenotypic mean for Ip, Fu, and S) likely indicates that QTL alleles for robust bone phenotypes are present in both the B6 and C3H mouse strains. This is not surprising, because these inbred strains were not developed under specific selection for these bone phenotypes.(18)
Measures of the three biomechanical phenotypes (U, Fu, and S) and the structural phenotype Ir all link to the telomeric portion of chromosome 13. No other study of structural measures in mice(15) or humans(10) reported linkage to this chromosomal region. A highly significant QTL for the biomechanical measures Fu and U was detected on chromosome 10 near the 20-cM position in our sample of C3H × B6 F2 mice. Li et al.(16) also reported a QTL for Fu in MRL × SJL F2 mice; however, their QTL peak was a substantial distance away from our finding, at the 50-cM position. Linkage to the U phenotype was not detected by these authors on chromosome 10.(17)
We find evidence that loci contribute to quantitative variation in all three of the measured aspects of bone properties, and the loci seem to contribute to variation in only one phenotype or to phenotypes representing only one of the three aspects of bone phenotypes measured in this study. Based on current knowledge of bone biology, we can speculate regarding the types of genes that might be underlying these QTLs. Variation in genes coding for structural proteins composing the bone matrix (e.g., collagens and cross-linking proteins) might be expected to have effects on bone structure, density, and biomechanical properties. Genes involved in bone turnover could be expected to have effects similar to pharmacologic agents currently in use in humans, increasing BMD and improving biomechanical strength independent of BMD. Finally, genes with more specific effects might affect one of the aspects uniquely. For example homeobox (HOX) genes, proposed to play a role in structural arrangement of vertebrate tissues, might contribute to structural variability, with little effect on density or biomechanical properties. This would be consistent with the observation that HOX genes can have substantial effects on specific bones from widely divergent developmental pathways.(25)
Several of the chromosomal regions nominated in this study have also been linked in human studies of related phenotypes. Unfortunately, each of the linked regions in humans is quite large, often encompassing more than 30 megabases of genomic DNA. Unlike the mapping of simple Mendelian traits, recombination-based methods cannot be used to narrow the critical genetic interval. Therefore, it is particularly important that some of the animal models have identified QTLs that are syntenic to a number of the chromosomal regions nominated in previous human studies. On the hypothesis that the same locus is contributing to phenotypic variability in both the human and the mouse, strategies such as congenic breeding protocols can be undertaken, which are impossible in the human and which can rapidly narrow the chromosomal region of interest containing the QTL.(26)
Our derived principal component measures explain a substantial majority of the overall phenotypic variability, with nearly two-thirds of this variability accounted for by the first two principal components. QTLs linked to these components can be interpreted, based on the loadings of each component on the original phenotypes, as containing genes underlying variation in all or nearly all of the measured density, structure, and biomechanical measures for the first principal component, and genes uniquely affecting Ip, Ir, and S or uniquely affecting variability in femoral midshaft vBMD, for the second principal component. We could have examined the LOD score results for all combinations of phenotypes in each chromosomal region to identify coincident linkage. However, the use of the first and second principal components in the linkage analyses ensured that an unbiased, consistent approach to pleiotropic detection was used. Our highly significant linkage findings with these principal component measures strongly support the hypothesis that the QTLs detected have pleiotropic effects. An alternate explanation for these findings might be multiple, but closely linked genes, within our QTL regions, acting on the phenotypes showing linkage in that region.(27) This could be evaluated experimentally through the testing of congenics or could be determined definitively after the relevant genes in the QTL interval are identified. The QTLs on chromosomes 4 and 14, showing strong evidence of linkage to the first principal component, seem to have pleiotropic effects across a wide range of density, structure, and biomechanical phenotypes, whereas the chromosome 1, 8, 13, and 17 QTLs have major effects limited either to femoral midshaft vBMD or to affect structure and stiffness with little effect on density. The robust linkage finding with the second principal component on chromosome 17 adds additional support to the observation of phenotype-specific effects for this QTL and illustrates the benefits of performing multivariate analyses in addition to analysis of each phenotype separately. These findings may aid greatly in the progress toward understanding the genetic basis for variability in bone phenotypes, including the degree to which these phenotypes are under the influence of common genetic factors.
