• skeleton;
  • fractures;
  • biomechanics;
  • linkage genetics;
  • quantitative trait


  1. Top of page
  2. Abstract
  7. Acknowledgements

Fracture susceptibility depends jointly on bone mineral content (BMC), gross bone anatomy, and bone microarchitecture and quality. Overall, it has been estimated that 50-70% of bone strength is determined genetically. Because of the difficulty of performing studies of the genetics of bone strength in humans, we have used the HcB/Dem series of recombinant congenic (RC) mice to investigate this phenotype. We performed a comprehensive phenotypic analysis of the HcB/Dem strains including morphological analysis of long bones, measurement of ash percentage, and biomechanical testing. Body mass, ash percentage, and moment of inertia each correlated moderately but imperfectly with biomechanical performance. Several chromosome regions, on chromosomes 1, 2, 8, 10, 11, and 12, show sufficient evidence of linkage to warrant closer examination in further crosses. These studies support the view that mineral content, diaphyseal diameter, and additional nonmineral material properties contributing to overall bone strength are controlled by distinct sets of genes. Moreover, the mapping data are consistent with the existence of pleiotropic loci for bone strength-related phenotypes. These findings show the importance of factors other than mineral content in determining skeletal performance and that these factors can be dissected genetically.


  1. Top of page
  2. Abstract
  7. Acknowledgements

FRACTURES OCCURRING as a consequence of skeletal fragility are an important health problem.(1, 2) Yet, relatively little is known regarding the determination of bone strength. Most investigations have focused on bone mineral density (BMD), because this is a phenotype that is easily and reproducibly measured(1) and for which pharmacologic interventions are available.(2) However, although BMD possesses some predictive value with regard to human fracture risk, other risk factors have been identified, including past fracture history, maternal fracture history, visual acuity, height, level of physical activity, general health status, and use of steroids or anticonvulsants.(3) From epidemiological surveys, some of these additional risk factors appear to be comparable in importance with BMD but have not been studied in nearly as great detail.

Although the morbidity and mortality arising from skeletal fragility are suffered predominantly by the elderly, many aspects of bone strength are determined early in life.(4) Bone mass peaks at approximately age 30 years, with bone mineral loss occurring over the remainder of the life cycle. Both peak bone mass achieved and rate at which bone loss occurs are complex traits, determined by interactions between genetic constitution and environmental factors. Overall, it has been estimated that 50-70% of bone strength is determined genetically.(5, 6–15)

Because of the difficulty in performing studies of the genetics of bone strength in humans, we have used the HcB/Dem series of recombinant congenic (RC) mice to investigate this phenotype. The RC mice used in this report are inbred lines descended from third generation backcross brother-sister pairs in which strain C3H/DiSnA (C3H) was the background progenitor and C57BL/ScSnA (B10) was the donor progenitor.(16–18) Each of the 27 HcB/Dem strains carries an average of 12.5% B10 genome, but each has a different combination of donor and recipient strain alleles. The chromosome regions derived from each progenitor can be determined by genotyping for markers distributed throughout the genome. Because they are inbred, the results of mapping apply to all members of the strain and are cumulative. Phenotyping RC strains allows one to include multiple individuals as replicates, allowing generation of sample sizes that are sufficiently large to establish the presence of small differences between individual strains, just as is true of any inbred strain.

The experiments reported here were performed to achieve three objectives. First, we sought to describe the long bone properties of C3H, B10, and 24 of the HcB/Dem strains. Failure load, structural stiffness, failure stress, and modulus were determined through three-point bend tests of the animals' left humeri. Middiaphyseal inner diameter (ID) and outer diameter (OD) cross-sectional area (CSA), and moment of inertia (I) were determined through image analysis of radiographs. Ash percentage of femora from the same animals was determined gravimetrically. Body mass and body mass index (BMI) were determined by weighing and measuring animals at death. Second, these phenotypic data and the published HcB/Dem genotypic data(19, 20) were used to perform quantitative trait locus (QTL) linkage mapping of the genes that contribute to the measured phenotypes. Third, we investigated the interrelationships among the various bone properties by developing multivariate regression models of failure load and by examining patterns of pleiotropy among the QTLs discovered by linkage analysis. The ability to define components of biomechanical performance and map them individually contributes to a deeper understanding of how genetic factors determine bone strength than would the study of any single parameter. This comprehensive approach allows simultaneous estimation of the degree to which individual parameters share common genetic determinants and the contribution of each identified genomic region to biomechanical performance.


  1. Top of page
  2. Abstract
  7. Acknowledgements


The HcB/Dem strains were established and are maintained at the Netherlands Cancer Institute. Briefly, the HcB/Dem mice are inbred strains derived from arbitrary pairs of N3 backcross animals.(16–18) The HcB/Dem strains have been genotyped at 130 marker loci distributed over each of the autosomes.(19, 20) Because they are inbred, individuals from a single strain have the same genetic composition, except for new mutations and residual unfixed chromosome segments.(21, 22) Less than 5% of the genome was unfixed at the time of genotyping; residual heterozygosity is expected to be reduced by half in each generation of inbreeding. The mice described in this report were provided by Dr. Peter Demant and maintained at the Hospital for Special Surgery until 6-7 months of age under 12 h light-dark cycling and fed irradiated PICO 5058 rodent chow (Purina, St. Louis, MO, USA) and autoclaved tap water ad lib. Beamer and colleagues have reported that although mice continue to grow over the lifespan, peak bone mass is achieved at 4 months of age.(23) We examined strains HcB/1 through HcB/9, HcB11 through HcB/15, HcB/17, HcB/18, HcB/20 through HcB/23, HcB/25, HcB/26, HcB/28, and HcB/29. HcB/16 is extinct and HcB/27 is an unassigned number; so HcB/10, HcB/19, and HcB/24 were unexamined because these strains were unavailable at the time the experiments described here were performed. Between 4 and 10 animals of each strain were studied, yielding 179 HcB/Dem mice. Animals were allowed ad lib activity and were housed 1-5 animals/cage. Only females were studied, because inclusion of males would have introduced sex-dependent variability in the traits in addition to the strain-specific and environmental variability already encountered. At death, body mass and rostroanal length were measured. This work satisfied The Hospital for Special Surgery's requirements for the ethical use of laboratory research animals.

