Quantitative Trait Loci Affecting Body Weight and Fatness From a Mouse Line Selected for Extreme High Growth

Corresponding author: Gudrun A. Brockmann, Department of Molecular Biology, Research Institute for the Biology of Farm Animals, Wilhelm-Stahl-Allee 2, 18196 Dummerstorf, Germany., brockman{at}fbn-dummerstorf.de (E-mail).

Communicating editor: T. F. C. MACKAY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Quantitative trait loci (QTL) influencing body weight were mapped by linkage analysis in crosses between a high body weight selected line (DU6) and a control line (DUKs). The two mouse lines differ in body weight by 106% and in abdominal fat weight by 100% at 42 days. They were generated from the same base population and maintained as outbred colonies. Determination of line-specific allele frequencies at microsatellite markers spanning the genome indicated significant changes between the lines on 15 autosomes and the X chromosome. To confirm these effects, a QTL analysis was performed using structured F₂ pedigrees derived from crosses of a single male from DU6 with a female from DUKs. QTL significant at the genome-wide level were mapped for body weight on chromosome 11; for abdominal fat weight on chromosomes 4, 11, and 13; for abdominal fat percentage on chromosomes 3 and 4; and for the weights of liver on chromosomes 4 and 11, of kidney on chromosomes 2 and 9, and of spleen on chromosome 11. The strong effect on body weight of the QTL on chromosome 11 was confirmed in three independent pedigrees. The effect was additive and independent of sex, accounting for 21–35% of the phenotypic variance of body weight within the corresponding F₂ populations. The test for multiple QTL on chromosome 11 with combined data from all pedigrees indicated the segregation of two loci separated by 36 cM influencing body weight.

BODY weight is a composite trait comprising the weights of lean (protein) and fat compounds as well as those of bones and body fluids. As a quantitative trait, body weight reflects the effect of a complex net of gene actions under the influence of the environment. Among the high number of quantitative trait loci (QTL) known to affect body weight, some genes and their products are well known for their growth-stimulating effects on the metabolism and their influence upon differentiation. Candidate loci include the genes encoding growth hormone, insulin-like growth factors, and leptin, as well as their receptors and binding proteins. So far, 13 loci affecting 10 obesity syndromes have been mapped in man, five single gene mutations have been mapped in rodent models (Lepr, Cpe, Lep, Tub, A^y), and 24 QTL related to body weight or body fat have been identified from crossbreeding experiments in mouse, rat, and pig (reviewed by PERUSSE et al. 1997 ). Additionally, the hg locus, a line-specific spontaneous mutation that is characterized by a 30–50% increase in weight gain and mature body size, was mapped to chromosome 10 (HORVAT and MEDRANO 1995 ).

So far, in model animals all loci affecting growth performance or obesity have been mapped in crosses of specific inbred lines. Recently, analytical methods have been developed for detecting linkage in crosses between outbred lines such as long-term selected populations (HALEY et al. 1994 ). Long-term selection for high body weight is expected to fix growth-promoting QTL alleles, at least those contributing appreciably to the selection response. Hence, selected outbred lines offer the possibility of detecting the accumulated QTL affecting the selected trait. However, in contrast to inbred lines, usually not all loci in the genome are homozygous, but rather share common alleles with an unselected line.

Here we report results of a genome-wide QTL analysis for growth parameters in crosses between mouse line DU6, which had been selected for 70 generations for high body weight, and the unselected control line DUKs. The mating of individual mice as father or mother for the generation of F₂ pedigrees allows segregation analysis of each parental allele, comparable with common crosses of inbred lines. DU6 and DUKs differ in body weight by 106% and in abdominal fat weight by 100% at 42 days. Both lines were generated from the same base population and kept as outbred lines (SCHULER 1985 ). Examination of multilocus DNA fingerprints indicated line-specific banding patterns between DU6 and DUKs and revealed linkage between specific bands and the growth performance (BROCKMANN et al. 1993 ). Previous marker analysis of a restricted region of chromosome 11 harboring the growth hormone gene (Gh) showed an association between the high growth performance in line DU6 and alleles of marker D11Mit125, which is tightly linked to the Gh gene. As Gh is a major candidate gene, partial sequence analysis, restriction fragment length polymorphism, and single strand conformation polymorphism (SSCP) studies were carried out, but did not show any polymorphism in the examined gene structure (DAS et al. 1996 ). This led to the assumption that the suggested QTL is not the Gh gene, but rather a locus that either regulates the Gh gene or takes effect independently of it. Therefore, an extended QTL search was needed.

