Identification of Novel Genetic Loci for Bone Size and Mechanosensitivity in an ENU Mutant Exhibiting Decreased Bone Size


  • Apurva K Srivastava,

    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    2. Department of Medicine, Loma Linda University, Loma Linda, California, USA
    Search for more papers by this author
  • Sanjay Kapur,

    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    Search for more papers by this author
  • Suburaman Mohan PhD,

    Corresponding author
    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    2. Department of Medicine, Loma Linda University, Loma Linda, California, USA
    3. Department of Biochemistry, Loma Linda University, Loma Linda, California, USA
    • Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, CA 92357, USA
    Search for more papers by this author
  • Hongrun Yu,

    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    Search for more papers by this author
  • Sonia Kapur,

    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    Search for more papers by this author
  • Jon Wergedal,

    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    2. Department of Medicine, Loma Linda University, Loma Linda, California, USA
    Search for more papers by this author
  • David J Baylink

    1. Musculoskeletal Disease Center, Jerry L. Pettis Veterans Administration Medical Center, Loma Linda, California, USA
    2. Department of Medicine, Loma Linda University, Loma Linda, California, USA
    3. Department of Biochemistry, Loma Linda University, Loma Linda, California, USA
    Search for more papers by this author

  • The authors have no conflict of interest.


Using a dominant ENU mutagenesis screen in C57BL/6J (B6) mice to reveal gene function, we identified a mutant, 917M, with a reduced bone size phenotype, which is expressed only in males. We show that mutation results in osteoblasts with reduced proliferation, increased apoptosis, and an impaired response to in vitro mechanical load. The mutation is mapped to a novel locus (LOD score of 7.9 at 10.5 cM) on chromosome 4.

Introduction: Using a dominant ENU mutagenesis screen in C57BL/6J (B6) mice to reveal gene function, we identified a mutant, 917M, with a reduced bone size phenotype, which is expressed only in males. In this report, we show the chromosomal location of this mutation using linkage analysis and cellular characterization of the mutant phenotype.

Materials and Methods: The mutant mouse was bred to wildtype B6 to produce progeny for characterization of the bone size phenotype. Periosteal osteoblasts isolated from the tibia and femur of mutant and wildtype mice were studied for proliferation, differentiation, and apoptosis potential. To determine the chromosomal location of the mutation, a low-resolution linkage map was established by completing a genome-wide scan in B6C3H F2 male mice generated from intercross breeding of mutant mice.

Results and Conclusions: Mutant progeny (16 weeks old) displayed a total body bone area that was 10-13% lower and a periosteal circumference that was 5-8% lower at the femur and tibia midshaft compared with wildtype B6 mice. Periosteal osteoblasts from mutant mice showed 17-27% reduced cell proliferation and 23% increased apoptosis compared with wildtype controls. In addition, osteoblasts from mutant mice showed an impaired response to shear stress-induced proliferation rate, an in vitro model for mechanical loading. Interval mapping in B6C3H F2 males (n = 69) indicated two major loci affecting bone size on chromosome 1 at 45 cM (LOD 4.9) and chromosome 4 at 10.5 cM (LOD 7.9, genome-wide p < 0.01). Interval mapping using body weight as covariate revealed only one significant interval at chromosome 4 (LOD 6.8). Alleles of the chromosome 4 interval inherited from the B6 mutant strain contributed to a significantly lower bone size than those inherited from C3H. A pairwise interaction analysis showed evidence for a significant interaction between loci on chromosome 1 with the chromosome 4 quantitative trait loci. The 917M locus on chromosome 4 seems to be novel because it does not correspond with those loci previously associated with bone size on chromosome 4 in B6 and C3H/HeJ mice or other crosses.


The use of chemical (ethyl N-nitrosourea [ENU]) mutagenesis has provided an opportunity to develop and expand the repertoire of mutants affecting several disease models(1–5) for gene function studies. Phenotype-driven approaches have gained substantial attention in recent years because of their focus on mutagenesis procedures that emphasize recovery of new phenotypes relevant to clinical diseases without an assumption about the nature of the underlying genes or biological pathways involved. A major advantage of the large-scale ENU mutagenesis method is that each locus is randomly mutated, and mutations are expected to result in nonlethal phenotypes and altered protein function instead of complete abrogation, thereby increasing the opportunity to identify mutants that are informative for that locus. This approach has recently produced variants that mimic phenotypes commonly seen in humans in the clinical setting.(6–8)

