• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. Jan 2007; 175(1): 349–360.
PMCID: PMC1775020

Fine Mapping Reveals Sex Bias in Quantitative Trait Loci Affecting Growth, Skeletal Size and Obesity-Related Traits on Mouse Chromosomes 2 and 11

Abstract

Previous speed congenic analysis has suggested that the expression of growth and obesity quantitative trait loci (QTL) on distal mouse chromosomes (MMU) 2 and 11, segregating between the CAST/EiJ (CAST) and C57BL/6J-hg/hg (HG) strains, is dependent on sex. To confirm, fine map, and further evaluate QTL × sex interactions, we constructed congenic by recipient F2 crosses for the HG.CAST-(D2Mit329-D2Mit457)N(6) (HG2D) and HG.CAST-(D11Mit260-D11Mit255)N(6) (HG11) congenic strains. Over 700 F2 mice were densely genotyped and phenotyped for a panel of 40 body and organ weight, skeletal length, and obesity-related traits at 9 weeks of age. Linkage analysis revealed 20 QTL affecting a representative subset of phenotypes in HG2DF2 and HG11F2 mice. The effect of sex was quantified by comparing two linear models: the first model included sex as an additive covariate and the second incorporated sex as an additive and an interactive covariate. Of the 20 QTL, 8 were sex biased, sex specific, or sex antagonistic. Most traits were regulated by single QTL; however, two closely linked loci were identified for five traits in HG2DF2 mice. Additionally, the confidence intervals for most QTL were significantly reduced relative to the original mapping results, setting the stage for quantitative trait gene (QTG) discovery. These results highlight the importance of assessing the contribution of sex in complex trait analyses.

ONE approach to discovering genetic variation contributing to common human disease is the use of mouse models to identify quantitative trait loci (QTL) affecting traits of biomedical importance. There are several advantages to using mice for the analysis of complex traits such as a highly controlled environment, availability of inbred strains, and ability to produce genetically defined crosses, just to name a few (Abiola et al. 2003). More importantly, once identified, mouse QTL can be systemically fine mapped until the underlying gene(s) is discovered. Although a plethora of QTL have been discovered (Flint et al. 2005), the transition to the next more important step of identifying quantitative trait genes (QTG) has proven difficult. In spite of this, the number of QTG identified has steadily increased in recent years (Bodnar et al. 2002; Klein et al. 2004; Yalcin et al. 2004; Oliver et al. 2005; Wang et al. 2005; Clee et al. 2006).

In 2001, the identification of growth and carcass composition QTL was reported in an F2 cross between the low-body-weight (mature male body weight of ~15 g) CAST/EiJ (CAST) and high-body-weight (mature male body weight of ~40 g) C57BL/6J-hg/hg (referred to as high growth or HG) mouse strains (Corva et al. 2001). The HG strain lacks expression of the Socs2 gene (a negative inhibitor of growth hormone signaling) due to a spontaneously arising 500-kbp deletion on mouse chromosome (MMU) 10 (Horvat and Medrano 2001; Wong et al. 2002). In the above cross, MMU2 and MMU11 harbored QTL with significant affects on weight gain (Wg2 on MMU2 and Wg4 on MMU11), carcass ash (Cara1 on MMU2 and Cara2 on MMU11), and carcass protein (Carp1 on MMU2 and Carp2 on MMU11). Interestingly, the expression of Wg2 and Carp2 were altered in mutant (lacking Socs2) relative to wild-type (possessing a functional Socs2) F2 mice, indicating that these loci were modifiers of the high-growth (hg) phenotype. Specifically, Wg2 influenced weight gain in an overdominant fashion and was located at 60 cM in mutant mice, while in wild-type mice Wg2 was additive and peaked at 80 cM (Corva et al. 2001). On the other hand, CAST alleles at Carp2 decreased carcass protein in mutant but not wild-type F2 mice (Corva et al. 2001).

Congenic strains possess alleles within a defined chromosomal segment from an inbred donor strain on the genetic background of a second recipient strain (Silver 1995). To decrease the time required to generate congenics, the “speed congenic approach” utilizes genetic markers to increase the rate at which unlinked recipient strain alleles are eliminated (Markel et al. 1997; Wakeland et al. 1997). We have recently used this approach to isolate the aforementioned MMU2 and MMU11 QTL (Farber et al. 2006). Two speed congenic strains, HG.CAST-(D2Mit329-D2Mit457)N(6) (HG2D) and HG.CAST-(D11Mit260-D11Mit255)N(6) (HG11), were developed by introgressing CAST donor regions on an HG background. HG2D mice possessed CAST alleles on MMU2 from 74.9 to 181.2 Mbp and were developed, but not characterized due to reduced fertility in homozygous congenic mice. However, a second strain, HG.CAST-(D2Mit329-D2Mit490)N(6) (HG2M), nested within HG2D and extending from 74.9 to 138.6 Mbp, was characterized and displayed sex-specific alterations in adiposity, among other growth and skeletal length phenotypes (Farber et al. 2006). Similarly, the HG11 congenic (CAST alleles on MMU11 from 61.6 to 114.0 Mbp) displayed sex-specific differences in growth, body length, adiposity, and carcass composition (Farber et al. 2006).

To further dissect QTL × sex interactions apparent on both chromosomes, we chose to pursue the HG2D and HG11 strains in parallel using congenic-by-recipient F2 intercrosses. One of the advantages of this approach is the elimination of any phenotypic contribution of unlinked contaminating alleles, since these regions segregate randomly with respect to donor region genotype in the F2. Additionally, the power to resolve QTL and detect QTL × sex interactions is increased relative to a traditional whole-genome F2, since only QTL within the donor regions are segregating. Therefore, we have implemented this strategy with the objectives of (1) confirming that QTL effects are due to alleles within each donor region, (2) refining the genomic interval harboring each QTL, and (3) characterizing the effect of sex on QTL expression. Our results demonstrate a highly complex inheritance pattern for QTL affecting growth, skeletal size, and obesity-related traits on both chromosomes. This complexity involves multiple QTL affecting the same trait and varying degrees of QTL × sex interactions. These data provide insight into the sexually dimorphic genetic architecture of biomedically important traits and serve as the platform for QTG discovery.