Initially, our growth-differentiated outbred lines were analyzed for line-specific distribution of microsatellite alleles to characterize the heterogeneity within and between DU6 and DUKs and to find chromosomal regions, which might have been involved in the selection process. Subsequently, linkage between marker alleles and performance was evaluated in segregating pedigrees from crosses of DU6 x DUKs for body weight, abdominal fat weight, abdominal fat percentage, and the weights of liver, kidney, and spleen.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mouse lines:
The study was carried out on the outbred mouse line DU6, which had been selected for 70 generations for high body weight, and the randomly mated control line DUKs (BUNGER et al. 1983 ). Both lines descend from original crosses of four base (NMRI orig., Han:NMRI, CFW, CF1) and four inbred (CBA/Bln, AB/Bln, C57BL/Bln, XVII/Bln) populations in the Research Institute for the Biology of Farm Animals, Dummerstorf, Germany (SCHULER 1985 ). The population size in both lines was 80 pairs per generation. The litter size was standardized to nine. Offspring were weaned at 21 days. Animals were fed ad libitum with a breeding diet containing 12.5 MJ/kg metabolic energy with an average content of 22.5% crude protein, 5.0% crude fat, 4.5% crude fiber, 6.5% crude ash, 13.5% water, 48.0% N-free extract, vitamins, trace elements, amino acids, and minerals (Altromin diet 1314, Lage, Germany). The body weight at 42 days in the selected line DU6 (58.3 ± 6.1 g) was over twice that of the randomly mated control DUKs (28.7 ± 2.3 g) (BUNGER et al. 1990 ). The epididymal fat percentage in line DU6 (2.2 ± 0.8%) was twice as high as in the control (1.1 ± 0.4%) at the same age. Line-specific allele distribution patterns were determined by microsatellite analysis of 20 animals per line with every animal from a different litter.

Pedigree design:
QTL analysis was performed using pedigrees of F₂-intercross design. These were established by crossing four males of the high growth selected line DU6 to four females of the control line DUKs. Large pedigrees with a total of 54 F₁ and 715 F₂ offspring were generated by repeated mating of the parents and subsequently of F₁ offspring from the same parents. Mating was initially at the age of 10 wk and repeated after 6 wk and was performed simultaneously in all pedigrees. The structures of the individual pedigrees are summarized in Table 1. The largest pedigree (identity number 8) with 341 F₂ offspring was chosen for the genome-wide scan. The quantitative measurements used in the QTL analysis were body weight, abdominal fat, and the weight of the liver, kidney and spleen of all F₂ individuals. The abdominal fat measured in males was the gonadal fat and in females the perimetrial fat. The ratio of abdominal fat to body weight was defined as abdominal fat percentage. The phenotypic data were always recorded at 42 days of age; that is the time for the selection decision in all generations. This age corresponds to the end of the juvenile phase of ontogenesis.

Marker analysis:
Initially the distribution of marker alleles between the lines was tested for 84 loci. The markers were chosen to be distributed over all chromosomes, with an average distance of 16.6 cM between markers and the largest interval being 42 cM.

Markers were chosen for their high variability between appropriate inbred lines from the mouse genome database (MGD 1998 ). Mouse MapPair primers were purchased from Research Genetics (Huntsville, AL), D11Bhm* primers were a gift from THOMAS BOEHM (NEHLS et al. 1995 ), and D9Fbn1 was developed by RAINER FÜRBASS in our unit. Although the marker map resolution of the mouse genome is high, finding an informative marker set for our pedigrees generated from crosses of outbred lines was difficult because of shared alleles between parents. Over 450 microsatellite markers were tested for informative parental alleles in pedigree 8 before the parents, F₁, and F₂ were genotyped for 94 loci covering all chromosomes at an average spacing of 12.8 cM. Individuals of pedigrees 3, 4, and 10 were genotyped for informative microsatellites on chromosome 11.

DNA was extracted from mouse tail clips by selective binding of DNA to an ion exchange matrix after Proteinase K digestion using the QIAamp Tissue Kit (QIAGEN, Hilden, Germany). DNA was amplified with Taq polymerase (Appligene, Heidelberg, Germany) with a modified standard PCR protocol (DIETRICH et al. 1992 ): 20 ng of template DNA was used in a total reaction volume of 10 µl. After denaturation (96°, 3 min) followed by two rounds using 59° and 57° as annealing temperatures, DNA was amplified for 25 cycles comprising 15 sec at 94°, 1 min at 55°, and 1 min at 70°. DNA was automatically distributed for PCR with the Biomek 2000 system (Beckman, Fullerton, CA). The amplifications were carried out in 96-well microtiter plates on the PCR-System 9600 (Perkin Elmer, Norwalk, CT). PCR products were dried, redissolved in 3 µl formamide loading buffer and separated on 6% polyacrylamide gels, 40 cm in length, under denaturing conditions. The gels were stained with silver nitrate (RIESNER et al. 1989 ). After initial fixation in a 10% ethanol/0.5% acetic acid solution for 10 min, the gels were immersed for 15 min into a fresh 5.9 mM solution of silver nitrate and developed for 10 min in 375 mM NaOH, 2.3 mM NaBH₄, 0.125% formaldehyde (37% w/v). After two rinses in water, the gels were dried on Filtrak paper (Niederschlag, Germany).

On each gel, the parental marker alleles were loaded in separate lanes, serving as a standard for genotyping the F₁ and F₂ offspring. All genotyping results were scored twice and runs were repeated where there were discrepancies.

Test statistics:
Quantitative differences in the distribution of microsatellite alleles between the lines were assessed with the chi-square test (RASCH et al. 1978 ). The heterozygosity indices for the two mouse lines and the genetic distance between the lines were calculated according to GREGORIUS 1984 . A heterogeneity index >0.5 between the lines at a single locus is consistent with the significance threshold of P < 10^-4 of the chi-square test.

For linkage analysis, initially pedigree-specific marker maps were generated with the program CRIMAP (LANDER and GREEN 1987 ). As the maps were internally consistent with the published map, we used the marker distances of the consensus map of the mouse genome (DIETRICH et al. 1996 ) for the mapping of QTL.