Earlier, we established an ENU screen in mice for mutations affecting musculoskeletal and growth-related phenotypes(9) in an effort to improve our understanding of the genetic basis of osteoporotic disease. Genetic factors contributing to osteoporosis involve material properties such as BMC and structural components such as size, shape, and 3D structure. Therefore, our ENU screen consisted of an assessment of screens for BMD, BMC, bone area, radiographs, and serum biochemistry in 10- and 16-week-old mice. Inheritance testing of mice with abnormal phenotypes has confirmed the presence of several robustly inherited mutant phenotypes, including those affecting body weight, BMD, bone size, and bone markers. One of the ENU mutants (named 917M) confirmed in the ENU screen had a reduced bone size phenotype identified in a dominant screen in C57BL/6J (B6) mice. Three of the main parameters that assess bone size, specifically total body bone area, total body BMC, and bone perimeter (periosteal circumference) at the midshaft tibia and femur, were all significantly lower in affected mice, implying that the affected genes might play a role in bone size regulation. In addition, the differences in bone size were expressed only in the male progeny from the 917M mice.(10) The main aims of this study were to further characterize affected cell types and to identify the mutant locus affecting decreased bone size in 917M mouse.

Our efforts on the cell type characterization was primarily focused on periosteal cells from long bones for the following reasons: (1) bone size is dependent on periosteal expansion mediated by increased bone formation and/or by decreased bone resorption(11,12); and (2) important factors that regulate bone size such as sex hormones and exercise influence periosteal expansion by increasing periosteal bone formation and/or by decreasing periosteal bone resorption.(13,14)


Generation of mutant progeny

Mice were maintained and used in accordance with protocols established by the Institutional Animal Care and Use Committee of this facility. Generation and dominant screening of progeny from ENU-injected B6 male mice has been described previously.(9) The 917M male mouse was identified in the dominant screen with a low total body bone area and BMC (22% low compared with wildtype [WT] control) phenotype measured by DXA (PIXImus; Lunar Corp., Madison, WI, USA). The 917M male was first bred with two to three WT B6 female mice to produce ∼20 first-generation progeny. We determined the tibia midshaft bone area in these progeny by in vivo pQCT, which showed ∼8% lower periosteal circumference in male progeny compared with age- and sex-matched WT control mice. Because animals were not genotyped at this stage, phenotype distribution was the only means for differentiating the mutants from their unaffected littermates. To avoid potential breeding of an unaffected progeny because of classification errors in distinguishing mutants, we bred only extreme scoring mice with WT female B6 mice for generating second and third generations of progeny and excluded those mice that were on the borderline of our criteria to define mutants as described earlier.(9)

Screening of mutant progeny

Bone size was measured using DXA, which reports bone size in area measurements per units of centimeter squared. In addition, bone size was also measured at the midshaft tibia using pQCT as described earlier.(9,15) Body weight was used to normalize all size traits in the mutant progenies as described earlier.(9) Baseline data on WT B6 mice were collected from ∼50-80 male and female control mice, DXA and pQCT were performed when animals were 10 and 16 weeks of age, respectively (age varied by ±2 days).

Isolation of periosteal osteoblasts from long bones of mutant mice

The periosteal osteoblasts were isolated from the femur and tibias of WT and 917M mutant B6 male and female mice and were propagated in culture. The 10-week-old mice were killed using carbon dioxide, femurs and tibias were dissected out, and soft tissue was removed with minimum scraping of the bones. These bones were used for isolation of periosteal cells by collagenase (1 mg/ml in DMEM; Sigma Chemicals, Perth, Australia) digestion for 90 minutes at 37°C. Cells were plated at a density of 100,000 cells/well in 6-well tissue culture plates and grown to confluence in 10% FBS/DMEM/antibiotics. Periosteal osteoblasts at passages 2-3 were used to study cell proliferation, differentiation, and apoptosis. Alkaline phosphatase (ALP) staining was used to confirm the presence of osteoblasts. Cells isolated in identical manner from age- and sex-matched WT B6 mice were used as controls.

In vitro proliferation, differentiation, and apoptosis of periosteal osteoblasts from mutant mice

