Fine mapping of a QTL region with large effects on growth and fatness on mouse chromosome 2
Nancy C. Jerez-Timaure1,
Eugene J. Eisen2 and
Daniel Pomp1
1 Department of Animal Science, University of Nebraska, Lincoln, Nebraska
2 North Carolina State University, Raleigh, North Carolina
 |
ABSTRACT
|
---|
We combined the use of a congenic line and recombinant progeny testing (RPT) to characterize and fine map a previously identified region of distal mouse chromosome 2 (MMU2) harboring quantitative trait loci (QTL) with large effects on growth and fatness. The congenic line [M16i.B6-(D2Mit306-D2Mit52); MB2] was created using an inbred line (M16i) derived from a line that had undergone long-term selection for rapid weight gain (M16) as the recipient for an
38-cM region on MMU2 from the inbred line C57BL/6J. A large F2 cohort (1,200 mice) originating from a cross between MB2 and M16i was created, and 40 F2 males with defined recombinations within the QTL region were used to produce 665 segregating progeny. Linkage analysis of the F2 population detected QTL with very large effects on body weight, body fat, lean tissue mass, bone mineral density, and liver weight. Confidence intervals of the QTL were narrowed to regions of 1.54.5 cM. Analysis of progeny of the recombinant F2 males confirmed the existence of the QTL and further contributed to localization of their map positions. These efforts confirmed the presence of QTL with major effect on MMU2, narrowed the estimated region harboring the QTL from 38 to 12 cM, and further characterized phenotypic effects of the QTL, effectively culminating in a significantly decreased pool of positional candidate genes potentially representing these genes controlling predisposition to growth and fatness.
fine mapping; quantitative trait locus; obesity; congenic; recombinant progeny testing
 |
INTRODUCTION
|
---|
BODY WEIGHT, BODY COMPOSITION, AND ENERGY stored in adipose tissue are determined by a complex combination of genetic and environmental factors (6, 44). Identification of genes underlying predisposition to these traits may help to elucidate the fundamental mechanisms of growth and obesity. In humans, such knowledge would be valuable in the development of diagnostic and therapeutic approaches to treat obesity and comorbidities. In food-producing animals, it could enable manipulation of growth and body composition through marker-assisted selection or by transgenic strategies.
Lines of mice that have undergone long-term selection provide a uniquely powerful model to dissect and understand the complex polygenic architecture of quantitative traits with relevance to biomedicine and animal agriculture. The M16 line was the result of 27 generations of selective breeding for rapid growth rate (15), resulting in correlated responses of polygenic obesity and symptoms of type II diabetes (2). Full-sib mating within M16 has led to a fully inbred derivative denoted as M16i (21).
One method for identifying genes that regulate complex traits is quantitative trait locus (QTL) mapping. Many successful efforts to map QTL for growth and obesity in mice (33) and humans (43) have been published, including large-scale studies employing the M16i line (37, 38). However, a deficiency of the QTL approach is that resolution of mapping includes broad genomic confidence intervals encompassing hundreds or even thousands of genes, rendering positional cloning an extremely difficult proposition.
Several strategies have been proposed to increase the physical resolution of QTL detection (1) including fine mapping, using interval-specific congenic strains (11, 12), genome-tagged mice (19, 20), and chromosome substitution strains of mice (42). Congenic strains differ from their founder strains only for the genomic region containing a specific, well-defined chromosomal interval. Other strategies such as recombinant progeny testing have been successfully incorporated into fine-mapping paradigms to increase resolution (8, 10, 18, 26, 27, 52). The basis of progeny testing is to identify recombinant individuals in the interval where the putative QTL is located, propagate progeny representing the alternative allelic forms in this region, and determine whether there is a phenotypic difference among allelic class means (26).
In our QTL studies using the M16i line, a common finding is that the distal region of chromosome 2 harbors one or more QTL with very large effects on traits related to growth, body weight, and fatness (33, 37, 38). To fine map these QTL, we developed a congenic line [M16i.B6-(D2Mit306-D2Mit52); MB2] using M16i as the recipient for a
38-cM region on mouse chromosome 2 (MMU2) from the inbred line C57BL/6J. The phenotypic characterization of the MB2 congenic line was described previously (21). Briefly, absence of M16i alleles in this 38-cM region on MMU2 resulted in a 60% decrease in adiposity and 15% decrease in body weight in MB2, confirming the existence of loci within this region that have large effects on these traits. Several other studies have found QTL in this region with effects on body weight and adiposity (34, 49), and recent reports have described characterization of congenic lines with a variety of donor and recipient backgrounds (13, 49) with relevancy to gene discovery on MMU2.
We now report on use of the MB2 congenic strain to fine map and better characterize the QTL on distal MMU2. Using a combination of a large F2 intercross between M16i and MB2, and extensive progeny testing of F2 males with defined recombinations within the QTL region, we have built a map of QTL in this region, estimated their locations within intervals of 1.44.5 cM, and further defined the phenotypic consequences of alternative QTL alleles. Results of this effort will be instrumental in facilitating positional cloning of the underlying polygenes in this region of the mouse genome that help regulate predisposition to growth and obesity.
 |
MATERIALS AND METHODS
|
---|
Mouse lines.
Mouse lines used in these studies included the polygenically obese M16i line and the MB2 congenic line [M16i.B6-(D2Mit306-D2Mit52)] derived from the M16i line as recipient for a
38-cM region of MMU2 from the inbred line C57BL/6J (C57) obtained from the Jackson Laboratory (Bar Harbor, ME). The M16 line was derived from an outbred population (ICR) selected for rapid 3- to 6-wk weight gain for 27 generations (15). Subsequently, over 20 generations of full-sib mating were implemented to produce the inbred M16i line. Details of development of MB2 and phenotypic comparison between MB2 and M16i were previously reported (21).
Mouse care and maintenance.
Fraternal litter size was standardized 1 day after birth to 10 pups per litter. Pups were weaned at 3 wk and housed three to four per cage (by sex) with ad libitum access to feed (Teklad 8640 Rodent Chow) and water. Mice were reared at 22°C and 30% relative humidity and with a 12:12-h light-dark cycle, with lights on at 0700. All procedures and protocols were approved by the University of Nebraska Institutional Animal Care and Use Committee.
Experimental design.