For the analysis of the pedigrees, initially the influences of sex, parity, and subfamily (i.e., F₂ animals from the same pair of F₁ parents) were estimated via variance analysis (SAS 1990 ) and found to be significant and so were included as fixed effects in the linkage analysis. For the analysis of chromosome 11, each pedigree was evaluated separately, followed by combined analyses over specific pedigrees. In the analysis over several pedigrees, a fixed effect of pedigree was also included in the model.

Data were analyzed by multiple regression as developed for the analysis of crosses between outbred lines (HALEY et al. 1994 ). Initially, the standard interval mapping model was used with a single QTL on the linkage group. The obtained estimates were revised by fitting background genetic effects on other linkage groups, as suggested by ZENG 1993 and JANSEN 1993 . Background genetic effects were included as cofactors stepwise, beginning with the locus showing the largest estimated effect, followed by the locus with the second largest effect, until no further QTL were detected at the suggestive level of significance. During that procedure a background effect was always dropped from the analysis in the case of analyzing its own position. The sex chromosome was divided into a pseudo-autosomal and an X-specific section. In additional analyses, body weight was introduced as a covariate in the analysis of abdominal fat weight and percentage to examine the dependence between both traits. The joint effect of all significant QTL for a trait (i.e., variance explained of the F₂) was estimated as reduction of the residual mean square in the one QTL analysis fitting all QTL positions as cofactors in comparison with no QTL fitted.

Levels for suggestive ( = 0.1), significant ( = 0.05), and highly significant ( = 0.01) linkage were used (LANDER and KRUGLYAK 1995 ). Chromosome-specific and genome-wide empirical threshold values of the test statistics from the regression analysis were estimated with the permutation test proposed by CHURCHILL and DOERGE 1994 . The suggestive level is equivalent to the level at which one false positive result is expected in a genome scan. In a genome of 20 independent pairs of chromosomes, this is approximately equivalent to using the chromosomal 5% significance level, so that on average one of the chromosomes is significant at this level in a scan of the entire genome. The chromosome wide levels vary between chromosomes depending on their length and the markers they contain. We used the two-LOD drop to provide a conservative estimate of the 95% confidence interval for loci of highly significant level of genetic linkage (VAN OOIJEN 1992 ).

Once a single QTL on a chromosome had been identified, the presence of a second QTL was investigated by performing a grid search at 2-cM intervals. The two-QTL model was accepted if there was a significant improvement over the best possible one-QTL model at P < 0.05 using a variance ratio (F) test with 3 d.f. (for the additional additive and dominance effect and position estimated for the second QTL). Map locations are given as genetic distance from the centromere in cM. Where a QTL was identified, the interaction of the QTL effects with pedigree, intrapedigree subfamily, and sex were tested and were accepted as significant at P < 0.05.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Line-specific allele distribution:
Initially the degree of heterogeneity between the two outbred mouse lines was tested. In the selected line DU6, more extreme allele frequencies and a reduced number of alleles were observed compared to DUKs. Thus, 35.4% of all alleles found were not detected in DU6, whereas 22.8% were not found in DUKs. The resulting heterozygosity index over all markers was reduced by one-fourth in the selected line (0.260 ± 0.025) compared with the random mated control (0.330 ± 0.026) (Table 2). The genetic distance of GREGORIUS 1984 between DU6 and DUKs was 0.463.

At 39 of 84 microsatellite loci we found significant differences of allele frequencies (P < 0.0001, chi-square test) between the growth-differentiated lines DU6 and DUKs. The fixation of line-specific marker alleles is accompanied by increased genetic distances between the lines at single loci. Table 3 contains all markers showing genetic distances higher than 0.5, which is equivalent with fixation of a single allele in the selected line DU6 and normal distribution of all detected alleles in DUKs. Marker loci with line-specific allele distribution patterns between DU6 and DUKs were found on 15 autosomes and the X chromosome. Chromosomes 8, 16, 18, and 19 did not reveal significant differences in allele frequencies of the analyzed markers.

Chromosomal parts with differences in allele frequencies between the lines are shown in Figure 1 as bold labeled regions indicating the 20-cM linkage intervals of corresponding marker locations. They mark regions in the genome which could be linked with a QTL for body weight, if these allele frequency differences are the result of selection. However, we must not neglect genetic drift as a potential reason for line-specific allele distribution.

View larger version (40K): In this window In a new window Download PPT slide   Figure 1. The map shows all markers which were informative and genotyped for linkage analysis in pedigree 8. Marker positions are given according to the mouse consensus map (DIETRICH et al. 1996 ). Bold labeled chromosomal regions indicate 20-cM linkage intervals of markers showing significantly different allele frequencies between DU6 and DUKs.

Pedigree characteristics:
The F₂-populations of the four generated pedigrees had mean body weights between 40 and 43 g (Table 4), which are very close to the average of the two parental lines, with mean values of 58 g in DU6 and 28 g in DUKs. The mean abdominal fat weights varied between 0.584 g and 0.914 g, corresponding with the abdominal fat percentage of 1.2–2.0% within the different F₂ populations. Pedigree 10 had a conspicuously higher body weight and abdominal fat weight compared with pedigrees 3, 4, and 8 (F test of differences between pedigrees, P < 0.001) (Table 4). Body weight and abdominal fat weight were highly correlated, at r = 0.55 to r = 0.74, in the different pedigrees. High correlations were also evident between body weight and the weights of liver and kidney, as shown for pedigree 8 in Table 5.

