Mouse genome-wide association mapping needs linkage analysis to avoid false-positive Loci

Abstract

We carried out genome-wide association (GWA) studies in inbred mouse strains characterized for their lung tumor susceptibility phenotypes (spontaneous or urethane-induced) with panels of 12,959 (13K) or 138,793 (140K) single-nucleotide polymorphisms (SNPs). Above the statistical thresholds, we detected only SNP rs3681853 on Chromosome 5, two SNPs in the pulmonary adenoma susceptibility 1 (Pas1) locus, and SNP rs4174648 on Chromosome 16 for spontaneous tumor incidence, urethane-induced tumor incidence, and urethane-induced tumor multiplicity, respectively, with the 13K SNP panel, but only the Pas1 locus with the 140K SNP panel. Haplotype analysis carried out in the latter panel detected four additional loci. Loci reported in previous GWA studies failed to replicate. Genome-wide genetic linkage analysis in urethane-treated (BALB/cxC3H/He)F2, (BALB/cxSWR/J)F2, and (A/JxC3H/He)F2 mice showed that Pas1, but none of the other loci detected previously or herein by GWA, had a significant effect. The Lasc1 gene, identified by GWA as a functional element (Nat. Genet., 38:888-95, 2006), showed no genetic effects in the two independent intercross mouse populations containing both alleles, nor was it expressed in mouse normal lung or lung tumors. Our results indicate that GWA studies in mouse inbred strains can suffer a high rate of false-positive results and that such an approach should be used in conjunction with classical linkage mapping in genetic crosses.

Figure 1. Spontaneous lung tumor incidence correlates with both urethane-induced lung tumor multiplicity (green) and incidence (red) in mouse inbred strains.

Incidence is given as mean percentages, whereas multiplicity is mean number of tumors/mouse. See Table 1 for phenotype values.

Figure 2. Genome wide scans for haplotype association with urethane-induced lung tumor multiplicity in mouse inbred strains using F-test for window size of 3 SNPs against the marker map plot.

Threshold (in dotted green line) p value (α = 0.10) was calculated according to Bonferroni's criterion.

Figure 3. Genome-wide genetic linkage analysis of loci affecting urethane-induced lung tumor multiplicity.

(A) (BALB/c×C3H/He)F2 cross detects the Pas1 locus at LOD score = 18.4. (B) (A/J×C3H/He)F2 cross detected the Pas1 locus at LOD score = 18.7. Red curves indicate the results of the composite interval mapping, whereas black curves indicate the results of genome scan using the Kras genotype as covariate (conditioning on the Pas1 alleles). Horizontal lines indicate the threshold values (α = 0.05) of the LOD score. The Clas2 locus (Chromosome 4) showed no significant linkage, despite the presence of the claimed functional polymorphism (D102E) in both crosses. No other locus detected by whole-genome strain survey showed significant linkage.

Figure 4. Absence of Lasc1 gene expression in mouse normal lung and lung tumors of (A/J×C57BL/6J)F1 mice.

The ethidium bromide-stained gel shows the RT-PCR results: lanes 1–3, normal lung derived from adult mice; lanes 4–6, lung tumors derived from urethane-treated mice; lane M, DNA size marker; lane DNA, genomic DNA. Only genomic DNA (top panel) and the Itpr2 housekeeping gene (bottom panel) were amplified.