European Journal of Human Genetics (2006) 14, 721–728 & 2006 Nature Publishing Group All rights reserved 1018-4813/06 $30.00 www.nature.com/ejhg REVIEW The genetic basis of emotional behaviour in mice Saffron AG Willis-Owen*,1 and Jonathan Flint1 1Wellcome Trust Centre for Human Genetics, Roosevelt Drive, Headington, Oxford, UK The last decade has witnessed a steady expansion in the number of quantitative trait loci (QTL) mapped for complex phenotypes. However, despite this proliferation, the number of successfully cloned QTL has remained surprisingly low, and to a great extent limited to large effect loci. In this review, we follow the progress of one complex trait locus; a low magnitude moderator of murine emotionality identified some 10 years ago in a simple two-strain intercross, and successively resolved using a variety of crosses and fear-related phenotypes. These experiments have revealed a complex underlying genetic architecture, whereby genetic effects fractionate into several separable QTL with some evidence of phenotype specificity. Ultimately, we describe a method of assessing gene candidacy, and show that given sufficient access to genetic diversity and recombination, progression from QTL to gene can be achieved even for low magnitude genetic effects. European Journal of Human Genetics (2006) 14, 721–728. doi:10.1038/sj.ejhg.5201569 Keywords: mouse; emotionality; quantitative trait locus; QTL Introduction tude genetic effects and their interactions (both with other Emotionality is a psychological trait of complex aetiology, genetic loci (ie epistasis) and nongenetic (environmental) which moderates an organism’s response to stress. Beha- factors) ultimately producing a quasi-continuously distrib- vioural evidence of emotionality has been documented uted phenotype. As a combined consequence of the small across a wide range of taxa from amphibians to rodents, individual genetic effects and their multifaceted patterns of and higher mammals; both in terms of a fear-like moderation, genetic effects which contribute to variation avoidance of perceived threats (eg predator–prey interac- in emotionality have, as in other complex traits proved tions and other environmental dangers),1 and enduring remarkably difficult to identify. personality-like variation in sensitivity to stress.2 While The mouse exhibits a number of attributes, which may evidence has only recently begun to emerge in favour of a be useful to genetic research. These include a short capacity for non-human animals (including rodents) to gestation period, an early puberty, a short oestrus cycle experience3 positive mood states, it is now accepted that a and a tendency to produce large litters. The mouse genome variety of species are likely possess an evolutionarily is well characterised,4 containing around 22 000 predicted conserved capacity for fear and anxiety. genes, about 80% of which have a single identifiable Emotionality, like other complex traits, is thought to human orthologue. These factors, along with a capacity for result from the cumulative superimposition of low magni- directed mating and rigid environmental control render the mouse an invaluable tool for complex trait dissection, *Correspondence: Dr SAG Willis-Owen, Wellcome Trust Centre for with additional potential relevance towards human disease Human Genetics, Roosevelt Drive, Headington, Oxford, OX3 7BN, UK. pathology. This relevance is illustrated through observa- Tel: þ 44 1865 287510; Fax: þ 44 1865 287501; tions of mouse–human quantitative trait loci (QTL) E-mail: [email protected] Received 16 August 2005; revised 9 November 2005; accepted 10 concordance across a range of (primarily physiological) November 2005 phenotypes (such as high-density lipoprotein (HDL) The genetic basis of emotional behaviour in mice SAG Willis-Owen and J Flint 722 cholesterol concentrations5). While it is less likely that behaviour across these tests can be divided into a small behavioural QTL will colocalise across species (not least number of genetically separable dimensions. due to difficulties in phenotype translation), it remains A recent principal components analysis of over 100 plausible that a small number of behavioural mechanisms, behavioural phenotypes in nearly 1700 mice identified five which carry a significant survival advantage may be genetically separable composite measures of anxiety. These conserved through evolution, and as such may depend, include: (i) activity suppression in ‘safe areas’, (ii) avoid- at least in part, on the same underlying genes. ance of anxiogenic areas, (iii) suppression of rearing, (iv) latency to enter novel areas, and (v) autonomic activity in novel environments. Between 4 and 6 QTL were found Emotionality in mice to contribute towards variance