The general experimental approach for fine mapping the MMU2 QTL involved creation of an F2 intercross population between M16i and MB2, followed by progeny testing of recombinant F2 males (Fig. 1). Work was divided into three specific phases: 1) initial selective genotyping of F2 with extreme fatness phenotypes to narrow the region containing the QTL, 2) QTL analysis of the entire F2 population to obtain additional mapping resolution, and 3) recombinant progeny testing of recombinant F2 males to evaluate specific and defined segments for the presence of QTL.

View larger version (30K):
[in this window]
[in a new window]
|
Fig. 1. Overall experimental approach for quantitative trait locus (QTL) fine mapping. Three phases were used: 1) selective genotyping to define significant intervals within the congenic region, 2) full genotyping and QTL analysis within the significant intervals, and 3) recombinant progeny testing. Reps, replicates.
|
|
Creation of the M16i x MB2 F2 intercross population.
A large F2 population (n = 1,200 mice) was created over four consecutive time periods (replicates) from a cross between MB2 and M16i. Matings in periods 1 and 3 were repeated in periods 2 and 4, respectively. The resource population was created by reciprocal crossing of five males with five females representing each line to produce the F1 generation, and then intermating the resulting F1 males and F1 females to produce the F2 population.
Recombinant progeny testing.
Thirty-five males carrying a recombinant segment of chromosome 2 between markers D2Mit304 and D2Mit52 were used in the F2 population as sires for recombinant progeny testing (RPT). These males possessed one evident recombinant (based on marker genotyping) in at least one of seven possible marker-to-marker intervals of the congenic segment. The progeny test involved crossing each recombinant male to four C57 and two M16i females, generating 424 and 241 F3 mice, respectively.
Measurement of phenotypes.
The same phenotypic variables were measured in both the F2 and RPT populations. Body weight was measured at 3, 6, 9, 12, and 15 wk of age. At 15 wk, all mice were killed by cervical dislocation after 1.5 h of fasting. The head was removed and trunk blood was collected, with serum subsequently separated by centrifugation and stored at 80°C. The subcranial body was then scanned using dual-energy X-ray absorptiometry (DEXA; Lunar, Madison, WI) to measure total body fat (FAT), total lean tissue (LEAN), and overall bone mineral density (BMD). These DEXA traits were evaluated for replicates 3 and 4 of the F2 population and all RPT progeny. For all mice, the right-side epididymal (male) or perimetrial (female) fat pad (PAD) and liver (LIV) were dissected, weighed, snap frozen in liquid nitrogen, and stored at 80°C. Serum glucose was measured with the glucose oxidase method, using the SureStep Blood Glucose Monitoring System (LifeScan).
Genotyping.
DNA lysate was prepared from toe clips (used to identify mice at 12 days of age) or from tail clip samples, following a general method described previously (35). Aliquots of 5 µl of DNA lysate were dispensed into 384-well plates, dried at room temperature, and stored at 40°C. Genotypes were determined by PCR using infrared dye labeling, followed by electrophoresis and analysis on the LI-COR Model 4200 IR2 system (LI-COR, Lincoln, NE). Gels were analyzed using Gene ImagIR analysis software (LI-COR) to determine allele sizes of each individual.
Genotyping was performed in three phases. Initially, selective genotyping was employed based on weight of epididymal and perimetrial PAD as percentages of body weight, selecting the highest and lowest (48 mice in each of the high and low groups, with approximately equal representation of each sex) from replicates 1 and 2. These 96 samples were scored for 16 markers covering the complete congenic segment of MMU2. Second, all F2 males from replicates 3 and 4 were genotyped for eight markers [between markers D2Mit304 and D2Mit52 (Table 1), as determined by analysis of the selective genotyping results]. This step identified males with specific recombinations to be used for RPT. Finally, all individuals in the F2 and RPT populations were subsequently genotyped for these eight markers.
Statistical analysis: selective genotyping.
Replicates 1 and 2 of the F2 population were used for these analyses. Deviation from Mendelian expectations (1:2:1) of individual markers within selectively genotyped mice was evaluated using two chi-square analyses. The first determined whether genotypic frequency for the given marker within the high or low selected phenotypic groups differed from the 1:2:1 Mendelian ratio expected for the null hypothesis of no linkage of the given marker to a hypothesized QTL in the region. The second analysis tested for marker allele frequency equivalence between the high and low selected phenotypic groups (High-Low), which is also expected under the null hypothesis of no linkage of the given marker to a QTL. Effects of individual marker genotypes were evaluated by analysis of variance, using the PROC MIXED procedure of SAS (39). Fixed effects included sex, marker genotype, and genotype-by-sex interaction, with dam as a random effect.
Statistical analysis: QTL detection.
A total of 18 traits were analyzed for the presence, most likely position, and effects of QTL in the complete F2 population (n = 1,200). Descriptive statistics and distributions for each trait were calculated using the MEANS and UNIVARIATE procedures of SAS (39). Various fixed effects were tested for each trait, including sex, replicate, family, and possible two-way interactions among these, using the GLM procedure of SAS (39). Traits displaying a significant departure from a normal distribution (PADP, FATP, LEANP; see RESULTS) were normalized, using a log transformation. Phenotypic correlations among the traits, adjusted for sex, were estimated using multivariate analysis of variance (PROC MANOVA; SAS).
To test for segregation distortion, marker genotype frequencies in the F2 were compared with the expected Mendelian ratio of 1:2:1 using a chi-square test. For all markers, the null hypothesis of expected Mendelian segregation was not rejected using a Bonferroni protected significance threshold of P > 0.006. This analysis also enabled evaluation of systematic genotyping errors, and none were found. Marker linkage maps were generated with MAPMAKER/EXP and marker distances calculated in this specific cross (Table 1) were used in the subsequent QTL analysis. Positions of markers and QTL (with their respective confidence intervals) were estimated in the physical map, using the Ensembl Mouse Genome Server.
Residuals from the fixed effects models were utilized with QTL Cartographer (v.1.15; Ref. 4) to first perform simple interval mapping (SIM) analysis (23). To obtain better resolution and precision of QTL location and effect, a composite interval mapping (CIM) analysis (53, 54) was performed. A forward-backward stepwise regression procedure (4) with a 0.01 threshold for the addition and elimination of new markers was used to select background factors for CIM, and a 0.5-cM window size was adopted. Experiment-wise significance thresholds were established with at least 1,000 permutations (4, 9). Additive (a) and dominance (d) effects of each QTL were estimated. Effects of QTL were estimated as the percentage of residual phenotypic variance explained by the QTL, and confidence intervals (CI) for each QTL were determined using a 1-LOD drop from the QTL peak (23).
