Skip to main content
NCBI home page
Journal of Bone and Mineral Research logoLink to Journal of Bone and Mineral Research
. 2008 Apr 21;23(9):1529–1537. doi: 10.1359/JBMR.080414

Quantitative Trait Loci for BMD in an SM/J by NZB/BlNJ Intercross Population and Identification of Trps1 as a Probable Candidate Gene

Naoki Ishimori 1, Ioannis M Stylianou 1, Ron Korstanje 1, Michael A Marion 1, Renhua Li 1, Leah Rae Donahue 1, Clifford J Rosen 1, Wesley G Beamer 1, Beverly Paigen 1, Gary A Churchill 1
PMCID: PMC2586053  NIHMSID: NIHMS77498  PMID: 18442308

Abstract

Identification of genes that regulate BMD will enhance our understanding of osteoporosis and could provide novel molecular targets for treatment or prevention. We generated a mouse intercross population and carried out a quantitative trait locus (QTL) analysis of 143 female and 124 male F2 progeny from progenitor strains SM/J and NZB/BlNJ using whole body and vertebral areal BMD (aBMD) as measured by DXA. We found that both whole body and vertebral aBMD was affected by two loci on chromosome 9: one with a significant epistatic interaction on distal chromosome 8 and the other with a sex-specific effect. Two additional significant QTLs were identified on chromosome 12, and several suggestive ones were identified on chromosomes 5, 8, 15, and 19. The chromosome 9, 12, and 15 loci have been previously identified in other crosses. SNP-based haplotype analysis of the progenitor strains identified blocks within the QTL region that distinguish the low allele strains from the high allele strains, significantly narrowing the QTL region and reducing the possible candidate genes to 98 for chromosome 9, 31 for chromosome 12, and only 2 for chromosome 15. Trps1 is the most probable candidate gene for the chromosome 15 QTL. The sex-specific effects may help to elucidate the BMD differences between males and females. This study shows the power of statistical modeling to resolve linked QTLs and the use of haplotype analysis in narrowing the list of candidates.

Key words: high-fat diet, peripheral DXA, quantitative trait gene, haplotype analysis

INTRODUCTION

Osteoporosis is a common disease in the elderly, and its prevalence is increasing.(1) BMD is an important component of human bone strength and can affect the risk of osteoporosis along with other environmental factors. Identification of genes that regulate BMD will enhance our understanding of osteoporosis and could provide novel molecular targets for treatment or prevention. Quantitative trait locus (QTL) analysis of mouse inbred line crosses previously identified a number of loci with effects on both volumetric BMD measured by pQCT and areal BMD (aBMD) measured by DXA.(2,3) Several of these studies have noted differences in genetic factors affecting BMD between females and males that are not well understood.(4,5)

Human epidemiological evidence links osteoporosis with hyperlipidemia and atherosclerosis,(6) and a cross in which these traits have been measured might give additional information on this relationship. Measurements of BMD in inbred mouse strains(7–9) show that strain SM/J (SM) has low BMD and strain NZB/B1NJ (NZB) has relatively high BMD. Previously, we reported QTL for plasma high-density lipoprotein (HDL)-cholesterol levels,(10) phospholipid transfer activity,(11) atherosclerosis,(12) and obesity(13) in a (SM × NXB)F2 intercross population that was maintained on a high-fat diet for 16 wk. Here we report QTLs for whole body and vertebral BMD and characterize the epistatic interactions. We find sex-specific effects for some QTLs, which may help to elucidate the BMD differences between males and females. We resolve a previously reported QTL(14,15) on chromosome (Chr) 9 into two distinct loci and further narrowed some QTL regions through haplotype analysis.(16) The ability to narrow the Chr 15 QTL to only two genes shows the power of our approach.

MATERIALS AND METHODS

Animals and diet

SM/J (SM) and NZB/BlNJ (NZB) inbred mouse strains were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). SM females were mated to NZB males to produce the F1 progeny, which were intercrossed to produce F2 progeny, and 267 of these (124 males and 143 females) were phenotyped for BMD. Mice were housed in a temperature- and humidity- controlled environment with a 14-h:10-h light-dark cycle. After weaning, mice were maintained on a chow diet (Old Guilford 234A; Guilford) until 8 wk of age with free access to food and water throughout the experiment. Starting at 8 wk of age, mice were fed a high-fat diet for 16 wk containing (wt/wt) 15% dairy fat, 50% sucrose, 20% casein, 1.0% cholesterol, and 0.5% cholic acid, as well as cellulose, vitamins, and minerals. The source of chemicals and the diet have been described previously.(17,18) The Jackson Laboratory's Animal Care and Use Committee approved all experiments.

Quantitative phenotype measurements

At 24 wk of age, mice were necropsied, and whole body aBMD and lumbar vertebrae aBMD (vertebral aBMD) were measured using DXA (PIXImus; GE-Lunar, Madison, WI, USA).(19) PIXImus measurements of whole body and site-specific aBMD correlate well with mineral content of phantom standard of known density (r 2 = 0.997).

Genotyping

The mice were genotyped with 154 simple sequence length polymorphic (SSLP) markers (Research Genetics, Huntsville, AL, USA) spaced ∼8.3 cM apart. Methods used for DNA isolation, PCR amplifications, and subsequent gel electrophoreses have been described previously.(10) Reported genetic map positions were retrieved from the Mouse Genome Informatics database (http://www.informatics.jax.org).

QTL analysis

A logarithmic transformation of individual aBMD values was performed to correct for the slightly skewed data. QTL analysis was conducted as described below to search for main effect QTL and pairwise epistatic interactions. We applied linear model based likelihood ratio tests to resolve multiple QTLs and integrated all of the QTLs into a multilocus genetic model. Analyses were carried out using R/qtl(20) (ver. 1.05–2 available at http://www.rqtl.org/) and Pseudomarker(21) (ver. 2.03 available at http://www.jax.org/staff/churchill/labsite) software packages.