MATERIALS AND METHODS

Mice:

The creation of the HG2D and HG11 speed congenic strains has been previously described (Farber et al. 2006). For this study, congenic-by-recipient F2 crosses were generated by mating a single congenic male from each strain (HG2D and HG11) to multiple recipient C57BL/6J-hg/hg (HG) females. The resulting F1 mice were intermated to generate HG2DF2 (n = 278) and HG11F2 (n = 454) mice. The average litter size was 7.4 ± 0.3 pups for HG2DF1 and 6.9 ± 0.3 pups for HG11F1 females. F2 mice were provided with a normal chow diet (Purina 5008; 23.5% protein, 6.5% fat, 3.3 kcal/g) and water ad libitum. All mice were housed in polycarbonate cages under controlled conditions of temperature (21° ± 2°), humidity (40–70%), and lighting (14 hr light/10 hr dark, lights on at 7 am), and managed according to the guidelines of the American Association for Accreditation of Laboratory Animal Care.

Phenotypic analysis:

F2 mice were weighed at 2, 3, 6, and 9 weeks of age to the nearest 0.1 g. At the time of sacrifice (9 weeks ± 5 days), nonfasted mice were anesthetized under isoflurane and nasal–anal length, nasal–tail length, and tail length (nasal–tail − nasal–anal = tail length) were measured to the nearest millimeter. Blood was collected via the retroorbital sinus from anesthetized mice in separator tubes (Microtainer, Becton Dickinson) containing 8 μl of 0.5 m EDTA. Plasma was isolated by centrifugation for 5 min at 8000 × g and stored at −70° until analysis. Anesthetized mice were sacrificed by decapitation and exsanguination. Fat pads (femoral, gonadal, mesenteric, and retroperitoneal), whole brain, liver, spleen, heart, and kidney were removed and weighed to the nearest milligram. Femurs were also removed from all HG11F2 carcasses, cleaned, and measured to the nearest 0.5 mm using calipers.

QTL influencing plasma lipid levels have been identified on distal MMU2 between B6 and CAST mice (Mehrabian et al. 1998, 2000; Estrada-Smith et al. 2004). Therefore, to identify QTL for these traits in our cross, HG2D plasma profiles for total cholesterol (TC), high-density lipoprotein cholesterol (HDL), cholesterol esters, unesterified cholesterol, free fatty acids, and triglycerides were assayed as described in Mehrabian et al. (1993). Very-low-density and low-density lipoprotein cholesterol levels were calculated by subtracting HDL from TC. A Beckman Glucose Analyzer 2 (Beckman Instruments, Fullerton, CA) was used to measure plasma glucose by the glucose oxidase reaction.

Genotyping:

HG2DF2 and HG11F2 mice were genotyped using published microsatellite markers (Table 1) (Dietrich et al. 1996). DNA was isolated from 1.0- to 2.0-mm tail clips by digesting with proteinase K (Fisher) at 55° in a buffer composed of 0.45% NP40 (Sigma, St. Louis), 0.45% Tween 20 (Fisher), and 1× PCR buffer (Promega, Madison, WI). The product of this digestion was diluted (1:10) in sterile H2O and used for genotyping without further purification. Microsatellite genotyping was performed using standard PCR and gel electrophoresis protocols. Reaction conditions for each marker are listed in Table 1.

TABLE 1
Microsatellite markers typed in HG2DF2 and HG11F2 mice

Statistical analysis:

Basic statistical analysis was preformed using the Rcmdr package of R (Ihaka and Gentleman 1996; Fox 2005). Standard descriptive statistics were computed for all traits to test normality assumptions and to identify putative data entry errors. A simple linear regression model was used to determine the relationship between body weight and fat mass in HG2DF2 mice. A χ2 test was used to determine adherence to Mendelian segregation ratios for all markers.

QTL analysis:

QTL scans within each cross were implemented using the R/qtl package of R (Broman et al. 2003). Initially, sex-averaged genetic maps (using a Kosambi map function) were created using the “est.map” function. Conditional genotype probabilities were then calculated using the “calc.genoprob” function across each congenic donor region at 0.5-cM intervals. As a first approximation, maximum-likelihood (EM algorithm) interval mapping was used to screen for QTL in each sex separately via the “scanone” function, with age at sacrifice and litter size as additive covariates (data not shown) (Lander and Botstein 1989). These analyses, along with trait means for nonrecombinant mice (data not shown), suggested strong sex effects for many traits. Therefore, to account for the effect of sex, we reanalyzed each trait using methods outlined in Solberg et al. (2004). The approach consisted of fitting two separate models using the complete data set; Model_Additive (Model_A) was a linear model with sex, age at sacrifice, and litter size as additive covariates; Model_SexInteraction (Model_SI) used the same additive covariates but included sex as a covariate interacting with QTL genotype. Model_SI accounted for QTL × sex interactions arising from loci that are sex specific (QTL influencing a trait in only one sex), sex biased (QTL with an allelic effect that is more pronounced in one sex), or sex antagonistic (QTL with allelic effects going in opposite directions between the sexes). The difference in LOD score (ΔLOD) between the models (LOD Model_SI–LOD Model_A) was calculated using the “arithscan” function and used to estimate the magnitude of QTL × sex interactions. A small ΔLOD may be biologically meaningful; however, the overall threshold for declaring a significant QTL was higher using Model_SI, due to an increase in the number of estimated parameters. Therefore, we arbitrarily chose ΔLOD ≥ 1 as the cutoff for declaring a QTL to be sex biased. To declare sex specificity, we followed an approach similar to the one used in Solberg et al. (2004). They required the ΔLOD to exceed a nominal P < 0.05 threshold for QTL significance. In our case, the average significance threshold for all traits was LOD = 3.0 (P < 0.05). Therefore, QTL with a ΔLOD ≥ 3.0 were deemed sex specific or sex antagonistic. Significance thresholds for all models were calculated by permuting the observed data 1000 times using the “n.perm” function (Churchill and Doerge 1994). Additionally, 1.5-LOD confidence intervals (C.I.) were calculated using the “lodint” function.

