The authors state that they have no conflicts of interest.
Published online on December 29, 2008
A genome-wide linkage analysis to identify quantitative trait loci (QTLs) for bone phenotypes was performed in an F2 intercross of inbred spontaneously type 2 diabetic GK and normoglycemic F344 rats (108 males and 98 females). The aim of the study was to locate genome regions with candidate genes affecting trabecular and cortical bone and to investigate the effects of sex and reciprocal cross. pQCT was used to determine tibial bone phenotypes in the F2 rats, comprising reciprocal crosses with divergent mitochondrial (mt) DNA. Sex and reciprocal cross-separated QTL analyses were performed followed by assessment of specific interactions. Four genome-wide significant QTLs linked to either cortical vBMD, tibia length, body length, or metaphyseal area were identified in males on chromosomes (chr) 1, 8, and 15. In females, three significant QTLs linked to cortical BMC or metaphyseal total vBMD were identified on chr 1 and 2. Several additional suggestive loci for trabecular and cortical traits were detected in both males and females. Four female-specific QTLs on chr 2, 3, 5, and 10 and four reciprocal cross-specific QTLs on chr 1, 10, and 18 were identified, suggesting that both sex and mt genotype influence the expression of bone phenotypes.
Osteoporosis is a multifactorial bone disease associated with reduced BMD and characterized by increased bone fragility and fracture risk. Although BMD is a major component determining the ability of bone to resist fracture, other traits such as bone size and skeletal macrostructure, bone morphology, cortical geometry, and trabecular microarchitecture affect bone strength and may therefore influence susceptibility to fracture through different mechanisms.(1)
Osteoporosis has been associated with a number of conditions, including diabetes mellitus. It has been suggested that insulin has anabolic effects on bone, resulting in a higher BMD.(2) Hence, in the hyperinsulinemic phase of type 2 diabetes, the effects on bone may be anabolic.(3,4) Conversely, type 1 diabetes and the late hypoinsulinemic phase of type 2 diabetes can be associated with an increased fracture risk.(5–7) Furthermore, recent evidence suggests that bone participates in the regulation of glucose homeostasis and insulin sensitivity through osteocalcin, an osteoblast-specific secreted protein.(8) Hence, it is of interest to further explore the genetic regulation of bone in the presence of diabetes.
About 70% of the variance in BMD in human populations is genetically determined, and several studies have suggested that genes regulating BMD act in a sex- and site-specific manner.(9,10) However, despite extensive research using linkage analyses and association studies in humans, identification of bone regulatory genes has proven a difficult task.
One alternative approach is to use animal models to identify genomic segments to discover novel candidate genes. Genetic analyses of inbred strains of mice have identified several genome regions linked to different skeletal phenotypes,(11,12) with some reports suggesting a possible existence of sex-specific genetic regulation of BMD.(13,14) Rat models are less commonly used in genetic studies of bone,(15–17) but offer advantages over mice because of larger bones, allowing more precise structural and biomechanical measurements. The GK rat is a well-established model of multifactorial type 2 diabetes that can be used for studies of bone and diabetes. Genetic linkage analyses have identified and characterized several genome regions affecting diabetes-related phenotypes in the GK rat.(18–20) Furthermore, bone changes (e.g., increased moment of inertia and loss of trabecular volumetric BMD in tibia) have been detected in GK, making this rat well suited for studies of interactions between bone and type 2 diabetes.(21,22) The nondiabetic F344 rat strain develops osteopenia similar to humans and is therefore useful for studying genetic influences on bone strength and structure.(23,24) Bone-related phenotypes of the two parental strains have been characterized and are well suited for identification of segregating traits.(19,22,25)
Mitochondria play a key role in energy metabolism and require coordinated expression of nuclear and mitochondrial genes. Defects in the components responsible for the cross-talk between the nuclear and mitochondrial genomes can contribute to diseases in energy homeostasis. Pravenec et al.(26) showed that mitochondrial genotype can influence risk factors for type 2 diabetes. A nuclear quantitative trait locus (QTL) interacting with mtDNA in hearing impairment in mice was reported using reciprocal crosses carrying divergent mtDNA.(27) Varanasi and Datta(28) detected high levels of mtDNA deletions in cortical bone from middle-aged and elderly humans, suggesting a possible involvement of mitochondria in bone loss. However, reports evaluating the effects of mitochondrial genotype and its interaction with the nuclear genome are lacking for bone phenotypes.
