Little epistasis for anxiety-related measures in the DeFries strains of laboratory mice Jonathan Flint, 1 John C. DeFries, 2 Norman D. Henderson 3 1 Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive, Oxford, OX3 7BN, UK 2 Institute for Behavioral Genetics, University of Colorado, Boulder, Colorado 80309, USA 3 Oberlin College, Oberlin, Ohio 44074-1086, USA Received: 15 July 2003 / Accepted: 10 October 2003 Abstract Recent advances in methodologies for testing epi- static interactions, combined with several successes in demonstrating genetic interaction effects in ani- mal and human genetics, have rekindled interest in the role of epistatic influences on complex traits. It has even been suggested that the unacknowledged presence of epistasis vitiates the genetic dissection of human and animal behavior. Here we report a genome-wide interaction analysis of 1636 F 2 mice to show that epistasis is of minimal importance in an animal model of anxiety. By using a sufficiently large sample of F 2 animals, we provide evidence that interaction effects between any two loci contribute less than 5% to the total phenotypic variance in multiple tests of anxiety. We conclude that interac- tions between loci do not necessarily vitiate the ge- netic analysis of behavior in at least one animal model of anxiety. The extent to which epistasis contributes to varia- tion in complex traits is currently unknown; but a number of experimental results in different organ- isms suggests that it may be an important compo- nent of genetic variance. Analysis of nearly isogenic lines identified epistatic interactions in tomatoes (Eshed and Zamir 1996); molecular dissection of variation in bristle number in Drosophila indicated substantial interaction between quantitative trait loci (QTL) (Gurganus et al. 1999; Long et al. 1995); in the mouse, the use of recombinant congenic lines detected large numbers of interactions between loci that modify lung tumor incidence (Fijneman et al. 1996; van Wezel et al. 1996, 1999); using crosses between inbred mouse strains, Cheverud and col- leagues have reported evidence for epistatic interac- tions contributing to variation in adiposity, maternal performance, and mandible size (Cheverud et al. 2001; Leamy et al. 2002; Peripato et al. 2002; Workman et al. 2002a, 2002b); a QTL analysis of circadian rhythm in mice reported non-additive in- teractions at two loci (Shimomura et al. 2001). Nei- ther locus had much effect on the phenotype when assessed on its own, but some allelic combinations resulted in highly significant effects. An emerging position, that complex genetic traits are likely to involve substantial epistatic components (Frankel and Schork 1996), together with the view that genetic influences on behavioral traits are likely to be highly complex, has led to the conclusion that epistasis may be common in be- havioral traits (Grigorenko 2003). It has even been argued that the unacknowledged presence of epista- sis vitiates the genetic dissection of human and an- imal behavior (Feldman and Otto 1995). Lack of empirical support for this position can be attributed to the failure of most behavioral research to look for genetic interaction effects, or to do so with experiments that lack sufficient power to de- tect such effects. Small sample size severely limits the power to detect epistasis: a hundred animals generated from a backcross will contain only 25 in- dividuals of each genotype for two unlinked QTL, and only 12 individuals for three unlinked QTL. The statistical challenge in the analysis of epistasis arises from the potential number of combinations that have to be analyzed: for a genome scan with 100 markers, there are 4950 two-way interactions to consider. We decided to determine whether epistasis was an important contributor to genetic variation in an animal model of anxiety, using a sufficiently large F 2 DOI: 10.1007/s00335-003-3033-x • Volume 15, 77–82 (2004) •Ó Springer-Verlag New York, Inc. 2004 77 Correspondence to: J. Flint; E-mail: jf@molbiol.ox.ac.uk