Due to the presence of common marker alleles in DU6 and DUKs, the parental genotypes had to be checked. Within pedigrees, those markers showing line-specific allele distribution were tested and complemented by others where necessary. For the analysis of pedigree 8 we found 94 fully informative markers, which covered 92.7% of the genome at average intervals of 12.8 cM. Of these markers, 35 overlap with the marker set used for the determination of allele frequencies in the two lines. Using all genotyping information, specific marker maps were built for pedigree 8 and a map of chromosome 11 was also built for pedigrees 3, 4, and 10. The maps were in good agreement with the consensus map of the mouse genome (DIETRICH et al. 1996 ). Hence, marker locations and distances from the consensus map were used for linkage analyses. The map containing information markers used for the linkage analyses in pedigree 8 is presented in Figure 1. Additional informative markers used for the linkage analyses in pedigree 8 is presented in Figure 1. Additional informative markers were chosen for the extended analysis of chromosome 11 in pedigrees 3, 4, and 10; these markers are shown in Figure 2.

View larger version (37K): In this window In a new window Download PPT slide   Figure 2. The graphs plot the F-ratio testing the hypothesis of a single QTL in a given position on the chromosome. Marker positions are given in centimorgans from the centromere. The dotted horizontal line represents the significance threshold equivalent to P < 0.05.

Genome-wide QTL mapping:
The genome-wide significance levels for body weight were determined by permutation analysis to be 7.65 for the 5% level and 9.49 for the 1% level. The chromosome-wide 5% significance level and hence the genome-wide suggestive level were taken to vary with chromosome (Table 7). Pedigree 8 with 341 F₂ offspring was evaluated for the presence of detectable QTL influencing body weight, abdominal fat weight, abdominal fat percentage, and the weights of liver, kidney, and spleen. Table 6 presents the net effects of the selected cofactors for the different traits. The estimates for the most likely positions and effects of QTL detected in pedigree 8 are given in Table 7. QTL at the highly significant level of linkage (P < 0.01; F > 9.49) were mapped for body weight on chromosome 11, for abdominal fat weight on chromosomes 4, 11, and 13, for abdominal fat percentage on chromosomes 3 and 4, for the weights of liver on chromosomes 4 and 11, of kidney on chromosomes 2 and 9, and of spleen on chromosome 11. The inclusion of body weight as a covariate in the analyses indicated that some QTL influencing body weight have an effect on fat proportional to the overall phenotypic association between body weight and fatness. Thus, inclusion of body weight as a covariate in the analysis of the fat traits removed all evidence of QTL for fat traits on chromosomes 9, 11, and 17. Because the two parents of the pedigree come from outbred lines, they are often heterozygous at a specific marker locus and hence marker genotypes varied between different subfamilies in a pedigree. Despite this, within pedigree 8 there were no significant interactions between a detected QTL and subfamily; thus there is no evidence that QTL genotypes differed between subfamilies.

Positive estimates of genetic effects indicate that alleles from the selected line (DU6) increase the trait. As expected, the great majority of estimated genetic effects are in this direction. Most QTL contribute to the trait difference between genotypes by additive genetic effects; for body weight, only the QTL on chromosome 1 shows significant dominance. For fatness, the QTL on chromosome 11 has significant dominance for the low allele.

The X-linked QTL affecting body weight and the weight of kidneys act significantly differently (P < 0.05) in males and females, with an effect in females, but almost no influence in males. A QTL for body weight on chromosome 13 increases the body weight by an additive effect of the DU6 allele in males and by dominance in females. Among the loci responsible for abdominal fat weight, the QTL on chromosome 11 mainly influences fat accumulation in females, while the QTL on chromosomes 17 and 19 have major effects on males.

The net effect of all detected QTL for body weight explains 37% of the phenotypic variance in the F₂ population; QTL influencing abdominal fat weight explain 35% of the variance in the F₂ population. The joint net effects were estimated as reduction of the residual mean square in the one QTL analysis fitting all QTL positions as cofactors in comparison with no QTL fitted.

Estimating the total genetic variance in the F₂ is not straightforward as the lines crossed were not inbred. However, we can obtain an approximate estimate of the genetic variance for body weight due to the QTL alleles that explain the line difference and hence contribute to an increased variance in the F₂ compared with the F₁. This estimate is Var(F₂) - Var(F₁) = 9.15; the heritability due to these effects in the F₂ is thus estimated as 9.15/16.84 = 0.54. The effects of the identified QTL for body weight account for 68% of the estimated genetic variance in the F₂ population.

The largest effect on growth performance was detected on chromosome 11, contributing 23% to the corresponding F₂ variance. This chromosome had effect on body weight and on the highly correlated weights of liver and spleen, and on abdominal fat weight.

