This guest post is by Bin He on two preprints, Genetic Complexity in a Drosophila Model of Diabetes-Associated Misfolded Human Proinsulin and Effect of Genetic Variation in a Drosophila Model of Misfolded Human Proinsulin, arXived here and here, respectively. This is a cross-post from Bin’s blog
Here we describe a pair of papers, both of which have been posted by Joe on this blog in the past month. But since they are intimately connected, we would like to write an additional post to explain the rationales behind them and the major findings therein.
The central questions in these two papers concern the genetic architecture of complex traits, such as those in human common disorders. We took a model organism approach in order to complement human studies, which are getting more and more powerful because of the successful community collaboration, but are still limited in several aspects, including mapping resolution and the ability to perform experimental validations.
Another important thinking underlying this project is the idea that decanalization of a trait may have caused a release of genetic variation, which subsequently contributed to the disease variability we see today. To this end, our fly model of misfolded human proinsulin may be viewed as an external perturbation, which, by exhausting the organism’s buffering capacity, reveals normally cryptic genetic variation. Under this view, our model will have general relevance in many human disorders.
To perform this study, we first established a fly model of a disease-associated human mutant proinsulin, which was the subject of our first paper “Genetic Complexity in a Drosophila Model of Diabetes-Associated Misfolded Human Proinsulin”.
We’d like to bring out several points. First, regarding the etiology of the disease phenotype in our fly model, we believe it is mainly due to the physical property of the mutant protein, rather than the biological function of the human proinsulin. Although Drosophila also has insulin-like proteins, their sequence similarity and functions differ substantially from the human homolog. Consistent with this view, when we made a transgenic fly expressing the wild-type human proinsulin, what we observed is that, at both phenotype and transcription level, expressing the wild-type human proinsulin in developing eye and other imaginal discs do not cause any visible changes. We thus propose that our fly model is for a general class of human disease associated with unfolded or misfolded protein.
In the first paper, we also described the phenomenon of variable phenotypic severity when put on different wild-derived genetic background. A series of experiments ruled out possible confounding factors, such as correlations induced by natural variability in eye size, or different levels of transgene expression.
We were exploring the idea of using natural variation in the fly to identify associated loci underlying a complex disease trait. We did so by crossing the transgenic, Mendelian disease carrying line to a panel of wild-derived inbred lines, and asked whether the severity of the disease is dependent on the genetic background. The answer is a definite yes: the range of phenotype quantified by the size of the eye span from 10% to 80% of wildtype (the mutant human proinsulin was expressed in the eye disc during development, causing neurodegeneration. We used eye because it is dispensable in lab conditions, and easy to measure the phenotype). We then conducted a GWAS, which led to the identification of sfl, as described above, and also the HS biosynthetic pathway by genetic test. One unique advantage of our system is its ultra-high resolution in mapping: we localized the association signal to ~400bp LD block within one of the introns of sfl, allowing us to test specific hypotheses about the molecular mechanisms of the associated variants. Pyro-sequencing analysis revealed allele-specific expression difference due to the intronic variation, but also highlighted the genetic heterogeneity even within that locus, with additional cis-variants present to influence the expression level. Overall, we believe that our fly model system is a powerful complementary approach to the genetic study of complex traits. Its high mapping resolution and rich molecular/genetic toolkits allow faster and in-depth characterization of disease-associated variation, which is a unique advantage.
Bin Z. He
Kreitman Lab, Dept of Ecology and Evolution, University of Chicago
current address: O’Shea Lab, FAS Center for Systems Biology, Harvard University / HHMI