Carroll on evolution and gene regulation

Sean Carroll is a colleague of mine here at UW, author of the recent book, Endless Forms Most Beautiful: The New Science of Evo Devo and the Making of the Animal Kingdom, and expert on the evolution of developmental genes and their regulation. He has a new review article on PLoS Biology outlining some of his current thoughts about the evolution of gene regulation, with particular relevance to human evolution (via Gene Expression).

After a long introduction concerning the history of genomic analysis, Carroll states his major assumption:

While the agnostic, "wait and see" position would appear safer, that would not at all be in keeping with the bold spirit of the pioneers who first wrestled with the question. Moreover, I argue that a trend is evident, and that that trend should, of course, inform ongoing and future work. Based upon (i) empirical studies of the evolution of traits and of gene regulation in development, (ii) the rate of gene duplication and the specific histories of important developmental gene families, (iii) the fact that regulatory proteins are the most slowly evolving of all classes of proteins, and (iv) theoretical considerations concerning the pleiotropy of mutations, I argue that there is adequate basis to conclude that the evolution of anatomy occurs primarily through changes in regulatory sequences.

It is fair to point out that there is not universal agreement on this assumption; there are still those who would argue that coding sequence evolution was a major (if not the major) kind of evolutionary change. I see little point to arguing it, clearly both kinds of changes are sometimes important. Carroll's point is to make a case for significant regulatory changes, and to succeed in this, he needs an example.

For this, he gets down to business. How does one infer that gene regulation has been the target of selection, instead of coding sequence evolution? When is evolution of the coding sequence not sufficient to explain the pattern of change? To answer these questions, he daringly chooses one of the highest profile genes: FoxP2.

Considering this gene as well as MYH16, he begins with this cautionary note:

My concern here is not whether these specific associations did or did not play a role in human evolution; rather, my concern is the exclusive focus, by choice or by necessity, on the evolution of coding sequences in these and more genome-wide population genetic surveys of chimphuman differences [63].
There exists some disconnect between what studies in model species have underscored -- the ability or sufficiency of regulatory sequences to account for the evolution of physical traits -- and which models of evolution are implicitly or explicitly being tested when only coding sequence divergence is considered.

Carroll notes the facts: the human and gorilla copies of FoxP2 differ at just two amino acid coding substitutions, and flanking sequence indicates that FoxP2 has undergone a selective sweep within the past 200,000 years (Enard et al. 2002). Then he adds some other facts that might not ordinarily be considered relevant:

FOXP2 is expressed at multiple sites, not just in the brain, but in the lungs, heart, and gut as well [64,65]. Patients with the FOXP2 mutation do have multiple neural deficits [66]. And, because FOXP2 is expressed in different organs and different regions of the brain, it is certain to possess multiple regulatory elements. Furthermore, it is an enormous, complex locus, spanning some 267 kb. Based upon a simple average base pair divergence of 1.2%, there should be over 2,000 nucleotide differences between chimps and humans in this span. Because there is much more potential for functional divergence in non-coding sequences, there is no specific reason to favor coding sequence divergence over regulatory sequence divergence at FOXP2.

He then notes that the regulation of FoxP2 appears to differ in song-learning birds vs. non-learners, whereas the amino acid sequences appear not to be related to vocal communication in different vertebrates. His conclusion is that these facts are all consistent with regulatory evolution of FoxP2 in humans, rather than simple amino acid sequence evolution. If so, then the evolutionary changes themselves have not yet been characterized.

The contrast between the negative conclusions drawn from the analysis of coding sequences and the fascinating correlation revealed by the comparative study of gene regulation in vivo highlights the general inadequacies of, and potential error in, the exclusive analysis of coding regions when considering the evolution of anatomy. But that inadequacy applies more broadly than just to the evolution of form. While standard population genetic tests have been used to search human protein sequences for statistical evidence of positive selection [63,69], several examples of positive selection on cis-regulatory sequences of physiological genes are documented [7072]. This includes the very clear case of the erythroid-specific loss of expression of the Duffy antigen chemokine receptor in populations resistant to Plasmodium vivax malaria [73]. This loss is due to a regulatory mutation that affects an erythroid cis-regulatory sequence but has no effect on receptor expression elsewhere in the body [74].

As an aside, I find two things interesting. First is just how much research has been done in the past three years on FoxP2 and its expression in different vertebrate taxa. Comparative biology really seems to turn on a dime when it comes to examining the effects of genes on different taxa. Second, is a related fact: almost none of this research deals with the function of the gene. What people can figure out very quickly is how the sequence of a gene compares among taxa, and (with a bit more difficulty) how the gene is expressed in the tissues of different animals. These kinds of analysis require DNA samples, genetic probes, and tissue libraries for different taxa, many of which have already been made. It's quick-and-dirty comparative biology at its simplest: as if Richard Owen were still at work in the genomic age.

What seems to be much harder is finding out just how these kinds of genetic mechanisms work. Although he doesn't say it directly, that seems to be what Carroll is getting at: understanding the mechanism of change requires some understanding of the function itself -- especially as it relates to the function of the same gene in different tissues, all of which must continue to function properly for the organism to survive. The quick-and-dirty analyses would seem to fall in the category that Carroll calls, "low-hanging fruit," and for good reason, since it is comparatively much easier to find and characterize such changes.

But as correct as Carroll may be, I wonder to what extent his message is essentially nihilistic with respect to human evolution. How likely is it that we will unravel all the interactions among genes that allow the development of complex brain structures? What kind of depth of understanding must one have of the development of the brain itself in order to find the key genetic changes? And how likely is it that changes in gene regulation can be placed into an evolutionary sequence and correlated with events in the fossil and archaeological record? The answers to these questions have to be attractive to a theorist like me, because they introduce many different dimensions of complexity. But they don't bode well for any solid answers to interesting questions.


Carroll SB. 2005. Evolution at two levels: on genes and form. PLoS Biol 3:e245. Free full text