Ash percentage

Bone mineral fraction was calculated by comparison of dry, defatted bone weight to ash weight of homogenized tissue.(24, 25) We chose to use entire bones rather than bones from which the epiphyses and marrow have been removed because the former technique is more reproducible in our hands. No ash data were obtained from the HcB/6 strain because of a laboratory accident.

Radiographic analysis

Image analysis of fine focus contact radiographs of dissected humeri was performed as described and used to calculate CSA and I.(25) Humeral length was defined as the distance along the diaphysis from the trochlea to the humeral head's most distant point. ODs and IDs were measured in orthogonal projections just distal to the deltoid tuberosity, perpendicular to the diaphyseal axis. CSA was calculated according to the elliptical approximation.(25)

  • equation image

where OR is the outer radius and IR is the inner radius in either the mediolateral (ML) or anteroposterior (AP) projection. I also was calculated according to the elliptical approximation(26)

  • equation image

Image analysis was performed with SigmaScan (Jandel Scientific) image analysis software. All radiographs included stepped aluminum densitometric phantoms. Radiographic images were digitized with a digital camera (Kodak, Rochester, NY, USA), with the photographic field including a length scale.

Biomechanical testing

Quasi-static three-point bend testing was performed on left humeri using posts designed and machined in-house with the MTS apparatus (MTS Systems Corp., Eden Prairie, MN, USA) and Instron (Canton, MA, USA) electronics as described.(27, 28) Humeri were oriented with the deltoid tuberosity downward and the specimens were oriented with the central post adjacent to the distal end of the deltoid tuberosity. This orientation corresponds to the ML axis being parallel to the applied force. Posts were separated by 3.75 mm except for the B10 specimens, which were tested at a post separation of 3.0 mm.

Biomechanical data were analyzed following several important assumptions. First, we assumed that bone strength is determined entirely by the cortical bone in the middiaphysis. Second, we assumed that the humeral diaphysis is an ellipse with its major axis lying in the ML plane and its minor axis lying in the AP axis. Calculated biomechanical parameters were obtained according to the following standard formulas for three-point bending of ellipses(26): stress (σ; MPa) = FLc/4I, where F is force, L, is length, and c is ML OR; strain (ϵ; mm/mm) = 12cd/liter2, where c is ML OR, d is displacement, and L is length; and Young's Modulus (E; MPa) = (F/day)(L3/48I).

Linkage analysis

Linkage mapping was carried out using the QTL cartographer software suite,(29, 30) using sib-recombinant inbred as the cross type. The breeding scheme of RC mice is equivalent to that for recombinant inbred mice except for the fact that full-sib matings are started after the third backcross rather than the second filial generation. This breeding scheme results in multiple opportunities for recombination to occur before fixation of the genotype. Consequently, linkage relationships are weakened according to the relationship R = 4r/(1 + 6r), where R is the observed fraction of recombinant genotypes and r is the single generation recombination fraction between a pair of loci.(21, 22) Briefly, data were first analyzed for normality and those traits for which Fisher's cumulant test for normality(31) was significant at the 5% level were log-transformed (failure stress, modulus, body mass, and BMI) before further analysis. Transformed data were distributed normally by this criterion. Interval mapping was carried out to generate the final model.(32, 33) A permutation test using 1000 simulated data sets was performed to estimate empirically significance levels.(34, 35) Linkage maps were plotted with Gnuplot.(36) Linkage maps were generated for strain-averaged data.

Other statistical analysis

All values are shown as the mean ± SD. Continuous variables were analyzed by analysis of variance (ANOVA) ƒ test. Phenotypes that were not distributed normally were log-transformed before analysis. In the specific case of BMI, the transform was log(1 + BMI). In general, the α-level was 0.05 with adjustment for multiple comparisons. We used linear regression with stepwise addition and backward elimination to generate the multivariate model for failure load as a function of the other parameters, using an initial inclusion criterion of p < 0.047 for adding parameters and a criterion of p < 0.05 for their retention.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Size and skeletal morphology

The body masses, humeral CSAs, and Is are shown in Fig. 1. The parental strains do not differ with respect to body mass or CSA, with C3H mice having a body mass of 23.2 ± 1.5 g and a CSA of 0.53 ± 0.03 mm2 and B10 mice having a body mass of 25.4 ± 1.4 g and a CSA of 0.55 ± 0.04 mm2. However, they differ markedly with respect to I (0.044 ± 0.005 mm4 for C3H and 0.089 ± 0.007 mm4 for B10). Although only I differs between the parental strains, each of these parameters varies markedly among the HcB/Dem strains. Minimum and maximum values for the various parameters and the corresponding strains are shown in Table 1. Also included in Table 1 are summary data for the diaphyseal diameters used to calculate CSA and I. The distribution of values among the HcB/Dem strains is consistent with segregation between the parental strains contributing to each of these phenotypes. Body masses of the HcB/Dem strains ranged between 18.2 ± 0.5 g (HcB/13) and 27.9 ± 1.3 g (HcB/6). Most HcB/Dem strains had larger CSAs than the parental strains, ranging from a minimum of 0.49 ± 0.06 mm2 (HcB/8) to a maximum of 0.72 ± 0.05 mm2 (HcB/14).

Table Table 1.. Parental Phenotypes and Phenotypic Ranges
Thumbnail image of
thumbnail image

Figure FIG. 1.. Body masses, humeral CSAs, and humeral I's for the HcB/Dem strains. Each bar graph shows the average ±1 SD for C3H/DiSnA, C57BL/10ScSnA, and each of the HcB strains tested. Sample sizes range between 4 and 10 animals/strain.

Download figure to PowerPoint

The magnitude of the differences in CSA and I is shown in Fig. 2. Although CSA is equal between the parental strains, the source of their marked difference in I is apparent from this figure, which shows the striking dissimilarity of their diaphyseal diameters. Thus, although C3H mice have a thick diaphyseal cortex and a small diaphyseal diameter, B10 mice have the converse phenotype. This is reflected in I, which is proportional to the product of CSA and the square of the distance of that CSA from the bending axis for the three-point bending test. The divergence of CSA among the HcB/Dem strains arises as a result of dissociation of diaphyseal diameter from cortical bone thickness. HcB/14 has both a thick diaphyseal cortex and a large overall diameter, while HcB/8 is characterized by a thin cortex and a small overall diameter.

thumbnail image

Figure FIG. 2.. Humeri of the parental strains. ML contact radiographs of (A) C3H/DiSnA and (B) C57BL/10ScSnA are shown.

