The Genetic Architecture of Complex Traits: Starvation Resistance in Drosophila melanogaster

Abstract

In nature, animals are often subjected to periods of sub-optimal food resources. Characteristic responses to starvation stress have been observed in bacteria, nematodes and yeast: they alter their morphology, become quiescent, and suspend reproduction until adequate food resources become available. Studies of laboratory and natural populations of Drosophila reveal a surprising amount of genetic variation for starvation tolerance. The presence of this genetic variation is an evolutionary puzzle, as variability would not be expected in a key trait related to individual survival. While starvation resistance has been positively correlated with lifespan and other stresses, it is often negatively correlated with fecundity, suggesting that a trade-off between reproduction and individual survival might be present. In order to evaluate this hypothesis, the suite of genes affecting starvation resistance and their properties must be known. Three complementary methods were used to identify genes affecting starvation resistance: a P-element insertional mutagenesis screen, which directly identifies candidate genes involved in the response to starvation stress; deficiency complementation mapping, which reveals small genomic regions contributing to variation in starvation resistance; and transcriptome analysis using microarrays, which has the potential to identify both types of genes.

The starvation tolerance phenotype was assessed for 933 P-element insertion lines in two isogenic backgrounds: Canton-S and Samarkand. 383 insertions had a significant effect on starvation tolerance. The effect of the P-element inserts was generally negative and often sex-specific. Only 31 insertions significantly increased starvation tolerance. Significant insertions tag genes that are putatively involved in the starvation stress response.

Deficiency complementation mapping was used to fine-map broad genomic regions (quantitative trait loci, or QTL) previously identified for starvation resistance. The five original QTL fractionated into thirteen smaller QTL, six of which had sex-specific effects. From these fine-mapped regions 26 genes were chosen for mutation complementation testing. Twelve of the 26 genes showed a significant effect on variation in starvation resistance between the two wild-type strains, Oregon-R and 2b.

Transcriptome analysis was performed on a subset of the recombinant inbred mapping population used to identify broad QTL affecting starvation resistance: two lines resistant to starvation, two lines susceptible to starvation, and the two parental lines, Oregon-R and 2b. RNA samples were obtained in both the unstarved and starved states for these lines. Many genes were involved in the response to starvation stress: 3,528 unique probe sets exhibited significant differences in transcript abundance between the unstarved and starved states. 217 probe sets were identified as QTL that may affect variation in starvation resistance. 47 of these probe sets fell within the original QTL regions; nine probe sets fell within fine-mapped QTL found from the deficiency complementation tests. Further analysis revealed substantial epistasis at both the transcript level and the level of the phenotype, adding unexpected complexity to the attempt to map QTL using microarrays.

Many of the genes implicated in this study have known phenotypes in cell fate specification/proliferation, feeding behavior, oogenesis, and metabolism, suggesting extensive pleiotropy. Mutational and transcriptional effects were often sex-specific. The large numbers of genes identified in this study suggest that a balance between mutation and selection may maintain variation in starvation resistance, rather than a trade-off among life history traits.