Gene duplication and the adaptive evolution of a classic genetic switch
Hittinger and Carrol, Nature (449), 2007
This paper provides two particularly striking examples of evolutionary issues. To set up the first, the authors provide the wonderful reminder that the process of duplication, degeneration and complementation (DDC) itself is not necessarily adaptive. This seems obvious, but while there are examples of DDC the adaptive advantage is not always clear. One key part of Hittinger and Carroll’s method that I found simple and quite impressive was the wonderfully sensitive assay for competition between 2 strains of yeast, allowing them to compete any two genetically manipulated strains.
Basically, they used flow cytometry to count the number of GFP and BFP tagged cells from either strain. This allowed them to better quantify fitness differences between mutants that would not normally be apparent with the common complementation assays done in yeast. The short story for DDC here is that since S. cerevisiae diverged from K. lactis, there was a whole genome duplication in the cerevisiae lineage and this allowed for the “escape from adaptive conflict” for a previously bifunctional GAL1/3 gene. Basically this gene’s two functions, as a co-inducer (GAL3 in S. cerevisiae) and galactokinase (GAL1 in S. cerevisiae), are best adapted with very different ranges of regulation. It is advantageous to have a very high amount of GAL1 to utilize galactose when it is around and GAL1 goes from completely off to 1,000 fold increase in transcription when induced. The co-inducer GAL3 on the other hand has basal level of expression and is only induced 3-5 fold.
In the ancestral bifunctional enzyme these totally different ranges of transcription represented an adaptive conflict that could then be resolved following genome duplication. I think this is a great example of DDC and how it can actually be adaptive. Of course, there’s probably many questions that come up from reading my simple explanation of this (and there is of course more to it), and the authors did a great job of taking advantage of their highly genetically manipulatable organism and sensitive assay to do all the experiments one could imagine. Though someone did note that contemporary K. lactis is not the same as the ancestral genome.
The second important point to this paper was showing what happened to these paralogous regulatory regions to give them this drastic difference in regulation. Simply put, it turned out that the helical phase, or spacing, between the GAL4 binding sites was sufficient to explain this difference. When 3 adjacent GAL4 dimer binding sites are spaced such that they are on the same side of DNA, they seem to cooperate to BOTH activate higher levels of transcription and inhibit transcription better (with GAL80). Again, DNA shape relative to transcription factors actually matters. Pretty cool, but again, can we just figure this out from sequence info? In this case, so much is already known about this particular genetic switch that sequence info seemed sufficient, but I suppose in other systems knowledge of how transcription factors interact with and potentially shape DNA is needed to predict what other potential transcription factor binding sites are actually relevant.