Judging from fossils and studies on DNA, the common ancestor of humans, chimpanzees, and bonobos lived roughly six million years ago. Hominids inherited the genome of that ancestor, and over time it evolved into the human genome. A major force driving that change was natural selection: a mutant gene that allowed hominids to produce more descendants than other versions of the gene became more common over time. Now that scientists can compare the genomes of humans, chimpanzees, mice, and other animals, they can pinpoint some of the genes that underwent particularly strong natural selection since the dawn of hominids. You might think that at the top of the list the scientists would put genes involved in the things that set us apart most obviously from other animals, such as our oversized brains or our upright posture. But according to the latest scan of some 13,000 human genes, that's not the case. Natural selection has been focused on other things--less obvious ones, but no less important. While the results of this scan are all fascinating, one stands out in particular. The authors of the study argue that much of our evolution is the result of a war we are waging against our own cells.
It's possible to reconstruct the history of natural selection thanks to a quirk in DNA. Genes carry the code for making proteins, but it's possible to change the code without changing the resulting protein. Consider for example how cells stick new amino acids at the end of growing proteins. The nucleotides in a gene can't have a one-to-one correspondence to amino acids, since there are only four nucleotides in DNA and twenty amino acids. Instead, the cell reads three nucleotides at a time from a gene, and then chooses its amino acid. The triplet CUU makes the amino acid leucine, for example. But so does CUC and CUA. In many cases, the last nucleotide in a triplet is irrelevant.The most intriguing result of this study is that we appear to be in an intense war with our own cells.
If a hominid's genes mutated such that CUC became CUA, the mutation would have no effect for good or bad on that hominid, because the mutation didn't change its proteins. Scientists have found that mutations to "non-coding" DNA can slowly spread through an entire species thanks merely to chance. If you compare a particular gene from a human and a chimpanzee and a gorilla, you'll see that each species has picked up some silent mutations since it split off from the common ancestor of all three species.
But mutations that actually change a protein's structure are a different kettle of fish. Many of them turn out to be outright disasters, leading to diseases, spontaneous abortions, and so on. These mutations tend to be weeded out by natural selection. On the other hand, mutations that change proteins in an adaptive way can spread quickly. And if a protein is under intense natural selection, a whole series of mutations to coding DNA may build up in its gene.
One sign that a gene has undergone intense natural selection in the past is the ratio of mutations to its coding DNA to mutations to its non-coding DNA. If the coding mutations significantly outnumber the non-coding ones, it's a safe bet that this ratio is the result of intense natural selection. There are other methods for detecting natural selection, but what I've described here is the basic idea behind the new PLOS Biology paper. Expanding on an earlier scan, the researchers looked for genes that showed signs of significant natural selection by comparing their sequences in humans and chimps. They then sorted these genes according to which organs they are active in the most, and made a "Top-50" list of the genes that have undergone the most intense natural selection.
The human brain, remarkably enough, shows no sign of harboring a lot of fast-evolving genes compared to other organs. "In fact," the authors write, "genes expressed in the brain seem to be among the most conserved genes with the least evidence for positive selection." Instead, they suggest, our unparalleled brains may have evolved through adaptive changes in relatively few genes, or perhaps by borrowing existing genes that were active elsewhere in the body (I've blogged about this gene-borrowing here).
So where did all the intense selection take place? Some of it turns up in the immune system, which must battle a rapidly evolving army of parasites. Some of it turns up in the nose, possibly in order to sniff out dangerous foods or possibly to recognize suitable mates. Some of it seems to be involved in how sperm and egg recognize one another. But the most fascinating set of fast-evolving genes do something else altogether: they control the way cells kill themselves.
Suicide is essential for a healthy body. Cells kill themselves for many good reasons--to protect other cells if they are infected with a dangerous pathogen, for example, or to stop the growth of an organ once it reaches the right size. Our hands would look like webbed duck feet if the cells between our fingers didn't commit suicide.
Sperm turn out to be a particularly suicidal bunch. Three-quarters of potential sperm cells kill themselves. Some researchers have suggested that they are so prone to suicide because their population needs to be kept in balance with the other cells in the testes that nourish them. The death of the individual sperm benefits the entire population--and thus the man who carries them.
On an evolutionary level, this creates a conflict between sperm and man. If one of the cells should mutate in such a way that it could escape suicide, it could reproduce madly while other sperm cells dutifully destroyed themselves. These mutant sperm would then be more likely to reach an egg, and as a result the mutant suicide gene would become more common.
While this kind of mutation may favor an individual sperm, it may do harm to its owner. His overall sperm production might suffer as a result of this mutiny down below, for example. It might even increase his risk of cancer. After all, one of the hallmarks of cancer is the mutation of suicide genes, allowing cancer cells to grow rapidly into tumors. Once a sperm fertilized an egg, its suicide-escaping genes would wind up in every cell of the resulting person, raising their chance of turning cancerous. (See this post for more on the intersection of evolution and cancer.)