Download figure to PowerPoint

Biomechanical performance and ash analysis

We performed quasi-static three-point bending tests of the left humeri for each animal. In this test, the fracture is initiated in the plane of contact between the specimen and the central post of the apparatus. This protocol allows the central post to be placed at a recognizable anatomical site and measurements of the bone can be made at the same site. Results of these studies are summarized in Table 1 and Fig. 3. Failure load is one of the most basic measures of structural strength—the force needed to fracture the bone. The parental strains do not differ in this parameter, with C3H mice having a failure load of 7.4 ± 0.3 N and B10 mice having a failure load of 7.7 ± 1.4 N. Nevertheless, a wide range of failure loads was observed over the HcB/Dem strains, from HcB/13's value of 6.1 ± 1.0 N to HcB/14's value of 10.4 ± 1.2 N. A similar pattern is seen for structural stiffness as well, with C3H having a value of 17.9 ± 3.7 N/mm and B10 having a value of 21.2 ± 12.9 N/mm. The most divergent strains are HcB/18 (17.2 ± 5.2 N/mm) and HcB/14 (40.8 ± 8.5 N/mm).

thumbnail image

Figure FIG. 3.. Biomechanical performance of the HcB/Dem strains. Each bar graph shows the average ±1 SD for C3H/DiSnA, C57BL/10ScSnA, and each of the HcB strains tested. Sample sizes range between 4 and 10 animals/strain.

Download figure to PowerPoint

It is useful to distinguish between structural and material properties in considering biomechanical testing data. Structural properties characterize the specimen as a whole, encompassing both its anatomy and its material. The structural properties are measured directly during the performance of the test. Material properties, in contrast, are calculated from the measured structural properties and the measured anatomical parameters. Failure stress is the material analogue to failure load and corrects for a specimen's CSA and diameter at the break point. In contrast to the data for failure load, the parental strains have widely divergent failure stresses, 173 ± 14 MPa for C3H and 87 ± 11 MPa for B10. B10's failure stress is the minimum for the strains studied and HcB/8's value of 192 ± 36 MPa is the maximum. Young's modulus is the material measure of stiffness; the parental strains' difference in failure stress is reflected by their moduli as well, with C3H having a modulus of 3570 ± 1160 MPa and B10 having a modulus of 1130 ± 700 MPa. B10 has the least stiff bone tissue observed in this study, whereas HcB/26, with a modulus of 5340 ± 880 MPa, has the stiffest bone tissue observed. These data show that both structural and material strength and stiffness vary widely among the HcB/Dem strains.

As a first step in characterizing the contribution of material properties to overall biomechanical performance, we performed ash analysis of femora from the same animals used for biomechanical testing. Ash percentage reflects the mineral content of the bone tissue and is the parameter most closely related to clinical measurements of BMD or bone mineral content (BMC). The results are summarized in Table 1 and Fig. 3 and show that C3H, with a value of 69.1 ± 1.1%, and B10, with a value of 64.2 ± 1.9%, differ markedly for this parameter. B10 has the minimum ash percentage of the strains studied and HcB/23, with a value of 70.2 ± 1.5%, has the maximum ash percentage. Thus, substantial variations exist in the mineral content of bones from the HcB/Dem strains.

Contributions of the measured parameters to bone strength

Data presented in the previous section show that marked variations in animal size, anatomy as reflected by diaphyseal diameter and I, and BMC exist among the HcB/Dem strains and each of these are expected to contribute to differences in biomechanical performance among them. As an initial step in understanding the contribution of each of these parameters to biomechanical performance, we performed a simple correlation analysis of each of the parameters in a pairwise fashion, after log-transforming data for parameters with skewed distributions. Results of this analysis are summarized in Table 2.

Table Table 2.. Correlations Among Bone Parameters
Thumbnail image of

We used these data to guide successive addition/elimination and elimination only stepwise regression analyses of failure load as a linear function of the other traits (Table 3). Such models express a single variable, in this case failure load, as a linear function of a subset of the other measured variables and an intercept given by the constant term of the equation. These other variables are included in the model only if they add to the model's explanatory power, as reflected by an increase in R2. A perfect model has an R2 = 1.0 and the R2 value gives the fraction of the variation in failure load that is attributable to variation in the other traits.

Table Table 3.. Stepwise Regression Models of Failure Load
Thumbnail image of

In the first model, we used all other measured parameters as possible independent variables. This model found that over 90% of the failure load could be accounted for as a function of log(failure stress), CSA, and I. The model's R2 = 0.906 and its equation is

  • equation image

This model, although accounting for over 90% of the failure stress, is limited by the fact that failure stress is a function of load and I.

In the second model, we allowed only variables that were measured directly rather than calculated from measured quantities. This model had poorer predictive ability, accounting for nearly 57% of failure load. Its R2 = 0.568 and its equation is given by

  • equation image

The majority of the explanatory power of this model came from the structural stiffness, which by itself explained 50% of the failure load. In both models, the same results were obtained by stepwise addition/elimination and elimination-only analyses.

Mapping of loci contributing to the phenotypes

Previously published genotypic data were used with the phenotypic analysis described previously to map loci contributing to each of the studied traits. The mapping data for selected chromosomes are summarized in Fig. 4. Mapping data for other traits (data not shown) reveal that LOD score graphs for all the diaphyseal diameters closely parallel each other. This also is true of CSA, I, failure stress, and modulus, which are calculated using the diaphyseal diameters. Because these parameters are codependent, only I is included in Fig. 4 along with the independently measured traits of failure load, structural stiffness, ash percentage, and body mass.

thumbnail image

Figure FIG. 4.. Linkage maps for five traits. Linkage maps for failure load (red diamonds), structural stiffness (green + signs), ash percentage (blue squares), I (magenta X's), and body mass (navy triangles) are shown for each autosome in which the maximum LOD for any trait exceeded 1.0. The X dimension measures position in morgans (1 M = 100 cM) from the most centromeric marker. No “tails” flanking the outermost markers are included in the maps. The Y dimension measures LOD score. Marker positions are indicated below the x axis.

Download figure to PowerPoint

We performed a 1000-iteration permutation test to estimate experiment-wide significance levels for each trait as summarized in Table 4. As shown in Fig. 4, none of the strain-averaged QTL peaks achieved experiment-wide statistical significance,(37, 38) regions on multiple chromosomes had LOD scores in excess of 1.7. An LOD ≥ 1 threshold is commonly used for exploratory genome scans of complex traits in a staged searching strategy.(37) However, given that we performed linkage mapping for five independently measured traits, it is appropriate to raise the exploratory threshold to the Bonferroni-corrected value of LOD ≥ 1.7.