The most likely locations of the identified QTL are in good agreement with those chromosomal regions showing specific marker alleles in DU6 in comparison with DUKs. Only the QTL for body weight on chromosome 13, for abdominal fat weight on chromosome 19, for liver weight on chromosome 10, and for spleen weight on chromosome 17 do not map in these regions. Figure 3 displays the chromosomal position of the identified QTL responsible for differences in body weight and abdominal fat weight in our mouse cross on the mouse map of known putative candidate genes for the complex traits that have been proved to contribute to defects in the endocrine system or mutations in growth or obesity in mice (MGD 1998 ).

View larger version (34K): In this window In a new window Download PPT slide   Figure 3. The map gives the most likely position of identified QTL relative to known candidate genes (MGD 1998 ).

QTL on chromosome 11:
The strong influence of a QTL located on chromosome 11 responsible for high body weight was confirmed in pedigrees 3, 4, and 10. Pedigree-specific estimates for the most likely QTL position and the QTL effect are given in Table 8. The F-value curves for the probability of linkage between genotype and phenotype as a function of chromosomal location pertaining to body weight, abdominal fat weight, abdominal fat percentage, and liver weight are displayed in Figure 2. In the analyses of all pedigrees together, the most likely position of the QTL for body weight was estimated at a peak F value of 64.7 at 42 cM from the centromere, with a two-LOD support interval of 14 cM. Due to the high significance in all pedigrees, this QTL for body weight was named Bw4. The allele effect was additive, with dominance effects of significantly different from zero (not shown). This locus accounted for 21–35% of the variance within the corresponding F₂ populations. The mice that were genotypically identical to the DU6 parent at the neighboring marker D11Mit36, which was informative in all pedigrees, were on average 4.5 g (pedigree 4) to 6.7 g (pedigrees 8, 10, and 3) heavier than mice that were of the DUKs genotype. The influence of Bw4 on body weight was sex independent, in that no significant interaction of the QTL with sex was found (results not shown).

Although the analyses of single pedigrees did not provide significant evidence for the segregation of multiple QTL, in the analysis of these pedigrees together, a model with two linked QTL on chromosome 11 was a significantly better fit than a model with a single QTL. We therefore conclude that, in addition to Bw4, a second QTL affecting body weight is located at 14 cM. Bw4 was placed in this analysis at 50 cM.

In addition to body weight, an effect on abdominal fat weight and the weight of liver was also found with highly significant estimates in the overall analysis, but not in every individual pedigree. Abdominal fat weight and abdominal fat percentage showed similar results. The effect of chromosome 11 on abdominal fat weight at a peak F value of 17.8 at 34 cM was evident with data from all pedigrees (Table 8). The two-LOD support interval covers a large region between 26 and 47 cM from the centromere. The loci on chromosome 11 account for 6–16% of the variance of abdominal fat weight within the F₂ offspring of the different pedigrees. The estimates for the location of the QTL and the effect on abdominal fat weight were pedigree dependent. For example, in pedigree 4, with a large effect on body weight in the distal chromosomal region, there was a maximal F value for abdominal fat weight near Bw4. On the other hand, the curve in pedigree 10, which was characterized by the highest correlation between body weight and abdominal fat weight (r = 0.74), might indicate two loci influencing abdominal fat weight. However, there was no statistical evidence for two QTL influencing fat in any analysis of single pedigrees or combined analyses.

Sex influenced the expression of the abdominal fat weight QTL significantly in all pedigrees (the interaction of the QTL effect and sex was significant, P < 0.002). Male mice always showed a negative dominance effect (d = -0.753 ± 0.046), while the genetic effect on abdominal fat weight in females was additive (a = 0.186 ± 0.029). Within pedigree 10 there was evidence of an influence of the subfamilies on the QTL affecting accumulation of abdominal fat (P < 0.05).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have analyzed a single pedigree from a cross of extreme individuals of an outbred high growth-selected mouse line with a randomly mated control and identified single entities contributing to the complex traits growth and fatness. The test statistics indicate nine chromosomes containing genes with influence on body weight and eight chromosomes affecting fat accumulation. The biggest effect on body weight, which also had an influence on fat, was found on chromosome 11. Previously undescribed QTL for fat were mapped to chromosomes 13 and 19.

Of the 31 QTL identified with effects significant at the suggestive level on body weight, abdominal fat weight, and weights of liver, kidney, and spleen, 27 were mapped to regions with significant differences in marker allele frequencies between DU6 and DUKs. This shows that the line-specific distribution of marker alleles in outbred lines that were generated from a common base population may be a valuable hint on chromosomal regions containing QTL, even if there is no genotypic information available from the base population.

We have uncovered in our line DU6 a strong QTL affecting high body weight on chromosome 11 and named it Bw4. This QTL was detected in pedigree 8 and confirmed in three more pedigrees. It accounted for up to 35% of the variance of body weight in four F₂ populations. In the high body weight-selected line DU6, the fixation of the allele at marker D11Mit120 (47 cM) which is closely linked to Bw4, combined with the fact that the DU6 allele of Bw4 increased body weight in all pedigrees, indicated that the identified QTL allele might have been fixed during selection in the high growth-selected line. The estimated position of the QTL was at 42 cM, with the two-LOD support interval from the single QTL analysis between 36 and 50 cM in the consensus mouse map (WATKINS-CHOW et al. 1996 ). This chromosomal region does not coincide with the distal region harboring the Gh gene. Potential candidate genes affecting growth in the chromosomal region surrounding Bw4 are Myhs (myosin heavy chain skeletal muscle) and Glut4 (glucose transporter-4, muscle and fat). The df (Ames dwarf) mutation is on chromosome 11 but is a 25 cM and hence does not map to the estimated support interval.