The aim of this study is to identify genome regions linked to trabecular and cortical bone phenotypes assessed by pQCT in an F2 intercross of inbred GK and F344 rats and to determine whether sex and reciprocal cross influence the phenotypic manifestations.
MATERIALS AND METHODS
GK/KyoSwe and F344/DuCrlSwe rats were maintained by brother-sister breeding. Two separate F2 intercrosses were generated: one originating from grandmaternal GK and grandpaternal F344 (cross 1; GK female × F344 male; F1) and the other from grandmaternal F344 and grandpaternal GK (cross 2; F344 female × GK male; F1). The two groups of reciprocal F1 progeny were mated separately to yield two reciprocal F2 populations. From a genetic standpoint, the reciprocal crosses differ by their mitochondrial genotypes as inherited from the founding female; all cross 1 progeny carry GK mitochondrial genotype, and F2 males carry the Y chromosome (chr) from F344, whereas females can only be homozygous for GK alleles located on chr X; all cross 2 progeny carry F344 mitochondrial genotype, and F2 males carry the Y chromosome from GK, whereas females can only be homozygous for F344 alleles on chr X.
All rats with free access to tap water and standard rodent chow were kept in the same room with controlled temperature and humidity and a 12-h light/dark cycle with the same number of rats per cage. To mimic the development of ailments associated with the consumption of a high-fat diet, the F2 progeny were fed a modified fat-enriched diet (commercial rodent chow supplemented with 2% cholesterol, 20% olive oil, and 0.5% bile acid; Lactamin AB, Linköping, Sweden) during their last 3 mo of life. At a mean age of 215 days, the progeny were killed, and body length from nose tip to tail base and body weight were recorded. For skeletal phenotypes, the left tibia was dissected free of fat and muscle, placed in 70% ethanol, and stored at room temperature. Permit was obtained from the local Animal Ethics Committee.
Left tibias from 108 male and 98 female F2 rats were measured by pQCT using a Stratec XCT Research M (Norland, Fort Atkinsson, WI, USA) with a voxel resolution of 70 μm modified for use on small bone specimens (Software 5.40B). The length of the tibia was measured using a caliper. Two pQCT scans were performed: one proximal metaphyseal and one mid-diaphyseal. For trabecular bone analysis, one scan at a distance equal to 5% of the length in the distal direction from the proximal tibia growth plate was performed (proximal metaphysis). The trabecular bone region was defined as the inner 45% of this area and generated trabecular vBMD (g/cm3), total metaphyseal vBMD (g/cm3), and cross-sectional area (CSA, mm2) of the whole proximal metaphyseal. Regarding trabecular vBMD, the voxel size exceeded the trabecular size; hence, we chose to exclude it for the calculations. Cortical diaphyseal bone was measured at a position 40% of the total bone length in the distal direction from the proximal tibia growth plate and provided the following phenotypes: cortical vBMD (g/cm3), cortical thickness (mm), cortical BMC (mg/mm), cortical CSA (mm2), periosteal (PC, mm), and endosteal circumferences (EC, mm). Biomechanical strength was estimated from the cortical diaphyseal data: cross-sectional moment of inertia (IP, mm4) and cross-sectional moment of resistance (RP, mm3). Precision and repeatability of the pQCT measurements were determined by conducting three repeat measurements after repositioning on each of five randomly chosen rats. The measurements were analyzed by the same operator, and the percent CV (%CV) was calculated for each phenotype: total metaphyseal BMD, 0.5%; trabecular vBMD, 2.8%; metaphyseal CSA, 3.0%; cortical vBMD, 0.2%; cortical thickness, 0.8%; cortical BMC, 0.6%; cortical CSA, 0.7%; periosteal area, 0.2%; endosteal area, 0.5%; IP, 1.1%; RP, 1.0%.
DNA isolation and microsatellite marker genotyping
Genomic DNA was purified from rat liver using the QIAmp DNA mini kit (QIAGEN, Valencia, CA, USA). Genotyping of the F2 progeny was made by microsatellite markers.(29) Genotypes in the F2 progeny were tested for Mendelian segregation. A genetic linkage map of the 20 rat autosomes was generated using MAPMAKER/EXP,(30) and a map of the X chromosome using R/qtl.(31) The total genetic map length was 1784.2 cM (autosomes, 1697.9 cM; 86.3 cM, X chromosome). The average distance between markers was 9.3 cM.