Table Table 4.. Experimental-Wide Significance Levels
Thumbnail image of

On several chromosomes, LOD peaks for different traits map to the same genomic locations. Overlapping mapping assignments are consistent with the existence of pleiotropic (affecting multiple traits) QTLs at these locations. The degree of overlap among the mapping assignments is variable. Failure load and structural stiffness linkage maps are nearly parallel throughout the genome. A lesser degree of overlap is seen between failure load and I, and even less is observed between failure load and ash percentage. The linkage data suggest that there are distinct but overlapping sets of QTLs that contribute to each of the bone strength-related traits shown in the HcB/Dem system. This is well illustrated in comparing the maps of chromosomes 1 and 10 (Fig. 4). On chromosome 1, a peak centered at D1Mit10 includes failure load, structural stiffness, I, and body mass, but not ash percentage. Conversely, the peak on chromosome 10 centered on D10Mit3 includes failure load, structural stiffness, and ash percentage but not I or body mass.


  1. Top of page
  2. Abstract
  7. Acknowledgements

This report presents bone phenotypes for 24 of the 27 HcB/Dem RC strains. These strains span nearly a 2-fold range of failure loads and vary widely for multiple related traits. Many phenotype pairs, even excluding those that are trivially related, display significant correlations. A multivariate linear regression model including only traits that were measured directly accounts for more than half the difference in failure load among the strains. The phenotypic data were used to generate the QTL maps for five independently measured traits in the HcB/Dem strains shown in Fig. 4.

The linkage maps reveal that potential QTLs for distinct traits sometimes colocalize to a single position in the genome, which we interpret as the presence of a single gene with pleiotropic effects (affecting multiple traits) at each multitrait LOD peak. An alternative explanation for colocalization is that distinct but closely linked QTLs are responsible for each of the phenotypes. Although the pleiotropy and multiple linked gene models cannot be distinguished by our data, the pleiotropy model is both more economical and consistent with present understanding of bone strength. Our data reveal coincident LOD peaks for distinct subsets of the traits at several positions in the genome. There is prior evidence relating these traits to failure load.(26, 28, 39–44) The existence of pleiotropic QTLs would provide a biological basis for the observed clinical correlations between body mass and bone strength. The biological importance of these putative pleiotropic loci is reinforced by the fact that they are based on very different, independently measured phenotypes. Mechanistically, pleiotropy most likely reflects the many developmental and biochemical steps separating the products of the putative pleiotropic loci and the measured phenotypes. If one accepts the pleiotropy interpretation, it also is worth noting that colocalization of LOD peaks for multiple independent traits mitigates lack of statistical significance for linkage assignments. Permutation tests determine the frequency with which peaks of a given LOD threshold will occur anywhere in the genome by chance alone. Because there are many genomic locations where artifactual LOD peaks may occur, coincidence of peaks for multiple, independently measured traits can be interpreted more plausibly as representing biology rather than statistical accident. On this basis, we would predict that further experiments in this system are more likely to find significant QTLs on chromosomes 1, 10, and 11 than on chromosomes 2, 8, and 12.

In the linkage analysis, LOD scores may be increased because of epistasis between specific locus pairs. In previous work, Demant and colleagues showed large epistatic interactions between tumor susceptibility loci.(45, 46) The small number of genotypes studied here will tend to magnify the contributions of individual loci, because main effects cannot be distinguished from interactions. These limitations of the data presented here can best be addressed by performing additional crosses; these are in progress.

Genetic differences are believed to account for a substantial portion of the variability in the bone strength of both mice and humans. In mice, Beamer and colleagues have measured volumetric BMD and cortical thickness by quantitative computed tomography (CT) scanning in a panel of 11 inbred mouse strains, noting that C57BL/6J and C3H/HeJ are the most extreme strains in their sample for these parameters.(23) These strains are related closely to the progenitors of the HcB/Dem RC strains. These investigators have begun to map volumetric BMD in crosses between C57BL/6J and CAST/EiJ(47) and between C57BL/6J and C3H/HeJ.(48) In these crosses, a highly significant QTL centered at D1Mit15 falls in a region that may overlap with our LOD peak centered at D1Mit10. Two groups have studied the genetic basis of peak bone mass in the SAM/P6 senescence-accelerated, osteoporotic mouse. Shimizu and colleagues(49) used a midfemoral cortical thickness index as a measure of bone mass in a cross between SAMP6 and SAMP2, reporting a highly significant QTL on chromosome 11 between D11Mit90 and D11Mit59 that may overlap our peak on that chromosome. Benes et al. have performed linkage mapping of areal BMD in crosses of SAMP6 with SAMR1 and AKR/J.(50) These workers also found a significant QTL in the corresponding chromosome 11 region. Klein and colleagues analyzed areal (i.e., projected) BMD in the B × D recombinant inbred strains as measured by dual-energy X-ray absorptiometry.(51) This study, like ours, was exploratory in nature because of the limited number of genotypes examined. These related investigations have identified genomic regions that overlap partially with our results, providing independent evidence that LOD score peaks noted in the HcB/Dem system reflect true QTLs.

It is important to note two important differences between our experiments and the work performed by the other groups. First, although the phenotypes investigated are related, they are not the same and therefore may be under somewhat different genetic control. Second, the progenitors in each investigation are different. Consequently, in each system, a different set of loci and alleles segregates. These caveats are particularly important in relating the data presented here with other mapping studies. Colocalization of putative QTLs from different experiments, although providing evidence that a relevant gene is present, still requires caution if the phenotypic measures differ, as is true for chromosomes 1 and 11. Conversely, it is not surprising that some distinct loci contribute to bone strength in each experiment, given the diversity of traits and strains studied.

A human locus controlling bone mass (HBM) has been mapped to chromosome 11 q12-13(52) and there is evidence that this locus may account for variation in bone mass in a broader-based population as well.(53) The murine homologue to the human HBM locus is predicted to map to near the centromere of chromosome 19.