This work was supported by National Institutes of Health Grants R01AR046530 (CHT), R01AR043618 (WGB), and P01AG018397 (CHT, DLK, TF).
- 11994 Reduced femoral neck bone density in the daughters of women with hip fractures: The role of low peak bone density in the pathogenesis of osteoporosis. J Bone Miner Res 9:739–743., , , ,
- 21995 Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med 332:767–773., , , , , , , ,
- 31996 The heritability of bone mineral density, ultrasound of the calcaneus and hip axis length: A study of postmenopausal twins. J Bone Miner Res 11:530–534., , , ,
- 41993 Long-term fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res 8:1227–1233., , , ,
- 51996 Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 312:1254–1259., ,
- 61995 Better discrimination of hip fracture using bone density, geometry and architecture. Osteoporos Int 5:167–173., , , , ,
- 71995 Hip axis length and osteoporotic fractures. Study of Osteoporotic Fractures Research Group. J Bone Miner Res 10:506–508., , , , ,
- 82002 Genetics of osteoporosis. Endocr Rev 23:303–326., , ,
- 92000 Genome screen for QTLs contributing to normal variation in bone mineral density and osteoporosis. J Clin Endocrinol Metab 85:3116–3120., , , , , , , , , , , ,
- 102001 Genome screen for quantitative trait loci underlying normal variation in femoral structure. J Bone Miner Res 16:985–991., , , , , , , , , , ,
- 111999 Quantitative trait loci for bone density in C57BL/6J and CAST/EiJ inbred mice. Mamm Genome 10:1043–1049., , , , , ,
- 122001 Quantitative trait loci for femoral and lumbar vertebral bone mineral density in C57BL/6J and C3H/HeJ inbred strains of mice. J Bone Miner Res 16:1195–1206., , , , , , ,
- 132001 Confirmation and fine mapping of chromosomal regions influencing peak bone mass in mice. J Bone Miner Res 16:1953–1961., , , , , , ,
- 141998 Quantitative trait loci affecting peak bone mineral density in mice. J Bone Miner Res 13:1648–1656., , , ,
- 152002 Mapping quantitative trait loci that influence femoral cross-sectional area in mice. J Bone Miner Res 17:1752–1760., , , , , , , ,
- 162002 Genetic dissection of femur breaking strength in a large population (MRL/MpJ x SJL/J) of F2 Mice: Single QTL effects, epistasis, and pleiotropy. Genomics 79:734–740., , , , ,
- 172002 Chromosomal regions harboring genes for the work to femur failure in mice. Funct Integr Genomics 1:367–374., , , , , , ,
- 181996 Genetic variability in adult bone density among inbred strains of mice. Bone 18:397–403., , ,
- 191993 Basic biomechanical measurements of bone: A tutorial. Bone 14:595–608.,
- 201987 MAPMAKER: An interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174–181., , , , , ,
- 211999 Estimating the genetic architecture of quantitative traits. Genet Res 74:279–289., ,
- 221996 Permutation tests for multiple loci affecting a quantitative character. Genetics 142:285–294.,
- 231998 First-stage autosomal genome screen in extended pedigrees suggests genes predisposing to low bone mineral density on chromosomes 1p, 2p and 4q. Eur J Hum Genet 6:151–157., , , , , , , , , , , , , ,
- 241999 A genome-wide scan for loci linked to forearm bone mineral density. Hum Genet 104:226–233., , , , , , , , ,
- 251995 MHox and vertebrate skeletogenesis: The long and the short of it. Bioessays 17:750–753.
- 262002 Construction of a BAC contig for a 3 cM biologically significant region of mouse chromosome 1. Genetica 114:1–9., , , , , , , ,
- 272002 Two closely linked interactive blood pressure QTL on rat chromosome 5 defined using congenic Dahl rats. Physiol Genomics 8:81–86.,