Genetic divergence and the genetic architecture of complex traits in chromosome substitution strains of mice

Abstract

Background: The genetic architecture of complex traits strongly influences the consequences of inherited mutations, genetic engineering, environmental and genetic perturbations, and natural and artificial selection. But because most studies are under-powered, the picture of complex traits is often incomplete. Chromosome substitution strains (CSSs) are a unique paradigm for these genome surveys because they enable statistically independent, powerful tests for the phenotypic effects of each chromosome on a uniform inbred genetic background. A previous CSS survey in mice and rats revealed many complex trait genes (QTLs), large phenotypic effects, extensive epistasis, as well as systems properties such as strongly directional phenotypic changes and genetically-determined limits on the range of phenotypic variation. However, the unusually close genetic relation between the CSS progenitor strains in that study raised questions about the impact of genetic divergence: would greater divergence between progenitor strains, with the corresponding changes in gene regulation and protein function, lead to significantly more distinctive phenotypic features, or alternatively would epistasis and systems constraints, which are pervasive in CSSs, limit the range of phenotypic variation regardless of the extent of DNA sequence variation?

Results: We analyzed results for an extensive survey of traits in two new panels of CSSs where the donor strains were derived from inbred strains with more distant origins and discovered a strong similarity in genetic and systems properties among the three CSS panels, regardless of divergence time.

Conclusion: Our results argue that DNA sequence differences between host and donor strains did not substantially affect the architecture of complex traits, and suggest instead that strong epistasis buffered the phenotypic effects of genetic divergence, thereby constraining the range of phenotypic variation.

Figure 1

Cumulative action of combined, significant and non-significant CSS effects for multigenic traits. For each trait that differed significantly between the parental strains, we calculated the cumulative signed phenotypic effect (summed across the CSSs) and the corresponding SEM. The absolute value of the cumulative phenotypic effect is shown in rank order for each trait. For additive effects, the cumulative phenotypic effect should approach ~100% (dashed horizontal line at 100%). Traits were termed ‘epistatic’ (indicated in red) if the cumulative phenotypic effect exceeded 100% by more than the SEM. The analysis was repeated for only those CSSs whose phenotypic difference from the host strain that achieved statistical significance, those that fell short of statistical significance, and all CSSs combined. A. PWD, and B. MSM

Figure 2

Frequency distribution of phenotypic effects in two CSS panels. For traits that differed significantly between the parental strains, phenotypic effects for CSSs that differed significantly from the C57BL/6J host strain are indicated in red, and those that did not differ significantly from B6 are indicated in blue. Phenotypes were normalized so that C57BL/6J = 0% and A/J = 100%. A. PWD CSS panel. B. MSM CSS panel