Thus far, mapping studies of bone properties have been heavily weighted toward radiographically determined mineral content as the phenotype. There are good reasons for this, most notably the existence of several well-characterized precise methods for measuring this parameter. However, as illustrated by the data presented here and consistent with clinical experience in humans, mineral content is only one among several factors contributing to overall bone strength and fracture risk.(3, 40 43 54) Our observations also suggest that material properties of bone that were not investigated contribute to differences of failure stress and Young's modulus as well as failure load. These might include crystal size and morphology, degree of crystal perfection, degree of substitution of carbonate in the mineral, degree or pattern of collagen cross-linking, and differences in noncollagenous proteins. In this study, we have exploited the genetic homogeneity of inbred mice to allow biomechanical data to be used as a phenotype in a preliminary mapping study. The HcB/Dem RC strains each contain a distinct complement of B10 alleles on a background of the C3H genome. Both progenitors are “wild type,” lacking mutations affecting skeletal structure, function, or development in an obvious fashion. Yet, the cumulative effects of allelic differences between these strains lead to quite dramatic differences in the cortical bone properties of adult mice.

Recently, increased attention has been focused on how anatomy affects bone strength in humans. Myers and colleagues(55) related load to fracture for cadaveric human forearms to BMD, BMC, CSA, and I. They found that both CSA and I were more predictive of biomechanical performance than either BMD or BMC measured by dual-energy X-ray absorptiometry, consistent with our phenotypic data. Several groups have suggested that a longer hip axis length (HAL) increases the risk of femoral neck fractures.(56–66) Bell and colleagues have reported cross-sectional data suggesting that loss of cortical bone mass in the anteroinferior-posterosuperior axis of the femoral neck increases the risk of hip fracture.(67) They hypothesize that formation of “giant” Haversian canals in the femoral neck contributes to the specific loss of cortical bone mass.(68) Duan and colleagues assessed the contributions of reduced bone size and reduced volumetric BMD to vertebral fracture risk.(69) These investigators found in a cross-sectional sample that both contributed to fracture risk and that reduced vertebral body size accounted for 16% of the areal BMD deficit of fracture patients relative to controls. This bone size effect accounts for the apparent discrepancy between the work of Beamer and colleagues, who find C57BL/6J to be a low BMD strain, and that of Klein and colleagues, who find C57BL/6J to be the high BMD progenitor in the B × D recombinant inbred system. The Beamer group follows volumetric BMD whereas the Klein group follows areal BMD. Like C57BL/10ScSnA, the progenitor of the HcB/Dem RC series, C57BL/6J is characterized by a large diaphyseal diameter,(70) resulting in a bone volume-dependent increase in areal BMD. As in different strains of mice, humans display ethnic differences in bone volume(10, 11, 61, 71–74) and bone size is a highly heritable trait.(5, 6-9, 12-15, 75) Areal BMD's better performance in predicting fracture risk compared with volumetric BMD thus is seen to be the result of its inherently compound nature—it includes a measure of bone mineralization and superimposes an anatomic factor that reflects I.(73, 76–81) Mapping QTLs that allow resolution of the anatomic and material aspects of bone strength therefore is a notable step toward unraveling the complexities of fracture risk.

Work to date has made it apparent that family relationships among genes have been conserved over evolution. One practical result of these findings is that genes' functions are similar in humans to their functions in model organisms. Moreover, we have learned that the organization of chromosome structure in these two species is highly conserved, with the preservation of so-called syntenies, or groups of physically linked genes that have remained together over the course of evolution. Only about 200 major chromosome rearrangements are thought to have occurred since the human and murine lineages diverged.(82–85) Therefore, a second practical consequence of the genome project's progress is that genetic mapping data from the mouse can be used to predict the locations of corresponding human genes. The existence of conserved synteny relationships will allow testing of the roles of genetic loci identified in the mouse in human populations.

The data presented previously are limited in several important ways. First, sample sizes are relatively small and are limited to 6-month-old female mice. Beamer and colleagues have shown that bone mass in mice does not remain constant throughout adulthood and that change in BMC follows different time courses in different strains.(23) Loss of bone mass during aging also has been established by the SAMP6 mouse, which develops osteoporosis as it ages.(86) Our data do not address any aspect of bone turnover over an individual's lifetime.

Second, there are important differences in bone metabolism between mice and humans. The mouse skeleton undergoes virtually no osteonal remodeling,(87) whereas such remodeling is a central feature of human bone. Although mice grow primarily during the first 4 months of life, their epiphyses remain open throughout their lives. The murine estrous cycle is quite dissimilar to human menses and mice do not undergo menopause in midlife. Each of these differences limits the applicability of the findings reported here to human bone properties.

Third, important limitations accompany the choice of the HcB/Dem system. Only 24 strains were examined, limiting the power of the linkage study to an exploratory level. Mapping was performed with strain-averaged data, obscuring intrastrain variability and weighting the individuals from the strains with the smallest sample sizes most heavily. Genotypic data for the HcB/Dem strains only include approximately 130 markers and their genotypes include some relatively large untyped segments.(19, 20) The RC breeding scheme aggravates the impact of gaps in the HcB/Dem strain distribution pattern. There are multiple opportunities for crossing over to occur during the inbreeding process, so that the extent of the genome spanned by each marker in RC lines is only approximately one-fourth that spanned by markers in a single-generation experiment.(21, 22, 88) There are likely to be additional small, unidentified differential segments between the two parental strains. Moreover, there may be segregating QTLs that contribute to the phenotypes examined that this experiment was unable to detect. Incorporation of additional markers to the HcB/Dem strain distribution pattern will allow this issue to be addressed.

Fourth, although biomechanical tests are more closely related to fracture risk than is BMC, the fractures generated in these tests are still artifactual with regard to fracture site and fracture mechanism. Three-point bending of the mouse humerus assesses cortical bone strength, because the site is virtually devoid of trabecular bone. The stresses that lead to human fragility fractures and hip fractures in particular probably differ from the fractures generated in our biomechanical tests in important ways.

Fifth, not all the phenotypes studied are of equivalent status. Failure load, structural stiffness, body mass, BMI, ash percentage, and the various diameters are measured directly. Failure stress, modulus, CSA, and I, in contrast, are calculated from the directly measured phenotypes. Although all measurements are subject to potential errors, only the calculated parameters are subject to compounded errors.

Sixth, the LOD scores of potential QTLs are systematically overestimated as discussed previously, primarily as a consequence of the small number of independent genotypes examined. This means that QTL mapping assignments based on either recombinant inbred or RC strain data must be confirmed in an independent breeding experiment. This limitation applies equally to the data presented here and those presented by Klein and coworkers.(51)

These limitations notwithstanding, the experiments reported here are a comprehensive analysis of the biomechanical properties of humeri and of their structural, material underpinnings and a preliminary investigation of their genetic basis in the HcB/Dem RC system. The virtue of combining QTL mapping with comprehensive phenotypic analysis is that insight is gained not only regarding the chromosomal locations of the genes contributing to bone strength, but also regarding the mechanisms by which strength is achieved.


