Thursday, October 26, 2006

Mice, men, and man's best friend   posted by p-ter @ 10/26/2006 04:41:00 PM

In 2004, a group based out of Bruce Lahn's lab published a paper arguing that nervous system genes had experienced accelerated evolution in primates, particularly on the human lineage. Their approach was based on the following observations: humans and macaques diverged about 20 million years ago, as did mice and rats. But the changes in brain morphology are strikingly different in the two taxa (see their figure on the left). So they compared the evolution of a set of genes expressed in the nervous system in rodents with the evolution of the same genes in primates. Their set of 200-some genes was chosen based on a number of criteria, including expression studies, literature searches, and the inclusion of genes mutated in known neurological diseases.

As a sort of control, they did the same analyses on a number of "housekeeping" genes-- genes that perform important cellular processes, and thus are likely to be evolving under strong evolutionary constraint. Their results were consistent with their predictions-- nervous system genes, in general, evolved faster in primates than in rodents, while housekeeping genes evolved at about the same rate in both taxa. They then further broke down their group of nervous system genes into those involved in neurodevelopment, those involved in physiology, and those that they couldn't classify. Strikingly, it seemed that the difference between primates and rodents was most pronounced in neurodevelopmental genes. The group then nominated genes with particularly accelerated evolution in primates as candidates worthy of further study. 17 of these 24 are involved in regulating brain size or behavior, a particularly interesting finding. Of course, they then followed up on some of them.

It seems an odd time to attack this study as unjustified, but that's exactly what a new paper attempts to do. They claim the Lahn paper has four major shortcomings; I present them here in bold, with my comments afterwards:

1. "First, they compared only 24 nervous system genes between human and chimpanzee--the most relevant species pair for studying evolution of the human brain"

It's true the original paper is based on a human-macaque comparison (they then compare their 24 outliers with chimps), but they justify this as follows:
Recent discussions surrounding the genetic origin of humans have placed a great emphasis on human-chimpanzee comparative genomics. Undoubtedly, this approach has revealed-and will continue to reveal-genetic differences that might underlie the biological distinctions between these two sister species. Because of the exceedingly high degree of sequence identity between human and chimpanzee genomes, however, comparative studies often lack statistical power, and in many cases would overlook genetic differences that bear biological relevance. The issue of weak statistical power in human-chimpanzee sequence comparisons has been noted before (Shi et al., 2003) and is supported by our simulation studies showing that the average stochastic variance in Ks as a fraction of the true underlying mutation rate is about twice in human-chimpanzee comparison as it is in human-macaque comparison (our unpublished data). Relative to human-chimpanzee comparisons, our approach offers two important advantages. First, the use of a more distant primate species for comparison with humans provides the much needed statistical power for determining the evolutionary significance of sequence changes.
Further, while human-chimp comparisons are certainly valuable, it's worth noting that we are extremely similar to chimps, and the evolutionary trends leading to modern humans are likely to have been present long before the human-chimpanzee divergence.

2. "Second, their list of nervous system genes was manually compiled and might thus be incomplete or biased"

The new paper claims to make an unbiased complilation of brain expressed genes and show that, in this compilation, there is no acceleration for nervous system genes. I don't feel like I'm really in a position to judge this claim, except to note that, well, the "small, biased gene set" used by Lahn's group contained a lot of genes that have later been confirmed to be under selection. There's certainly a difference in goals between the two groups: Lahn's group sought to identify candidates to follow up, thus they used specific hypotheses about which genes to include, while this group looks to comment on broad patterns in human evolution. That is, they want to say that many amino acid changes in many genes are not involved in human brain evolution. Lahn identified 24 of them; maybe that's not enough to claim "a lot", but it's certainly an interesting set of genes to examine.

3. "Third, they used house-keeping genes as controls in some of the analyses, which seems inappropriate because tissue-specific genes and house-keeping genes are expected to have different evolutionary patterns"

Yes, tissue-specific genes and housekeeping genes are supposed to have different evolutionary patterns, in that houskeeping genes are supposed to be under more constraint. That is, indeed, the point of using them as a control.

4. "Fourth, a recent comparison between the dog and mouse genomes found that 18 nervous system genes that evolved faster in primates than in rodents also evolved faster in carnivores than in rodents, suggesting that the findings of Dorus et al. [the Lahn paper] might partially be due to rodent deceleration rather than primate acceleration"

I'm glad this point came last, because it's particularly interesting. So the domestic dog has experienced an acceration of nervous system genes relative to rodents. The authors seems to think this implies a deceleration of evolution in rodents. The other possibility, of course, is that evolution was accelerated on the dog lineage as well as on the primate lineage. I don't know too much about the ancestral species leading to dogs, but it's worth noting that the domestic dog has been under intense artificial selection for behavioral traits in the last 100,000 years. So perhaps the molecular substrates underlying dog domestication and human domestication (because essentially, our recent evolution has been a "domestication") are similar. I find this hypothesis much more enticing.