Genetic influences on brain gene expression in rats selected for tameness and aggression

Abstract

Interindividual differences in many behaviors are partly due to genetic differences, but the identification of the genes and variants that influence behavior remains challenging. Here, we studied an F2 intercross of two outbred lines of rats selected for tame and aggressive behavior toward humans for >64 generations. By using a mapping approach that is able to identify genetic loci segregating within the lines, we identified four times more loci influencing tameness and aggression than by an approach that assumes fixation of causative alleles, suggesting that many causative loci were not driven to fixation by the selection. We used RNA sequencing in 150 F2 animals to identify hundreds of loci that influence brain gene expression. Several of these loci colocalize with tameness loci and may reflect the same genetic variants. Through analyses of correlations between allele effects on behavior and gene expression, differential expression between the tame and aggressive rat selection lines, and correlations between gene expression and tameness in F2 animals, we identify the genes Gltscr2, Lgi4, Zfp40, and Slc17a7 as candidate contributors to the strikingly different behavior of the tame and aggressive animals.

Keywords: RNA sequencing; brain gene expression; genetics of tame and aggressive behavior; natural variation; quantitative trait locus (QTL) mapping in outbred rats.

Figure 1

Tameness QTL mapping. Genome-wide statistical support for QTL affecting tameness in analyses assuming fixation of causal alleles within the founder populations (HKR, blue line) as well as analyses allowing segregation of causal alleles among founder animals (FIA, red line). Horizontal dashed lines indicate a 2% FDR significance threshold; QTL exceeding the threshold are indicated by * (FIA) and ^‡ (HKR). Allele effect plots of the FIA tameness QTL on chromosomes 1 and 10 are shown below. Circles indicate the point estimate and lines the standard errors of the allellic effects. The numbers of individuals that carry a respective allele are printed to the right of the point estimates. For clarity, the results for alleles with <20 observations are shown in light gray.

Figure 2

Brain expression QTL mapping. Locations of eQTL affecting brain gene expression in an F₂ intercross between two rat lines selected for tame and aggressive behavior toward humans. Positions of genes are plotted against the positions of eQTL identified by FIA.

Figure 3

Genome-wide overlap of QTL affecting tameness and eQTL affecting gene expression in the brain. Red solid line, FIA scores for tameness QTL mapping; light red rectangles, confidence intervals for tameness QTL locations. FIA eQTL at a 5% FDR are shown as dots. Blue, local eQTL; green, distant eQTL; red dashed horizontal line, 2% FDR score threshold for tameness QTL.

Figure 4

Enrichment of highly significant eQTL at the Tame-1 locus. The proportion of local and distant eQTL overlapping with Tame-1 (red line) at a given significance threshold (x-axis) is compared to the total number of all eQTL reaching the given level of significance (dashed black line).

Figure 5

Correlations between allele effects of eQTL and tameness QTL along the genome. Green, significant (P < 0.05, Benjamini–Hochberg corrected) correlations; blue, nonsignificant correlations. FIA mapping scores for tameness are shown as the solid red line and the confidence intervals for significant tameness QTL are shown as light red rectangles. Red dashed horizontal line, 2% FDR score threshold for tameness QTL. (A) Local eQTL that overlap a tameness QTL tested at the peak of the corresponding tameness QTL. (B) Allele effect correlations at all local eQTL peak positions. Note that 18 of 23 significant allele effect correlations occur at local eQTL that overlap with a tameness QTL.

Figure 6

Genes that show evidence for influencing tameness and aggression. Shown are the number and names of genes that meet at least one of five criteria for involvement in tameness: (i) positional overlap between eQTL and tameness QTL (green oval); significant allele effect correlations with tameness QTL at either (ii) eQTL peaks (blue oval) or (iii) tameness QTL peaks (red circle); (iv) differential expression in RNA-seq frontal cortex or in whole brain microarray between the tame and aggressive populations (yellow oval); (v) significant correlation with tameness in F₂ animals [gene names in italics, * indicates significance across all genes, and (*) indicates significance only for genes whose eQTL overlap tameness QTL]. The numbers in parentheses indicate the number of genes that fulfilled the respective criterion. Gene names given on the left are color coded corresponding to those in the figure and are based on the criteria the given gene matches. Note that the overlap between the two scans for allele effect correlations is based on strict significance criteria in both scans (Materials and Methods), resulting in a conservative estimate of the intersection.