on each measure, with only A range of behavioural phenotypes are currently utilised as about 20% of loci contributing 42% of phenotypic informative measures of emotional reactivity in mice. variance to any given trait. These observations are These measures are founded on principals of avoidance, consistent with previous multivariate mapping studies autonomic activation, and behavioural inhibition (the which have shown that a proportion of QTL in both 11 12 discontinuation of species-typical behaviours such as mice and rats exhibit characteristic ethological profiles grooming, exploration and consumption in anxiogenic which transcend test type, and can be replicated across environments). These measures can be derived from intercrosses. a variety of paradigms, the most widely used of which is the open-field apparatus; a circular white, brightly lit and fully enclosed arena, within which behaviour can be remotely monitored. Methods of mapping emotionality QTL in mice Defecation and ambulation in the open-field exhibit a Inbred mouse strains exhibit large differences in emo- 10,13,14 negative correlation.6,7 Since these two measures are not tionality. These differences can be exploited for the thought to be under the control of the same peripheral purposes of quantitative trait mapping through the nervous system, it follows that their consistent inverse construction of crosses between phenotypically divergent relationship is likely to reflect the action of a central progenitor strains, and the subsequent use of directed coordinating mechanism or psychological trait (such as breeding strategies with or without phenotype selection. anxiety).8 Based on observations that intense fear can As a general principal these approaches aim to maximise result in defecation, urination and immobility in humans, the level of genetic diversity at informative loci, and those animals found to demonstrate a combination of increase the number of recombination events intervening behavioural inactivity and heightened defecation have between genotyped markers and the QTL. Given an infinite historically been considered to be more fearful or emo- marker density, crosses which contain higher levels of tionally reactive. Consistent with this interpretation, progenitor diversity, and which are separated from the repeated test exposures in the open-field result in a parental strains by a larger number of meioses will provide reduction in defection and decreased avoidance of the the highest level of mapping resolution. central (anxiogenic) areas9 (ie a habituation). Several variations on this strategy exist, including the backcross and F2 intercross, recombinant inbred lines The emotionality profile and phenotype (RILs), chromosome substitutions, recombinant congenics heterogeneity and heterogeneous stocks (HS). The backcross and F2 In addition to the open field, a number of other intercross are the most widely used of these strategies. behavioural tests are available for the assessment and/or Two inbred strains (usually, but not necessarily with characterisation of an animal’s emotionality profile. Those contrasting phenotypes) are crossed, the offspring geno- which we will pursue in this review include the elevated typed and a test of the association between genotypic plus maze (EPM) and light–dark box (two environments and phenotypic variation is performed. RILs involve a consisting of contrasting anxiogenic (light and open) and cross between two inbred strains, but in this case the F2 safe (dark and enclosed) spaces); fear conditioning (evalu- generation are randomly intercrossed and inbred for ating an animal’s behavioural response to cues or contexts around 20 generations yielding a diverse panel of inbred previously associated with a negative experience (either in lines containing varying quantities of the two progenitor terms of avoidance or immobility)); the forced swim and genomes. QTL can then be mapped by comparing the tail suspension tests (measures of immobility in response to phenotypes of these lines, and establishing specifically unavoidable aversive situations), and acoustic startle which chromosomal regions they have in common. response. Although the measures derived from these tests The chromosome substitution method of phenotype are broadly consistent with one another, with for example analysis represents an alternative first stage approach behaviour in the EPM successfully predicting performance towards QTL detection. Single chromosomes from one in other models of anxiety,10 there is some evidence that inbred strain are introgressed into the background of European Journal of Human Genetics The genetic basis of emotional behaviour in mice SAG Willis-Owen and J Flint 723 another by selective breeding, yielding a panel