We used linear regression to address the relationship between total fat pad mass (FAT) and weight at 9 weeks of age (9WK). FAT was highly correlated with 9WK in females but not in males (Figure 1), illustrating the sex-dependent relationship between these two traits. Several recent studies have suggested that due to the lack of equality in this relationship between the sexes (Stylianou et al. 2006) and because of induced spurious correlation between the traits (Lang et al. 2005), the use of ratios like adiposity index (fat pad mass/body weight × 100) in QTL analysis may result in statistical artifacts. Therefore, to circumvent potential problems, we chose to adjust FAT (in the HG2DF2 cross) and gonadal fat pad weight (GFP) (in the HG11F2 cross) by including weight at sacrifice (WSAC) and WSAC × sex terms as additive covariates in both Model_A and Model_SI. We also used this same approach to identify QTL affecting organ weight independent of body size.

Figure 1.
Linear regression of FAT on 9WK weight illustrates the strong relationship between fat and body weight in female but not male HG2DF2 mice. R2 is the coefficient of determination and indicates the percentage of variance in FAT explained by 9WK.

A number of traits, primarily in the HG2D cross, appeared to be regulated by multiple QTL. To test this statistically, we used the “fitqtl” function to evaluate two-QTL models. Model fitness was evaluated after adjusting the data for the effects of sex. The two-QTL model was accepted if both QTL terms were significant at the P < 0.05 level.

RESULTS

Development of the HG2DF2 and HG11F2 crosses:

Over 700 congenic-derived F2 mice (n = 278 for HG2DF2 and n = 454 for HG11F2) were produced to characterize QTL on MMU2 and MMU11. In each cross, only CAST donor regions on MMU2 (HG2DF2) and MMU11 (HG11F2) were segregating and each represented <3.5% of the genome. The inheritance of HG and CAST alleles was followed using microsatellite markers. A total of 19 and 11 markers were genotyped in the HG2DF2 and HG11F2 crosses, respectively (Table 1). The average marker spacing was 2.7 cM for HG2DF2 and 3.0 cM for HG11F2. The HG2DF2 congenic spanned 104.4 Mbp or 51.5 cM and the HG11F2 congenic spanned 52.1 Mbp or 33.8 cM (Table 1). P-values for tests of Mendelian segregation are presented in Table 1. All MMU2 markers were inherited in the expected 1:2:1 ratio. In contrast, nearly all HG11F2 markers showed evidence (nominal P-values <0.10) suggestive of skewed inheritance, although none were significant after correcting for multiple comparisons. The underlying cause was a deficiency of CAST/CAST homozygotes in both sexes and may indicate a slight prenatal selection against these mice, although a small number of mice that were not genotyped did not live to sacrifice and may have contributed to the observed deficiency.

QTL analysis:

Measurements on >40 traits were collected in both crosses. Linkage analysis was preformed on all traits; however, we report only results for the traits with the most significant LOD scores. We used this strategy since many of the traits within the four trait categories [growth (body weight), organ weight, skeletal length, and obesity] were highly correlated (data not shown) and reflect pleiotropic effects of a single QTL. For example, in HG2DF2 mice, weight at 6 weeks of age (6WK) and at 9WK and weight gain from 2–6, 2–9, 3–6, 3–9, and 6–9 weeks were affected by the same QTL; therefore, to eliminate redundancy we have presented only results for the traits (6WK and 9WK) with the most significant LOD score peaks. Table 2 lists the statistics for all reported QTL. We have provided the approximate physical location of each QTL by cross-referencing the centimorgan and megabase-pair position of each marker in Table 1. QTL have been named according to Mouse Genome Informatics guidelines as determined by the International Committee on Standardized Genetic Nomenclature for Mice (http://www.informatics.jax.org/mgihome/nomen/index.shtml).

TABLE 2
Growth, organ weight, skeletal length, and obesity-related QTL identified in HG2DF2 and HG11F2 mice

In HG2DF2 mice, the LOD score profiles for 6WK, 9WK, kidney weight (KID), FAT, and TC suggested a minimum of two QTL. To gain statistical support for the presence of distinct loci, we analyzed two-QTL models. Both QTL terms were significant (P < 0.05) in models for 6WK and TC (data not shown), but not for the remaining traits (although the rest were suggestive with P < 0.10). The lack of support for the other traits could be due to the tight linkage between the putative QTL. In light of this, we still chose to report the second QTL for all traits. Our decision is supported by the observation that in all cases, except for the second TC QTL, either the second QTL was sex biased when the first was not or the second QTL was not sex biased when the first was (Table 2). We also have evidence from a separate analysis of B6.CAST MMU2 subcongenics on a wild-type (pure C57BL/6J) background, clearly demonstrating the presence of multiple QTL for weight and adiposity within the 2D donor region (C. R. Farber and J. F. Medrano, unpublished results).

The 1.5-LOD C.I., in centimorgans, is reported for the QTL with the most significant LOD score per trait (Table 2). In addition, it should be noted that C.I.'s for traits with multiple putative QTL may be inflated due to closely linked QTL with similar peak LOD scores [as an example, see LOD score plot for 6WK in HG2DF2 mice (Table 2; Figure 2A)]. In the following sections and figures, the three genotypes are abbreviated as C (CAST/CAST), H (HG/CAST), and B (HG/HG).

Figure 2.
LOD score profiles and genotypic contrasts for body weight in HG2DF2 and HG11F2 mice. LOD score plots are shown with peak position delineated by a horizontal solid bar underneath the QTL name. A dotted line represents the significant LOD threshold for ...

HG2DF2 growth traits:

Two QTL, weight gain QTL 5 (Wg5) and weight gain QTL 6 (Wg6), influencing both 6WK and 9WK, were identified in HG2DF2 mice (Table 2; Figure 2A). Wg5 and Wg6 altered only postnatal growth without influencing weight at 2 and 3 weeks of age (data not shown). Wg5 and Wg6 had peak LOD scores for both traits at 9.0 and 27.5 cM, respectively. Wg5 was unaffected by sex (Table 2; Figure 2B); in contrast, Wg6 displayed male-biased expression (6WK ΔLOD = 1.5; 9WK ΔLOD = 2.0) (Table 2; Figure 2C). Both QTL, especially Wg6, demonstrated a higher dominance effect in males than in females (Table 2). This is presented graphically in Figure 2C, which shows genotypic means for the marker (D2Mit260) most closely linked to Wg6. These results are concordant with data from the original mapping experiment in which weight gain 2 (Wg2) demonstrated a high level of dominance in HG but not in B6 mice (Corva et al. 2001). The additive effects for both QTL were negative, indicating that C alleles decreased body weight.