The authors of the study point out that many of the genes that end up near the top of their list have long been known to be involved in cancer. Perhaps, they suggest, many cases of cancer are the result of this pressure on sperm to escape suicide. And if their hypothesis is right, then you'd expect that a mutation that can stop these renegade sperm from wreaking havoc might be favored by natural selection. There are a number of genes that are crucial for suppressing tumors, and--as predicted--they are also among the fastest-evolving genes. In fact, some of these fast-evolving tumor suppressing genes are only active in the testes, where they may be keeping sperm in check.
This sort of two-level evolution may seem bizarre, but biologists are documenting a growing number of cases of it. It was particularly important, for example, in the evolution of multicellular animals from singe-celled protists some 700 million years ago. But it's hardly ancient history, this new study suggests. Every time cancer strikes, it makes its presence known.
Update, 7pm: PZ Meyers offers a detailed tour of the Top 50.
I would take genome wide dn/ds studies with a grain of salt. Using ds as a control presents two major challenges: 1. You must assume that ds is an accurate representative of the neutral evolutionary rate; and 2. With closely related species (eg humans and chimps) ds won't be saturated, but there may have not been enough substitutions to accurately estimate dn, ds, or both (this is analogous to not flipping a coin enough times to realize the probability of heads and tails). This study is statistically rigrorous (the authors are accomplished in statistical genetics) and includes some polymorphism data as well. Both of these features add weight to the conclusions.
I'll just let you know that I was looking over a dataset of my own in which I though the sequences were too similar to each other to efficiently detect selection with someone extremely familiar with the human-chimp divergence data. He said he was envious of the amount of divergence I had between my sequences (i.e., much more than the chimp data), so I can only imagine how hard it is to perform comparative analyses on humans and chimps.
A point on the brain gene issue - check out this paper.
Dorus S, Vallender EJ, Evans PD, Anderson JR, Gilbert SL, Mahowald M, Wyckoff GJ, Malcom CM, Lahn BT. Accelerated evolution of nervous system genes in the origin of Homo sapiens. Cell. 2004 Dec 29;119(7):1027-40
They compared human, macacque, rat and mouse lineages - a far more extensive study than the human/chimp one you're talking about here. They found that between primate and rodent, and also between human and macacque, that the major selective effects were on brain developmental genes. Possibly human and chimpanzee are just too similar to pick up this effect?
Turning to the human/chimp comparison, there are a couple of other reasons why testis genes may be highly selected. In particular, they end up in the haploid state in spermatids. This uncovers all the recessive alleles, exposing them to selection. Thus, you'd expect spermatid genes to show a higher rate of evolution *anyway*, even if the selective advantage of any particular allele isn't very high.
Secondly, I wonder what the chromosomal distribution of these genes is - I wouldn't be at all susprised to see an over-representation of X and Y chromosome genes in the list. These evolve extra fast due to the lack of recombination (especially the Y), and also end up enriched for testis genes.
Finally, you're quite right to point out the conflict between the germ cells - this has slightly wider ramifications than you cover. Again, due to the fact that spermatids have a haploid gene content, there is the potential for conflict between the two copies of the diploid genome, not just between the individual and the herd. This may be why the majority of germ cell apoptosis occurs in spermatogonia and spermatocytes, not in spermatids.
I think the overall conclusion is that we have to be quite careful when looking at rates of protein change - fast rates of evolutionary change don't necessarily imply strong selective pressure, particularly for testis genes and/or genes on the sex chromosomes.
Turning to the human/chimp comparison, there are a couple of other reasons why testis genes may be highly selected. In particular, they end up in the haploid state in spermatids. This uncovers all the recessive alleles, exposing them to selection. Thus, you'd expect spermatid genes to show a higher rate of evolution *anyway*, even if the selective advantage of any particular allele isn't very high.
That assumes that these genes are expressed in the spermatids. Many of the genes are probably only expressed in testes (which contain many diploid cells responsible for spermatogenesis) and not in the spermatids.
The observation of faster evolution of testes specific genes is not rare; in fact, it's the norm. Evidence from Drosophila (and other studies as well) has suggested this for a long time. The evidence for faster X evolution is not as conclusive.
Also, the point about what causes faster rates of evolution is an important one. It is assumed that dN/dS1 implies positive selection. The authors identify genes with dN/dS significantly greater than 1, allowing them to infer positive selection. Unless there is strong selective constraint on synonymous substitutions at those loci, I believe they have strong support for positive selection.
One other issue that may confound the sperm data is the differences in chimp and human mating systems. Chimp testes (compared to body mass) are significantly larger than human testes and one potential reason is that chimps do not maintain monogamous pair bonds and as such there may be significantly more sperm-sperm competition. I don't know how this may fit into the differences but it may be worth consideration.
Where do I find the paper? Did I miss a reference?
RPM: Agreed that testis expression does not necessarily imply spermatid expression, but there have been a number of studies out in the last few years suggesting that the majority of genes *specifically* expressed in testis are in meiotic and post-meiotic (spermatid) germ cell stages.
The one that comes top of the dataset, protamine, is a case in point. It would be good to go through the list and check, at the least.
Agreed they have good evidence for positive selection - my point is rather that depending on the manner in which genes are expressed, the degree of positive selection may not be as strongly correlated with reproductive fitness as you'd expect.
I didn't realize I had linked to the summary instead of the paper itself. It's here.The link in the article is also fixed.