We carried out single locus scans to identify main effect QTLs. LOD scores were computed at 2-cM intervals across the genome using the “EM” method in R/QTL. We included sex as an additive covariate in all single and pairwise QTL scans to account for overall differences in BMD between the sexes. Main effect QTLs were deemed significant (suggestive) if they exceeded the 95% (37%) genome-wide adjusted threshold based on 1000 permutations.(22) The significant and suggestive thresholds were 3.6 and 2.2 for these genome scans, respectively. CIs for QTL positions were computed from the posterior probability density, as described previously.(21) A second set of single-locus scans was carried out to identify loci with sex-specific effects by including an interaction between sex and the putative QTL at each locus. The significant and suggestive thresholds for these scans were 4.7 and 3.1, respectively. The difference in LOD scores (ΔLOD) between these two scans, with and without the QTL by sex interaction, constitutes a test for sex-specific effects. We applied a significance threshold of ΔLOD >2.0 corresponding to p < 0.01, based on the two degrees of freedom χ2 distribution of the log likelihood ratio for QTL-by-sex interactions.(23) We carried out genome-wide all-pairs scans to search for pairwise epistatic effects. These scans also included sex as an additive covariate. The significant and suggestive thresholds for the pairwise scans were 11.0 and 5.0, respectively. In addition, we required that the interaction alone should account for at least a 4.0 LOD increase over a two locus additive model to declare a pairwise epistatic effect.

For some of the QTLs, the shape of the LOD profile from the single-locus genome scans suggested that there might be two or more linked QTLs present. To resolve linked QTLs, we carried out linear model based multilocus scans on a per chromosome basis. In these scans, we propose a model that may include one or more QTLs, a sex effect, sex by QTL, and QTL by QTL interactions. The QTLs are scanned over all possible positions in 2-cM steps along a single chromosome. The maximum LOD score attained is assigned to the model, and the best QTL locations are recorded. Statistical tests are computed as likelihood ratios and converted to LOD scores. The difference in LOD scores between two models that differ in a single term is a test for the significance of that term. Significance levels for these tests were based on the asymptotic distribution of the log-likelihood ratio.

To summarize the simultaneous effects of sex, QTLs, and interactions, we carried out a multilocus analysis in which all significant and suggestive QTLs and interactions were incorporated in a common multiple regression model. Any terms that did not meet the nominal 0.01 level in the regression analysis were eliminated in backward stepwise fashion, with the exception that main effect terms involved in a significant interaction were retained. The adjusted proportion of variance explained by each QTL is based on this multilocus model.

QTLs were named in accordance with the International Committee on Standardized Genetic Nomenclature for Mice (http://www.informatics.jax.org/mgihome/nomen) and the Complex Traits Consortium.(24) QTLs were named if they were significant or if they were suggestive but confirmed previously reported QTLs. They were given the same name if the crosses identifying them shared at least one common parental strain and a new name if the crosses identifying them involved no common strains.

Haplotype analysis

Recent evidence showed that most genetic variation in inbred mouse strains predates the derivation of the common laboratory strains.(25) It follows that co-localized QTLs found in multiple crosses using different strains are likely to represent shared alleles and that QTLs can be effectively narrowed by comparing haplotypes of the parental strains throughout the QTL region.(15,25,26) SNPs used in haplotype analysis were obtained from the Mouse Phenome Database (www.jax.org/phenome).

RESULTS

Inheritance of whole body aBMD and vertebral aBMD

Whole body aBMD and vertebral aBMD for the parental, reciprocal F1, and F2 mice were measured at 24 wk of age after mice had been fed the high-fat diet for 16 wk (Table 1). Vertebral aBMD in females was higher than in males for each group. Whole body and vertebral aBMDs of NZB females were significantly higher (p < 0.001) than those of SM and reciprocal F1 females. Whole body and vertebral aBMDs of the reciprocal F1 females were intermediate to those of the parental females, and aBMDs of SM females were significantly lower (p < 0.001) than those of (NZB × SM)F1 females but not significantly different from those of (SM × NZB)F1 females. Whole body and vertebral aBMDs of NZB males were significantly higher (p < 0.001) than those of SM and (SM × NZB)F1 males. Whole body and vertebral aBMDs of the (SM × NZB)F1 males were intermediate to those of the parental males, and whole body aBMD of SM males was significantly lower (p < 0.05) than those of (SM × NZB)F1 males. Distributions of whole body and vertebral aBMDs of the F2 mice are shown in Figs. 1A–1D. Mean aBMD positions of the progenitors and reciprocal F1 mice are shown in the figures, indicated by arrows. Analysis of F2 progeny showed a significant correlation between whole body aBMD and vertebral aBMD (r = 0.84, p < 0.0001).

Table 1.

Whole Body BMD and Vertebral BMD of SM, NZB, Reciprocal F1, and F2 Progeny

Mice Whole body aBMD (mg/mm2)
Vertebral aBMD (mg/mm2)a
Female Male Female Male
NZB (n = 19) 0.643 ± 0.008*†‡ 0.632 ± 0.007*† 1.109 ± 0.041*†‡ 1.004 ± 0.031*†
SM (n = 20) 0.494 ± 0.007‡ 0.493 ± 0.007§ 0.644 ± 0.020‡ 0.627 ± 0.013
(SM × NZB)F1 (n = 9) 0.525 ± 0.010 0.533 ± 0.017 0.754 ± 0.080 0.714 ± 0.028
(NZB × SM)F1 (n = 9) 0.565 ± 0.006 ND 0.909 ± 0.027 ND
(SM × NZB)F2 F (n = 143); M (n = 124) 0.569 ± 0.004 0.567 ± 0.003 0.854 ± 0.013 0.822 ± 0.009
Open in a new tab

Data are presented as the mean ± SE.