  1. Top of page
  2. Abstract
  7. Acknowledgements

The authors gratefully acknowledge Dr. Peter Demant for generously providing the mice for these experiments. The authors thank Drs. Adele Boskey, Cory Brayton, Nancy Camacho, Elizabeth Myers, Eleftherios Paschalis, Marjolein van der Meulen, and Timothy Wright for many helpful discussions and instruction in the performance of various assays. In addition, they thank Rajarsi Gupta, Meredith Sobel, Jorge Oldan, Darlene Grillo, and Nolan James for technical assistance. This work was supported by National Institutes of Health Multipurpose Arthritis and Musculoskeletal Disease Center grant P50 AR3850, subproject 0028, a research grant from the New York Chapter of the Arthritis Foundation, a research grant from the American Federation for Aging Research, and a fellowship from the Children's Brittle Bone Foundation (all to R.D.B.) and a research grant from the Patricia Grossman Fund (R.S.B.).


  1. Top of page
  2. Abstract
  7. Acknowledgements
  • 1
    Kleerekoper M, Nelson DA 1997 Which bone density measurement? J Bone Miner Res 12:712714.
  • 2
    Eastell R 1998 Treatment of postmenopausal osteoporosis. N Engl J Med 338:736746.
  • 3
    Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM, Ensrud KE, Cauley J, Black D, Vogt TM 1995 Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med 332:767773.
  • 4
    Newton-John HF, Morgan DB 1968 Osteoporosis: Disease or senescence. Lancet 1:232233.
  • 5
    Lutz J 1986 Bone mineral, serum calcium, and dietary intakes of mother/daughter pairs. Am J Clin Nutr 44:99106.
  • 6
    Smith DM, Nance WE, Kang KW, Christian JC, Johnston CC Jr 1973 Genetic factors in determining bone mass. J Clin Invest 52:28002808.
  • 7
    Sowers MR, Burns TL, Wallace RB 1986 Familial resemblance of bone mass in adult women. Genet Epidemiol 3:8593.
  • 8
    Pokock NA, Eisman JA, Hopper JL, Yeates MG, Sambrook PN, Ebert S 1987 Genetic determinants of bone mass in adults. J Clin Invest 80:706710.
  • 9
    Dequeker J, Nijs J, Verstraeten A, Geusens P, Gevers G 1987 Genetic determinants of bone mineral content at the spine and radius: A twin study. Bone 8:207209.
  • 10
    Liel Y, Edwards J, Spicer KM, Gordon L, Bell NH 1988 The effects of race and body habitus on bone mineral density of the radius, hip, and spine in premenopausal women. J Clin Endocrinol Metab 66:12471250.
  • 11
    Pollitzer WS, Anderson JJB 1989 Ethnic and genetic differences in bone mass: A review with a hereditary vs. environmental perspective. Am J Clin Nutr 50:12441259.
  • 12
    Christian JC, Yu P-L, Slemenda CW, Johnston CC Jr 1989 Heritability of bone mass: A longitudinal study in aging male twins. Am J Hum Genet 44:429433.
  • 13
    Seeman E, Hopper JL, Bach LA, Cooper ME, Parkinson E, McKay J, Jerums G 1989 Reduced bone mass in daughters of women with osteoporosis. N Engl J Med 320:554558.
  • 14
    Lutz J, Tesar R 1990 Mother-daughter pairs: Spinal and femoral bone densities and dietary intakes. Am J Clin Nutr 52:872877.
  • 15
    Seeman E, Tsalamandris C, Formica C, Hopper JL, McKay J 1994 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 9:739743.
  • 16
    Demant P, Hart AA 1986 Recombinant congenic strains—a new tool for analyzing genetic traits determined by more than one gene. Immunogenetics 24:416422.
  • 17
    Moen CJ, Van Der Valk MA, Snoek M, Van Zutphen BF, von Deimling O, Hart AA, Demant P 1991 The recombinant congenic strains—a novel genetic tool applied to the study of colon tumor development in the mouse. Mamm Genome 1:217227.
  • 18
    Van Zutphen LF, Den Bieman M, Lankhorst A, Demant P 1991 Segregation of genes from donor strain during the production of recombinant congenic strains. Lab Anim 25:193197.
  • 19
    Groot PC, Moen CJ, Dietrich W, Stoye JP, Lander ES, Demant P 1992 The recombinant congenic strains for analysis of multigenic traits: Genetic composition. FASEB J 6:28262835.
  • 20
    Stassen AP, Groot PC, Eppig JT, Demant P 1996 Genetic composition of the recombinant congenic strains. Mamm Genome 7:5558.
  • 21
    Taylor BA 1978 Recombinant inbred strains: Use in gene mapping. In: MorseHC (ed.) Origins of Inbred Mice. Academic Press, New York, NY, USA, pp. 423438.
  • 22
    Bailey DW 1981 Recombinant inbred strains and bilineal congenic strains. In: FosterHL, SmallJD, FoxJG (eds.) The Mouse in Biomedical Research. Academic Press, New York, NY USA, pp. 223239.
  • 23
    Beamer W, Donahue L, Rosen C, Baylink D 1996 Genetic variability in adult bone density among inbred strains of mice. Bone 18:397403.
  • 24
    Donnelly R, Bockman R, Di Carlo E, Betts F, Boskey A 1993 The effect of gallium nitrate on healing of vitamin D- and phosphate-deficient rickets in the immature rat. Calcif Tissue Int 53:400410.
  • 25
    Camacho NP, Rimnac CM, Meyer RAJ, Doty S, Boskey AL 1995 Effect of abnormal mineralization on the mechanical behavior of X-linked hypophosphatemic mice femora. Bone 17:271278.
  • 26
    Turner CH, Burr DB 1993 Basic biomechanical measurements of bone: A tutorial. Bone 14:595608.
  • 27
    Simske SJ, Luttges MW, Wachtel H 1990 Age Dependent Development of Osteopenia in the Long Bones of Tail-Suspended Mice ISA, vol. 90-014, pp. 8794.
  • 28