Statistical analysis: RPT.
To further fine map QTL, two approaches were used: the recombinant ancestral haplotype analysis (17) and the contrast mapping method (48).
Phenotypic averages of haplotypes or offspring groups were calculated and contrasted to derive a specific comparison for each marker interval (48). The expectation is that the interval-specific contrast will be maximal in the true interval; therefore, the maximum P value (more significant) observed among the observed contrasts is assumed to contain the QTL. Data were analyzed using the GLM procedure of SAS (39) with the fixed effect of haplotype, sex, and sire. Haplotype effects were analyzed by defining contrasts to test specific allele substitution effects to determine the QTL effect at each specific interval.
 |
RESULTS
|
---|
Phenotypic characterization of the M16i x MB2 F2 population.
Means, standard deviations, and ranges for all traits averaged across males and females in the M16i x MB2 F2 population are presented in Table 2. These traits include body weight at 6 (WT6), 9 (WT9), 12 (WT12), and 15 wk (WT15); body weight gain between 3 and 6 (GAIN36), 6 and 9 (GAIN69), 9 and 12 (GAIN912), and 12 and 15 wk (GAIN1215); liver weight (LIV); liver weight expressed as percent kill weight (LIVP); epididymal (males) or perimetrial (females) fat pad weights (PAD); fat pad weight expressed as percent kill weight (PADP); total fat weight (FAT); total fat weight expressed as percent kill weight (FATP); total lean weight (LEAN); total lean weight expressed as percent kill weight (LEANP); BMD; and blood glucose. A very large variation in phenotypes was created in the F2 population, with mice ranging from 34.47 to 76.60 g (mean 56.16, standard deviation 7.45) for WT15, from 2.56 to 20.66 g (mean 9.12, standard deviation 2.18) for FAT, and from 5.88 to 36.51% (mean 17.05, standard deviation 3.98) for FATP.
View this table:
[in this window]
[in a new window]
|
Table 2. Means (averaged across gender) and basic statistics for body weight at different ages, weight gain, liver weight, and body composition traits in the M16i x MB2 F2 population
|
|
Phenotypic correlations between WT15, body composition traits and glucose in the F2 population are presented in Table 3. As expected, WT15 is highly correlated with PAD, LIV, FAT, LEAN, and BMD. However, these correlations were much smaller in magnitude when composition traits were expressed as a percentage of body weight. Both LIV and LIVP are positively correlated with LEAN and LEANP and negatively correlated with all fat traits. The high correlation between PAD and FAT (r = 0.832) demonstrates that PAD is a relatively good predictor of FAT in this population. All fat traits had high negative correlations with lean traits. Correlations between weight and composition traits with glucose were low (0.10.2) but significantly (P < 0.01) positive.
Selective genotyping of the M16i x MB2 F2 population.
Observed segregation ratios for 16 markers covering the congenic region (6199 cM) of MMU2 in mice with the highest and lowest phenotypes for PADP in the M16i x MB2 F2 population did not deviate from expected (1:2:1) in the high group, but differences were observed in the low group and between the two groups (High-Low) for markers located between 71 and 99 cM (Table 4). Eight markers mapping between 73 and 99 cM on chromosome 2 showed potential linkage (P < 0.05) with putative QTL regulating adiposity (Table 4). These eight markers were thus genotyped in the entire F2 population, and all subsequent description and discussion of results (including all figures) relate to this reduced region (D2Mit304 to D2Mit52) and not the full congenic region (D2Mit306 to D2Mit52).
QTL analysis.
Positions of markers based on linkage analysis in the F2 population and comparison with those found in the Mouse Genome Database (MGD; http://www.informatics.jax.org/) and with the physical map from the Ensembl Mouse Genome Server (MGSC; http://www.ensembl.org/Mus_musculus/, accessed October 2004, build v.25.33a.1) are presented in Table 1. In general, our linkage map agreed with the linkage map from the MGD; the largest discrepancy is for the interval between D2Mit106 and D2Mit411 (8.4 vs. 2.0 cM). A discrepancy between our linkage map and the physical map from the MGSC was observed for markers D2Mit411 and D2Mit343. On the basis of the MGSC, marker D2Mit411 is located at 160.2 Mb, distal to marker D2Mit285, and marker D2Mit343 is located at 169.9 Mb, distal to marker D2Mit52. Clearly, knowing the true locations of the markers used in this study is critical in resolving QTL positions based on fine mapping. These discrepancies will likely not be adequately clarified until additional sequencing is completed across several mouse strains and a consensus physical map is generated with very high confidence. Because the linkage map generated in this study is based on a very large number of meioses, we consider this order and spacing of markers to be suitable for use in interpreting fine-mapping data within this population.
SIM analysis detected loci influencing WT9, WT12, WT15, GAIN69, PAD, PADP, LIV, LIVP, FAT, FATP, LEANP, and BMD (data not shown). The experiment-wise 5% LOD for SIM, based on permutation analysis of all traits, ranged between 2.8 and 3.5. For simplicity, the average of 3.15 was used as the significant threshold for all traits.
The location of, peak LOD score for, and effect of QTL, using CIM, are presented in Table 5. The 5% experiment-wise significant threshold (LOD) determined from 1,000-permutation testing ranged from 3.0 to 4.5 across the traits measured (displayed as horizontal dotted lines in Figs. 26). Most of the QTL map to positions that are similar to those determined in earlier studies; thus new symbols representing these QTL are not provided, and reference should be made to the QTL symbols proposed by Rocha et al. (37, 38).
View this table:
[in this window]
[in a new window]
|
Table 5. Locations and effects of QTL controlling body weights and body composition-related traits estimated from the M16i x MB2 F2 population
|
|
Figure 2 graphically displays CIM results for WT9, WT12, WT15, and GAIN69. A single QTL flanked by markers D2Mit194 and D2Mit343, but more closely linked to D2Mit194, appears to explain the observed variation in these four weight traits. The CI for this QTL is between 136.5 and 147.5 Mb (78.182.4 cM). The proportion of residual phenotypic variance explained by this body weight QTL ranged from 2.5 to 3.8%. A QTL detected for BMD and another for LEANP mapped to slightly different positions but still close to marker D2Mit194 (Fig. 3). The most likely positions for each of these QTL were essentially within each of their respective CIs (Table 5), which might imply a single pleiotropic QTL. The QTL for BMD and that for LEANP explained 8.5 and 11.7% of the phenotypic variance for each trait, respectively.

