The conditional nature of genetic interactions: the consequences of wild-type backgrounds on mutational interactions in a genome-wide modifier screen

Abstract

The phenotypic outcome of a mutation cannot be simply mapped onto the underlying DNA variant. Instead, the phenotype is a function of the allele, the genetic background in which it occurs and the environment where the mutational effects are expressed. While the influence of genetic background on the expressivity of individual mutations is recognized, its consequences on the interactions between genes, or the genetic network they form, is largely unknown. The description of genetic networks is essential for much of biology; yet if, and how, the topologies of such networks are influenced by background is unknown. Furthermore, a comprehensive examination of the background dependent nature of genetic interactions may lead to identification of novel modifiers of biological processes. Previous work in Drosophila melanogaster demonstrated that wild-type genetic background influences the effects of an allele of scalloped (sd), with respect to both its principal consequence on wing development and its interactions with a mutation in optomotor blind. In this study we address whether the background dependence of mutational interactions is a general property of genetic systems by performing a genome wide dominant modifier screen of the sd(E3) allele in two wild-type genetic backgrounds using molecularly defined deletions. We demonstrate that ~74% of all modifiers of the sd(E3) phenotype are background-dependent due in part to differential sensitivity to genetic perturbation. These background dependent interactions include some with qualitative differences in the phenotypic outcome, as well as instances of sign epistasis. This suggests that genetic interactions are often contingent on genetic background, with flexibility in genetic networks due to segregating variation in populations. Such background dependent effects can substantially alter conclusions about how genes influence biological processes, the potential for genetic screens in alternative wild-type backgrounds identifying new loci that contribute to trait expression, and the inferences of the topology of genetic networks.

Figure 1. Genetic background effects influence sd^E3, and are used for a dominant modifier screen.

A) Outline of the modifier screen employed in this study (illustrated for 2^nd chromosome deletions). Using the DrosDel and Exelixis Deletion collections, male deletion–bearing (denoted with -()- ) flies were crossed to females homozygous for the sd^E3 mutation from each wild-type genetic background, Samarkand (blue) and Oregon-R (red). Male offspring that were hemizygous for the sd^E3 allele and heterozygous at all other loci, including the deletion, were compared between the two genetic backgrounds. Thus we were scoring male flies hemizygous for sd^E3, and heterozygous for the deletion. The co-isogenic progenitor wild-type strains was used for control crosses. Each grey rectangle represents a chromosome (X, 2 & 3 from left to right), with centromeres (black dots), and balancer chromosomes (brown rectangles). Yellow represents the sd^E3 mutation and closely linked genomic region on the X chromosome. B) The effect of genetic background on the phenotypic expression of the sd^E3 allele, and examples of suppression and enhancement of this allele in each background. Letters beside each image represent the semi-quantitative scores assigned to wings (all figures taken at 40× magnification).

Figure 2. The majority of autosomal modifiers of sd^E3 are background-dependent.

A) Proportion of deletions that modify the sd^E3 phenotype in a background-dependent or -independent manner, by chromosome arm and deletion collection. DD = DrosDel collection. EX = Exelixis collection. Numbers at the bottom of each bar indicate whether the effects are in autosomal chromosome two or three, while the letters L and R represent whether the effects are found on the left or right chromosome arms, respectively.

Figure 3. Genomic distribution of background-dependent and independent modifiers of sd^E3.

A) Example of the distribution of background-dependent and -independent modifiers of sd^E3 on the left arm of chromosome 3 for each deletion collection. The cytological location (61–80) of all deletions on the left arm of chromosome 3 are shown. Regions with no coverage are left blank (white). While there are several locations that show co-enhancement or suppression for Samarkand (SAM) or Oregon-R (ORE), most show an effect in only one background, and occasionally opposite effects (i.e. between 61–62 in DrosDel), consistent with sign epistasis. In a given collection where there were two deletions with overlapping genomic locations (or were nested), the regions in the figure are divided vertically to show the effect of each deletion. The remaining chromosome arms are shown in Figure S4. B) Evidence for background dependent interactions for apriori known interacting loci. For deletions that covered the known interacting factors of sd, we show the background dependent effects . Unlike the finding for the genome as a whole, there appears to be more synthetic enhancers in Samarkand than Oregon-R.

Figure 4. Background dependence is partially a consequence of strain specific sensitivity to genetic perturbation.

Quantitative effects of a subset of 44 deletions on the modification of the sd^E3 phenotype are shown. The deletions are rank-ordered based on wing size in the Oregon-R background. Enhancement and suppression of the sd^E3 phenotype is much greater in the Oregon-R background, relative to Samarkand, in both absolute (shown) and relative terms (not shown). Solid and stippled lines (blue and red) represent the mean and 95% confidence interval, respectively, for wing size in the control sd hemizygous males (sd^E3/Y). Circles represent deletions with an a priori expectation of modification based on the initial semi-quantitative screen, while triangles represent deletions with no observed effect in that screen. Filled symbols represent a significant observed effect in the quantitative screen. The Y axis shows a measure of wing size using centroid size (see methods).