HG11F2 growth traits:

A single QTL, weight gain QTL 7 (Wg7), was localized in HG11F2 mice (Table 2; Figure 2D). This QTL was the only locus demonstrating sex-antagonistic expression in either cross (ΔLOD = 3.0). Wg7 increased 9WK by 1.3 g in male mice and decreased 9WK by 1.0 g in female C mice (Table 2; Figure 2E). Linkage was not detected using a strictly additive model (Model_A); however, when a QTL × sex interaction term was included (Model_SI), a significant LOD of 3.3 (P < 0.05) was obtained for 9WK (Figure 2D). 6WK and most weight gain periods also showed similar LOD score profiles but did not reach statistical significance (data not shown). Wg7 was located at 18.0 cM and explained 5.2% of the phenotypic variance in 9WK.

HG2DF2 organ weights:

To identify QTL influencing growth independent of systemic changes in body weight or adiposity, we measured weights of the five major organs: liver (LIV), spleen (SPL), heart (HRT), KID, and brain (BRN). In each case, organ weights were adjusted for differences in WSAC by including WSAC and WSAC × sex terms as additive covariates. In HG2DF2 mice, highly significant QTL for SPL (spleen weight QTL 6; Swg6) and KID (kidney weight QTL 7 and 8; Kwq7 and Kwq8) weight were discovered (Table 2; Figure 3A). Swq6 had a peak LOD score of 11.0 at 30.0 cM and explained 16.7% of the variance in SPL. Interestingly, it affected SPL in an additive fashion in males, but displayed a large dominance effect in females (Table 2; Figure 3B). Kwq7 and Kwq8 had peak LOD scores of 9.8 and 7.9 at 12.5 and 41.0 cM, respectively. Kwq8 was expressed equally in both sexes; however, Kwq7 was male biased (ΔLOD = 1.4). In females, both loci were inherited additively; however, male mice heterozygous for Kwq7 had the highest KID weight (Table 2; Figure 3C).

Figure 3.
LOD score profiles and genotypic contrasts for organ weights in HG2DF2 and HG11F2 mice. See the legend of Figure 2 for a detailed description of the figure. (A) LOD scores for SPL and KID in HG2DF2 mice. (B) Effect of D2Mit262 on SPL in HG2DF2 mice. (C) ...

HG11F2 organ weights:

Significant QTL regulating the size of the BRN, LIV, and KID, independent of body mass, were localized in HG11F2 mice (Table 2; Figure 3D). Brain-weight QTL 1 (Brwq1) was located at 18.5 cM and accounted for 9.5% of the phenotypic variance in BRN (Table 2). Female mice with C alleles at Brwq1 had an ~25-mg reduction in brain weight, relative to B females (Figure 3E). Liver-weight QTL 9 (Lwq9) was a male-specific locus (ΔLOD = 4.5) (Figure 3F) located at 10.5 cM with a LOD of 7.0 and explained 11.0% of the phenotypic variance in LIV. LIV was unaffected in females but C males had significantly elevated LIV, relative to B and H mice, which did not differ. Kidney-weight QTL 9 (Kwq9) displayed a similar LOD score profile, relative to Lwq9, with a peak LOD score of 8.0 at 11.5 cM, possibly suggesting a common QTL (Figure 3D). However, instead of increasing KID in C mice, it elicited a decrease (Figure 3G), arguing for two distinct loci. In further support for separate genes, the expression of Kwq9 was independent of sex.

HG2DF2 skeletal length traits:

In the same manner as described for organ weights, we identified QTL influencing skeletal length independent of changes in body weight. However, unlike organ weights, QTL for skeletal length traits were relatively unaffected by body weight. The LOD score profiles for nasal–anal body length (NA) and tail length (TAIL) in HG2DF2 mice and TAIL and femur length (FEM) in HG11F2 mice were nearly identical whether or not WSAC and WSAC × sex interactions were included as additive covariates (data not shown). Therefore, due to lower significance thresholds, we report QTL only for measures of skeletal and body length unadjusted for body weight.

Tail-length QTL 7 (Tailq7) mapped with a peak LOD score of 6.7 at 12.5 cM in HG2DF2 mice (Table 2; Figure 4A). Although Tailq7 overlaps with Wg5, the decrease in TAIL is not likely a pleiotropic effect of weight, since, as indicated above, weight was a minor contributor to the differences in TAIL. In addition, Wg5 displayed a high level of apparent dominance and Tailq7 was additive (Table 2; Figure 4B).

Figure 4.
LOD score profiles and genotypic contrasts for skeletal length in HG2DF2 and HG11F2 mice. See the legend of Figure 2 for a detailed description of the figure. (A) LOD scores for TAIL and NA in HG2DF2 mice. (B) Effect of D2Mit17 on TAIL in HG2DF2 mice. ...

Interestingly, TAIL and NA were not controlled by the same QTL in HG2DF2 mice. In contrast to Tailq7, body-length QTL 7 (Bdlnq7) mapped at 33.0 cM with a peak LOD score of 5.3 (Table 2; Figure 4A). The ΔLOD for Bdlnq7 was only 0.7, providing insufficient statistical evidence to classify it as sex biased on the basis of our criteria (Table 2). Bdlnq7 also displayed a large dominance effect in females (Table 2; Figure 4C). Bdlnq7 accounted for 8.4% of the variance in NA (Table 2).

HG11F2 skeletal length traits:

The tail-length QTL 8 (Tailq8) mapped with a peak LOD score of 8.5 at 19.0 cM (Table 2; Figure 4D). Tailq8 was male biased (ΔLOD = 2.3), having only a minor, nonsignificant effect on TAIL in females (Figure 4E). Tailq8 was additive and explained 13.1% of the variance in TAIL. The most significant QTL in either cross was femur-length QTL 4 (Feml4), a female-biased QTL (ΔLOD = 2.7) with a peak LOD of 16.5 at 5.0 cM (Table 2; Figure 4D). Feml4 explained 24.0% of the variance in FEM. Feml4 decreased FEM by nearly 0.6 mm (4%) in C genotype females (Figure 4F).