View larger version (25K):
[in this window]
[in a new window]
|
Fig. 2. LOD score plots obtained by composite interval mapping (CIM) for growth traits. Arrows indicate approximate QTL positions. WT9, 9-wk body wt; WT12, 12-wk body wt; WT15, 15-wk body wt; GAIN69, body wt gain between 12 and 15 wk.
|
|

View larger version (20K):
[in this window]
[in a new window]
|
Fig. 3. LOD score plots obtained by CIM for total lean as percent body wt (LEANP) and bone mineral density (BMD). Arrows indicate approximate QTL positions.
|
|
For traits representing measures of adiposity, evidence for up to three different QTL was detected (Figs. 4 and 5). A QTL influencing fat traits unadjusted for body weight (PAD and FAT) is located near marker D2Mit194, with a CI between 142.4 and 153.3 Mb (80.484.2 cM) that partially overlaps with the CI for body weight QTL. Effects of this QTL on PAD and FAT explain 13.3 and 18.4% of their residual phenotypic variances, respectively. Very strong evidence for a QTL influencing fat traits adjusted for body weight (PADP and FATP) was detected between markers D2Mit285 and D2Mit263, with a CI between 157.3 and 161.3 Mb (88.491.6 cM) This QTL explains 8.7 and 10.8% of the phenotypic variation of PADP and FATP, respectively. A third QTL for PADP was detected between D2Mit343 and D2Mit285, with a CI between 154.0 and 156.9 Mb (84.686.4 cM). This QTL explains 8.3% of the phenotypic variance in PADP.

View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4. LOD score plots obtained by CIM for total fat wt (FAT) and total fat as percent body wt (FATP). Arrows indicate approximate QTL positions.
|
|

View larger version (18K):
[in this window]
[in a new window]
|
Fig. 5. LOD score plots obtained by CIM for epididymal/perimetrial fat pad wt (PAD) and epididymal/perimetrial fat expressed as percent body wt (PADP). Arrows indicate approximate QTL positions.
|
|
Figure 6 shows a QTL influencing LIV and LIVP mapping close to D2Mit106, in an interval between D2Mit304 and D2Mit106. This QTL accounts for
4.0% of the residual phenotypic variance for LIV, with a CI between 128.5 and 134.5 Mb (73.076.6 cM).

View larger version (18K):
[in this window]
[in a new window]
|
Fig. 6. LOD score plots obtained by CIM for liver wt (LIV) and liver as percent body wt (LIVP). Arrows indicate approximate QTL positions.
|
|
The mode of gene action for all QTL detected in this study was almost entirely additive in nature (Table 5). Estimates of additive effects for each of the detected QTL demonstrate that the C57 allele always decreases body weight, fat, and BMD, while increasing lean tissue, relative to the M16i allele.
RPT.
Progeny were grouped into distinct paternal multi-marker haplotypes, with haplotypes differing as to the location of single recombination events. Details of each haplotype class, genotype, and markers present in each interval are presented in Table 6.
The P value from an interval-specific contrast was used as the criterion for determining QTL effect. Table 6 shows the defined contrasts and the P values for the parental haplotypes. The maximum P value observed between contrasts is assumed to represent the interval harboring the QTL. Recombinant progeny with haplotype MH2 (interval 75.677.6 cM) were significantly different from those with MH4 (interval 81.484 cM) for WT12 (Fig. 7). Furthermore, the reciprocal comparison (haplotypes HM4 vs. HM2) was also significantly different for WT9 and WT12. In both groups, mice with homozygous M16 genotypes at D2Mit411 and at D2Mit194 were heavier than mice that had heterozygous genotypes at those markers. The contrast between recombinants from M16i dams showed significant phenotypic effects upon allele substitution in two adjacent marker intervals (D2Mit106D2Mit411 and D2Mit194D2Mit343); however, no mice with the intermediate haplotype HM3 or MH3 were obtained in this study. As a result, a precise determination of which of the two intervals is more likely to harbor the QTL is not possible from the current data.

View larger version (20K):
[in this window]
[in a new window]
|
Fig. 7. Phenotypic contrasts for significant recombinant haplotype classes. A: means of WT9 and WT12. B: means of epididymal/perimetrial PAD and epididymal/perimetrial PADP. C: means of FAT and FATP. Details of the haplotype classes are presented in Table 6.
|
|
As was seen for body weight, mice with recombinations in interval 2 (75.677.6 cM) had significantly different body composition phenotypes vis a vis those with recombinations in interval 4 (81.484 cM). Mice with haplotype HM4 (interval 4) possessed more PAD, FAT, and FATP (Fig. 7) than those with HM2 (interval 2). The interval-specific contrast was maximized in terms of P value within the marker interval D2Mit343D2Mit285 (Table 6). Haplotype MH5 (interval 84.286.0 cM) was different from MH6 (interval 86.092 cM) in terms of PAD, PADP, FAT, and FATP phenotypes. Progeny with two copies of the M16i allele at marker D2Mit285 had more PAD and FAT than those with one copy. When comparing recombinants from the C57 background, the interval from 81.4 to 84.2 cM also appeared to harbor a QTL. The haplotype with recombinants in the interval 81.484 cM (HC4) had lower PAD, PADP, FAT, and FATP than the haplotype with recombinants in the interval 84.286.0 cM (HC5) (Fig. 7). No significant contrast differences were detected for LIV or BMD (data not shown).
Interestingly, nearly all contrasts found to be significant (P < 0.01) within recombinant haplotype classes were found in the subset of data generated by M16i dams. The parallel contrasts in the subset of data generated by the C57 dams for the most part approached significance, so it is likely that this observation reflects differences in magnitude or scale of QTL effects in the two genetic backgrounds.
 |
DISCUSSION
|
---|
Whole genome scans have proven to be extremely useful for coarse mapping of the chromosomal locations of QTL. However, proceeding from initial QTL mapping to eventual cloning of the underlying responsible gene(s) remains a very difficult and elusive task. Further mapping using congenic lines and subsequently subcongenic lines (or recombinant progeny testing) offers a means for more precise QTL localization, and establishes definitive limits for a QTL-containing region (31). The combination of interval-specific congenic strains and recombinant progeny testing has led to successful reduction of QTL intervals, and positional identification of at least one confirmed QTL with very large effect has resulted from such efforts (3).