* Significant difference (p < 0.001, by ANOVA) vs. SM.

† Significant difference (p < 0.001, by ANOVA) vs. (SM × NZB)F1.

‡ Significant difference (p < 0.001, by ANOVA) vs. (NZB × SM)F1.

§ Significant difference (p < 0.005, by ANOVA) vs. (SM × NZB)F1.

Because it is the distribution and not the mean among the F2 population that is most important for detecting genetic linkage to a phenotype, we did not test for significant differences between F2 progeny and either the parental strains or reciprocal F1.

ND, not determined.

FIG. 1.

FIG. 1

Open in a new tab

Distributions for aBMD. The 267 (NZB × SM)F2 progeny were fed a high-fat diet for 16 wk. Arrows indicate the positions of mean aBMD for NZB, SM, and reciprocal F1 progeny. The bell-shaped line depicts the theoretical normal distribution. Whole body aBMD for females (A) and males (B) and vertebral aBMD for females (C) and males (D).

QTL analysis: main effects

Single-locus genome scans for whole body and vertebral aBMD phenotypes were carried out on the 267 (SM × NZB)F2 progeny to detect QTLs with main effects. These scans included sex as an additive (Figs. 2A and 2B) or an interactive (Figs. 2C and 2D) covariate to account for overall differences in phenotype between the sexes. For both phenotypes, we found a significant QTL on Chr 9 near 50 cM, which consists of two linked QTLs at 34 and at 50 cM as we show below. We also identified a significant QTL on Chr 12 for whole body aBMD at 16 cM and for vertebral aBMD at 44 cM, but the shape of the LOD score curves in Figs. 3E and 3G indicates that both QTLs are present for both phenotypes. We identified a suggestive QTL on Chr 8 for both aBMD traits (peak at 71 cM) and three additional suggestive loci for vertebral aBMD on Chrs 5 (77 cM), 15 (29 cM), and 19 (23 cM). These QTLs, the LOD scores, and 95% CIs are summarized in Table 2. For most QTLs, the NZB allele was associated with increased BMD (Fig. 3), but SM contributed the allele for increased BMD for Chr 5, 8, and 19.

FIG. 2.

FIG. 2

Open in a new tab

Genome-wide scans for aBMD. These scans are based on 267 (NZB × SM)F2 progeny fed a high-fat diet for 16 wk. Chrs 1 through X are represented numerically on the ordinate. The relative width of the space allotted for each chromosome reflects the relative length of each chromosome. The abscissa represents the LOD score, the traditional metric of genetic linkage. The significant (p < 0.05) and suggestive (p < 0.63) levels of linkage were determined by permutation testing. Whole body aBMD with sex as an additive covariate (A) or with sex as an interactive covariate (C); vertebral aBMD with sex as an additive covariate (B) or sex as an interactive covariate (D).

FIG. 3.

FIG. 3

Open in a new tab

Genome-wide scans (solid lines) and posterior probability densities (broken lines) for the main effect QTLs on chromosomes 9 and 12. (A, B, E, and F) QTL, posterior probability density, and 95% CI. (C and G) Allele effects at the peak of the vertebral aBMD QTL (allele effects for whole body aBMD are almost identical). The posterior probability densities are a likelihood statistic that give rise to the 95% CIs indicated by gray bars. Homozygosity for NZB alleles is represented by N/N, homozygosity for SM alleles by S/S, and heterozygosity at a locus by N/S. Marker locations for each QTL are in parentheses. y-axes represent mean values of vertebral aBMD (C and F). Error bars represent SE. The pairwise genome scans detected an effect of gene interaction between the Chr 8 and Chr 9 locus (D). Homozygosity for NZB alleles is represented by N/N, homozygosity for SM alleles by S/S and heterozygosity at a locus by N/S. y-axes show mean values of whole body aBMD (A) and vertebral aBMD (B). Error bars represent SE.

Table 2.

QTLs Identified for a Single Gene Genome- Wide Scan of 267 (SM × NZB)F2 Progenies

Chr Locus name Sex additive
High allele Mode of inheritance Nearest marker Overlapping QTLs
LOD score Peak (cM) 95% CI (cM)
Whole body aBMD 8 2.8 71 49–71 SM Recessive D8Mit42
9 Bmd40 12.2 51 45–57 NZB Recessive D9Mit196 Bmd10, (13) Tbbmd4 (14)
12 Bmd41 6.5 16 14–48 NZB Additive D12Mit60 Bmd12,(13) Unnamed QTL(14)
Vertebral aBMD 5 2.3 77 61–84 SM Dominant D5Mit284
9 Bmd40 13.7 49 24–60 NZB Recessive D9Mit196 Bmd10,(13) Tbbmd14 (14)
12 Bmd41 5.9 44 14–58 NZB Additive D12Mit5 Bmd12,(13) Unnamed QTL(14)
15 Bmd42 2.3 29 22–58 NZB Recessive D15Mit26 Bmd4,(40) Unnamed QTL(41)
19 2.7 23 15–43 SM Dominant D19Mit11
Open in a new tab

QTL analysis: epistasis

Pairwise genome scans, with sex as an additive covariate, showed a significant epistatic interaction between a Chr 8 locus at 70 cM and the Chr 9 QTL at 50 cM (whole body aBMD: LODfull = 18.0, LODint = 2.61; vertebral aBMD: LODfull = 20.0, LODint = 4.11). The shape of the interaction suggests that heterozygosity on distal Chr 8 masks the effect of the Chr 9 locus. Among the animals that are homozygous (NN or SS) on distal Chr 8, the Chr 9 effect is additive, and higher BMD is associated with N alleles (Fig. 3D).