HG2DF2 obesity traits:

One of the most clearly sexually dimorphic traits in either cross was obesity in HG2DF2 mice (Table 2; Figure 5A). Two QTL, total fat pad mass QTL 1 and 2 (Fatq1 and Fatq2), influenced FAT (adjusted for WSAC) with peaks at 20.3 and 37.0 cM. Fatq1 was female biased (ΔLOD = 2.2), decreasing FAT by 28% in C females compared to only a 7% decrease in C males (Figure 5B). Fatq2 decreased FAT in both sexes, although its effect size was much smaller.

Figure 5.
LOD score profiles and genotypic contrasts for obesity-related traits in HG2DF2 and HG11F2 mice. See the legend of Figure 2 for a detailed description of the figure. (A) LOD scores for FAT and TC in HG2DF2 mice. (B) Effect of D2Mit223 on FAT in HG2DF ...

Although the peak locations were not identical, considerable overlap was observed among Fatq1, Fatq2, and the two QTL controlling TC, suggesting pleiotropy (Table 2; Figure 5D). Total plasma cholesterol QTL 1 and 2 (Tcq1 and Tcq2) were located at 16.5 and 32.0 cM and regulated TC equally in both sexes. Tcq2, the more significant of the two loci, decreased TC by 12% in C mice of either sex (Figure 5C).

HG11F2 obesity traits:

Only small changes in adipose mass were observed in HG11F2 mice. A minor QTL primarily altering GFP mass, GFP QTL 1 (Gfpq1), was located at 32.5 cM with a peak LOD of 3.5 (Table 2; Figure 5D). This was the location of the peak; however, the LOD score profile suggests that GFP may be influenced by more than one gene, each with very small effects (Figure 5D). Given its low peak LOD scores, we elected not to consider each peak a distinct QTL. All other fat pad weights and FAT also demonstrated similar peak LOD scores but did not reach statistical significance. Gfpq1 was expressed in both sexes and increased GFP mass in C mice (Figure 5E).

DISCUSSION

Genetic fine mapping is a critical step in the progression from QTL to QTG. In this study we used congenic-derived F2 intercrosses to dissect two regions of the mouse genome, MMU2 and MMU11, harboring growth, skeletal size, and obesity-related QTL. Previous studies have implicated sex as a modifier of growth and adiposity in these two chromosomal regions (Farber et al. 2006); therefore, an important objective of our study was to clarify the nature and the extent of QTL × sex interactions. To our surprise, the interactions were more widespread than originally expected and, of the 20 total QTL identified, 40% were influenced by sex. Moderate sex differences were seen for most of the remaining QTL; however, on the basis of our criteria QTL × sex interactions could not be statistically established. In addition to the influence of sex, multiple QTL affecting the same trait and nonadditive gene action added to the genetic complexities of the MMU2 and MMU11 regions.

In HG2DF2 mice, significant decreases in body weight were observed in mice with C alleles at Wg5 and Wg6. This was exclusively due to a slower rate of postnatal gain (data not shown). The difference in body weight was also reflected by concordant decreases in HRT and LIV (data not shown). In contrast, three QTL were identified, independent of body weight, which increased SPL (Swq6) and KID (Kwq7 and Kwq8) weight in C mice. This indicates the existence of genes on MMU2 that influence organ size through pathways independent of systemic changes in weight.

Two QTL were detected for both FAT and TC in HG2DF2 mice. Although the peaks were not perfectly coincident, there is substantial overlap between the loci, suggesting pleiotropy. A similar study of B6 and CAST identified similar QTL on MMU2 for subcutaneous fat mass and percentage of body lipids, as well as HDL cholesterol (Mehrabian et al. 1998). We did not detect significant linkage with HDL, although suggestive peaks were observed. It should be noted that plasma lipids in our study were measured in nonfasted mice.

In HG11F2 mice, 9WK (and most other growth traits, albeit to a lesser degree) was regulated by Wg7, the only sexually antagonistic QTL identified in either cross. Wg7 increased 9WK in C genotype male mice while decreasing weight in females of the same genotype. The phenomenon of sexually antagonistic QTL has also been recently described for fat mass in a separate mouse cross (Wang et al. 2006).

The trait affected most significantly in HG11F2 mice was skeletal length. Both Tailq8 and Feml4 accounted for substantial portions of the variance in their respective traits. Both loci were complex; Tailq8 was male biased while Feml4 was female biased. Although both QTL control measures of skeletal dimension, they were linked in repulsion. Male mice with C alleles at Tailq8 possessed a longer tail; however, if the same mice also possessed C alleles at Feml4, femur length was decreased (Figure 4, E and F).

MMU2 is a hotspot for growth and obesity QTL. Over 30 QTL have been identified (Corva and Medrano 2001; Perusse et al. 2005), of which several have been subsequently fine mapped using congenic strains and congenic-derived crosses (Diament et al. 2004; Estrada-Smith et al. 2004; Warden et al. 2004; Jerez-Timaure et al. 2005). A high percentage of previously identified MMU2 QTL and those uncovered in this study overlap and may represent the same QTG. More importantly, three human chromosomal regions share significant intervals of synteny with the HG2D donor region. MMU2 from 84 to 110 Mbp is syntenic with human chromosome (HSA) 11 from 26 to 57 Mbp, MMU2 from 111 to 126 Mbp is syntenic with HSA15 from 30 to 49 Mbp, and MMU2 from 128 to 181 Mbp is syntenic with the entire HSA20. Genetic linkage and SNP association studies have identified correlations between genotype and obesity [mainly body mass index (BMI); BMI was analyzed in the HG2DF2 cross and its LOD score profile was nearly identical to WK9 and therefore was not reported] in each of these human regions, supporting the potential correspondence between human and mouse QTG (Perusse et al. 2005).

Of particular interest is HSA20, which, like its MMU2 syntenic region, is an obesity hotspot. Numerous linkages between HSA20 and obesity-related traits have been identified (Lembertas et al. 1997; Hunt et al. 2001; Dong et al. 2003; Gorlova et al. 2003; Collaku et al. 2004). Additionally, the Mendelian syndromes Bardet–Biedl syndrome 6 (Katsanis et al. 2001) and Albright hereditary osteodystrophy (Patten et al. 1990), which are associated with obese states, are located on HSA20. To the best of our knowledge, no evidence of sex-specific obesity QTL have been found on HSA20. Fatq1, Fatq2, Tcq1, and Tcq2 will be our top priority for future investigations, given the important implications that their identification may have.