    Ferretti JL, Spiaggi EP, Capozza R, Cointry G, Zanchetta JR 1992 Interrelationships between geometric and mechanical properties of long bones from three rodent species with very different biomass: Phylogenetic implications. J Bone Miner Res 7:S433S435.
  • 29
    Basten CJ, Weir BS, Zeng Z-B 1994 Zmap—a QTL cartographer. In: Smith C, Gavora JS, Benkel B, Chesnais J, Fairfull W, Gibson JP, Kennedy BW, Burnside EB (eds.) Fifth World Conference on Genetics Applied to Livestock Production: Computing Strategies and Software, vol. 22, Guelph, Ontario, Canada, pp. 6566.
  • 30
    Basten CJ, Weir BS, Zeng Z-B 1999 QTL cartographer, 1.13 ed. North Carolina State University, Raleigh, NC.
  • 31
    Kanji GK 1993 100 Statistical Tests. SAGE Publications, London, UK, pp. 4244.
  • 32
    Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE, Newburg L 1987 MAPMAKER: An interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174181.
  • 33
    Lander ES, Botstein D 1989 Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185199.
  • 34
    Churchill GA, Doerge RW 1994 Empirical threshold values for quantitative trait mapping. Genetics 138:963971.
  • 35
    Doerge RW, Churchill GA 1996 Permutation tests for multiple loci affecting a quantitative character. Genetics 142:285294.
  • 36
    Williams T, Kelley C, Lang J, Kotz D, Campbell J, Elber G 1986–1993 Gnuplot, 3.0 ed.
  • 37
    Lander E, Schork N 1994 Genetic dissection of complex traits. Science 265:20382048.
  • 38
    Lander E, Kruglyak L 1995 Genetic dissection of complex traits: Guidelines for interpreting and reporting linkage results. Nat Genet 11:241247.
  • 39
    Hayes W 1986 Basic biomechanics of the skeleton. In: UlthoffHK, JaworskiZFG (eds.) Current Concepts of Bone Fragility. Springer-Verlag, Heidelberg, Germany, pp. 318.
  • 40
    Cummings SR 1996 Treatable and untreatable risk factors for hip fracture. Bone 18:165S167S.
  • 41
    Nguyen TV, Eisman JA, Kelly PJ, Sambrook PN 1996 Risk factors for osteoporotic fractures in elderly men. Am J Epidemiol 144:255263.
  • 42
    Ross PD, Davis JW, Epstein RS, Wasnich RD 1991 Pre-existing fractures and bone mass predict vertebral fracture incidence in women. Ann Intern Med 114:919923.
  • 43
    Ross PD, Davis JW, Wasnich RD 1993 Bone mass and beyond: Risk factors for fractures. Calcif Tissue Int 53(Suppl 1):S134S138.
  • 44
    Wasnich R 1993 Bone mass measurement: Prediction of risk. Am J Med 95:6S10S.
  • 45
    Fijneman RJ, de Vries SS, Jansen RC, Demant P 1996 Complex interactions of new quantitative trait loci, Sluc1, Sluc2, Sluc3, and Sluc4, that influence the susceptibility to lung cancer in the mouse. Nat Genet 14:465467.
  • 46
    Van Wezel T, Stassen AP, Moen CJ, Hart AA, Van Der Valk MA, Demant P 1996 Gene interaction and single gene effects in colon tumour susceptibility in mice. Nat Genet 14:468470.
  • 47
    Beamer WG, Shultz KL, Churchill GA, Frankel WN, Baylink DJ, Rosen CJ, Donahue LR 1999 Quantitative trait loci for bone density in C57BL/6J and CAST/EiJ inbred mice. Mamm Genome 10:10431049.
  • 48
    Beamer WG, Rosen CJ, Donahue LR, Frankel WN, Churchill GA, Shultz KL, Baylink DJ, Pettis JL 1998 Location of genes regulating volumetric bone mineral density in C57BL/6J (low) and C3H/HeJ (high) inbred strains of mice, ASBMR-IBMS Second Joint Meeting, Bone, vol. 23, San Francisco, CA, USA, p. S162.
  • 49
    Shimizu M, Higuchi K, Bennett B, Xia C, Tsuboyama T, Kasai S, Chiba T, Fujisawa H, Kogishi K, Kitado H, Kimoto M, Takeda N, Matsushita M, Okumura H, Serikawa T, Nakamura T, Johnson TE, Hosokawa M 1999 Identification of peak bone mass QTL in a spontaneously osteoporotic mouse strain. Mamm Genome 10:8187.
  • 50
    Benes H, Weinstein RS, Zheng W, Thaden JJ, Jilka RL, Manolagas SC, Shmookler Reis RJ 2000 Chromosomal mapping of osteopenia-associated quantitative trait loci using closely related mouse strains. J Bone Miner Res 15:626633.
  • 51
    Klein RF, Mitchell SR, Phillips TJ, Belknap JK, Orwoll ES 1998 Quantitative trait loci affecting peak bone mineral density in mice. J Bone Miner Res 13:16481656.
  • 52
    Johnson ML, Gong G, Kimberling W, Recker SM, Kimmel DB, Recker RB 1997 Linkage of a gene causing high bone mass to human chromosome 11 (11q12-13). Am J Hum Genet 60:13261332.
  • 53
    Koller DL, Rodriguez LA, Christian JC, Slemenda CW, Econs MJ, Hui SL, Morin P, Conneally PM, Joslyn G, Curran ME, Peacock M, Johnston CC, Foroud T 1998 Linkage of a QTL contributing to normal variation in bone mineral density to chromosome 11q12-13. J Bone Miner Res 13:19031908.
  • 54
    Wasnich RD, Davis JW, Ross PD 1994 Spine fracture risk is predicted by non-spine fractures. Osteoporos Int 4:15.
  • 55
    Myers ER, Hecker AT, Rooks DS, Hipp JA, Hayes WC 1993 Geometric variables from DXA of the radius predict forearm fracture load in vitro. Calcif Tissue Int 52:199204.
  • 56
    Boonen S, Koutri R, Dequeker J, Aerssens J, Lowet G, Nijs J, Verbeke G, Lesaffre E, Geusens P 1995 Measurement of femoral geometry in type I and type II osteoporosis: Differences in hip axis length consistent with heterogeneity in the pathogenesis of osteoporotic fractures. J Bone Miner Res 10:19081912.
  • 57
    Faulkner KG, Cummings SR, Black D, Palermo L, Gluer CC, Genant HK 1993 Simple measurement of femoral geometry predicts hip fracture: The study of osteoporotic fractures. J Bone Miner Res 8:12111217.
  • 58
