Mark Rebeiz,1 John E. Pool,2,3 Victoria A. Kassner,1 Charles F. Aquadro,4 Sean B. Carroll1,*
Summary of paper
In Africa, drosophila dorsal abdominal pigmentation is correlated with altitude. Ebony is a candidate gene for involvement in this phenotypic gradient, as it codes for an enzyme in a pathway that produces a tannish brown pigment. A previous study (Pool and Aquadro 2007) produced evidence for a selective sweep at the ebony locus, further strengthening its potential as a candidate gene. The goal of Rebeiz et al (2009) was to confirm that the ebony locus was responsible for this phenotypic variation and to determine the specific nucleotide variation causing it.
Three types of experiments were performed. The first was a series of in-situ hybridizations to show that ebony expression was correlated with the amount of abdominal pigmentation. The second was to determine the boundaries of the regulatory elements that control spatial expression of this gene, with a specific goal of determining the element that controlled expression in the abdomen.
The third set was a series of transgenic reporter assays to determine which of the nucleotides that varied between light and dark drosophila lines in the abdominal regulatory element were responsible for the phenotypic variation. Five SNPs were found to be casual: three that affected an enhancer element and two that affected a repressor element.
Finally, the paper surveyed drosophila populations from throughout Africa to determine if these SNPs are standing genetic variation or if they arose de novo in the higher-altitude (i.e. darker) lines. The three differences found in the enhancer element were also found in several other populations, indicating that they existed in the ancestral population of the higher altitude lines and then were swept to fixation due to selection for the darker phenotype. The SNPs found in the repressor element, however, only exist in the high altitude populations, which suggests they arose de novo.
In conclusion, it appears that the derived darker phenotype results from a combination of sorting of standing genetic variation and de novo mutations.
Significance of work
This paper represents one of the most complete understandings of the genetic basis of natural variation in an (almost certainly) adaptive phenotype.
Rebeiz et al discuss their work in the context of whether coding or regulatory changes are the primary locus of evolution. They (among others) argue that changes resulting in adaptive phenotypic differences are more likely to occur in parts of the genome that regulate the amount, location and timing of a gene’s expression than changes that cause differences in protein coding. Their arguments are summarized below:
Protein evolution is constrained
- Proteins often have pleiotropic functions.
- The triplet genetic code cannon accommodate most insertions or deletions.
- Proper folding is essential – random amino acid replacements are usually destabilizing.
- Order of amino acid replacement is constrained: relatively few evolutionary pathways are accessible.
- There are few mutational targets, as many proteins only have one active site or a few binding domains.
Regulatory evolution is less constrained
- Enhancers are modular: disruption or change of one will not affect the others.
- Not affected by constraints of the triplet genetic code.
- Not constrained by requirements folding structure.
- Chimeric enhancers indicate that multiple evolutionary pathways are accessible.
- Enhancers contain multiple binding sites so there are multiple mutational targets.
These arguments are logical, but as significant as this study is, it can only offer limited support for their conclusion. There is a straightforward correlation of ebony expression with dark pigmentation – so we expect mutations affecting gene expression are more likely to be selected for in such a phenotypic gradient. In other words, the type of mutation underlying the phenotypic variation being studied may be strongly influenced by the phenotype. To come up with a definitive answer in the regulatory versus coding debate will require surveying many phenotypes in many taxa as precisely and exhaustively was done in this study.
This study is best viewed not in the context of the regulatory versus coding debate, but in the more fundamental context of how genetic variation is produced and sorted to cause adaptive phenotypic evolution.