QTL analysis: sex-specific effects

To identify sex-specific QTLs, we carried out a second set of single-locus genome scans that included sex as an interactive covariate (Fisg. 2B and 2D) as previously described.(23) The difference in LOD scores (ΔLOD) between the scans with sex as an additive covariate and with sex as an interactive covariate constitutes a test for sex-by-QTL interaction. A change >2 LOD between the sex additive and sex interactive scores indicates a locus at which the QTL effects differ between the sexes. The major QTL on Chr 9 presents strong evidence for sex-specific effects on whole body (ΔLOD = 2.42; p = 0.0004) and vertebral aBMD (ΔLOD = 3.22; p = 0.0001; Table 3, test between model a and b). The sex-specific effect resulted in more pronounced differences in females compared with males (Fig. 3C). An additional suggestive QTL that interacts with sex was identified for vertebral BMD on Chr 5 at 1 cM (Table 2).

Table 3.

Models and Test Statistics Applied to Resolve Chromosome 9 QTLs

Whole body aBMD
Vertebral aBMD
Test Df ΔLOD p ΔLOD p
Y ∼ Sex + Q9a (a) c-a 2 2.20 0.0096 2.11 0.0078
Y ∼ Sex + Q9a + Q9a × Sex (b) d-b 2 2.37 0.0038 2.20 0.0063
Y ∼ Sex + Q9a + Q9ab (c) b-a 2 2.42 0.0004 3.22 0.0001
Y ∼ Sex + Q9a + Q9ab + Q9ab × Sex (d) d-c 2 3.71 0.0002 3.93 0.0001
Y ∼ Sex + Q9a + Q9ab + Q9b × Sex + Q8 (e) e-d 2 1.90 0.013 1.49 0.032
Y ∼ Sex + Q9a + Q9b + Q9b × Sex + Q8 + Q8 × Q9b (f) g-e 4 4.53 0.0003 6.83 2.5 × 10−6
Y ∼ Sex + Q9a + Q9b + Q9b × Sex + Q8 + Q8 × Q9a (g) f-e 4 2.80 0.012 4.68 0.0002
Open in a new tab

The phenotype Y is expressed as a linear model with terms shown on the right. All models include an additive term for sex. QTL genotypes are denoted as Q with chromosome indicated in the subscript. Each QTL is scanned in 2-cM steps across the across the chromosome to find the locations(s) that maximize the LOD score. Differences in the maximized LOD scores between models are used to test the significance of model terms.

Test refers to contrasts between models. The degrees of freedom (df) correspond the difference in the numbers of free parameters between the models being compared. Change in LOD scores (ΔLOD) and p value.

QTL analysis: resolution of the linked loci

To resolve the Chr 9 QTLs, we fit a series of multiple QTL models, listed in the left half of Table 3. Maximum LOD scores were obtained from each model by scanning the QTLs simultaneously to find their optimal locations. Changes in LOD scores between models were used to construct tests for various features of the QTLs (right half of Table 3). We constructed tests for one QTL versus two QTLs both with and without an interaction term for sex (c-a and d-b). We constructed tests for the sex interaction with both one and two QTL models (b-a and d-c). All of the above tests were significant (ΔLOD > 2), leading us to conclude that there are two distinct QTLs on Chr 9 and an interaction with sex. We tested the addition of Chr 8 as a main effect and found that it held up with only modest significance (d-e). Interaction between QTLs on Chrs 8 and 9 was tested using two models. In one model, the interactions with Chr 8 and sex are with the same locus on Chr 9 (f). In a second model, the interactions are with distinct loci on Chr 9 (g). The latter model yielded the highest LOD score, particularly for vertebral aBMD (g-e compared with f-e). The QTL-by-QTL interaction was localized to Chr 8 at 70 cM and Chr 9 at 32 cM, and the QTL-by-sex interaction localized to Chr 9 at 50 cM. The pairwise scan analysis (Fig. 4) confirmed the presence of two QTLs on Chr 9 and that the interaction with Chr 8 involves the proximal locus.

FIG. 4.

FIG. 4

Open in a new tab

Chromosome maps for the 95% CI regions on Chr 9, 12, and 15. Number of genes according to Ensembl (build 36) is indicated before and after interval-specific haplotype analysis.

Chr 12 also seems to harbor two distinct QTLs. The proximal peak at 16 cM is stronger for whole body aBMD (Fig. 2A) and the peak at 42 cM is stronger for vertebral aBMD (Fig. 2D), but both peaks are present in scans for both traits. Neither trait has a significant ΔLOD for one versus two QTLs (ΔLOD for whole body aBMD is 1.54; ΔLOD for vertebral aBMD is 1.63), despite the fact that the two peaks are distinct. However, the QTL for whole body aBMD and the QTL for vertebral aBMD are well separated, so we report these as two distinct QTLs, one for each trait.

QTL analysis: multilocus modeling

To provide an integrated summary of the genetic architecture, we constructed multiple QTL models (Table 4). The proportion of the total phenotypic variance accounted for in the F2 population is 41.2% for whole body aBMD and 49.7% for vertebral aBMD. These models confirm the importance of the QTL-by-sex and QTL-by-QTL interactions that individually account for between 3.5% and 5.9% of the total variance. The multilocus model for whole body aBMD includes the two Chr 9 loci, their interactions with Chr 8 and sex, and a main effect contribution from Chr 12 at 16 cM. The multilocus model for vertebral aBMD includes the same terms involving Chr 9 and a Chr 12 main effect at 46 cM. In addition, the multiple QTL model confirms the contribution to the vertebral aBMD of the minor QTLs on Chrs 8, 15, and 19 that account for 7.7%, 2.2%, and 3.8% of the total variance, respectively. The suggestive loci detected on Chr 5 in the genome scans did not contribute significantly to the multilocus model.

Table 4.