All phenotypes were normally distributed or log-transformed to a normal distribution. To compare the trabecular and cortical bone phenotypes between males and females and between the reciprocal crosses, one-way ANOVA was used. The level of significance was set at p < 0.05. Unless stated, p values are nominal. Phenotypes were adjusted for cross, age, litter size, and body weight using regression analysis. Residuals were checked for normality and, after removing one outlier, used in the QTL analysis. To account for bone quality differences between sexes, the residuals were computed separately for each sex.
Linkage analysis of quantitative traits
Genetic linkage analysis was performed for each sex separately. To identify possible interaction differences between loci in the nuclear genome and reciprocal cross, the sex separated F2 progeny was separated on the basis of reciprocal cross. QTLs on autosomes were identified using MAP MANAGER/QTX v. b20.(32) The X chromosome was also included in our linkage analysis using R/qtl.(31) Mapping QTLs on the X chromosome was performed on each sex separately without further separation. Permutation tests were performed to establish genome-wide significance levels by randomization of the phenotypic data in relation to genotypic data.(33) Significant (i.e., genome-wide false-positive rate of 5%) and suggestive (i.e., genome-wide false-positive rate of 63%) linkage was used to establish genome-wide thresholds(34,35): the likelihood ratio statistic (LRS) for suggestive linkage range 11.1–11.2 (LOD = 2.4) and for significant linkage LRS range 17.9–18.3 (LOD = 3.9–4.0). Co-localization of pQCT phenotypes to the same genome regions was defined as one locus and named Pqctm1–6 for males and Pqctf 1–12 for females.
Evaluation of sex- and cross-specific QTLs
To identify sex-specific QTLs, the LOD score differences between males and females across the genome were assessed (ΔLODsex score). We applied a permutation method to evaluate sex-specific QTLs, where thresholds for sex-specific QTLs were established using two randomly selected equal sized subsets of males and females.(36) The randomization was conducted within each cross. Subsequently, bone phenotypes in the two subsets were permutated to calculate ΔLODsex scores across the genome. The genetic markers on the X chromosome were not included in the permutation tests. Genome-wide suggestive significance (i.e., genome-wide false-positive rate of 63%) for sex specificity was met if the ΔLODsex score at a locus was higher than the expected maximum ΔLODsex score based on 1000 permutations of the ΔLODsex score across the 20 autosomes: the average ΔLODsex score for a suggestive sex specific QTL was 2.4.
Within each sex, subsequent cross-separated linkage analyses were conducted to identify cross-specific QTLs. The LOD score differences between cross 1 and cross 2 (ΔLODcross score) across the genome were examined within each sex. Thresholds of the cross-specific QTL were computed by permutation using two randomly selected equal-sized cross 1 and cross 2 subsets within each sex. Genome-wide suggestive significance for cross-specificity was met if the ΔLODcross score (i.e., the LOD score differences between cross 1 and cross 2) at a locus was higher than the expected maximum ΔLODcross score for the random subsets, which was calculated by the maximum ΔLODcross score across the 20 autosomes for each of the 1000 permutations: the average ΔLODcross score for suggestive cross-specific QTL was 2.5.
For further evaluation of sex-specific and cross-specific QTLs identified with the ΔLODsex/ΔLODcross method, likelihood ratio tests were performed comparing a full model with a QTL × sex interaction term/cross interaction term (full model 1) and a reduced model (reduced model 1) without the interaction term, using both male and female data for sex interaction and each sex separately for cross interaction.
where Y is the phenotype; β0 is the mean; C is a vector of regression coefficients for cross/- (i.e., sex-specific QTLs only), age, litter size, and body weight; Z is a matrix of regression variables for cross/-, age, litter size, and body weight; β1 is a regression coefficient for sex/cross; X is the regression variable for sex/cross; β2 is a regression coefficient for QTL; QTL is a regression variable for QTL; β3 is a regression coefficient for the QTL-by-sex interaction/QTL-by-cross interaction; QTL × X is the regression variable for the QTL-by-sex interaction/QTL-by-cross interaction; and e is the residual error.
Residuals of each phenotype were examined for normality with normal probability plots. The level of significance for a specific QTL interaction with sex or cross was set at p < 0.05.
We conducted the statistical power calculation using the method of Lynch and Walsh(37) under different assumptions regarding the sample size. We assumed that QTL acts additively in this power calculation. The fraction of phenotypic variance explained by QTLs (i.e., R2) was considered as effect size of QTL. The observed values of R2 for the four traits (metaphyseal total vBMD, metaphyseal total CSA, cortical PC, and EC) for the reciprocal cross-specific suggestive linkages in females span a narrow range of values (i.e., 0.20–0.28; average, 0.23). When a LOD score of 2.4 is used to control false-positive detection of linkage, the calculations indicate that a sample size of 52 would be necessary to achieve 80% statistical power for detecting a QTL with an R2 value of 0.25.