Bw4 denotes a chromosomal region affecting both body weight and fat accumulation. Candidate genes for elevated fat accumulation in the detected region might be Ebf (early B-cell factor, tissue-specific transcriptional activator protein), and Idd4 (insulin dependent diabetes susceptibility-4).

As mouse line DU6 had been selected for high body mass at 42 days of life, the increased body fat content, including abdominal fat weight, is a by-product of this selection process. The estimate for the realized heritability of body weight was 0.43 (BUNGER et al. 1992 ). Due to the conceivable physiological cross-relationship between various traits, some of the loci identified in our analyses may have correlated effects on other analyzed traits. Associated effects of Gh-gene activity and obesity was demonstrated, e.g., in bovine growth hormone transgenes (POMP et al. 1996 ). The possible simultaneous effect of the QTL on chromosome 11 on increased body weight and elevated fat accumulation, as well as of those on chromosomes 4, 5, and 13, highlights the often observed genetic predisposition of linked inheritance of high growth rates and fatness in livestock production. The generation of congenic mouse strains may help to resolve the map positions of single genes in the identified chromosomal regions and to decide whether the QTL represents a high performance variant of a single gene affecting both traits, or several genes. To exploit Bw4 for efficient trait selection, it will be necessary, however, to detect the primary polymorphisms in the functional genes. The region around Bw4 on murine chromosome 11 is conserved throughout evolution, since its human equivalent resides as a single block on human chromosome 17 (WATKINS-CHOW et al. 1996 ).

Additional support for the existence of QTL on growth found in this linkage analysis is provided by results from other QTL mapping studies. In a cross of SM/J x LG/J inbred lines, QTL responsible for age-specific growth were found on 16 of the 19 autosomes, with the marker D11Mit14 on chromosome 11 (31 cM) explaining 3.4% of the variance of weight at 6 wk of age (CHEVERUD et al. 1996 ). Another genome scan of mouse lines divergently selected for high and low body weight from a basic cross between two inbred strains revealed line-specific distributions of marker alleles on 10 autosomes; the QTL with the largest estimated effect was found on chromosome 7; for marker D11Nds16 on chromosome 11 (47 cM) an effect of 0.24 phenotypic standard deviations was estimated (KEIGHTLEY and BULFIELD 1993 ; KEIGHTLEY et al. 1996 ). Although the QTL effects on chromosome 11 detected by CHEVERUD et al. 1996 and KEIGHTLEY et al. 1996 are far below the cross-specific results of Bw4 in our analysis, the same QTL with different allelic variants might be responsible for the growth differences. In crosses from inbred lines, a QTL for body weight in the middle part of chromosome 13 (KIRKPATRICK et al. 1998 ) and a QTL with sex-specific influence on body weight on the distal part of chromosome X (RANCE et al. 1994 ) were proved. On the X chromosome, a locus influencing body length has been mapped in backcross progeny of C57BL/6J and Mus spretus (LEMBERTAS et al. 1996 ). Our results confirm mapping data for fat QTL on chromosomes 4 and 9 (WEST et al. 1994 ). The obesity QTL on mouse chromosome 2 reported by LEMBERTAS et al. 1997 may correspond to the QTL on body weight that we mapped to the same chromosome. Compared to previously reported QTL, those affecting abdominal fat weight on chromosomes 13 and 19 are specific to our cross.

The estimates for the best positions of QTL affecting body weight or abdominal fat give a valuable hint on the chromosomal position and, via gene maps, on the likely nature of candidate genes, known for their effects either on growth and/or obesity or on defects of the endocrine system, as shown in Figure 3. The QTL on chromosome 1, for instance, with a dominant effect on body weight and no influence on fat accumulation, maps close to the Mstd (myostatin) gene, where a dominant mutation causes muscle hypertrophy in mice and cattle (GROBET et al. 1997 ; MCPHERRON et al. 1997 ). Muscle hypertrophy with increased muscle diameter was found for high body weight in line DU6 after 40 generations of selection (REHFELD and BUNGER 1990 ). However, muscle characteristics were not evaluated in the F₂ animals of our cross. Similarly, positions of the QTL influencing body weight and fat on chromosome 4 as detected in our line DU6 support the proposed role of the Lepr (leptin receptor) gene on obesity in DU6. The Lep (leptin) gene itself, residing on chromosome 6, was not found to be linked to QTL affecting abdominal fat weight or growth rate. Likewise, we found an influence on body weight of the region on chromosome 15 harboring the gene encoding the receptor for growth hormone (Ghr), while no significant influence of the gene encoding growth hormone (Gh) was found. Although genes responsible for various types of obesity have been assigned to a number of chromosomes, until now no effects could be attributed to mouse chromosomes 11, 13, and 19 (WARDEN et al. 1995 ; YORK et al. 1996 ). For the fat QTL on chromosome 13 near the centromere, there is no candidate gene known influencing fat accumulation. The genes Vldlr and Ide encoding the very low density lipoprotein receptor and the insulin-degrading enzyme map close to the QTL region identified on chromosome 19.