Using the M16i model of polygenic obesity, Pomp (33) reported two QTL (Pfat1, Pfat2) in an interval of
12 cM on MMU2 affecting both fat and body weight. Rocha et al. (37, 38) also reported QTL for body weight and fat-related traits in a second cross utilizing M16i, providing significantly more details on location, effects, and gene action of the QTL. Many other studies (7, 16, 25, 29, 47, 50), using a variety of different mouse strains, have also reported QTL for body weight and/or fatness within or near this same interval, supporting the significance of distal MMU2 for the genetic regulation of growth and body composition. This region of MMU2 is uniquely conserved relative to HSA20 (55), which harbors QTL for obesity and non-insulin-dependent diabetes mellitus phenotypes based on multiple reports.
In a previous study (21), we described development and characterization of the MB2 congenic line [M16i.B6-(D2Mit306-D2Mit52)], consisting of the M16i line as host harboring
38 cM of distal MMU2 from C57BL/6J. The replacement of M16i alleles with C57 alleles in the congenic region on MMU2 resulted in a decreased adiposity of 60% and a decreased body weight of 15%, as well as decreased levels of plasma leptin, insulin, and glucose. In this study, we used a cross between M16i and MB2 to further confirm the presence in this region of M16 alleles with large effects on growth and fatness, and to fine map their locations.
Initially, selective genotyping based on percentage epididymal/perimetrial fat (PADP) in the F2 population allowed the chromosomal interval harboring QTL to be reduced from 38 to 26 cM. Genotypic frequencies deviating from Mendelian expectation, which suggests linkage to a QTL, were observed only within the low selected group. This result strongly suggests a strong relationship of the low fat phenotype with the absence of the M16 allele in the distal MMU2. There may be more genetic heterogeneity within the high fat extreme relative to the low fat extreme, which has been observed previously in a different mouse model of polygenic obesity (40). Using the full F2 population, we next delineated QTL effects to smaller, marker-defined regions, effectively reducing the QTL intervals for body weight, total lean, total fat, and BMD to regions of 1.54.5 cM. This also allowed us to definitively determine that the large QTL effects on MMU2 observed in previous crosses using M16i (37, 38), and confirmed in the characterization studies of the MB2 line (21), are actually the result of multiple linked genes as previously suspected (13, 33, 37, 38) for this region. The general finding that an initially strong QTL peak is represented by the presence of more than one underlying gene confirms previous theory (33) and supports similar recent evidence from other regions (45).
The close proximity of a significant QTL [likely representing Epfq1 and Scfq1, from Rocha et al. (37, 38)] associated with fat, lean, and BMD to a QTL affecting body weight [likely representing W10q1, from Rocha et al. (37)] suggests strong pleiotropic effects of a single underlying gene. The strong correlation structure between these traits supports this theory. Further evaluation such as multiple-trait QTL analysis may help test this hypothesis of the presence of a single QTL with pleiotropic effect on body weight and body composition, but definitive proof will require identification of the underlying gene and its function.
Epididymal (or perimetrial) fat is a reliable predictor of total body fat (r = 0.82). The expectation in this study was to find the same QTL for both traits. However, we found one QTL influencing PADP but not (simultaneously) influencing percentage of total body fat. Mehrabian et al. (29) found differential genetic effects on measures of regional body fat. The proximal (Mob7) and central (Mob6) QTL on MMU2 influenced subcutaneous fat, whereas a more distal QTL (Mob5) affected percent total body fat. A similar pattern was found in this study. Our distal QTL mostly affected total body fat, whereas the proximal QTL influenced percentage of epididymal/perimetrial fat.
As suggested above, the fact that the QTL affecting fat-related traits (PAD and FAT) are located at the same marker interval as the QTL for body weight (D2Mit194D2Mit343) is probably due to variation in body weight. However, the second peak located between D2Mit343 and D2Mit263 [likely representing Epfq2, from Rocha et al. (37, 38)] appears to influence adiposity independently of body weight. Together, these two QTL regions explain
30% of the variation in the F2 population, which is in line with expectations based on the heritability of body fat levels.
A QTL was found to affect liver weight within the interval D2Mit304D2Mit106 (3.2 cM), similar to what was found (Lvrq1; Ref. 37) using the M16i x L6 cross. Moody et al. (30) reported seven QTL affecting liver weight, each accounting for 25% of residual variance, but none of these QTL mapped to MMU2. Leamy et al. (24), using a backcross between Cast/Ei and M16i, reported a QTL for liver weight close to marker D2Mit164, which is 2 cM distal to that found in this study. Most likely, the QTL from the present study and from Leamy et al. represent the same underlying gene.
BMD is a predictor of skeletal fragility underlying osteoporosis (36). Masinde et al. (28) reported nine QTL on chromosomes 1, 2, 4, 9, 11, 14, and 15 for BMD in a mouse intercross. One of the two QTL on chromosome 2 reported by Masinde et al. coincides with the QTL reported in this study, also supporting additional previous findings (5, 22).
The recombinant progeny testing element of this study involved identifying recombinants in the congenic interval, propagating these recombinants into progeny, and then performing contrast analysis to determine the existence of phenotypic differences among marker (or haplotype) class means (8, 26). Because the alleles at the QTL segregate, marker allelic means should give rise to phenotypic contrasts that will be large or small depending on where the QTL lies with respect to a marker-interval recombination point. For all traits except BMD, analysis of the segregation of parental recombinant haplotypes confirmed the existence of QTL but did not further increase mapping resolution relative to the large F2 intercross. Considered independently, both methods provided similar results and were equally valuable in this fine-mapping effort, but their benefits were not additive or synergistic. The absence of recombinant progeny in interval 3 impeded making gains in mapping resolution. The distance between markers D2Mit194 and D2Mit343 is 2.8 cM, so that additional recombinants within that interval would help reduce the size of the QTL region. Further genotyping within the RPT population will be useful in this regard and may thus provide benefits beyond those obtained from the F2 mapping. Denser genotyping will also help resolve the actual placement of recombinations in haplotype classes where there are two or more males, which would aid in achieving finer mapping of the QTL.