Multiple Regression Analysis for All QTLs

Whole body aBMD df Percent variance p (F) Vertebral aBMD df Percent variance p (F)
Sex 3 6.0 2.2 × 10−5 Sex 3 5.5 1.1 × 10−5
Chr8@70 6 6.3 2.4 × 10−4 Chr8@70 6 7.6 4.7 × 10−6
Chr9@32 6 5.8 5.9 × 10−4
Chr9@50 4 9.2 2.4 × 10−7 Chr9@50 4 10.7 1.2 × 10−9
Chr12@16 2 8.8 2.9 × 10−8
Chr12@46 2 6.6 2.8 × 10−7
Chr15@29 2 2.2 4.9 × 10−3
Chr19@23 2 3.8 1.4 × 10−4
Chr9@32 6 6.6 2.8 × 10−5
Sex: Chr9@50 2 5.9 6.0 × 10−6 Sex: Chr9@50 2 5.2 6.0 × 10−6
Chr8@70: Chr9@32 4 3.5 6.0 × 10−3 Chr8@70: Chr9@50 4 5.5 4.2 × 10−5
Open in a new tab

Haplotype analysis reduces the Chr 15 QTL region

For those QTLs that co-localized with previously reported QTLs (Table 2), we identified genomic regions with shared common ancestry as inferred by common SNPs. The strains carrying the high allele for BMD were NZB, MRL, and B6 and the low allele strains were SM, SJL, and C3H for Chr 9, and for the Chr 12 QTLs, the strains carrying the high allele for BMD were NZB, SJL, and B6, and the low allele strains were SM, MRL, and C3H. The high allele strains were B6 and NZB and the low allele strains were C3H/HeJ, DBA/2J, and SM/J for the Chr 15 QTLs. Initial haplotype analysis, which searched for genomic regions where the high alleles were shared and differed from the low alleles, reduced the Chr 9 QTLs from 188 to 98 genes, the Chr 12 QTL from 500 to 31 genes, and the Chr 15 QTL from 527 to just 2 genes (Fig. 4). The Chr 9 and 12 QTLs need some additional means of reducing the large number of genes, but we considered the two genes in the Chr 15 QTL further.

The two candidate genes are Trps1 (zinc finger transcription factor for tricho-rhino-phalangeal syndrome type I protein) and Eif3h (eukaryotic translation initiation factor 3). However, Eif3h was identified as a candidate because of the nucleotide pattern at a single SNP, and the strains involved did not share a haplotype pattern surrounding that SNP. Nonetheless, that one SNP could be responsible for the QTLs. In addition, nothing in the tissue distribution, literature, or the gene ontologeny (GO) terms suggests that this gene might be particularly important in the bone. In contrast, the haplotype region of the Trps1 gene contains 259 SNPs; the high allele strains, C3H/HeJ, DBA/2J, and SM, fall into one haplotype group and the low allele strains, B6 and NZB, fall into a second haplotype group, and these haplotype groups are consistent throughout the large gene. Trps1 contains one nonsynonomous SNP (rs32398060), which changes a valine in B6 to a leucine in C3H and DBA/2. The Trps1 gene codes for four transcripts in the mouse, and this SNP is present in all four transcripts. The change from valine to leucine does not change polarity or the acid-base properties, but it does seem to be conserved across species and lies in a conserved region of the protein. An analysis of the homology of Trps1 gene transcripts from 21 species showed that all species (except fish) have leucine (Fig. 5) at this amino acid, and only some mouse strains, including B6 with the high allele for BMD, have valine. Whether this SNP and the resultant change in amino acid are important to the function of the gene transcripts is unclear. The SNP is not in a recognized binding domain. Trps1 does seem to have some function in the bone because the syndrome lacking this gene has skeletal abnormalities.(27)

FIG. 5.

FIG. 5

Open in a new tab

Species comparison of the amino acid sequence. Analysis of the homology of Trps1 gene transcripts from 21 species shows that all species (except fish) have leucine at this amino acid and only some mouse strains, including B6 with the high allele for BMD, have valine.

DISCUSSION

A QTL analysis of (NZB × SM)F2 female and male mice identified two main effect QTLs that determine whole body and vertebral aBMD: Bmd40 on Chr 9 and Bmd41 on Chr 12. The pairwise genome scans showed a significant QTL–QTL interaction between BMD40 and a locus on Chr 8 common to whole body aBMD and vertebral aBMD. Epidemiological studies have shown that sex exerts a profound effect on the skeleton phenotype,(28,29) suggesting that it is important to differentiate between two sexes in QTL analysis for BMD. Analyzing the entire population with and without a sex-by-QTL interaction is most powerful, because analyzing males and females separately might result in failure to detect common QTL effects in one or both subsets of data.(10,23) We found evidence for a sex-specific effect of Bmd41 and a suggestive sex-specific QTL for vertebral BMD near D5Mit146 on Chr 5.

Human epidemiological evidence has linked osteoporosis with hyperlipidemia and atherosclerosis.(6) The same may be true in mice: Parhami et al.(30) and Tintut et al.(31) found that an atherogenic diet contributes to osteoporosis in mice by blocking osteoblastic differentiation and increasing osteoclastic bone resorption in vivo. In this study, however, no QTLs for BMD co-localized with plasma lipids, obesity QTLs,(10,13) or atherosclerosis QTLs identified in a backcross between SM and NZB.(12)