Basal cell proliferation was studied by two assays: (1) CyQuant-GR cell proliferation assay kit (Molecular Probes, Eugene, OR, USA) and (2) [3H]thymidine incorporation into cell DNA. Briefly, osteoblast cells were incubated for 24 h in serum-free media, and [3H]thymidine (1.5 μCi/ml) was added during the final 6 h of the incubation. Cell differentiation was measured by observing changes in the specific activity of ALP using paranitrophenylphosphate as substrate.(16) Caspase activities were measured by Homogeneous Caspase Assay kit (Roche Diagnostics, Indianapolis, IN, USA) to determine the rate of apoptosis. Briefly, the assay was based on the ability of activated caspases to cleave a fluorogenic peptide containing a caspase-specific cleavage site, consequently releasing the fluorogenic molecule rhodamine. The fluorogenic tetrapeptide used in this assay is a substrate for caspases 2, 3, 6, 7, 8, 9, and 10. Apoptosis was also analyzed by flow cytometry using Annexin-V conjugated to FITC-labeled dUTP to label apoptotic cells and a FACScan flow cytometer (Becton Dickinson, San Jose, CA, USA) to identify fluorescent-labeled cells. The percentage of cells that underwent apoptosis was determined after electronic subtraction of signal caused by background fluorescence, which was determined using cells incubated with FITC-labeled dUTP, but without terminal deoxynucleotidyl transferase. A minimum of 5000 cells was analyzed.

Fluid flow shear stress as an in vitro model of mechanical loading

Periosteal osteoblasts isolated from WT B6 mice and ENU mutant male and female mice were plated on glass slides (75 × 38 mm) at 5 × 104 cells/slide in DMEM containing 10% bovine calf serum. After the cultures became ∼80% confluent, the cells were serum-deprived for 24 h and subjected to shear stress as previously described.(16) Briefly, slides containing osteoblast cells were placed in a chamber inside the Cytodyne flow loop apparatus (San Diego, CA, USA), which was under constant hydrostatic pressure, exposing the cells to the steady laminar fluid flow and a well-defined fluid shear stress of 20 dynes/cm2 for 30 minutes. The system was always maintained at 37°C, and the medium was continuously bubbled with 5% CO2-95% air. The static controls were performed on cells grown in identical conditions but were not exposed to the shear stress.

Genetic analysis

Genomic DNA was isolated from tail clips using mouse DNAeasy kits (Qiagen, Valencia, CA, USA). DNA samples were quantified, and the quality was determined by measuring their absorbance at 260 and 280 nm. A genome-wide genotyping scan using 60 microsatellite markers (Invitrogen, Carlsbad, CA, USA) was undertaken using the ABI Linkage Mapping Set 2. PCR reactions and running conditions allowed from five to six microsatellite markers to be multiplexed in a single electrophoretic lane. The pooled products were analyzed for fragment size on an ABI Model 3100 DNA Analyzer, and Genescan software was used to size alleles (Applied Biosystems, Foster City, CA, USA). Allele calls and edits were done using Genotyper software (Applied Biosystems) and exported as tab-delimited tables. Preliminary analysis involved using four markers on chromosomes 1, 2, and 4 and three markers for each of the rest of the autosomes. Subsequent analysis involved 10-12 additional markers for chromosomes 1 and 4.

Statistical analyses

The total body bone areas obtained from DXA were normalized for body weight as described earlier.(10) The z scores were calculated using following formula: z score = (bone area of progeny − mean bone area of control male mice)/SD of bone area of control male mice (n = 40-80). Results from cell culture experiments are shown as mean ± SE. The statistical significance of the differences between groups was determined by two-way ANOVA, using sex and phenotype as variables. The interaction between sex and phenotype variable was calculated to indicate that males are more affected than females. A p value of <0.05 was considered for significant interactions.

Genotype data were initially analyzed using a MAPQTL (5.0) program (Kyazma, Wageningen, The Netherlands). MAPQTL interval mapping was used for quantitative trait loci (QTL) mapping, and the LOD score significance thresholds were calculated using 1000 permutation tests.(17) Because body weight is a strong predictor of bone size traits, we used multivariate analysis with body weight as the covariant to perform interval mapping using the Pseudomarker (obtained from MAINSCAN program written for the MATLAB (Mathworks, Natick, MA, USA) programming environment. This allows us to delineate the bone size trait from growth-related QTLs. To study genome-wide interactions between QTLs, we used the Pseudomarker PAIRSCAN algorithm.(17) This program analyzes not only the phenotypic effect of each marker or marker interval taken singly (MAINSCAN) but also the phenotypic effects of pairs of markers or intervals taken jointly (PAIRSCAN) for their effects on the trait. The PAIRSCAN allows a genome-wide search for epistasis. For PAIRSCAN, we tested the combined (or full model) effects on the trait of a marker pair, which reflects the main effects of both markers plus their interaction. The threshold for genome-wide error was set at 5%, which was estimated by a 500 permutation test carried out on the F2 data. When the combined effect of the marker pair was significant, the interaction was tested at a significance level of p < 0.05 as described earlier.(17)