Several potential candidate genes that may underlie these fine-mapped QTL can be identified based on their map positions and their associations with known biological mechanisms of growth and/or fatness. There are also still many other genes in this region with functions that are as yet unknown. And furthermore, the nature of a QTL may be regulatory and may not represent genetic variation within primary genes of a pathway. Microarray analysis can be an important tool in candidate gene selection, especially if the array included essentially all transcripts residing in the QTL interval. However, utility of expression analysis in this regard assumes existence of detectable differences in steady-state mRNA levels at the temporal and spatial coordinates selected for study. A previous transcriptional profiling revealed differential expression of several genes in liver and adipose tissue between the M16i and MB2 lines (21), several of which map within the CIs of the QTL for liver, body weight, lean, and fat (Table 7). On the basis of this differential expression and on their map position, these genes become viable positional candidates for the QTL fine mapped in the present study, although none of them maps in close proximity to any of the QTL for body weight. In particular, somatostatin receptor 4 (Smstr4, located at 148 Mb based on the MGSC) maps directly on top of the peak for a QTL impacting fat and lean tissue mass. Additional examples of differentially expressed genes mapping in close proximity to QTL detected in this study include the acetyl-CoA synthetase-like (Acas2), MAPK phosphatase (Dusp15), and growth differentiation factor-5 (Gdf-5) loci (Table 7). Strategies for determining that a gene underlies a QTL effect have been recently summarized (1).
While detailed description of every candidate gene is not feasible within the context of this discussion, it is worth noting that the diminutive locus (dm), discovered on the basis of a spontaneous mouse mutant that was shorter than average with low body weight (45), maps to 80 cM on MMU2, which coincides with one of the fine-mapped intervals harboring a body weight and fatness QTL in this study. It is thus possible that this QTL represents a hypomorph of the dm spontaneous mutation, the underlying gene for which has yet to be identified.
By identifying expression QTL that coincided with a QTL peak for, and had correlations with, subcutaneous fat mass, Schadt et al. (40, 41) proposed the NM_025575 and NM_15731 genes as candidates for the distal MMU2 QTL for dietary-induced obesity, previously described (25). These candidate genes are located at
165 Mb based on the MGSC, thus falling outside the CIs for any of the QTL localized using the M16i selection line of mice.
The QTL characterized in this study are all harbored in the region of 130160 Mb, based on the MGSC. The adiposity QTL recently fine mapped (49) using the UW-H3b we Pax1un at/Sn (B10.UW) congenic strain is found in this same general region of MMU2. In addition, the B6.S-D2Mit194-D2Mit311 congenic strain was recently used to isolate a QTL for percent body lipid within the same interval (13). In each of these cases, a C57-derived strain was used as recipient in the congenic line, and loss of C57 alleles led to a decrease in adiposity. This is in direct contrast to our work, where C57 was the donor strain for the congenic line, and where inclusion of C57 alleles dramatically decreased body weight and fatness. When the MB2 line was originally developed, a reciprocal congenic line (M16i as donor and C57 as recipient) was also initiated but did not reach fruition due to reproductive problems. The selection of C57 germplasm as a suitable contrast for M16 was based on two criteria. First, C57 is a widely used inbred line in biomedical (and obesity) research, and second, C57 mice have been shown to be relatively lean when fed a diet that was not high fat in nature (51). The opposite effects of C57 alleles in the present study relative to previous studies (13, 49) may be attributable to our use of the polygenically obese M16i strain as recipient in the MB2 congenic line, or other genetic background interactions.
Extensive sequencing of genes within this region has not been carried out within the M16i line. Such information would be extremely valuable to identify possible polymorphisms between M16i and C57BL/6J. It would also provide increased power to the method of identifying candidate genes based on the pattern of polymorphisms across different strains (32).
In summary, identification of genes underlying quantitative traits is a major challenge. Congenic strains and recombinant progeny testing, among other resources and strategies, are valuable tools for fine mapping of QTL. The polygenic obese M16 selection line is an excellent resource population to initiate positional cloning of novel genes with effects on growth and fatness that have been previously localized to distal MMU2, affecting growth at different ages, fat, lean, and BMD. In this study, the presence of several QTL on distal MMU2 affecting fatness and growth was confirmed and their locations mapped to intervals of 1.54.5 cM. Although recombinant progeny testing validated the existence and location of the QTL for most of the traits, little gain in resolution was obtained relative to mapping in the F2 population. However, further addition of markers within the significant marker intervals will likely increase the mapping precision of the QTL using RPT. Several potential candidate genes were identified based on their map position relative to the QTL and their association with growth and fat metabolism.
 |
GRANTS
|
---|
N. Jerez-Timuare was supported by fellowships from La Universidad del Zulia of Venezuela. This research is a contribution of the University of Nebraska Agricultural Research Division (Lincoln, NE; Journal Series No. 14686) and the North Carolina Agricultural Research Service and was supported in part by funds provided through the Hatch Act. This research was also partially based on work supported by the National Science Foundation under grant no. 0091900 (Nebraska EPSCOR Infrastructure Improvement Grant).
 |
ACKNOWLEDGMENTS
|
---|
We are appreciative of the efforts of Sara Olberding, Jeryl Hauptman, Mark Allan, and Stephanie Wesolowski in assisting with mouse husbandry and phenotypic data collection. We thank Dr. John Rocha and Dr. James Specht for assistance with QTL analysis. This manuscript was significantly improved on the basis of insightful revisions suggested by two anonymous reviewers.
Present address of N. Jerez-Timuare: La Universidad del Zulia, Facultad de Agronomía, Zulia, Venezuela.
 |
FOOTNOTES
|
---|
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: D. Pomp, Dept. of Animal Science, Univ. of Nebraska, Lincoln, NE 68583-0908 (e-mail: dpomp{at}unl.edu).
doi:10.1152/physiolgenomics.00256.2004.
 |
REFERENCES
|
---|
- Abiola O, Angel J, Avner P, Bachmanov A, Belknap J, Bennett B, Blankenhorn E, Blizard D, Bolivar V, Brockmann GA, et al. The nature and identification of quantitative trait loci: a communitys view. Nat Rev Genet 4: 911916, 2003.[CrossRef][ISI][Medline]
- Allan MF, Eisen EJ, and Pomp D. The M16 mouse: an outbred animal model of polygenic obesity and obesity induced diabetes (diabesity). Obesity Res 12: 13971407, 2004.[Abstract/Free Full Text]
- Bachmanov AA, Li X, Reed DR, Ohmen JD, Li S, Chen Z, Tordoff MG, de Jong PJ, Wu C, West DB, Chatterjee A, Ross DA, and Beauchamp GK. Positional cloning of the mouse saccharin preference (Sac) locus. Chem Senses 26: 925933, 2001.[Abstract/Free Full Text]
- Basten CJ, Weir BS, and Zeng ZB. QTL Cartographer (v.1.15). Raleigh, NC: North Carolina State Univ., 2001.
