QTL analysis of dietary obesity in C57BL/6byj X 129P3/J F2 mice: diet- and sex-dependent effects

Abstract

Obesity is a heritable trait caused by complex interactions between genes and environment, including diet. Gene-by-diet interactions are difficult to study in humans because the human diet is hard to control. Here, we used mice to study dietary obesity genes, by four methods. First, we bred 213 F2 mice from strains that are susceptible [C57BL/6ByJ (B6)] or resistant [129P3/J (129)] to dietary obesity. Percent body fat was assessed after mice ate low-energy diet and again after the same mice ate high-energy diet for 8 weeks. Linkage analyses identified QTLs associated with dietary obesity. Three methods were used to filter candidate genes within the QTL regions: (a) association mapping was conducted using >40 strains; (b) differential gene expression and (c) comparison of genomic DNA sequence, using two strains closely related to the progenitor strains from Experiment 1. The QTL effects depended on whether the mice were male or female or which diet they were recently fed. After feeding a low-energy diet, percent body fat was linked to chr 7 (LOD=3.42). After feeding a high-energy diet, percent body fat was linked to chr 9 (Obq5; LOD=3.88), chr 12 (Obq34; LOD=3.88), and chr 17 (LOD=4.56). The Chr 7 and 12 QTLs were sex dependent and all QTL were diet-dependent. The combination of filtering methods highlighted seven candidate genes within the QTL locus boundaries: Crx, Dmpk, Ahr, Mrpl28, Glo1, Tubb5, and Mut. However, these filtering methods have limitations so gene identification will require alternative strategies, such as the construction of congenics with very small donor regions.

Figure 1. Experimental design.

In Experiment 1, B6 x 129 F₂ mice were fed a low- and then a high-fat diet. Body composition was measured at the end of each diet period, and QTL analyses were conducted for each diet condition. To narrow down a list of candidate genes that could account for the QTL, genotype association mapping (Experiment 2) was conducted using inbred strains and 4 million imputed SNPs. Differential gene expression analysis of tissues using microarrays (Experiment 3) indicated which genes in the QTL boundaries were differentially expressed between the parental B6 and 129 strains in liver, muscle, and adipose tissue. DNA genomic sequences from the QTL regions were compared between the parental strains (Experiment 4) to identify variants that might affect gene function. Results of all four experiments were compared to highlight genes identified by multiple methods.

Figure 2. Genome-wide scans for the percentage of body fat (% body fat) of C57BL/6ByJ x 129P3/J F₂ male and female mice fed low- and then high-energy diet (Experiment 1).

Horizontal lines represent suggestive linkage thresholds determined by 1000 permutation tests [threshold = 3.08 when mice were fed the low-energy diet (dashed lines) and 3.77 when fed the high-energy diet (solid lines)]. (A and B) Sex-dependent QTLs are determined by the difference between the LOD scores when sex is used as an additive versus a combined additive and interactive covariate. Significant differences between the additive (dashed line) and additive plus interactive (solid line) models after the low-energy (A) and high-energy (B) diets are indicated by arrows. (C) Diet-dependent effects were determined by the difference in LOD scores between each diet period (arrows). The y-axis lists chromosome numbers. All analyses included age and litter size as covariates, and the diet-specific analysis in C was conducted using sex as an interactive covariate. Details about individual QTLs are shown in Table 2 .

Figure 3. Interval maps of chr 7 (low-energy diet) and chr 9, 12, and 17 (high-energy diet), which each contained a QTL identified in the genome-wide scan linked to percent body fat in C57BL/6ByJ x 129P3/J F₂ mice (Experiment 1).

LOD score peaks are marked with arrows, and the associated locus boundaries are indicated by black bars below the peaks. Genetic locations (cM) are based on the experimental map from this genetic cross. Analysis of one- and two-QTL models indicated there are two peaks on chr 9. Significance thresholds are denoted by horizontal lines (for details, see Figure 2 ).

Figure 4. Percent body fat of F₂ mice by sex and genotype (Experiment 1).

Figure 5. Percent body fat of F₂ mice by genotype of interacting loci detected in the pairwise genome scan (Experiment 1): chr 2 and 16 (upper left), chr 5 and 9 (upper right), and chr 3 and 13 (females, lower left; males, lower right).

The x-axis shows genotype: BB, homozygous for C57BL/6ByJ alleles; PP, homozygous 129P3/J strain alleles; BP, heterozygous. Values are mean ± SEM. Groups that do not share a common subscript (a-f) significantly differed by post hoc testing.

Figure 6. Results from genotype association mapping (Experiment 2), differential gene expression (Experiment 3), and strain genome comparison (Experiment 4) for the QTL boundaries on chr 7, 9, 12, and 17.

The x-axis is Mb along the chromosome (markers mapped to GRCm38). Details of the top hits in the genotype association and differential gene expression analyses are listed in Table S5. Brown squares indicate results from the gene expression analysis (Gene Exp). Orange dots indicate results from the genotype association mapping (GAM). When gene expression and genotype association mapping results converged to a single location, the results are marked by blue stars. Some genes are marked twice by blue stars if they are differentially expressed in two tissues, e. g., adipose and muscle. When stop codons and gene expression converged to the same gene, the location is marked by red stars. The point of convergence of a stop codon and genotype association mapping results is marked by a purple star. Glo1 is highlighted by a green star because it was previously nominated as a candidate gene in dietary obesity using similar methodologies.