Together, all loci with proven genome-wide significant effects contribute about one-third of the phenotypic variance of body weight and abdominal fat weight in the F₂ population. We concluded that the alleles of the identified QTL were homozygous in the parents of pedigree 8, because there was no hint of any influence of subfamilies on the QTL effect. However, it might be that we did not detect QTL because alleles were not fixed in one of the lines, reducing the contrast between alleles from the two lines and hence power of detection. It seems reasonable to assume fixation of QTL alleles for the selected DU6 parent, although within-line segregation at QTL cannot be excluded for the randomly bred DUKs parent. However, evidence that some QTL contributing to high body weight were not fixed in DU6 comes from experimental results revealing a reduction of the mean body weight after relaxed selection (G. A. BROCKMANN, unpublished data). Differences in trait means and variances were observed between pedigrees derived from crosses between DU6 and DUKs. These differences reflect heterogeneity between the individual parental genotypes contributing to each pedigree and lead to differentiated, pedigree-specific results in the linkage analysis of chromosome 11. These were evident for both location and effects of QTL responsible either for body weight variation or for abdominal fat weight. The subfamily effect on fatness in pedigree 10 may be a hint for heterozygous QTL genotypes of founder parents.

The observed sex-dependent influence on abdominal fat weight of three QTL is likely to be due to the different fat pads measured in the two sexes, in males the gonadal fat and in females the perimetrial fat, which may be regulated in a sex-specific manner.

The results obtained in this study provide a partial chromosomal dissection of the complex traits of body weight and fat. Genetic marker analysis found that in some cases a QTL for fatness was associated with a QTL for body weight. The exact identification and mapping of fat-growth QTL in mice is an important step in the way to identify genes that play an important role in the control of growth and fat accumulation. The present data permit a direct marker-assisted generation of congenic lines carrying those chromosomal regions of DU6 harboring the QTL.

	ACKNOWLEDGMENTS

Excellent technical assistance was provided by HANNELORE TYCHSEN for DNA preparation and genotyping. This work was supported by the German Research Foundation, Grants no. BR 1285/1-2 and BR 1285/4-1 and the H. Wilhelm Schaumann Stiftung. C.S.H. acknowledges support from the Biotechnology and Biological Sciences Research Council (BBSRC), Ministry of Agriculture, Fisheries and Food (MAFF), U.K., and the European Community (EC); S.A.K. is grateful for funding from the Royal Society.

Manuscript received March 31, 1998; Accepted for publication June 12, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BROCKMANN, G., J. BUITKAMP, L. BÜNGER, J. T. EPPLEN and M. SCHWERIN, 1993 DNA-fingerprinting of trait-selected mouse lines and linkage analysis in reference families, pp. 403–410 in DNA Fingerprinting: Approaches and Applications, edited by R. CHAKRABORTY, J. T. EPPLEN, A. J. JEFFREYS and S. D. J. PENA. Birkhäuser Verlag, Basel, Switzerland.

BÜNGER, L., L. SCHÜLER, B. KUPATZ, and U. RENNE, 1983  Selection for growth in laboratory mice. II. Direct selection response. Arch. Tierz. 26:281-293.

BÜNGER, L., G. HERRENDÖRFER and U. RENNE, 1990 Results of long term-selection for growth traits in laboratory mice. Proceedings of the 4th World Congress on Genetics Applied to Livestock Production, Univ. Edinburgh, Edinburgh, Scotland 13: 321–324.

BÜNGER, L., U. RENNE, and G. DIETL, 1992  Selection for body weight at 42 days in laboratory mice with and without litter size standardization—direct response and correlated effects on litter size. Arch. Tierz. 35:305-319.

CHEVERUD, J. M., E. J. ROUTMAN, F. A. M. DUARTE, B. VAN SWINDEREN, and K. COTHRAN et al., 1996  Quantitative trait loci for murine growth. Genetics 142:1305-1319[Abstract].

CHURCHILL, G. A. and R. W. DOERGE, 1994  Empirical threshold values for quantitative trait mapping. Genetics 138:963-971[Abstract].

DAS, P., G. BROCKMANN, L. MEYER, U. RENNE, and G. FREYER et al., 1996  The effect of a restricted region of chromosome 11 on body weight in mice under special consideration of the growth hormone gene locus. Arch. Tierz. 39:185-194.

DIETRICH, W., H. KATZ, S. E. LINCOLN, H.-S. SHIN, and J. FRIEDMAN et al., 1992  A genetic map of the mouse suitable for typing intraspecific crosses. Genetics 131:423-447[Abstract].

DIETRICH, W. F., J. MILLER, R. STEEN, M. A. MERCHANT, and D. DAMRON-BOLES et al., 1996  A comprehensive genetic map of the mouse genome. Nature 380:149-152[Medline].

GREGORIUS, H.-R., 1984  A unique genetic distance. Biom. J. 26:13-18.

GROBET, L., L. J. R. MARTIN, D. PONCELET, D. PIROTTIN, and B. BROUWERS et al., 1997  A deletion in the bovine myostatin gene causes the double-muscled phenotype in cattle. Nature Genet. 17:71-74[Medline].

HALEY, C. S., S. A. KNOTT, and L. M. ELSEN, 1994  Mapping quantitative trait loci in crosses between outbred lines using least squares. Genetics 136:1195-1207[Abstract].

