Epidermal growth factor receptor and transforming growth factor-beta signaling contributes to variation for wing shape in Drosophila melanogaster

Abstract

Wing development in Drosophila is a common model system for the dissection of genetic networks and their roles during development. In particular, the RTK and TGF-beta regulatory networks appear to be involved with numerous aspects of wing development, including patterning, cell determination, growth, proliferation, and survival in the developing imaginal wing disc. However, little is known as to how subtle changes in the function of these genes may contribute to quantitative variation for wing shape, per se. In this study 50 insertional mutations, representing 43 loci in the RTK, Hedgehog, TGF-beta pathways, and their genetically interacting factors were used to study the role of these networks on wing shape. To concurrently examine how genetic background modulates the effects of the mutation, each insertion was introgressed into two wild-type genetic backgrounds. Using geometric morphometric methods, it is shown that the majority of these mutations have profound effects on shape but not size of the wing when measured as heterozygotes. To examine the relationships between how each mutation affects wing shape hierarchical clustering was used. Unlike previous observations of environmental canalization, these mutations did not generally increase within-line variation relative to their wild-type counterparts. These results provide an entry point into the genetics of wing shape and are discussed within the framework of the dissection of complex phenotypes.

Figure 1.—

The wing blade of Drosophila melanogaster. (A) Image of a wing from D. melanogaster, with the nine landmarks used in this study superimposed as solid circles. In addition to studying variation across the entire wing, regional variation was also examined for the anterior (B), central (C), and posterior (D) areas of the wing. (B) Variation in landmark position after procrustes superimposition for all samples used in this study.

Figure 2.—

Statistical association and effect size for mutations used in this study. (A) The effects of the mutation on wing shape for the whole wing blade, ordered by significance. (B–D) MANOVA for the anterior (B), central (C), or posterior (D) subregions of the wing. Horizontal axis: each mutation used in this study. Left vertical axis: negative log of the P-value from the MANOVA (Wilk's Λ). Right vertical axis: procrustes distance (PD) between the mean configurations of mutant from its wild type. Correcting for multiple contrasts using Bonferroni correction maintains α = 0.05 at −Log(P) = 3.0, represented by the horizontal line, while a nominal P = 0.05 is at −Log(P) = 1.3. * represents the alternative allele for that gene. Procrustes distances for both background and sex effects (shaded circles) are included at the end of each graph for comparative purposes.

Figure 3.—

Genotypic effects on shape are not sensitive to allometric scaling with size. Variance explained by genotype without centroid size in the model (labeled G, solid bars) or with centroid size as a covariate in the regression model (labeled G with covariate, hatched bars) is shown. For the majority of the mutations examined, including centroid size as a covariate in the model has negligible effects on the proportion of variation explained on the basis of Goodall's test on procrustes distance. In addition, the proportion of variation explained by genotype for the uniform components is shown (G uniform components, shaded bars), demonstrating that the shape change of many of the mutations is due to the effects of linear transformations.

Figure 4.—

The effects of mutations in the Egfr, Hedgehog, and TGF-β pathways on wing shape. The magnitude of the vectors describing the shape change is multiplied five times to facilitate visual examination of the shape change. For all illustrations, black represent the mean shape of the mutation, while gray represents the mean shape for the wild-type siblings from the relevant crosses. Solid segments represent estimated connections between landmarks sampled in this study. The dashed lines are used to illustrate the remaining wing morphology and are for illustrative purposes only.

Figure 5.—

No general increase in within-line variation for wing shape due to genotype. Using two measures of multivariate variance, the total variance (A) and a measure of the generalized variance (B), there is evidence for an increase in variation due to the presence of the mutations relative to the wild type. (A) Interestingly, the amount of variation around landmarks is greater in the proximal–distal axis, relative to the anterior–posterior axis. (B) The generalized variance as measured by for the first six eigenvalues (λ_i) shows a similar picture to that of the total variance. (C) The increase in variation is not due to a general increase in within-line variation, as can be seen by examining each mutation separately for males in the Ore-R background. In this instance, only 2 mutations, omb and bs, of the 50 used in this study are observed to increase the within-line variance for shape.