To identify traits modified by sex, we first analyzed nonrecombinant HG2DF2 and HG11F2 mice and preformed linkage analysis in each sex separately (data not shown). From this analysis most traits demonstrated a dependency on sex. Our next step was to quantify these effects by evaluating the fit of two linear models relative to a null model. The approach that we used was similar to the one described in Solberg et al. (2004). The first model, Model_A, included sex as an additive covariate and the second, Model_SI, was equivalent to Model_A plus a QTL × sex interaction term. The use of multiple linear models proved effective, most importantly because QTL that were not identified in the sex-specific analysis were detected [e.g., Wg7 was detected only when Model_SI was used (Table 2; Figure 2D)]. Therefore, this is an effective strategy for evaluating QTL × covariate interactions (in our case sex was included as a covariate, but other covariates could also be used) and for disentangling complex interactions critical to the proper assessment of complex traits.

The sex-specific genetic architecture of human obesity and lipid levels (Weiss et al. 2006) as well as femoral bone structure (Peacock et al. 2005) have been reported. Similarly, several studies in mice have also demonstrated the effect of sex on the segregation of QTL influencing complex traits (Turner et al. 2003; Wang et al. 2005). In development of the HG2D and HG11 congenic strains, sex was not expected to alter QTL expression, since QTL × sex interactions were not detected in the original mapping experiment (Corva et al. 2001). However, the loci captured in congenic strains did exhibit differences between the sexes (Farber et al. 2006). In our previous congenic analysis, MMU2 was comprehensively examined by generating strains with identical donor regions on both B6 and HG backgrounds with the aim of evaluating QTL × hg, but not QTL × sex interactions. One critical finding was the observation that QTL expression was sexually dimorphic in some HG strains, but not in the corresponding B6 strains (Farber et al. 2006). Sex effects specific to the hg locus have been reported; hg is recessive in females and partially additive in males, with respect to body weight (i.e., heterozygous +/hg males display an intermediate weight between +/+ and hg/hg males) (Horvat and Medrano 1995). Therefore, it is likely that the observed sex × QTL interaction effects are a result of the hg mutation.

One explanation for the finding that CAST MMU2 and MMU11 chromosomes on an HG background harbor sex-biased/-specific QTL, but not other chromosomes, may be the result of the clustering of genes involved in growth hormone (Gh) signaling. HG mice lack Socs2 gene expression (Horvat and Medrano 2001) and SOCS2 protein is a negative regulator of growth hormone (Gh) signaling both in vitro and in vivo (Greenhalgh et al. 2005). QTL on MMU2 and MMU11 have been reported to interact with hg (Corva et al. 2001) and we have recently described the identification of Gh pathway genes located within these QTL and their sequence in CAST mice (Farber et al. 2006). Interestingly, the HG11 region near Wg7 (18.0 cM or near 98 Mbp), Brwq1 (18.5 cM or near 99 Mbp), and Tailq8 (19.0 cM or near 99 Mbp) is in close proximity to a region of MMU11, which is saturated with genes involved in the central intracellular pathway regulating the response to Gh, such as signal transducer and activator of transcription 5a and 5b (Stat5b at 100.5 Mbp and Stat5a at 100.6 Mbp) and the structual Gh gene itself at 106.1 Mbp. In addition, functional candidates responsive to Gh or known to be involved in Gh signaling, such as Stat3 (100.3 Mbp), somatostatin receptor 2 (Sstr2 at 113.4 Mbp), hepatic transcription factor 3 (Tcf2 at 83.6 Mbp), Solute carrier family 2, facilitated glucose transporter, member 4 (Slc2a4, also known as Glut4 at 69.7 Mbp), and insulin-like growth factor binding protein 4 (Igfbp4 at 98.9 Mbp), are all located within the confines of the HG11 strain. The sexually dimorphic nature of Gh signaling is well documented; therefore, the QTL × sex interactions may be regulated by Gh-signaling genes. Future studies will address this by characterizing the sequence and expression levels (in male and female mice) of these genes in congenic mice.

Our results also suggest that hg increases the degree of dominance, particularly for body and organ weight QTL. In the original intercross between HG and CAST (Corva et al. 2001), Wg2 was additive (with respect to body weight and growth rate) in wild-type F2 mice, but overdominant in hg/hg F2 mice. Similarly, in HG2DF2 heterozygous mice, most growth and organ weight QTL appeared not to be inherited in a strictly additive fashion. Thus, hg also appears to modulate gene action, which may also be mediated via Gh pathway genes.

Congenic-derived intercrosses are advantageous over other QTL fine-mapping methods such as subcongenic analysis for a number of reasons. First, they provide a quick confirmation of effects discovered in homozygous congenic strains. One criticism of congenic models is the lack of replication in fine-mapping experiments, after observing a phenotype in the initial characterization of the congenic model. In some (maybe most) cases, this is due to phenotypic contributions from unlinked contaminating alleles, which exist in the congenic initially but are eliminated during the backcrossing required for fine mapping. These effects are quickly identified in congenic-derived crosses since unlinked alleles segregate randomly with respect to donor region genotype. Second, relative to a whole-genome scan, the statistical power to detect and resolve QTL, not to mention to characterize environment effects (such as sex), is substantially increased. This is primarily due to the elimination of background genetic noise, consisting of the direct and interactive effects of tens or hundreds of QTL. This approach has been used to successfully fine map growth and obesity QTL within the MB2 congenic (Jerez-Timaure et al. 2005).

Originally, the 1.5-LOD C.I.'s for the MMU2 and MMU11 QTL were between 30 and 35 cM (Corva et al. 2001). In this study, the C.I.'s ranged from 5.3 to 37 cM, with a median C.I. of 15.5. Due to this reduction, we can now focus on each C.I. using targeted subcongenic strains and gene expression analysis to ultimately identify the underlying QTG. It is also worth noting that a number of QTL that we detected were not seen in the original study, highlighting the added power of detection afforded by the congenic-derived crosses.