Phenotypic differences between sexes
Strong sexual dimorphism was observed for all but one trait in the F2 progeny (Table 1). In females, the cross-sectional area of the tibial metaphyseal bone was 26% smaller (p < 0.0001), whereas the total BMD was 18% higher compared with males (p < 0.0001). When BMD measurement was restricted to the trabecular compartment of the metaphysis, BMD was 49% higher in females (p < 0.0001), indicating that the main BMD difference was observed in trabecular bone. The tibial diaphyseal bone was larger and thicker in males, (ranging from 13% [p < 0.0001] to 28% [p < 0.0001]). As a consequence, the biomechanical estimates IP and RP were considerably lower for female rats 53% (p < 0.0001) and 42% (p < 0.0001). Cortical vBMD was similar in the sexes.
Table Table 1.. pQCT Data From Tibia of Female and Male F2 Progeny Generated From GK and F344 in Two Reciprocal Crosses
Phenotypic differences between reciprocal crosses
Significant phenotypic differences were also observed between F2 progeny from the two reciprocal crosses (Table 1). In females, total vBMD were 6% (p < 0.002) lower in cross 2 progeny (F344 grandmaternal origin), whereas male cross 2 progeny had 5% (p < 0.001) lower total vBMD compared with cross 1 (GK grandmaternal origin). Cortical thickness was also significantly lower for cross 2 progeny of both sexes (females, 3.4% [p < 0.03]; males, 2.9% [p < 0.03]). In males, cortical vBMD was 1% lower (p < 0.02) for cross 2 progeny (Table 1). To study the mechanisms behind the observed sex and cross effects, we separated the data by sex, followed by cross, and performed linkage analyses.
QTLs in males
Results from the genome-wide QTL analysis in males are summarized in Table 2. Two separate QTLs on chr 1 regulated diaphyseal traits, where three traits attained genome-wide significance. The Pqctm1 locus linked to both length of tibia and body length, indicating that it mainly represents a locus influencing the size of the animal in general. Cortical vBMD, a more specific bone quality indicator, was also linked to the same locus (LOD = 4.4). The other locus (Pqctm2) was located in a broad genome region (36–93 cM). Several traits characterizing the diaphysis of tibia (cortical PC, cortical EC, cortical CSA, cortical BMC, IP, and RP) mapped to this region at the suggestive level. Also, tibia length co-mapped to Pqctm2 at the significant level (LOD = 4.3). The third genome-wide significant male QTL (Pqctm4) for tibia length was located on chr 8 (LOD = 4.1). Metaphyseal total cross-sectional area linked to Pqctm6 at the genome-wide significant level (LOD = 3.9), and the same trait linked to Pqctm3 at the suggestive level. Although several male-specific QTLs reached genome-wide suggestive significance with ΔLODsex evaluation (results not shown), none of them was validated for male-specific interaction in the likelihood ratio tests.
Table Table 2.. Tibial Bone-Related QTLs Identified in Male Rats
QTLs in females
Results from the linkage analysis of females are presented in Table 3. Several, mainly diaphyseal, phenotypes characterizing size and mechanical strength (tibia length, cortical BMC, cortical CSA, cortical PC, IP, and RP) co-localized to Pqctf1 on chr 1. Cortical BMC linked to this QTL at the significant level (LOD = 3.9). Two significant and three suggestive QTLs linked to metaphyseal total vBMD were identified on chr 1 (LOD = 4.0), 2 (LOD = 4.1), 3 (LOD = 3.6), 5 (LOD = 3.1), and 10 (LOD = 3.2). On chr 8 and 14, loci Pqctf8 and Pqctf11 were linked at the suggestive level to cortical PC, cortical EC, RP, and body length.
Table Table 3.. Tibial Bone-Related QTLs Identified in Female Rats
Several QTLs showed a first indication of female specificity in the evaluation of ΔLODsex (ΔLODsex > 2.4, result not shown). These loci were further evaluated for female-specific interaction with the likelihood ratio test. Metaphyseal total vBMD was linked to four loci with female-specific interaction: chr 2 (LR = 8.7, p = 0.013; Fig. 1A), chr 3 (LR = 7.3, p = 0.026; Fig. 1B), chr 5 (LR = 13.5, p = 0.001), and chr 10 (LR = 9.1, p = 0.01). Pqctf2 on chr 1 showed both female specificity with the ΔLODsex evaluation and significant QTL × sex interaction (LR = 12.1, p = 0.002), but diaphyseal phenotypes map to a partly overlapping locus (Pqctm2) in males. Consequently, the locus only exhibits female-specific linkage of this particular trait.
