How does gene duplication allow evolutionary innovation?
Genes with new functions do not magically appear from nowhere, or so most scientists assume. (If I thought that evolution, perhaps especially human evolution, was being guided by some supernatural individual or group, I would be looking for such â€śgenes from nowhereâ€? rather than whining that theories with no evidence should get equal time. Not that they want schools to teach all theories that lack evidence, of course, just ones favored by their particular religion or short-term economic interests.)
Random duplication of existing genes is often a key step, but there are at least two different ways in which gene duplication could facilitate evolutionary innovation. Once there are two copies of a gene, one could evolve a new function without interfering with the old geneâ€™s function. Or, a single gene could evolve two different functions, doing neither of them particularly well. Then, gene duplication would allow the two copies to evolve separately, each being optimized for a different function. This weekâ€™s paper shows that evolution has followed this second pathway at least once, and perhaps often. The paper also provides yet another example of how molecular methods are providing new details on how evolution works.
â€śEscape from adaptive conflict after duplication in an anthocyanin pathway geneâ€? was published in Nature by David L. des Marais â€“ apparently not the the David J. des Marais who has published on evolution of photosynthesis â€“ and Mark Rausher, at Duke University in North Carolina. They looked at genes whose enzyme products make pigments for flower colors.
They started by making a family tree for these genes, based on their DNA sequences. Different methods of doing this sometimes give slightly different trees, but in this case the three standard methods all gave the same result: the common ancestor of morning glories and potatoes had only one DFR gene (coding for the enzyme dihydroflavonol-4-reductases; arenâ€™t you glad you asked?) as does potato. But, somewhere on the pathway to morning glories, this gene was duplicated.
What happened next? The methods used to develop the tree can also predict the DNA sequence of ancestors along each branch. So they could compare the DNA sequences of the duplicated genes, in modern morning glories, with the DNA sequence of the last ancestor before the duplication. The authors reasoned that, if only one gene was evolving a new function, that copy would change more than the other. Furthermore, they could distinguish random changes from those driven by selection for a particular function. There are often two or more DNA triplets that code for the same amino acid. If DNA changes that actually change amino acids in the DFR enzyme are more common than expected by random chance, that suggests natural selection for a particular function, not random drift. (This is a widely used method for detecting natural selection.) The authors reasoned that, if only one copy was evolving a new function, they would see this pattern in only one copy. Instead, they saw clear evidence that both copies were evolving, but differently, under natural selection. This suggests that the ancestral enzyme was doing two jobs, neither of them particularly well. But, once there were two copies, they could evolve separately, each specializing in a different function.
But why speculate about function, when you can measure it? Using the DNA sequence of the ancestral gene, determined from the family tree, they made some of the ancestral enzyme. (See my earlier post about using similar methods to â€śresurrectâ€? proteins from ancient bacteria, measuring their optimum temperatures, and thereby inferring ancient temperatures.) Then they tested the resulting enzymes. Consistent with their hypothesis of two diverging functions, one copy works better than the ancestral version, while the other is worse.
The copy that got worse at one function (related making flower colors) presumably got better at something else. We know this because the DNA changes changed amino acids, and therefore actual enzyme function, more often than random. But we donâ€™t know what the new function is, yet. Stay tuned!