    Faulkner KG, McClung M, Cummings SR 1994 Automated evaluation of hip axis length for predicting hip fracture. J Bone Miner Res 9:10651070.
  • 59
    Flicker L, Faulkner KG, Hopper JL, Green RM, Kaymacki B, Nowson CA, Young D, Wark JD 1996 Determinants of hip axis length in women aged 10-89 years: A twin study. Bone 18:4145.
  • 60
    Gluer CC, Cummings SR, Pressman A, Li J, Gluer K, Faulkner KG, Grampp S, Genant HK 1994 Prediction of hip fractures from pelvic radiographs: The study of osteoporotic fractures. The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 9:671677.
  • 61
    Mikhail MB, Vaswani AN, Aloia JF 1996 Racial differences in femoral dimensions and their relation to hip fracture. Osteoporos Int 6:2224.
  • 62
    Nakamura T, Turner CH, Yoshikawa T, Slemenda CW, Peacock M, Burr DB, Mizuno Y, Orimo H, Ouchi Y, Johnston CJ 1994 Do variations in hip geometry explain differences in hip fracture risk between Japanese and white Americans? J Bone Miner Res 9:10711076.
  • 63
    O'Neill TW, Grazio S, Spector TD, Silman AJ 1996 Geometric measurements of the proximal femur in UK women: Secular increase between the late 1950s and early 1990s. Osteoporos Int 6:136140.
  • 64
    Reid IR, Chin K, Evans MC, Jones JG 1994 Relation between increase in length of hip axis in older women between 1950s and 1990s and increase in age specific rates of hip fracture. BMJ 309:508509.
  • 65
    Reid IR, Chin K, Evans MC, Cundy T 1996 Longer femoral necks in the young: A predictor of further increases in hip fracture incidence? N Z Med J 109:234235.
  • 66
    Van Der Meulen MC, Ashford MW Jr, Kiratli BJ, Bachrach LK, Carter DR 1996 Determinants of femoral geometry and structure during adolescent growth. J Orthop Res 14:2229.
  • 67
    Bell KL, Loveridge N, Power J, Garrahan N, Stanton M, Lunt M, Meggitt BF, Reeve J 1999 Structure of the femoral neck in hip fracture: Cortical bone loss in the inferoanterior to superoposterior axis. J Bone Miner Res 14:1111119.
  • 68
    Bell KL, Loveridge N, Power J, Garrahan N, Meggitt BF, Reeve J 1999 Regional differences in cortical porosity in the fractured femoral neck. Bone 24:5764.
  • 69
    Duan Y, Parfitt A, Seeman E 1999 Vertebral bone mass, size, and volumetric density in women with spinal fractures. J Bone Miner Res 14:17961802.
  • 70
    Akhter MP, Cullen DM, Pedersen EA, Kimmel DB, Recker RR 1998 Bone response to in vivo mechanical loading in two breeds of mice. Calcif Tissue Int 63:442449.
  • 71
    Bhudhikanok GS, Wang MC, Eckert K, Matkin C, Marcus R, Bachrach LK 1996 Differences in bone mineral in young Asian and Caucasian Americans may reflect differences in bone size. J Bone Miner Res 11:15451556.
  • 72
    Cummings SR, Cauley JA, Palermo L, Ross PD, Wasnich RD, Black D, Faulkner KG 1994 Racial differences in hip axis lengths might explain racial differences in rates of hip fracture. Study of Osteoporotic Fractures Research Group. Osteoporos Int 4:226229.
  • 73
    Seeman E 1998 Growth in bone mass and size—are racial and gender differences in bone mineral density more apparent than real? J Clin Endocrinol Metab 83:14141419.
  • 74
    Seeman E 1999 The structural basis of bone fragility in men. Bone 25:143147.
  • 75
    Niu T, Chen C, Cordell H, Yang J, Wang B, Wang Z, Fang Z, Schork NJ, Rosen CJ, Xu X 1999 A genome-wide scan for loci linked to forearm bone mineral density. Hum Genet 104:226233.
  • 76
    Katzman DK, Bachrach LK, Carter DR, Marcus R 1991 Clinical and anthropometric correlates of bone mineral acquisition in healthy adolescent girls. J Clin Endocrinol Metab 73:13321339.
  • 77
    Bonjour JP, Theintz G, Law F, Slosman D, Rizzoli R 1994 Peak bone mass. Osteoporos Int 4(Suppl 1):713.
  • 78
    Bass S, Delmas PD, Pearce G, Hendrich E, Tabensky A, Seeman E 1999 The differing tempo of growth in bone size, mass, and density in girls is region-specific. J Clin Invest 104:795804.
  • 79
    Zamberlan N, Radetti G, Paganini C, Gatti D, Rossini M, Braga V, Adami S 1996 Evaluation of cortical thickness and bone density by roentgen microdensitometry in growing males and females. Eur J Pediatr 155:377382.
  • 80
    Gilsanz V, Boechat MI, Gilsanz R, Loro ML, Roe TF, Goodman WG 1994 Gender differences in vertebral sizes in adults: Biomechanical implications. Radiology 190:678682.
  • 81
    Gilsanz V, Loro ML, Roe TF, Sayre J, Gilsanz R, Schulz EE 1995 Vertebral size in elderly women with osteoporosis. Mechanical implications and relationship to fractures. J Clin Invest 95:23322337.
  • 82
    Nadeau JH 1989 Maps of linkage and synteny homologies between mouse and man. Trends Genet 5:8286.
  • 83
    Eppig JT, Nadeau JH 1995 Comparative maps: The mammalian jigsaw puzzle. Curr Opin Genet Dev 5:709716.
  • 84
    DeBry RW, Seldin MF 1996 Human/mouse homology relationships. Genomics 33:337351.
  • 85
    Seldin MF 1997 Human/Mouse Homology Relationships, National Center for Biotechnology Information, Bethesda, MD, and Davis, CA, USA.
  • 86
    Takeda T, Hosokawa M, Higuchi K 1991 Senescence-accelerated mouse (SAM): A novel murine model of accelerated senescence. J Am Geriatr Soc 39:911919.
  • 87
    Frost HM, Jee WSS 1992 On the rat model of human osteopenias and osteoporoses. Bone Miner 18:227236.
  • 88
    Blank RD, Campbell GR, D'Eustachio P 1986 Possible derivation of the laboratory mouse genome from multiple wild Mus species. Genetics 114:12571269.