Bone size phenotype in 917M mice

We generated about 120 male and 120 female progeny from the original affected 917M male identified in the dominant screening. The bone size phenotypes, which included total body bone area and bone area at the tibia midshaft, were determined at 10 and 16 weeks of age in all inherited progeny. Data described below for distinguishing mutant progeny were taken from 16-week-old mice because we observed a lower variation and a higher difference between 917M and WT mice at this age. We used total body bone area (cm2) phenotype and −2.0 SD units as the cut-off to classify a mutant from the unaffected littermates. Among males, ∼42% of the mice were classified as affected 917M (with reduced bone size) and 58% were similar to WT, whereas females were largely unaffected (∼5% showed the low total body bone area phenotype as shown in Fig. 1). Both male and female progeny from 917M showed normal distribution for the low bone area trait indicating that about 2.5% of the females could fall below the 2 SD cut-off range. Based on these data, the phenotype was mainly expressed in males, and the observed ratio of mutants (∼0.8:1) in males was close to the 1:1 inheritance ratio expected for dominant mendelian traits. The bone size trait was not sex-linked (X-linked) based on the inheritance of affected males from WT females. The total bone area expressed as z scores is shown in Fig. 1. The mean z scores for bone area and body weight adjusted bone area for affected 917M mice were −3.3 ± 0.8 and −3.1 ± 1.0, respectively. The distribution of phenotype in affected, nonaffected, and WT male mice is shown in Fig. 2. The total bone area in affected 917M male progeny was 10-13% (p < 0.001) lower compared with unaffected 917M littermates or WT B6 males. The body weight-adjusted area was about 9-10% (p < 0.001) lower in the affected 917M mice compared with unaffected littermates or WT B6 males. The average total bone area of affected 917M male mice (8.80 ± 0.35 cm2) was significantly (p < 0.01) lower than the female littermates (9.25 ± 0.58 cm2), which was similar to WT B6 females (9.36 ± 0.45 cm2, p = not significant versus female littermates). The mean total body BMD (DXA) in the 917M mice (0.0483 ± 0.0032 g/cm2) was similar to WT B6 males (0.0488 ± 0.0026 g/m2). The periosteal circumference and other bone size parameters measured in vivo at the midshaft tibia (mean of three slices at 1.0 mm apart) are shown in Table 1. Figure 3 shows bone size data on excised femurs and tibias from 917M (n = 3) and WT control (n = 5) mice measured by pQCT at nine slices covering entire length of the bone. The femur or tibia length was not different in 917M mice compared with WT B6 control mice. The averaged total area was 10-15% lower, periosteal circumference was 4-8% lower (Fig. 3), and endosteal circumference was 6-11% lower in both femur and tibia compared with WT mice.

Table Table 1.. Differences in Bone Size Parameters at Tibia Midshaft in Affected 917M Progeny Compared With Nonaffected Littermates
original image
Figure FIG. 1..

Z scores calculated for total body bone area measured by DXA in 16-week-old progeny from 917M mice. The average z score of all affected 917M male mice was −3.45 and that of nonaffected male and female littermates was −0.69 and −0.55, respectively (z scores indicates differences in total body bone area in terms of SD units from control mice, n = 60). 917M AM, 917M affected male; 917M UM, 917M unaffected male littermates; 917M FM, 917M female littermates; WT, wildtype.

Figure FIG. 2..

Frequency histogram of total bone area (measured by DXA) phenotype measured in 16-week-old 917M affected mice (917M AM; n = 52), 917M unaffected littermates (917M UM; n = 72), 917M female littermates (917M FM; n = 122), and wildtype B6 males (WT B6; n = 59). The mean total bone area of the affected 917M AM progeny (8.80 cm2) was 11.7% lower (p < 0.0001) compared with 917M UM mice (9.96 cm2) and 13.4% lower compared with WT mice (10.16 cm2). Female littermates showed normal distribution and were largely unaffected by the mutation.

Figure FIG. 3..

Ex vivo bone area measured by pQCT in tibias (left) and femurs (right) of 16-week-old 917M male mice. Nine slices covering the entire length of the bone (slices 1 and 9 were excluded because of large variation) were scanned. Bone areas of affected 917M male mice (•) were consistently lower over the entire length of tibia and femur as shown with age- and sex-matched WT mice (○). *p < 0.05 vs. WT mice.

In vitro cell proliferation and differentiation potential of periosteal osteoblasts from 917M mice

The basal proliferation potential of 917M osteoblasts analyzed by CyQuant assay showed a 17% decrease (p < 0.05; 2608 ± 370 versus 3138 ± 371 fluorescence units for controls) in proliferation in 917M mice compared with WT. Similarly, [3H]thymidine incorporation assay showed a 27% decrease in proliferation rate in 917M males, as shown in Fig. 4. The two-way ANOVA revealed that osteoblasts from affected 917M mice had significantly reduced proliferation (p < 0.001 for interaction), whereas proliferation rate was not altered in cells from female littermates (Fig. 4) of 917M mice. No significant differences in the specific activity of ALP were observed between cells from WT (6.6 ± 0.9 mU/mg protein) and 917M mice (6.8 ± 0.8 mU/mg protein), indicating that 917M and wildtype osteoblasts had the same differentiation potential.

