Brain evolution: what do the genes say?

4 minute read

Reference: Dorus, S. et al.. 2004. Accelerated evolution of nervous system genes in the origin of Homo sapiens. Cell 119:1027-1040.

This paper examines “nervous system genes” in humans and other species to determine how many such genes may have undergone adaptive evolution during the course of human evolution. “Nervous system genes” were identified by amplifying cDNA (complementary DNA sequences from mRNA molecules) reflecting activity in brain tissue. The activity of these genes is by and large not known, although the authors do make a division between genes with a primarily homeostatic function and genes that may function in nervous system development.

The paper comes out of Bruce Lahn’s lab at the University of Chicago. They have done a substantial amount of work on finding genes responsible for brain development in humans, and have previously published papers on two genes that may have undergone adaptive evolution during hominoid or human evolution in association with changes in brain functioning, ASPM and MCPH1. In both of those cases, the evidence for adaptive evolution came from comparisons in substitution rate between living humans and other anthropoid species.

Here, the comparisons involve 214 candidate genes in four species: rats mice, macaques, and humans. First, the differences between rats and mice are compared to the differences between humans and macaques. Although they don’t put it this way, the idea is to test the null hypothesis that primates and rodents had the same rate of molecular change in these genes. They find that the primates had an apparently higher rate of molecular change, with a higher rate of protein change as opposed to synonymous change in the coding sequences of these genes.

Second, the comparison of the human lineage with the macaque lineage–rooted by the rodent outgroup–makes it clear that humans have had an especially high rate of adaptive evolution in these genes compared to other primates. This finding is supported by some comparisons with chimpanzees, which appear to show that human evolution itself during the past 7 million years has involved faster nervous system evolution than in chimpanzees, and presumably other primates.

An alternative hypothesis for a more rapid evolution of these genes in primates would be relaxed selective constraint on them. Finding that the rate of nonsynonymous changes was still low in these genes, the authors argue that this hypothesis is unlikely. To additionally test it, they compared a set of “housekeeping” genes that perform basic cellular functions that are likely to be conserved among species. I’m not convinced by the relevance of this particular comparison, since relaxed selection could certainly have occurred for nervous system genes–say, if learning causes greater phenotypic plasticity, the heritability of adaptive behaviors might decline and weaken the strength of selection on alleles correlated with adaptive behaviors. I don’t believe this story, just noting that continued selection on housekeeping functions probably doesn’t argue strongly for any hypothesis regarding nervous system genes.

A very interesting part is the focus on “developmentally biased” genes that are active especially during the formation of nervous system structures during development. They find that these genes have a higher rate of change in primates than other kinds of nervous system genes. Additionally, looking only at those 24 genes that had an exceptionally high number of amino acid changes during primate evolution, 7 of them affect brain size and 10 directly affect behavior in knockout mice. The authors conclude that this is evidence for a broad-based evolution of brain function and size, rather than any single critical genetic change being responsible.

On a few levels, the paper is sort of ugly. I wouldn’t be so quick to assume that the rat-mouse divergence time was around 16 million years, or roughly equivalent to the human-macaque divergence. There is a lot of debate about the timing of rodent divergences, with paleontologists typically supporting lower dates for the rat-mouse divergence, and some geneticists arguing for much more ancient dates than the Miocene. And the human-macaque divergence was at least Oligocene in age, probably substantially older than 30 million years. None of this is essential to their study, but it’s a little sloppy. Likewise, the use of encephalization quotient seems like an unnecessarily blunt instrument. They need a better argument for why a larger brain size should imply a substantial genetic difference rather than a simple one. But these are criticisms of form rather than of substance–just an observation that more interaction between bioinformaticians and morphologists would be fruitful and would add to studies like this one.

Basically, I think this paper is a great example of what someone can do with bioinformatics resources alone today. There is no reason why anyone else in the world couldn’t have put together an equivalent set of data, although Lahn’s team certainly has the advantages of familiarity with many of these genes from their work characterizing nervous system development. I expect to see many more syntheses like this one in the future.

That said, the cleverness is in matching the form to the hypothesis. Here, the use of the four species was essential because these four have large sequence sets available on GenBank. The comparison of the four was informative because the question was about the rate of change. The chimpanzee–which also has a lot of sequence available–was less useful because its similarity with humans makes it difficult to test for equality of rates. Other kinds of hypotheses are going to be tested differently, with different data and different combinations of things. Finding individual genes that caused events during the evolution of humans or other species is going to be hard or impossible. But statistical comparisons like this one will help to identify the kinds of events that are likely to happen, and that may have been important in the past.