In summary, we have gained considerable insight into the genetic architecture of growth, skeletal size, and obesity QTL segregating in two well-defined regions of the mouse genome. Our results highlight the importance of assessing the extent of QTL × sex interaction in linkage analyses. Additionally, it is likely that the extensive sex effects are a direct result of hg, which implicates genes involved in Gh function as putative QTG. These data serve as the platform for future QTG discovery and provide significant insight into the molecular mechanisms governing differences in growth and obesity between the genders.

Acknowledgments

The authors thank Vince De Vera, Alma Islas-Trejo, Ricardo Verdugo, Rodrigo Gularte, Gonzalo Rincon, James Chitwood, Kerri Morimoto, Karina Guevara, Lee Nguyen, and Anna Chen for assistance with mouse phenotyping. We also acknowledge Emily Farber for mouse genotyping and collection of the human–mouse synteny information. Plasma lipid measurements were kindly provided by Lawrence Castellani and Aldons J. Lusis at the University of California at Los Angeles Department of Medicine/Division of Cardiology. This work was supported by the National Research Initiative (grant no. 2005-35205-15453) of the U. S. Department of Agriculture Cooperative State Research, Education, and Extension Service. C. R. Farber was supported by the Austin Eugene Lyons fellowship.

References

  • Abiola, O., J. M. Angel, P. Avner, A. A. Bachmanov, J. K. Belknap et al., 2003. The nature and identification of quantitative trait loci: a community's view. Nat. Rev. Genet. 4: 911–916. [PMC free article] [PubMed]
  • Bodnar, J. S., A. Chatterjee, L. W. Castellani, D. A. Ross, J. Ohmen et al., 2002. Positional cloning of the combined hyperlipidemia gene Hyplip1. Nat. Genet. 30: 110–116. [PMC free article] [PubMed]
  • Broman, K. W., H. Wu, S. Sen and G. A. Churchill, 2003. R/qtl: QTL mapping in experimental crosses. Bioinformatics 19: 889–890. [PubMed]
  • Churchill, G. A., and R. W. Doerge, 1994. Empirical threshold values for quantitative trait mapping. Genetics 138: 963–971. [PMC free article] [PubMed]
  • Clee, S. M., B. S. Yandell, K. M. Schueler, M. E. Rabaglia, O. C. Richards et al., 2006. Positional cloning of Sorcs1, a type 2 diabetes quantitative trait locus. Nat. Genet. 38: 688–693. [PubMed]
  • Collaku, A., T. Rankinen, T. Rice, A. S. Leon, D. C. Rao et al., 2004. A genome-wide linkage scan for dietary energy and nutrient intakes: the Health, Risk Factors, Exercise Training, and Genetics (HERITAGE) family study. Am. J. Clin. Nutr. 79: 881–886. [PubMed]
  • Corva, P. M., and J. F. Medrano, 2001. Quantitative trait loci (QTLs) mapping for growth traits in the mouse: a review. Genet. Sel. Evol. 33: 105–132. [PMC free article] [PubMed]
  • Corva, P. M., S. Horvat and J. F. Medrano, 2001. Quantitative trait loci affecting growth in high growth (hg) mice. Mamm. Genome 12: 284–290. [PubMed]
  • Diament, A. L., P. Farahani, S. Chiu, J. Fisler and C. H. Warden, 2004. A novel mouse Chromosome 2 congenic strain with obesity phenotypes. Mamm. Genome 15: 452–459. [PubMed]
  • Dietrich, W. F., J. Miller, R. Steen, M. A. Merchant, D. Damron-Boles et al., 1996. A comprehensive genetic map of the mouse genome. Nature 380: 149–152. [PubMed]
  • Dong, C., S. Wang, W. D. Li, D. Li, H. Zhao et al., 2003. Interacting genetic loci on chromosomes 20 and 10 influence extreme human obesity. Am. J. Hum. Genet. 72: 115–124. [PMC free article] [PubMed]
  • Estrada-Smith, D., L. W. Castellani, H. Wong, P. Z. Wen, A. Chui et al., 2004. Dissection of multigenic obesity traits in congenic mouse strains. Mamm. Genome 15: 14–22. [PubMed]
  • Farber, C. R., P. M. Corva and J. F. Medrano, 2006. Genome-wide isolation of growth and obesity QTL using mouse speed congenic strains. BMC Genomics 7: 102. [PMC free article] [PubMed]
  • Flint, J., W. Valdar, S. Shifman and R. Mott, 2005. Strategies for mapping and cloning quantitative trait genes in rodents. Nat. Rev. Genet, 6: 271–286. [PubMed]
  • Fox, J. W., 2005. The R commander: a basic-statistics graphical user interface to R. J. Stat. Software 14: 1–42.
  • Gorlova, O. Y., C. I. Amos, N. W. Wang, S. Shete, S. T. Turner et al., 2003. Genetic linkage and imprinting effects on body mass index in children and young adults. Eur. J. Hum. Genet. 11: 425–432. [PubMed]
  • Greenhalgh, C. J., E. Rico-Bautista, M. Lorentzon, A. L. Thaus, P. O. Morgan et al., 2005. SOCS2 negatively regulates growth hormone action in vitro and in vivo. J. Clin. Invest. 115: 397–406. [PMC free article] [PubMed]
  • Horvat, S., and J. F. Medrano, 1995. Interval mapping of high growth (hg), a major locus that increases weight gain in mice. Genetics 139: 1737–1748. [PMC free article] [PubMed]
  • Horvat, S., and J. F. Medrano, 2001. Lack of Socs2 expression causes the high-growth phenotype in mice. Genomics 72: 209–212. [PubMed]
  • Hunt, S. C., V. Abkevich, C. H. Hensel, A. Gutin, C. D. Neff et al., 2001. Linkage of body mass index to chromosome 20 in Utah pedigrees. Hum. Genet. 109: 279–285. [PubMed]
  • Ihaka, R., and R. Gentleman, 1996. R: a language for data analysis and graphics. J. Comp. Graph. Stat. 5: 299–314.
