Pharyngula

So here I am at the IGERT Symposium on Evolution, Development, and Genomics, having a grand time, even if I did get called out in the very first talk. There were two keynote talks delivered this evening, both of which I was anticipating very much, and which represented the really good side of science: two differing points of view wrestling with each other for consensus and for testable, discriminating differences. They also had dueling t-shirts.

Here’s the argument in brief. The functional part of the genome can be roughly broken down into two components: the coding regions, or the actual bit of DNA that is transcribed and translated into working proteins, and the subset of the non-coding regions that are involved in regulating gene expression. The coding regions are obviously important — they’re what makes the actual meaty, bony, brainy part of the animal, or the leaves and roots and flowers of a plant — but the thing is, we primates share almost all of the same genes with a mouse, which makes one wonder what makes us different from a mouse. Evo-devo proponents have been arguing that the most important changes in evolution have been in the regulatory elements of the genome, in particular the stretches of DNA called cis-regulatory elements (CREs) that flank the coding regions. People and mice share the same keratin genes, and what makes us different is how they are switched on, with variations in regulation that produce furry mice and mostly hairless people. At its most extreme, we could argue that all of us mammals have basically the same coding genes, and all the interesting stuff in evolution has been variations in the CREs, which change how those genes are deployed.

But hang on, the non-evo-devo people rightly cry, while mice and humans might have very similar sets of genes, the coding regions of those genes are not identical, and it is premature to claim that those differences are relatively insignificant … and we also know of variations in coding sequences that make big differences in fitness. We’re grossly overemphasizing the importance of cis elements in evolution, and evo-devo is doing a disservice to the discipline with an excess of hype.

So that’s the context for this evening, with both sides represented in back-to-back keynote talks.

First up was Jerry Coyne, with the provocative title of “Give me just one cis-regulatory mutation and I’ll shut up!” He also showed off his t-shirt, which said “I’m no cis-sy”. Spoiling for a fight, he was…but actually it was a very thoughtful talk in which the first thing he did was reassure everyone that he is not the enemy of evo-devo he has been made out to be, an erroneous impression taken from the Hoekstra and Coyne paper. He cited several comments that put his criticisms in a particularly harsh light, yet at the same time he could quote the paper arguing that evo-devo has great potential, which has not yet been realized. He also mentioned my comment that he was a “rather significant detractor” of evo-devo, saying he was more of a critic. I accept the change. But I do have to say that when I wrote that, it was with both enthusiasm for and anticipation of his arguments.

Right away he made an unassailable case that the importance of CREs in evolution has been oversold. He put up quotes by Hughes, Carroll, Davidson, and Valentine that I have to agree were far too sweeping in their claims for an essentially exclusive pre-eminence of regulatory change (and I made a note to myself that I better be more cautious about my enthusiasms, or Jerry Coyne will catch me).

His approach in the talk was narrow but deep, dealing with specific cases. He marched through a laundry list of examples of cis-regulatory mutations that have been trotted out as examples of the importance of these changes in evolution by the evo-devo crowd, and more or less ripped them apart. Many of the purported effects were by association only — that is, the work hadn’t been done to show the actual mechanism by which the molecular difference was translated into a significant phenotypic change. Others lacked evidence that they actually had any adaptive significance; sure, the mutation changed something in development, but so what? It wasn’t clear that it made any difference to the organism. He also argued that many of the known examples are of morphological loss, which isn’t as impressive as finding morphological additions would be. By the end, he’d found serious weaknesses in all of the cases but one.

(The one, by the way, is Chris Cretekos’s work on mutations that affect limb length and led to the evolution of the bat forelimb. Chris is an old friend from my days in Utah, and he sent me that paper months ago, and it’s sitting on my desktop right now plaintively begging me to write it up. I guess now I better prioritize it!)

Anyway, the bottom line in Coyne’s talk is a demand for greater rigor. You can’t just see a mutation in a developmental gene that has an interesting effect and claim that therefore it is important in evolution. The full range of work in puzzling out the molecular mechanism, in identifying the specific genetic change, in showing the inheritance of the trait, and in demonstrating an adaptive evolutionary role has to be done. And he’s right, of course. Evo-devo has tantalizing bits and pieces ad a lot of promise, but the iron-clad story isn’t there yet.

The second keynote was by Greg Wray, and he took a very different approach. (Oh, and a different t-shirt: “Exons, Schmexons”.) He listed more examples of developmental genes and argued that the evidence of their importance in evolution was stronger than Coyne let on, but he also conceded that he wasn’t going to argue for the pre-eminence of CREs, and that this shouldn’t be an all-or-nothing debate: the evolution of both coding and non-coding regions is important. However, arguing over a list of examples isn’t going to answer the question, since there are going to be biases in ascertainment and inclusion. What he proposed instead is a genome-wide survey for regions that exhibit signs of positive selection. You might suspect that he would say this because he has been carrying out a genome-wide survey for such regions, and you’d be right.

There was a flurry of technical details which I did not get down in my notes — I’ll have to read the paper — but he’s been doing comparative analyses of macaque, chimpanzee, and human genomes to identify regions that show signs of selection, and categorizing those regions by whether they are in regulatory or coding regions of genes. The short answer is that regulatory regions win by a clear majority. The more interesting but complicated answer is that it varies.

Some of the most interesting data he showed was on which genes were showing what kind of changes. Genes of the immune system, for instance, showed little selection for changes in regulation, and lots of selection for changes in the coding regions, which makes sense. These are gene products that interact with a changing environment, so the effective mutations are ones that modify that interaction, and you wouldn’t really expect the developmental regulation between a chimpanzee and a human to differ that much. In developmental genes, on the other hand, two thirds of the variation was in regulation, and one third in coding sequence. Furthermore, the developmental genes that experienced the greatest structural variation were those involved in gametogenesis and maternal effect genes, while it was the zygotic genes that varied most in regulation.

This was all good stuff, but Coyne and Wray were orthogonal to each other and weren’t really hitting each other’s main points. Wray has developed some powerful tools to scan genomes for evolutionarily significant variations, but Coyne is also right that to nail the story down, you need to dig deep into the individual cases — what we’ve got are lots of candidate genes, and Wray has widened the spectrum of possibilities (and has some preliminary evidence that CREs have been very important in evolution), but now someone has to get to work and figure out mechanisms and genetics for each one.

So we had an evening of scientific dissent, with good arguments and good questions from both sides, and every one won, and the meeting is off to a great start. Now, tomorrow I see that the schedule starts at 9am and ends at 10pm, so I might well be worn out by the time it’s over … but for now I am well satisfied.

Comments

  1. #1 David Marjanovi?, OM
    April 5, 2008

    Regulatory bits, however, code pieces of RNA

    Most don’t. Most are just places where either some part of the transcription machinery binds (promoters, enhancers) or where proteins bind that prevent the transcription machinery from binding or from moving (operators, silencers). They don’t get transcribed themselves and are thus not coding. Genes for miRNA, siRNA and all that jazz are coding DNA.

  2. #2 David Marjanovi?, OM
    April 5, 2008

    ARRRGH! Next time when someone asks a basic mol. bio. question, I’LL JUST SHUT UP! I promise!