Most of the QTLs we found confirmed previously reported QTLs for BMD. Previously, Beamer et al.,(14) using a (B6 × C3H)F2 female intercross, found a vertebral volumetric BMD (vBMD) QTL on Chr 9 (48 cM) and named it Bmd10; the B6 allele conferred increased vertebral vBMD. Examination of the LOD curves reported in that study showed a bimodal LOD profile similar to that found in our study. Masinde et al.(15) used (MRL × SJL)F2 females to identify a femoral vBMD QTL on Chr 9 (49 cM) and named it Tbbmd4. Orwoll et al.(4) found a whole body aBMD QTL at the distal portion of Chr 9 (58.5 cM) in male recombinant inbred (RI) mouse strains derived from B6 and DBA/2 progenitors; we consider this to be a different QTL because B6 mice carry the allele for low BMD in contrast to the QTL reported here, in which B6 carries the allele for high BMD. Also, Beamer et al.,(14) using the (B6 × C3H)F2 female intercross, found a femoral vBMD QTL on Chr 12 (2 cM) and named it Bmd12. Masinde et al.(15) used the (MRL × SJL)F2 females to identify a total vBMD on Chr 12 (23 cM). In addition, several of the QTLs have concordance with QTLs for similar traits in rats and humans. The Chr 8 QTL is concordant with a QTL found in both rats(32) and humans,(33) as is the Chr 9 QTL(32,34) and the Chr 12 QTL.(33–35) The Chr 19 QTL is concordant with a human bone-related trait mapped to 8q24.3.(36)

Confirming QTLs in different crosses facilitates the use of haplotype analysis to narrow the candidate gene region. Using these methods, we effectively narrowed Bmd42 on Chr 15 and excluded all but two genes (Trps1 and Eif3h) in the region. Haplotype analysis may miss genes if the mutation in the QTL gene occurred after the strains were inbred; this seems unlikely for Bmd40 because this QTL has been found in three different crosses. Haplotyping may also miss genes if a region is inferred to be identical by descent but in reality contains some SNPs that do differ. This problem is more common if the SNP map is not dense; however, by the time the SNP map is based on such densely genotyped strains such as B6 and C3H, this is less likely. Eif3h encodes a subunit of the eukaryotic translation initiation factor 3 and seems to contribute to efficient translation initiation on 5` leader sequences.(37) Mutations in human TRPS1 cause the trichorhino-phalangeal syndromes, which is characterized by developmental defects of the hair, face, and selected bones.(38) Mice heterozygous for a deletion of the GATA domain in the gene have structural deficits in cortical and trabecular bones.(27) Because the literature shows that Trps1 is important in bone development, it may have a role in BMD as well. In addition, a recent genome-wide association study in humans identified an association between bone mass and SNPs close to TRPS1.(39) We suggest that Trps1 is the candidate gene for the chromosome 15 QTL in mice and the QTL at 8q24 in humans. Polymorphisms in Trps1 should be tested for association with BMD in humans.

This study showed the power of combining QTL crosses and using bioinformatics to focus on candidates whose further elucidation will enhance our understanding of skeletal biology.

ACKNOWLEDGMENTS

This work was funded by grants from the NIH (HL66611, GM070683, AR43618, and CA34196). The authors thank Ray A. Lambert, Jesse Hammer, and Jennifer L. Torrance for helping to prepare the manuscript.

Footnotes

The authors state that they have no conflicts of interest.

