Table 1.
Additional SNP markers for this study.
Table 2.
Disease activity index.
Figure 1.
Phenotypes used to refine the Gdac1 locus.
Panel A. A box-and-whisker plot of disease activity index of DKO mice grouped by strain background and Gdac1 genotype. Gdac1 alleles were based on typing at Rpusd2 (118.8 mbp) and Duox2 (122.1 mbp) and/or Slc30a4 (122.5 mbp). Statistics was evaluated with 1-way ANOVA (Kruskal-Wallis); Dunn’s post test; α = 0.05. Different letters above the boxes indicate means are different, where a>b>c. The number of mice in each group is 16, 74, 35, 38 and 233, from left to right. Panel B. A scatter plot of colon lengths. B6 DKO mice were not included as a reference because they and the B6 WT colon lengths fell in the middle of the collective129 data. This might be contributed by the smaller stature of B6 mice at weaning. Instead, Gpx1−/−Gpx2+/− littermates of the Gdac1 N7 DKO mice were used as the reference (129 N7 Het). One-way ANOVA; Tukey’s post test; N = 39, 56, 23, 43, 119. Different letters above the boxes indicate means are different, where a>b>c>d. Panel C. A box-and-whisker plot of colon pathology scores. One way ANOVA (Kruskal-Wallis), Dunn’s post test; N = 21, 52, 24, 38, 54. Different letters above the boxes indicate means are different, where a>b>c. Panel D. A scatter plot of Log10 CFU E. coli per gm cecal contents. One-way ANOVA (Kruskal-Wallis), Dunn’s post test; N = 14, 79, 24, 42, 31. Different letters above the boxes indicate means are different, where a>b.
Figure 2.
LOD plot for the distal colon pathology score phenotype across Chr 2 from 40.5 to 160 mbp.
Marker locations (mbp from centromere) are indicated on the X axis. Arrows demark the 95% CI. LOD plots for colon length, DAI and Log10 E. coli CFU/gm cecal contents are in Figure S1, S2, S3.
Figure 3.
Correlation between E. coli overgrowth and cecum pathology in 129 N7 Gpx1/2-DKO Gdac1B6/B6 mice (Panel A; N = 53) and Gdac1129/129 mice (Panel B; N = 25).
The slopes of the linear regression lines are not significantly different from 0. Data clustered at 6 for Log10 CFU have been offset on both the X and Y axes to show the points. Panel C shows the cecum pathology scores from Panel A and B in box-and-whisker format along with reference 129 N10 (N = 29). Different letters above the groups indicate significant differences (a>b; P<0.05; Kruskal-Wallis with Dunn’s multiple comparison test).
Table 3.
Gdac1 QTL candidate genes (shortlist in proximal to distal order).
Figure 4.
The subspecific origin of Gdac1 and the relationship to a human locus.
The upper portion of the figure shows chromosome ancestry analysis. The Gdac1 95% CI in B6 is derived from M.m. musculus (small section of indeterminate ancestry at proximal end; reference wild-derived M.m. musculus strain is PWK/PhJ). This contrasts with both 129 and C3H, where M.m. domesticus is the source of the chromosome (wild-derived WSB/EiJ is the reference M.m. domesticus strain). The boundaries of Gdac1 coincide with a portion of human chromosome 15q (brown), shown in the lower portion of the figure (data retrieved from CGD- http://msub.csbio.unc.edu/ −and MGD at the MGI website; 04/2012-mouse phylogeny viewer and mouse; human orthology map).
Figure 5.
Outline of the candidate categorization process.
The eliminated and discounted genes as well as the genes for 3 microRNAs and omitted genes are shown in dotted-black boxes. The viable candidates at each step are shown in solid blue boxes.
Figure 6.
e-QTL analysis by RT-PCR. Relative mRNA levels of Pla2g4e (A), Pla2g4f (B), Chac1 (C) and Dll4 (D) in the distal colons of 129 strain Gpx1/2-DKO Gdac1 mice by genotype and in WT B6 and WT 129.
Dll4 represents a common pattern among candidates, where the pattern suggests expression was affected due to pathology in the Gdac1129/129 mice. For all panels, N = 6, 6, 6, 6. An * indicates significant difference, α = 0.05, for pair-wise t-tests between parental strains or between the Gdac1 sets (B–D). Pla2g4e uses the non-parametric Mann-Whitney test, because the distributions were not Gaussian.