- Benes H, Weinstein RS, Zheng W, Thaden JJ, Jilka RL, Manolagas SC, and Reis RJS. Chromosomal mapping of osteopenia-associated quantitative trait loci using closely related mouse strains. J Bone Miner Res 15: 626633, 2000.[ISI][Medline]
- Bouchard C, Tremblay A, Nadeau A, Despres JP, Theriault G, Boulay MR, Lortie G, Leblanc C, and Fournier G. Genetic effect in resting and exercise metabolic rates. Metabolism 38: 364370, 1989.[CrossRef][ISI][Medline]
- Cheverud JM, Routman EJ, Duarte FA, Van Swinderen B, Cothran K, and Perel C. Quantitative trait loci for murine growth. Genetics 142: 13051319, 1996.[Abstract/Free Full Text]
- Christians JK and Keightley PD. Fine mapping of a murine growth locus to a 1.4-cM region and resolution of linked QTL. Mamm Genome 15: 482491, 2004.[CrossRef][ISI][Medline]
- Churchill GA and Doerge RW. Empirical threshold values for quantitative trait mapping. Genetics 138: 963971, 1994.[Abstract/Free Full Text]
- Cicila GT, Garrett RW, Lee SJ, Liu J, Dene H, and Rapp JP. High-resolution mapping of the blood pressure QTL on chromosome 7 using dahl rat congenic strains. Genomics 72: 5160, 2001.[CrossRef][ISI][Medline]
- Darvasi A. Interval-specific congenic strains (ISCS): an experimental design for mapping a QTL into a 1-centimorgan interval. Mamm Genome 8: 163167, 1997.[CrossRef][ISI][Medline]
- Darvasi A. Experimental strategies for the genetic dissection of complex traits in animal models. Nat Genet 18: 1924, 1998.[CrossRef][ISI][Medline]
- Diament AL, Farahani P, Chiu S, Fisler J, and Warden CH. A novel mouse Chromosome 2 congenic strain with obesity phenotypes. Mamm Genome 15: 452459, 2004.[CrossRef][ISI][Medline]
- Diament AL and Warden CJ. Multiple linked mouse chromosome 7 loci influence body fat mass. Int J Obes Relat Disord 28: 199210, 2004.
- Eisen EJ. Population size and selection intensity effects on long-term selection response in mice. Genetics 79: 305323, 1975.[Abstract/Free Full Text]
- Estrada-Smith D, Castellani LW, Wong H, Wen PZ, Chui A, Lusis AJ, and Davis RC. Dissection of multigenic obesity traits in congenic mouse strains. Mamm Genome 15: 1422, 2004.[CrossRef][ISI][Medline]
- Farnir F, Grisart B, Coppieters W, Riquet J, Berzi P, Cambisano N, Karim L, Mni M, Moisio S, and Simon P. Simultaneous mining of linkage and linkage disequilibrium to fine map quantitative trait loci in outbred half-sib pedigrees. Genetics 161: 275287, 2002.[Abstract/Free Full Text]
- Fehr C, Shirley RL, Belknap RK, Crabbe JC, and Buck KJ. Congenic mapping of alcohol and pentobarbital withdrawal liability loci to a <1 centimorgan interval of murine chromosome 4: identification of Mpdz as candidate gene. J Neurosci 22: 37303738, 2002.[Abstract/Free Full Text]
- Flier JS. Obesity wars: molecular progress confronts an expanding epidemic. Cell 116: 337350, 2004.[CrossRef][ISI][Medline]
- Iakoubova OA, Olsson CL, Dains KM, Ross DA, Andalibi A, Lau K, Choi J, Kalcheva I, Cunanan M, and Louie J. Genome-tagged mice (GTM): two sets of genome-wide congenic strains. Genomics 74: 89104, 2001.[CrossRef][ISI][Medline]
- Jerez-Timaure N, Kearney F, Simpson B, Eisen EJ, and Pomp D. Characterization of quantitative trait loci with major effects on fatness and growth on mouse chromosome 2. Obesity Res 12: 14081420, 2004.[Abstract/Free Full Text]
- Klein RF, Mitchell SR, Philips TJ, Belnak JK, and Orwoll ES. Quantitative trait loci affecting peak bone mineral density in mice. J Bone Miner Res 12: 16571659, 1998.[ISI]
- Lander ES and Botstein D. Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121: 185199, 1989.[Abstract/Free Full Text]
- Leamy LJ, Pomp D, Eisen EJ, and Cheverud JM. Pleiotropy of quantitative trait loci for organ weights and limb bone lengths in mice. Physiol Genomics 10: 2129, 2002.[Abstract/Free Full Text]
- Lembertas AV, Perusse L, Chagnon YC, Fisler JS, Warden CH, Purcell-Huyn DA, Dionne FT, Gagnon J, Nadeau A, Lusis AJ, and Bouchard C. Identification of an obesity quantitative trait locus on mouse chromosome 2 and evidence for linkage to body fat and insulin on the human homologous region 20q. J Clin Invest 100: 12401247, 1997.[Abstract/Free Full Text]
- Liu X, Oliver F, Brown SDM, Denny P, and Keightley PD. High-resolution quantitative trait locus mapping for body weight in mice by recombinant progeny testing. Genet Res 77: 191197, 2001.[CrossRef][ISI][Medline]
- Lyons PA, Armitage N, Argentina F, Denny O, Hill NJ, Lord CJ, Wilusz MB, Peternon LB, Wicker LS, and Todd JA. Congenic mapping of the Type 1 diabetes locus, Idd3, to 780-kb region of mouse chromosome 3: identification of candidate segment of ancestral DNA by haplotype mapping. Genome Res 10: 446453, 2000.[Abstract/Free Full Text]
- Masinde GL, Li X, Gu W, Wergedal J, Mohan S, and Baylink DJ. Quantitative trait loci for bone density in mice. Calcif Tissue Int 71: 421428, 2002.[CrossRef][ISI][Medline]
- Mehrabian M, Wen PZ, Fisler J, Davis RC, and Lusis AJ. Genetic loci controlling body fat, lipoprotein metabolism, and insulin levels in a multifactorial mouse model. J Clin Invest 101: 24852496, 1998.[Abstract/Free Full Text]
- Moody DE, Pomp D, Nielsen MK, and Van Vleck DL. Identification of quantitative trait loci influencing traits related in energy balance in selection and inbred lines of mice. Genetics 152: 699711, 1999.[Abstract/Free Full Text]
- Moore KJ and Nagle NL. Complex trait analysis in the mouse: the strengths, the limitations, and the promise yet to come. Annu Rev Genet 34: 653686, 2000.[CrossRef][ISI][Medline]
- Pletcher MT, McClurg P, Batalov S, Su AI, Barnes SW, Lagler E, Korstanje R, Wang X, Nusskern D, Bogue MA, Mural RJ, Paigen B, and Wiltshire T. Use of a dense single nucleotide polymorphism map for in silico mapping in the mouse. PLoS Biol 2: e393, 2004.[Medline]
- Pomp D. Genetic dissection of obesity in polygenic animal models. Behav Genet 27: 285306, 1997.[CrossRef][ISI][Medline]
- Pomp D, Allan MF, and Wesolowski S. Quantitative genomics: exploring the genetic architecture of complex trait predisposition. J Anim Sci 82: E300E312, 2004.[Abstract/Free Full Text]
- Pomp D and Murray J. Single day detection of transgenic mice by PCR of the toe-clips. Mouse Genome 89: 279280, 1991.