REFERENCES

  • 1.Eisman JA. Genetics of osteoporosis. Endocr Rev. 1999;20:788–804. doi: 10.1210/edrv.20.6.0384. [DOI] [PubMed] [Google Scholar]
  • 2.Korstanje R, Paigen B. From QTL to gene: The harvest begins. Nat Genet. 2002;31:235–236. doi: 10.1038/ng0702-235. [DOI] [PubMed] [Google Scholar]
  • 3.Rosen CJ, Beamer WG, Donahue LR. Defining the genetics of osteoporosis: Using the mouse to understand man. Osteoporos Int. 2001;12:803–810. doi: 10.1007/s001980170030. [DOI] [PubMed] [Google Scholar]
  • 4.Orwoll ES, Belknap JK, Klein RF. Gender specificity in the genetic determinants of peak bone mass. J Bone Miner Res. 2001;16:1962–1971. doi: 10.1359/jbmr.2001.16.11.1962. [DOI] [PubMed] [Google Scholar]
  • 5.Beamer WG, Shultz KL, Ackert-Bicknell CL, Horton LG, Delahunty KM, Coombs HF, III, Donahue LR, Canalis E, Rosen CJ. Genetic dissection of mouse distal Chromosome 1 reveals three linked BMD QTL with sex dependent regulation of bone phenotypes. J Bone Miner Res. 2007;22:1187–1196. doi: 10.1359/jbmr.070419. [DOI] [PubMed] [Google Scholar]
  • 6.Frye MA, Melton LJ, III, Bryant SC, Fitzpatrick LA, Wahner HW, Schwartz RS, Riggs BL. Osteoporosis and calcification of the aorta. Bone Miner. 1992;19:185–194. doi: 10.1016/0169-6009(92)90925-4. [DOI] [PubMed] [Google Scholar]
  • 7.Tordoff MG, Bachmanov AA. Survey of calcium & sodium intake and metabolism with bone and body composition data, MPD:103. Mouse Phenome Database (MPD) Web Site. 2005. Available online at http://www.jax.org/phenome.
  • 8.Naggert JK, Paigen B, Svenson KL, Petters LL. Diet effects on bone mineral density and content, body composition, and plasma glucose, leptin, and insulin levels, MPD:143. Mouse Phenome Database (MPD) Web Site. 2005. Available online at http://www.jax.org/phenome.
  • 9.Wergedal JE, Sheng M-C, Ackert-Bicknell CL, Beamer WG, Baylink DJ. Genetic variation in femur extrinsic strength in 29 different inbred strains of mice is dependent on variations in femur cross-sectional geometry and bone density. Bone. 2005;36:111–122. doi: 10.1016/j.bone.2004.09.012. [DOI] [PubMed] [Google Scholar]
  • 10.Korstanje R, Li R, Howard T, Kelmenson P, Marshall J, Paigen B, Churchill G. Influence of sex and diet on quantitative trait loci for HDL cholesterol levels in an SM/J by NZB/BlNJ intercross population. J Lipid Res. 2004;45:881–888. doi: 10.1194/jlr.M300460-JLR200. [DOI] [PubMed] [Google Scholar]
  • 11.Korstanje R, Albers JJ, Wolfbauer G, Li R, Tu AY, Churchill GA, Paigen BJ. Quantitative trait locus mapping of genes that regulate phospholipid transfer activity in SM/J and NZB/BlNJ inbred mice. Arterioscler Thromb Vasc Biol. 2004;24:155–160. doi: 10.1161/01.ATV.0000104241.44819.10. [DOI] [PubMed] [Google Scholar]
  • 12.Korstanje R, Eriksson P, Samnegard A, Olsson PG, Forsman-Semb K, Sen S, Churchill GA, Rollins J, Harris S, Hamsten A, Paigen B. Locating Ath8, a locus for murine atherosclerosis susceptibility and testing several of its candidate genes in mice and humans. Atherosclerosis. 2004;177:443–450. doi: 10.1016/j.atherosclerosis.2004.08.006. [DOI] [PubMed] [Google Scholar]
  • 13.Stylianou IM, Korstanje R, Li R, Sheehan S, Paigen B, Churchill GA. Quantitative trait locus analysis for obesity reveals multiple networks of interacting loci. Mamm Genome. 2006;17:22–36. doi: 10.1007/s00335-005-0091-2. [DOI] [PubMed] [Google Scholar]
  • 14.Beamer WG, Shultz KL, Donahue LR, Churchill GA, Sen S, Wergedal JR, Baylink DJ, Rosen CJ. Quantitative trait loci for femoral and lumbar vertebral bone mineral density in C57BL/6J and C3H/HeJ inbred strains of mice. J Bone Miner Res. 2001;16:1195–1206. doi: 10.1359/jbmr.2001.16.7.1195. [DOI] [PubMed] [Google Scholar]
  • 15.Masinde GL, Li X, Gu W, Wergedal J, Mohan S, Baylink DJ. Quantitative trait loci for bone density in mice: The genes determining total skeletal density and femur density show little overlap in F2 mice. Calcif Tissue Int. 2002;71:421–428. doi: 10.1007/s00223-001-1113-z. [DOI] [PubMed] [Google Scholar]
  • 16.DiPetrillo K, Wang X, Stylianou IM, Paigen B. Bioinformatics toolbox for narrowing rodent quantitative trait loci. Trends Genet. 2005;21:683–692. doi: 10.1016/j.tig.2005.09.008. [DOI] [PubMed] [Google Scholar]
  • 17.Nishina PM, Lowe S, Verstuyft J, Naggert JK, Kuypers FA, Paigen B. Effects of dietary fats from plant and animal sources on diet induced fatty streak lesions in C57BL/6J mice. J Lipid Res. 1993;34:1413–1422. [PubMed] [Google Scholar]
  • 18.Nishina PM, Verstuyft J, Paigen B. Synthetic low and high fat diets for the study of atherosclerosis in the mouse. J Lipid Res. 1990;31:859–869. [PubMed] [Google Scholar]
  • 19.Nagy TR, Clair AL. Precision and accuracy of dual-energy X-ray absorptiometry for determining in vivo body composition of mice. Obes Res. 2000;8:392–398. doi: 10.1038/oby.2000.47. [DOI] [PubMed] [Google Scholar]
  • 20.Broman KW, Wu H, Sen S, Churchill GA. R/qtl: QTL mapping in experimental crosses. Bioinformatics. 2003;19:889–890. doi: 10.1093/bioinformatics/btg112. [DOI] [PubMed] [Google Scholar]