HORVAT, S. and J. F. MEDRANO, 1995  Interval mapping of the high growth (hg), a major locus that increases weight gain in mice. Genetics 139:1737-1748[Abstract].

JANSEN, R. C., 1993  Interval mapping of multiple quantitative trait loci. Genetics 135:205-211[Abstract].

KEIGHTLEY, P. D. and G. BULFIELD, 1993  Detection of quantitative trait loci from frequency changes of marker alleles under selection. Genet. Res. 62:195-203[Medline].

KEIGHTLEY, P. D., T. HARDGE, L. MAY, and G. BULFIELD, 1996  A genetic map of quantitative trait loci for body weight in the mouse. Genetics 142:227-235[Abstract].

KIRKPATRICK, B. W., A. MENGELT, N. SCHULMAN, and I. C. A. MARTIN, 1998  Identification of quantitative trait loci for prolificacy and growth in mice. Mamm. Genome 9:97-102[Medline].

LANDER, E. and P. GREEN, 1987  Construction of multilocus genetic linkage maps in humans. Proc. Natl. Acad. Sci. USA 84:2363-2367[Abstract/Free Full Text].

LANDER, E. and L. KRUGLYAK, 1995  Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nature Genet. 11:241-247[Medline].

LEMBERTAS, A. V., J. S. FISLER, C. H. WARDEN, P.-Z. WEN, and Y.-R. XIA et al., 1996  A locus on the X chromosome is linked to body length in mice. Mamm. Genome 7:171-173[Medline].

LEMBERTAS, A. V., L. PÉRUSSE, Y. C. CHAGNON, J. S. FISLER, and 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[Medline].

MCPHERRON, A. C., A. M. LAWLER, and S.-J. LEE, 1997  Regulation of skeletal muscle mass in mice by a new TGF-ß superfamily member. Nature 386:83-90.

MGD, 1998 Mouse Genome Database, Mouse Genome Informatics, The Jackson Laboratory, Bar Harbor, ME, June 1998. (URL: http://www.informatics.jax.org/).

NEHLS, M., K. LÜNO, M. SCHORPP, D. PFEIFER, and S. KRAUSE et al., 1995  YAC/P1 contigs defining the location of 56 microsatellite markers and several genes across a 3.4 cM interval on mouse Chromosome 11. Mamm. Genome 6:321-331[Medline].

PÉRUSSE, L., Y. C. CHAGNON, F. T. DIONNE, and C. BOUCHARD, 1997  The human obesity gene map: the 1996 update. Obesity Res. 5:49-61[Medline].

POMP, D., A. M. OBERBAUER, and J. D. MURRAY, 1996  Development of obesity following inactivation of a growth hormone transgene in mice. Transgenic Res. 5:13-23[Medline].

RANCE, K. A., I. M. HASTINGS, W. G. HILL and P. D. KEIGHTLEY, 1994 Mapping of putative QTL influencing body weight on the X chromosome in mice. Proceedings of the 5th World Congress on Genetics Applied to Livestock Production, Univ. Guelph, Guelph, Ontario, Canada 21: 268–271.

RASCH, D., G. HERRENDÖRFER, J. BROCK and K. BUSCH, 1978 Verfahrensbibliothek Versuchsplanung und auswertung, Band 2. Deutscher Landwirtscheftsverlag, Berlin.

REHFELD, C. and L. BÜNGER, 1990  Effects of long-term selection of laboratory mice on parameters of muscle growth and muscle structure. Arch. Tierz. 33:507-516.

RIESNER, D., G. STEGER, R. A. OWENS, M. WAGENHÖFER, and W. HILLEN et al., 1989  Temperature-gradient gel electrophoresis of nucleic acids: analysis of conformational transitions, sequence variations, and protein-nucleic acid interactions. Electrophoresis 10:377-389[Medline].

SAS, 1990 SAS User's Guide: Statistics. SAS Inst. Inc., Cary, NC.

SCHÜLER, L., 1985  Mouse strain Fzt:Du and its use as model in animal breeding research. Arch. Tierz. 28:357-363.

VAN OOIJEN, J., 1992  Accuracy of mapping quantitative trait loci in autogamous species. Theor. Appl. Genet. 84:803-811.

WARDEN, C. H., J. S. FISLER, S. M. SHOEMAKER, P. Z. WEN, and K. L. SVENSON et al., 1995  Identification of four chromosomal loci determining obesity in a multifactorial mouse model. J. Clin. Invest. 95:1545-1552.

WATKINS-CHOW, D., M. ROLLER, M. M. NEWHOUSE, A. M. BUCHBERG, and S. A. CAMPER, 1996  Mouse chromosome 11. Mamm. Genome 6:201-220.

WEST, D. B., J. WAGUESPACK, B. YORK, J. GOUDEY-LEFEVRE, and R. A. PRICE, 1994  Genetics of dietary obesity in AKR/J x SWR/J mice: segregation of the trait and identification of a linked locus on chromosome 4. Mamm. Genome 5:546-552[Medline].

YORK, B., K. LEI, and D. B. WEST, 1996  Sensitivity to dietary obesity linked to a locus on chromosome 15 in a CAST/Ei x C57BL/6J F2 intercross. Mamm. Genome 7:677-681[Medline].

ZENG, Z.-B., 1993  Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc. Natl. Acad. Sci. USA 90:10972-10976[Abstract/Free Full Text].