- Ralston SH. The genetics of osteoporosis. Bone 25: 8586, 1999.[CrossRef][ISI][Medline]
- Rocha JL, Van Vleck LD, Eisen EJ, and Pomp D. A large-sample QTL study in mice. I. Growth. Mamm Genome 15: 8399, 2004.[CrossRef][ISI][Medline]
- Rocha JL, Van Vleck LD, Eisen EJ, and Pomp D. A large-sample QTL study in mice. II. Body composition. Mamm Genome 15: 100113, 2004.[CrossRef][ISI][Medline]
- SAS. SAS/STAT Users Guide. Cary, NC: SAS Institute, 1990.
- Schadt EE, Monks SA, Drake TA, Lusis AJ, Che N, Colinayo V, Ruff TG, Milligan SB, Lamb JR, and Cavet G, Linsley PS, Mao M, Stoughton RB, and Friend SH. Genetics of gene expression surveyed in maize, mouse and man. Nature 422: 297302, 2003.[CrossRef][ISI][Medline]
- Schadt EE, Monks SA, and Friend SH. A new paradigm for drug discovery: integrating clinical, genetic, genomic and molecular phenotype data to identify drug targets. Biochem Soc Trans 31: 437443, 2003.[CrossRef][ISI][Medline]
- Singer JB, Hill AE, Burrage LC, Olszens KR, Song J, Justice M, OBrien WE, Conti DV, Witte JS, Lander ES, and Nadeau JH. Genetic dissection of complex traits with chromosome substitution strains of mice. Science 304: 445448, 2004.[Abstract/Free Full Text]
- Snyder EE, Walts B, Perusse L, Chagnon YC, Weisnagel SJ, Rankinen T, and Bouchard C. The human obesity gene map: the 2003 update. Obesity Res 12: 369439, 2004.[Abstract/Free Full Text]
- Spiegelman BM and Flier JS. Obesity and the regulation of energy balance. Cell 104: 531543, 2001.[CrossRef][ISI][Medline]
- Stevens LC and Mackensen JA. The inheritance and expression of a mutation in the mouse affecting blood formation, the axial skeleton, and body size. J Hered 49: 506510, 1958.
- Stylianou IM, Christians JK, Keightley PD, Bunger L, Clinton M, Bulfield G, and Horvat S. Genetic complexity of an obesity QTL (Fob3) revealed by detailed genetic mapping. Mamm Genome 15: 472481, 2004.[CrossRef][ISI][Medline]
- Taylor BA and Phillips SJ. Obesity QTLs on mouse chromosome 2 and 17. Genomics 43: 249257, 1997.[CrossRef][ISI][Medline]
- Thaller G and Hoeschele I. Fine-mapping of quantitative trait loci in half-sib families using current recombinations. Genet Res 76: 87104, 2000.[CrossRef][ISI][Medline]
- Warden CH, Stone S, Chiu S, Diament AL, Corva P, Shattuck D, Riley R, Hunt SC, Easlick J, Fisler JS, and Medrano JF. Identification of a congenic mouse line with obesity and body length phenotypes. Mamm Genome 15: 460471, 2004.[CrossRef][ISI][Medline]
- Welch CL, Bretschger S, Wen PZ, Mehrabian M, Latib N, Fruchart-Najib J, Fruchart JC, Myrick C, and Lusis AJ. Novel QTLs for HDL levels identified in mice by controlling for Apoa2 allelic effects: confirmation of a chromosome 6 locus in a congenic strain. Physiol Genomics 17: 4859, 2004.[Abstract/Free Full Text]
- West DB, Boozer CN, Moody DL, and Atkinson RL. Dietary obesity in nine inbred mouse strains. Am J Physiol Regul Integr Comp Physiol 262: R1025R1032, 1992.[Abstract/Free Full Text]
- Yui MA, Muralidharan K, Moren B, Perrin GA, Wakeland EK, and Chestnut K. Production of congenic mouse strains carrying NOD-derived diabetogenic genetic intervals: an approach for the genetic dissection of complex traits. Mamm Genome 7: 331334, 1996.[CrossRef][ISI][Medline]
- Zeng ZB. Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc Natl Acad Sci USA 90: 1097210976, 1993.[Abstract/Free Full Text]
- Zeng ZB. Precision mapping of quantitative trait loci. Genetics 136: 14571468, 1994.[Abstract/Free Full Text]
- Zhu L, Swergold GD, and Seldin MF. Examination of sequence homology between human chromosome 20 and the mouse genome: intense conservation of many genomic elements. Hum Genet 113: 6070, 2003.[ISI][Medline]