  • Jerez-Timaure, N. C., E. J. Eisen and D. Pomp, 2005. Fine mapping of a QTL region with large effects on growth and fatness on mouse chromosome 2. Physiol. Genomics 21: 411–422. [PubMed]
  • Katsanis, N., S. J. Ansley, J. L. Badano, E. R. Eichers, R. A. Lewis et al., 2001. Triallelic inheritance in Bardet-Biedl syndrome, a Mendelian recessive disorder. Science 293: 2256–2259. [PubMed]
  • Klein, R. F., J. Allard, Z. Avnur, T. Nikolcheva, D. Rotstein et al., 2004. Regulation of bone mass in mice by the lipoxygenase gene Alox15. Science 303: 229–232. [PubMed]
  • Lander, E. S., and D. Botstein, 1989. Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121: 185–199. [PMC free article] [PubMed]
  • Lang, D. H., N. A. Sharkey, A. Lionikas, H. A. Mack, L. Larsson et al., 2005. Adjusting data to body size: a comparison of methods as applied to quantitative trait loci analysis of musculoskeletal phenotypes. J. Bone Miner. Res. 20: 748–757. [PMC free article] [PubMed]
  • Lembertas, A. V., L. Perusse, Y. C. Chagnon, J. S. Fisler, C. H. Warden et al., 1997. Identification of an obesity quantitative trait locus on mouse chromosome 2 and evidence of linkage to body fat and insulin on the human homologous region 20q. J. Clin. Invest. 100: 1240–1247. [PMC free article] [PubMed]
  • Markel, P., P. Shu, C. Ebeling, G. A. Carlson, D. L. Nagle et al., 1997. Theoretical and empirical issues for marker-assisted breeding of congenic mouse strains. Nat. Genet. 17: 280–284. [PubMed]
  • Mehrabian, M., J. H. Qiao, R. Hyman, D. Ruddle, C. Laughton et al., 1993. Influence of the apoA-II gene locus on HDL levels and fatty streak development in mice. Arterioscler. Thromb. 13: 1–10. [PubMed]
  • Mehrabian, M., P. Z. Wen, J. Fisler, R. C. Davis and A. J. Lusis, 1998. Genetic loci controlling body fat, lipoprotein metabolism, and insulin levels in a multifactorial mouse model. J. Clin. Invest. 101: 2485–2496. [PMC free article] [PubMed]
  • Mehrabian, M., L. W. Castellani, P. Z. Wen, J. Wong, T. Rithaporn et al., 2000. Genetic control of HDL levels and composition in an interspecific mouse cross (CAST/Ei × C57BL/6J). J. Lipid Res. 41: 1936–1946. [PubMed]
  • Oliver, F., J. K. Christians, X. Liu, S. Rhind, V. Verma et al., 2005. Regulatory variation at glypican-3 underlies a major growth QTL in mice. PLoS Biol. 3: e135. [PMC free article] [PubMed]
  • Patten, J. L., D. R. Johns, D. Valle, C. Eil, P. A. Gruppuso et al., 1990. Mutation in the gene encoding the stimulatory G protein of adenylate cyclase in Albright's hereditary osteodystrophy. N. Engl. J. Med. 322: 1412–1419. [PubMed]
  • Peacock, M., D. L. Koller, D. Lai, S. Hui, T. Foroud et al., 2005. Sex-specific quantitative trait loci contribute to normal variation in bone structure at the proximal femur in men. Bone 37: 467–473. [PMC free article] [PubMed]
  • Perusse, L., T. Rankinen, A. Zuberi, Y. C. Chagnon, S. J. Weisnagel et al., 2005. The human obesity gene map: the 2004 update. Obes. Res. 13: 381–490. [PubMed]
  • Silver, L. M., 1995. Mouse Genetics: Concepts and Applications. Oxford University Press, Oxford.
  • Solberg, L. C., A. E. Baum, N. Ahmadiyeh, K. Shimomura, R. Li et al., 2004. Sex- and lineage-specific inheritance of depression-like behavior in the rat. Mamm. Genome 15: 648–662. [PMC free article] [PubMed]
  • Stylianou, I. M., R. Korstanje, R. Li, S. Sheehan, B. Paigen et al., 2006. Quantitative trait locus analysis for obesity reveals multiple networks of interacting loci. Mamm. Genome 17: 22–36. [PubMed]
  • Turner, C. H., Q. Sun, J. Schriefer, N. Pitner, R. Price et al., 2003. Congenic mice reveal sex-specific genetic regulation of femoral structure and strength. Calcif. Tissue Int. 73: 297–303. [PubMed]
  • Wakeland, E., L. Morel, K. Achey, M. Yui and J. Longmate, 1997. Speed congenics: a classic technique in the fast lane (relatively speaking). Immunol. Today 18: 472–477. [PubMed]
  • Wang, S., N. Yehya, E. E. Schadt, H. Wang, T. A. Drake et al., 2006. Genetic and genomic analysis of a fat mass trait with complex inheritance reveals marked sex specificity. PLoS Genet. 2: e15. [PMC free article] [PubMed]
  • Wang, X., M. Ria, P. M. Kelmenson, P. Eriksson, D. C. Higgins et al., 2005. Positional identification of TNFSF4, encoding OX40 ligand, as a gene that influences atherosclerosis susceptibility. Nat. Genet. 37: 365–372. [PubMed]
  • Warden, C. H., S. Stone, S. Chiu, A. L. Diament, P. Corva et al., 2004. Identification of a congenic mouse line with obesity and body length phenotypes. Mamm. Genome 15: 460–471. [PubMed]
  • Weiss, L. A., L. Pan, M. Abney and C. Ober, 2006. The sex-specific genetic architecture of quantitative traits in humans. Nat. Genet. 38: 218–222. [PubMed]
  • Wong, M. L., A. Islas-Trejo and J. F. Medrano, 2002. Structural characterization of the mouse high growth deletion and discovery of a novel fusion transcript between suppressor of cytokine signaling-2 (Socs-2) and viral encoded semaphorin receptor (Plexin C1). Gene 299: 153–163. [PubMed]
  • Yalcin, B., S. A. Willis-Owen, J. Fullerton, A. Meesaq, R. M. Deacon et al., 2004. Genetic dissection of a behavioral quantitative trait locus shows that Rgs2 modulates anxiety in mice. Nat. Genet. 36: 1197–1202. [PubMed]

Articles from Genetics are provided here courtesy of Genetics Society of America

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...