Reciprocal cross-separated QTL analysis in males and females
The reciprocal cross-separated QTL analysis in each sex showed new significant and suggestive QTLs. In male F2 rats, one new locus (Pqctm5) was detected after separation (Table 2). This locus on chr 10 linked to cortical PC, cortical CSA, cortical BMC, IP, and RP only in cross 1 progeny and were not detected at all in the aggregated analysis with males from both reciprocal crosses.
In females, Pqctf1 on chr 1 linked to metaphyseal total vBMD at the genome-wide significant level in cross 2 carrying F344 mtDNA (LOD = 4.5). Pqctf5 (only diaphyseal phenotypes), Pqctf7, Pqctf8, and Pqctf9 showed higher linkage in females from cross 1 carrying GK mtDNA (Table 3).
In the ΔLODcross evaluation, several QTLs met the criteria for genome-wide suggestive significance (ΔLOD ≥ 2.5, results not shown), and the likelihood ratio tests validated a cross-specific interaction in females at four of them: metaphyseal total vBMD to Pqctf1 on chr 1 (LR = 7.0, p = 0.03; Fig. 1C), metaphyseal total CSA to Pqctf3 on chr 1 (LR = 8.3, p = 0.016), cortical PC (LR = 7.5, p = 0.023) and EC (LR = 8.1, p = 0.017) to Pqctf10 on chr 10, and metaphyseal total CSA to Pqctf12 on chr 18 (LR = 11.1, p = 0.004; Fig. 1D).
Linkage results using the aggregated data from all F2 progeny are shown in Tables 2 and 3 to allow comparison with the results from the separated groups.
Effect of GK allele
Male rats homozygous for the GK allele had higher genotypic mean values at the QTLs on chr 1, 8, 10, and 15, whereas the opposite was observed for Pqctm5 on chr 5. In females, the GK allele had an increasing effect on the genotypic mean values at the QTLs on chr 1, 2, 8, 10 (Pqctf10), and 14, whereas it had a decreasing effect on chr 3, 5, 7, and 10 (Pqctf9) (results not shown).
In this study, we identified chromosomal regions linked to trabecular and cortical bone phenotypes in a population of F2 rats that concomitantly segregate type 2–like diabetes. To our knowledge, this is the first study to show a specific interaction between reciprocal cross and QTL in the nuclear genome for bone-associated phenotypes.
The >100 variant positions in mtDNA from GK and F344 includes 12 nonsynonymous amino acid changes in proteins required for ATP synthesis and variants in the regulatory elements in the D-loop. Of these, nine nonsynonymous SNPs were found in the genes (i.e., five in ND2, three in ND4, and one in ND6) for the OXPHOS complex I, whereas the remaining were found in each of three genes encoding COX2, ATP6, and CYTB, respectively.(38) This supports the involvement of mtDNA as a major factor behind the observed reciprocal cross effect. There are additional genetic factors that in theory could explain phenotypic difference between two reciprocal crosses: QTLs on X and Y chromosomes and genomic imprinting.(39) In this study, an X chr–linked QTL is not a likely explanation to the cross effect because no locus was detected on chr X. The lack of detection of QTLs on the X chromosome may be related to sample size; however, other explanations are possible: two closely linked QTLs may act in opposite direction or epistatic interactions between genes on the X chromosome and QTLs on autosomes/mitochondrial genome can contribute to the phenotypic variations. Furthermore, the four QTLs interacting with the reciprocal cross were all found in females, excluding that an important Y-linked QTL could explain a substantial proportion of the cross effect. We cannot rule out a possible influence of genomic imprinting in the observed QTLs. This might, however, be possible to address with new statistical methods in the future. In the view of the large mtDNA sequence differences between the two rat strains and reported influence of mtDNA variation between inbred conplastic rat strains,(26) the most plausible explanation for the reciprocal cross effect resides in interactions between nuclear genes and mtDNA variation.
Our analysis of the reciprocal cross effect showed four QTLs with a significant cross-specific interaction (Pqctf1, Pqctf3, Pqctf10, and Pqctf12). These QTLs were identified in females and linked to metaphyseal bone phenotypes (Table 3). Interestingly, these QTLs were only expressed in cross 2, indicating that segregation of the QTLs can only be observed in the presence of F344 mtDNA. This result, together with the likelihood ratio test, provides a solid basis to conclude that unique interactions among nuclear genes and haplotypes of mtDNA occur and influence bone phenotypes at significant levels.