Figure FIG. 4..

Basal proliferation rate of periosteal osteoblasts isolated from femurs and tibias of 10-week-old B6 wildtype and 917M male (n = 3) and female mice (n = 3). Equal number of cells was seeded at a density of 2000 cells/well in a 96-well plate and grown in the presence of 10% FCS for 48 h. The differences between groups were analyzed by two-way ANOVA (overall p < 0.0001). *p < 0.05 for interaction between differences in proliferation and sex.#p > 0.05.

Increased apoptosis of periosteal osteoblasts from 917M mice

Caspase activity in periosteal osteoblasts from 917M affected male mice showed a 23% increase, whereas cells from female 917M littermates did not show significant differences in the caspase activity compared with those from WT (Fig. 5), as shown by two-way ANOVA with a sex-phenotype interaction p value of 0.0431. This indicates a caspase-dependent increase in apoptosis of osteoblast cells from 917M. FACS of periosteal osteoblasts from 917M mice indicated a higher (18.7%) number of apoptotic cells from 917M male mice compared with those from WT (13.3%; details not shown). There were no significant differences in total apoptosis in cells isolated from female littermates of 917M mice in FACS analysis.

Figure FIG. 5..

Caspase activity in periosteal osteoblasts isolated from 917M male and wildtype mice were grown under serum-depleted conditions for 48 h. WT F, wildtype females; 917M FM, female littermates of 917M; WT M, wildtype male mice; 917M AM, affected 917M male mice. *p < 0.05 for interaction between differences in proliferation and sex calculated by two-way ANOVA.#p > 0.05.

Reduced mechanosensitivity of periosteal osteoblasts from 917M mice

The application of a 30-minute steady flow shear stress at 20 dynes/cm2 on WT osteoblasts consistently caused a 60-80% (p < 0.01) increase in the [3H]thymidine incorporation compared with the control cells not subjected to shear stress (Fig. 6). In contrast, the shear stress-stimulated increase in proliferation was only 5% (of control) in cells from the 917M mice. The two-way ANOVA revealed that, in control mice, sex has no significant difference in response to stress (p = 0.3345), whereas in 917M littermates, the affected 917M mice had significantly reduced response to shear stress (p < 0.0001 for interaction).

Figure FIG. 6..

Reduced mechanosensitivity of osteoblasts isolated from 917M mice measured by [3H]thymidine incorporation in periosteal osteoblasts from normal and 917M male and female mice subjected to 30-minute steady fluid flow shear strain. The two-way ANOVA shows that, in control mice, sex has no significant differences in response to stress (p = 0.3345), whereas in 917M littermates, the affected 917M mice have significantly reduced response to shear stress (p < 0.0001 for interaction). *p < 0.001 vs. control;#p > 0.05 vs. control.

Linkage analyses for bone size traits in 917M B6C3H F2 mice

A total of 129 B6C3H F2 mice were generated from 3 917M male mice. Male and female mice were mapped separately. The bone areas of male F1 and F2 mice are shown in Fig. 7. The mean bone area in 917M B6C3H F1 male mice (Fig. 7) was 8-12% lower (p < 0.01 by ANOVA) compared with WT B6C3H F1 male mice, which shows that the low bone area phenotype was expressed in the C3H background. Similarly, the bone area in 917M B6C3H F2 male mice (Fig. 7) was significantly lower compared with WT B6C3H F2 male mice. The body weight-adjusted bone area of F2 males (n = 69) generated from both 917M and WT B6 mice was normally distributed. The results from whole genome linkage analysis using total bone area showed evidence for linkage at loci on chromosome 1 (LOD score 4.9 at 45 cM) and chromosome 4 (LOD score 7.9 at 10.5 cM) as shown in Fig. 8. The LOD scores for chromosome 1 and chromosome 4 QTLs were 3.4 (p > 0.05) and 6.8 (p < 0.01), respectively, when body weight was used as a covariate (data analyzed using Pseudomarker program). The LOD scores for chromosome 4 QTLs for femur bone area, periosteal circumference, and endosteal circumference when body weight was used as covariate were 6.2 (p < 0.01), 9.1 (p < 0.01), and 5.6 (p < 0.01), respectively. Genome-wide linkage analysis of B6C3H F2 females (n = 60) generated from 917M B6 male did not show any significant QTLs (data not shown).