  • 21.Sen S, Churchill GA. A statistical framework for quantitative trait mapping. Genetics. 2001;159:371–387. doi: 10.1093/genetics/159.1.371. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Churchill GA, Doerge RW. Empirical threshold values for quantitative trait mapping. Genetics. 1994;138:963–971. doi: 10.1093/genetics/138.3.963. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Solberg LC, Baum AE, Ahmadiyeh N, Shimomura K, Li R, Turek FW, Churchill GA, Takahashi JS, Redei EE. Sex- and lineage-specific inheritance of depression-like behavior in the rat. Mamm Genome. 2004;15:648–662. doi: 10.1007/s00335-004-2326-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Abiola O, Angel JM, Avner P, Bachmanov AA, Belknap JK, Bennett B, Blankenhorn EP, Blizard DA, Bolivar V, Brockmann GA, Buck KJ, Bureau JF, Casley WL, Chesler EJ, Cheverud JM, Churchill GA, Cook M, Crabbe JC, Crusio WE, Darvasi A, de Haan G, Dermant P, Doerge RW, Elliot RW, Farber CR, Flaherty L, Flint J, Gershenfeld H, Gibson JP, Gu J, Gu W, Himmelbauer H, Hitzemann R, Hsu HC, Hunter K, Iraqi FF, Jansen RC, Johnson TE, Jones BC, Kempermann G, Lammert F, Lu L, Manly KF, Matthews DB, Medrano JF, Mehrabian M, Mittlemann G, Mock BA, Mogil JS, Montagutelli X, Morahan G, Mountz JD, Nagase H, Nowakowski RS, O'Hara BF, Osadchuk AV, Paigen B, Palmer AA, Peirce JL, Pomp D, Rosemann M, Rosen GD, Schalkwyk LC, Seltzer Z, Settle S, Shimomura K, Shou S, Sikela JM, Siracusa LD, Spearow JL, Teuscher C, Threadgill DW, Toth LA, Toye AA, Vadasz C, Van Zant G, Wakeland E, Williams RW, Zhang HG, Zou F. The nature and identification of quantitative trait loci: A community's view. Nat Rev Genet. 2003;4:911–916. doi: 10.1038/nrg1206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Wade CM, Kulbokas EJ, III, Kirby AW, Zody MC, Mullikin JC, Lander ES, Lindblad-Toh K, Daly MJ. The mosaic structure of variation in the laboratory mouse genome. Nature. 2002;420:574–578. doi: 10.1038/nature01252. [DOI] [PubMed] [Google Scholar]
  • 26.Park YG, Clifford R, Buetow KH, Hunter KW. Multiple cross and inbred strain haplotype mapping of complex-trait candidate genes. Genome Res. 2003;13:118–121. doi: 10.1101/gr.786403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Malik TH, Von Stechow D, Bronson RT, Shivdasani RA. Deletion of the GATA domain of TRPS1 causes an absence of facial hair and provides new insights into the bone disorder in inherited tricho-rhino-phalangeal syndromes. Mol Cell Biol. 2002;22:8592–8600. doi: 10.1128/MCB.22.24.8592-8600.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Boling EP. Gender and osteoporosis: Similarities and sex-specific differences. J Gend Specif Med. 2001;4:36–43. [PubMed] [Google Scholar]
  • 29.Ruff CB, Hayes WC. Sex differences in age-related remodeling of the femur and tibia. J Orthop Res. 1988;6:886–896. doi: 10.1002/jor.1100060613. [DOI] [PubMed] [Google Scholar]
  • 30.Parhami F, Tintut Y, Beamer WG, Gharavi N, Goodman W, Demer LL. Atherogenic high-fat diet reduces bone mineralization in mice. J Bone Miner Res. 2001;16:182–188. doi: 10.1359/jbmr.2001.16.1.182. [DOI] [PubMed] [Google Scholar]
  • 31.Tintut Y, Morony S, Demer LL. Hyperlipidemia promotes osteoclastic potential of bone marrow cells ex vivo. Arterioscler Thromb Vasc Biol. 2004;24:e6–10. doi: 10.1161/01.ATV.0000112023.62695.7f. [DOI] [PubMed] [Google Scholar]
  • 32.Koller DL, Alam I, Sun Q, Liu L, Fishburn T, Carr LG, Econs MJ, Foroud T, Turner CH. Genome screen for bone mineral density phenotypes in Fisher 344 and Lewis rat strains. Mamm Genome. 2005;16:578–586. doi: 10.1007/s00335-004-2459-0. [DOI] [PubMed] [Google Scholar]
  • 33.Ioannidis JP, Ng MY, Sham PC, Zintzaras E, Lewis CM, Deng HW, Econs MJ, Karasik D, Devoto M, Kammerer CM, Spector T, Andrew T, Cupples LA, Duncan EL, Foroud T, Kiel DP, Koller D, Langdahl B, Mitchell BD, Peacock M, Recker R, Shen H, Sol-Church K, Spotila LD, Uitterlinden AG, Wilson SG, Kung AW, Ralston SH. Meta-analysis of genome-wide scans provides evidence for sex- and site-specific regulation of bone mass. J Bone Miner Res. 2007;22:173–183. doi: 10.1359/jbmr.060806. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Lee YH, Rho YH, Choi SJ, Ji JD, Song GG. Meta-analysis of genome-wide linkage studies for bone mineral density. J Hum Genet. 2006;51:480–486. doi: 10.1007/s10038-006-0390-9. [DOI] [PubMed] [Google Scholar]
  • 35.Alam I, Sun Q, Liu L, Koller DL, Fishburn T, Carr LG, Econs MJ, Foroud T, Turner CH. Identification of a quantitative trait locus on rat chromosome 4 that is strongly linked to femoral neck structure and strength. Bone. 2006;39:93–99. doi: 10.1016/j.bone.2005.12.009. [DOI] [PubMed] [Google Scholar]
  • 36.Karasik D, Cupples LA, Hannan MT, Kiel DP. Genome screen for a combined bone phenotype using principal component analysis: The Framingham study. Bone. 2004;34:547–556. doi: 10.1016/j.bone.2003.11.017. [DOI] [PubMed] [Google Scholar]
  • 37.Kim BH, Cai X, Vaughn JN, von Arnim AG. On the functions of the h subunit of eukaryotic initiation factor 3 in late stages of translation initiation. Genome Biol. 2007;8:R60. doi: 10.1186/gb-2007-8-4-r60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Giedion A, Burdea M, Fruchter Z, Meloni T, Trosc V. Autosomal-dominant transmission of the tricho-rhino-phalangeal syndrome. Report of 4 unrelated families, review of 60 cases. Helv Paediatr Acta. 1973;28:249–259. [PubMed] [Google Scholar]
  • 39.Kiel DP, Demissie S, Dupuis J, Lunetta KL, Murabito JM, Karasik D. Genome-wide association with bone mass and geometry in the Framingham Heart Study. BMC Med Genet. 2007;8(Suppl 1):S14. doi: 10.1186/1471-2350-8-S1-S14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Beamer WG, Shultz KL, Churchill GA, Frankel WN, Baylink DJ, Rosen CJ, Donahue LR. Quantitative trait loci for bone density in C57BL/6J and CAST/EiJ inbred mice. Mamm Genome. 1999;10:1043–1049. doi: 10.1007/s003359901159. [DOI] [PubMed] [Google Scholar]
  • 41.Drake TA, Schadt E, Hannani K, Kabo JM, Krass K, Colinayo V, Greaser LE, III, Goldin J, Lusis AJ. Genetic loci determining bone density in mice with diet-induced atherosclerosis. Physiol Genomics. 2001;5:205–215. doi: 10.1152/physiolgenomics.2001.5.4.205. [DOI] [PubMed] [Google Scholar]

Articles from Journal of Bone and Mineral Research are provided here courtesy of Oxford University Press

RESOURCES