The molecular nature of the interactions remains to be fully explored, but the identified cross-talk between mtDNA and the nuclear genome has enabled inclusion of a new group of genes—the mitochondrial proteins encoded by the nuclear genome—as candidate genes for bone disorders. Mitochondrial genes can be found in these four cross-interacting QTLs (e.g., a mitochondrial transcription factor B1 [Tfb1m] located within Pqctf1, two mitochondrial uncoupling proteins [Ucp2 and 3] and a mitochondrial ribosomal protein [Mrpl17] located within Pqctf3, and ATP synthase F0 [Atp5h], encoding a component in the fifth oxidative phosphorylation complex within Pqctf10.
Size-related phenotypes of cortical bone differed as expected between males and females. In females, the tibial diaphyseal bone was shorter and thinner than in males, and the cross-sectional area of the metaphysis was 26% smaller (Table 1). Accordingly, the identification of metaphyseal loci was slightly more powerful in females and vice versa for diaphyseal loci in males. The sex-separated QTL analysis identified three significant female-specific loci (Pqctf1, Pqctf2, Pqctf3). In males, the majority of significant QTLs were linked to bone size and cortical phenotypes: Pqctm1, Pqctm2, Pqctm4, and Pqctm6. Using the likelihood ratio tests, we could validate four QTLs for metaphyseal total vBMD showing sex-specific interaction, and all of them were identified in females. As for the mechanisms underlying sex differences in bone, several genetic and environmental factors might be involved. These include the presence of type 2–like diabetes in the F2 progeny, because males from intercrosses between GK and F344 have, on average, higher blood glucose and insulin concentrations than females.(40) Another possibility for the observed sex difference in bone could be the exposure to estrogen in females, which is known to stabilize trabecular bone.(41)
A broad region on chromosome 1 (10–93 cM) showed linkage to several parameters of cortical bone size in both sexes (CSA, BMC, IP, RP, and PC). The region overlaps several QTLs previously identified in (F344 × LEW) F2 rats.(16,24) The osteoporosis candidate genes TGFB1(42) and estrogen receptor α (ESR1)(43) map within this locus. Additionally, this region also overlaps QTLs for fasting glucose,(18,19) suggesting a potential genetic association between bone-related phenotypes and traits affecting type 2 diabetes.
The GK allele was associated with larger values in the majority of cortical and bone size phenotypes. On the other hand, the GK allele resulted in lower genotypic values for several metaphyseal phenotypes, suggesting that the risk of fracture in diabetes varies according to the specific subregions of bone. Interestingly, Ahmad et al.(21,22,44) observed similar bone changes, as well as IGF system abnormalities in the diabetic GK rat and suggested that the IGF system is associated with bone loss that seems to manifest differently in trabecular and cortical bone. In humans, diabetes mellitus has been associated with increased fracture risk in the metaphyseal region of hip, proximal humerus, and tibia.(7)
In conclusion, our study showed that mtDNA variation is the most plausible cause for the observed interactions between nuclear QTLs and reciprocal cross in bone phenotypes. Thus, genes required for the functional integrity of mitochondria that reside in interacting loci constitute a new group of candidate genes in bone fragility disorders. We also lend further support to the notion that QTLs interact specifically with sex.(45,46) This animal model for co-segregating type 2 diabetes and bone phenotypes can be used to study pathophysiological processes and to identify candidate genes for two increasingly common diseases. However, our findings pertain to this specific animal model and we cannot with certainty extend the conclusions to be valid for other species. Neither can it be generalized to the entire skeleton because we only measured specific regions of the tibial bone. Hence, follow-up studies with a large sample size are valuable to validate the findings.
The authors thank laboratory technician Anette Hansevi at the Department of Internal Medicine, Sahlgrenska University Hospital in Gothenburg, for the pQCT measurements. This study was supported by grants from the Swedish Research Council (K2006-72X-09109-17-3, 2006-73X-14691-04-3) and a Linné grant to Lund University Diabetes Center, Knut och Alice Wallenbergs Stiftelse, Malmö University Hospital Research Foundations, the Medical Faculty of Lund University, Novo Nordisk Foundation, Svenska Diabetesförbundets Forskningsfond, Albert Påhlsson Research Foundation, Greta and Johan Kock Foundation, and A Osterlund Foundation.