Figure FIG. 7..

Total body bone area of 10-week-old 917M (n = 62), wildtype (WT) B6 males (n = 58), WT C3H males (n = 82), 917M B6C3H F1 male hybrids (n = 5), WT B6C3H F1 male hybrids (n = 16), WT B6C3HF2 male hybrids (n = 69), and 917M B6C3H male F2 mice (n = 92). The box represents mean ± SE, and error bars represent SD. Bone area in 917M and 917M B6C3H F1 hybrids and 917M B6C3H F2 males were significantly (p < 0.05 by ANOVA) lower than the WTB6, WT B6C3H F1, and WT B6C3H F2 hybrids, respectively.

Figure FIG. 8..

Genome-wide main QTLs associated with total body bone area (DXA) in 10-week-old intercrossed B6C3H F2 males (n = 69) generated from 917M mice. The bottom plot shows QTLs associated with total body bone area in wildtype C3HB6 F2 males (n = 92). The horizontal lines indicate genome-wide error levels of p < 0.01 and p < 0.05. Chromosome 4 harbors a locus that significantly affects total bone size in F2 male mice generated from 917M; the bottom plot shows that this locus was not detected in F2 males from WT mice.

Interval mapping using 8 additional markers (total of 11 markers) for chromosome 4 and bone size traits revealed peak intervals on proximal chromosome 4 at 10.5 cM, with LOD scores of 4.0-8.2 (genome-wide error rate of p < 0.05; Fig. 9. However, the peak interval for periosteal circumference was shifted to 50 cM with a LOD score of 6.6 (genome-wide error rate of p < 0.01). Bone size QTLs on chromosome 4 explained 38-42% of the phenotypic variance in bone area of F2 male mice. Alleles in the chromosome 4 interval that are inherited from the B6 parental strain contribute to a significantly lower bone area than do alleles inherited from C3H (Fig. 10). Alleles of the chromosome 1 interval that contribute to low bone area are inherited from C3H mice.

Figure FIG. 9..

LOD scores of multiple bone size phenotypes for chromosome 4. Dashed horizontal line indicates significance threshold level used at p < 0.05 for genome wide significance. PC, periosteal circumference (pQCT); EC, endosteal circumference (pQCT); FBA, femur bone area (DXA); TBA, total bone area (DXA).

Figure FIG. 10..

The main effects of genetic alleles on the mean total body bone area for markers D4Mit193 (LOD score 6.74) and D1Mit132 (LOD score 3.4). Alleles of the chromosome interval that was inherited from the B6 parental strain (917M mutant strain) contribute to a significantly lower bone area than do alleles inherited from C3H for D4Mit193, but not for D1Mit132. Data are presented as mean ± SE, and number of animals for each group is indicated in parentheses.ap < 0.001 vs. C3H/C3H;bp < 0.01 vs. B6/B6 and p > 0.05 vs. C3H/C3H;cp < 0.05 vs. C3H/C3H and B6/B6 by ANOVA.

Linkage analyses for bone size in WT B6C3H F2 mice

A whole genome linkage search for loci affecting the body size traits (total body bone area and total bone area and periosteal circumference at midshaft tibia) in WT B6C3H F2 male (n = 92) and female (n = 85) mice did not reveal any suggestive or significant QTLs. A suggestive QTL for total bone area was observed on chromosome 14 (LOD score 4.7 at 5 cM) when body weight was used as a covariate (data not shown).

QTL-QTL interactions for total body bone area traits

Two loci show interactions when the genotype at one locus influences the effect of the other locus. Table 2 shows genome-wide significant interactions observed in B6C3H F2 mice for total body bone area. The interaction involving chromosome 1 and chromosome 4 was noteworthy because the chromosome 1 locus was the only other locus that was associated with any significant QTLs observed for bone size. The combined variance explained by the two interactive QTLs on chromosome 1 and chromosome 4 was 60% of the F2 population (calculated separately by Pseudomarker Fit QTL algorithm). Other significant interactions were noticed between the 10- and 80-cM region of chromosome 4. Both chromosome 1 and chromosome 4 harbor loci known to regulate BMD and bone biomechanical properties phenotypes. A significant interaction between these loci would suggest a complex interaction between genes that regulate skeletal phenotypes or suggest the presence of modifying loci influencing bone area trait.

Table Table 2.. List of Suggestive and Significant Marker Pair Showing Interaction From Genome-Wide Analysis of Bone Area in 917M B6C3H F2 Males
original image


Bone size is an independent determinant of bone strength.(18,19) Therefore, the identification of genes regulating bone size should reveal important determinants for fracture risk. Previous studies have used bone size differences in inbred strains of mice to identify QTLs that regulate bone size, geometry, and biomechanical properties.(20–26) We used a complementary phenotype-driven mutagenesis approach to reveal a locus that regulates bone size. The major advantage of this approach is that any locus identified in this study presumably represents one gene in contrast to QTLs identified by linkage studies, which are likely to represent multiple genes at each locus. Thus, positional cloning of the gene identified by the ENU mutagenesis approach is potentially easier.

The data presented in this study show that the point mutation generated by ENU results in decreased bone size at multiple anatomical sites, including decreased perimeter at the midshaft tibia and femur (without affecting the length) in the mutants compared with the wildtype mice. The phenotype was expressed in males with a dominant mode of inheritance. The potential mechanism whereby bone size could be genetically regulated includes factors affecting periosteal apposition or resorption.(11,12) Our data on the in vitro cell culture of osteoblasts suggest that a decrease in bone size could be attributed to an overall decrease in mature osteoblast cell numbers because of (1) a decreased cell proliferation rate; (2) increased apoptosis; or (3) a combination of both these factors. It could be inferred that the ENU mutation shifts the balance between undifferentiated and differentiated cells by decreasing the proliferation of immature cells and promoting apoptosis of more mature osteoblasts without affecting the differentiation potential and activity. Further studies will be required to explore the molecular basis of these differences in osteoblasts derived from mutant versus WT mice.

The finding that the mutation affects only males is consistent with the outcomes of linkage studies(20,21) that have implicated a variety of sex-specific chromosomal regions (and presumably genes) that regulate bone size in mice and humans.(26) There are different mechanisms in males and females for periosteal bone expansion leading to sex differences in bone size and geometry, which in part may explain the lower risk in males against age-related fracture incidences in humans.(27,28) In B6 mice, the bone size differences between males and females appear after puberty.(29) Our data (details not shown) on WT B6 male (n = 40-60) and female mice (n = 40-60) show that total body bone area is 12.9%, 11.5%, and 8.5% higher (p < 0.0001) in males compared with females at 6, 10, and 16 weeks, respectively. In 917M male mice, the total bone area is 0.3% and 4.9% lower at 10 and 16 weeks, respectively, compared with female littermates, indicating that the mutation is affecting a sex-related disparity in bone size. Two main factors, sex steroids and mechanical loading, have been proposed as important regulators of periosteal apposition (and presumably bone size) differences between males and females. Animal studies offer evidence for a positive effect of androgens and a negative effect of estrogens on periosteal bone formation rates. Thus, if the sex difference in bone size is caused by androgen/estrogen action as some animal studies suggest,(30) this lends us to speculate that the mutation identified in this study may mediate the effects of androgen/estrogen on bone size. Another finding that points toward the probable involvement of sex hormones is our data on a loss of stimulatory effect of mechanical stress on osteoblast proliferation in mutant mice. Mechanical force induces the expression of a variety of genes in the periosteum,(31) with a rapid transformation of quiescent periosteal surfaces to those on which bone formation occurs. In addition, it has been recently shown that the estrogen receptor (ER) is essential(32) for mediating the stimulatory effects of mechanical stress; thus, an impaired response to mechanical stimuli would indicate the involvement of the ER pathway. Further studies are needed to confirm whether the mutant gene is responsive to sex hormones and mechanical loading.

To localize ENU mutation, we used low-resolution linkage mapping covering the whole genome with markers spaced at intervals of ∼21 cM. Use of multivariate analysis with body weight as a covariate for bone size traits provides a more accurate assessment of bone size QTLs, which otherwise could be confounded by growth related QTLs. Interestingly, we identified only one strong candidate interval on proximal chromosome 4 that modulates bone size. No other chromosomal interval has a LOD score remotely as high as proximal chromosome 4. All bone size traits mapped to chromosome 4 to confirm specificity of this QTL, although peak interval for one bone size trait, periosteal circumference at the tibia midshaft (data on periosteal circumference is shown in Fig. 9), appear to be located at slightly distal region. It remains to be verified if the two peaks represent two distinct QTLs. As one would expect in point mutations, the locus identified on chromosome 4 accounts for as much as 40% of the phenotypic variance of all bone size traits, which is in contrast to ∼5-15% of the variance explained in QTL studies employing these two strains of mice.

Although our data on linkage analysis emphasizes a single interval where mutation could be located (on chromosome 4), there are differences in bone size between C3H and B6 mapping strains reflecting allelic variation at a variety of QTLs governing bone size and geometry. Thus, it is possible that the presence of these naturally segregating allelic variations will result in the appearance of QTL (background) confounding the search for a unique QTL that could be attributed to ENU mutation. Our finding of a single QTL is likely because of two factors: magnitude of the phenotype caused by mutation (as evident in Fig. 7) and robustness of the phenotype expressed in the background of C3H (mapping) strain of mice. As our data show, the effect of the mutant phenotypes is larger than most of the published bone size QTLs.(20–25) Our strategy to use a low-resolution map with a small sample size was designed to reduce the power to detect background QTLs, as shown in case of a WT control population (Fig. 9). Under these conditions, a QTL would arise only if a mutation occurred that significantly altered the phenotype distribution in the F2 population, which is the case in F2s generated from ENU mutant mice.

Several lines of evidence support our view that the locus on chromosome 4 harbors a mutation. First, the alleles on the chromosome 4 interval inherited from the B6 mutant strain contribute to a significantly lower bone area (including periosteal circumference) than alleles inherited from C3H, which is not the case with the QTL observed on chromosome 1. Second, we did not observe any significant or suggestive QTLs on chromosome 4 in interval mapping of the F2 population generated from the WT B6 mice. Additionally, previous linkage studies using male mice(20,21) have not identified any locus on chromosome 4 that regulates bone size. Although a linkage study(22) using the C3H and B6 strains of mice has shown a strong QTL (LOD score 16-21) on chromosome 4 for biomechanical properties of femurs, this locus was identified in female mice and was located in the 70- to 90-cM region. Based on these observations, the QTLs identified in this study represents a distinct linkage finding. One approach commonly used to identify mutant genes in an ENU-induced mutant mice is based on sequencing of a positional candidate gene. The potential candidate genes include those genes that regulate osteoblast activity, apoptosis, responsive elements to sex hormones, and those involved in mechanosensors such as gated ion channels. In this regard, potential candidate genes found on the chromosome 4 region are calb1 (calbindin), igfbpl1 (insulin-like growth factor binding protein-like I), Casp8ap2 (caspase 8 associated protein 2), Gabrr1 and Gabrr2 (γ-aminobutyric acid receptor, subunit), Cga (glycoprotein hormones, α subunit), Rragd (ras-related GTP binding D), Il11ra1 (interleukin 11 receptor, α chain 1), and Musk (muscle, skeletal, receptor tyrosine kinase). However, the QTL region identified in this study that regulates bone size represent a fairly large region, which encompasses many more important genes other than mentioned above that can contribute to the bone size phenotype. Therefore, in future studies, we propose to use differential displays of genes in 917M versus WT mice using microarray to identify key genes from the chromosome 4 QTL region that can be selected for sequencing to locate the point mutation(s).

We have also discovered several suggestive and significant interactions between QTLs on chromosome 4 with other loci on chromosomes 1 and 2. The locus on chromosome 1 is in the proximity of QTLs identified in earlier linkage studies,(33–35) indicating a wide range of pleiotropic effects across diverse phenotypes of BMD, bone size, and biomechanical properties. The interaction with the chromosome 1 loci was particularly significant because it may indicate that the main effect of the chromosome 1 locus (which had a suggestive LOD score of 4.9) was primarily caused by interaction effects with loci on chromosome 4. This was evident in a very high (60%) variance (of F2 population) explained by the interaction effects of these two loci together. Our finding on these interactions highlights the degree to which the phenotypes of BMD and bone size are under the control of common genetic factors.

In conclusion, we have provided evidence that the decrease in bone size observed in ENU mutant male mice could be caused by a mutation in gene-regulating osteoblast function in terms of proliferation, apoptosis, and response to mechanical stimulus. In addition, linkage analysis indicates that the point mutation is located on a novel allele of chromosome 4 because no mouse or human linkage study has described QTL in a corresponding position. Finally, this report shows for the first time the effectiveness of employing phenotype-driven ENU mutagenesis in the study of genetics of osteoporosis. It also shows the use of the ENU mutagenesis approach for generating models for use in genetic studies of osteoporosis where conventional knockout could be lethal.


This work was supported by the Army Assistance Award DAMD17-99-1-9571. The U.S. Army Medical Research Acquisition Activity (Fort Detrick, MD) 21702-5014 is the awarding and administering acquisition office for the DAMD award. The information contained in this publication does not necessarily reflect the position or the policy of the government, and no official endorsement should be inferred. All work was performed in facilities provided by the Department of Veterans Affairs. The authors thank Darcie Nagel, TaMarrah Oliver, Valerie Chest, and Cynthia Ganda for expert technical assistance.