When the chimpanzee genome sequence is published, sometime in the next few weeks, what everyone will want to know is how it compares with ours, and what genetic differences set us apart from our ape cousins. But looking for “the genes that make us human”, as people will doubtless be tempted to do, is not going to get us very far. As this special issue of 91av reveals, a simple gene-for-gene comparison is only the first step to finding out what makes humans so special.
For one thing, many of the key differences won’t be in genes at all. They will be in parts of the genome that don’t control which protein is produced, but rather where it is produced, how, and in what quantities. Robin Orwant explores this surprising idea in the article starting opposite.
What’s more, the sequence of DNA letters is only part of a bigger picture. It seems that our entire genome is uniquely innovative, creating far more opportunities for new genes and new functions to evolve than in other mammals. This creativity explains not just how we came to be different, but why, as Bob Holmes explains on page 40.
Advertisement
The story of what makes us special is written in our DNA — but not necessarily in our genes.
THE bubble burst sometime in February 2001. Before then, geneticists expected humans to have about 100,000 genes, reflecting how big and clever we are. Now we know better. The results of the human genome project suggest the figure is closer to 30,000 – only five times as many as your average bacterium and roughly the same as mice. To add insult to injury, most of our genes are virtually identical to those of chimps. If we have so much in common with the rest of the animal kingdom, what makes us unique?
The answer may lie in the vast regions of the genome that do not code for proteins, but instead control gene expression – in particular so-called cis-regulatory sequences (see Graphic), which regulate when, where and how much a gene is expressed. Many species, including humans, harbour large amounts of variation in these sequences. What is more, it is becoming clear that this variation can account for differences between species, too. One study, for example, has shown that humans express certain genes in the brain in a different way to chimps, which perhaps explains why we are cleverer than our primate cousins. The emerging conclusion seems to be it ain’t so much what you’ve got, as the way that you use it.
A hundred years after Darwin proposed his theory of natural selection, genes alone seemed to hold the key to understanding the differences between species. By tinkering with coding sequences, biochemists found they could sometimes alter an organism’s anatomy or physiology. This suggested that chance mutations in genes would also produce physical variation, providing fodder for natural selection. And so emerged the belief that if we examined the genomes of two different species, such as humans and chimps, we would find lots of coding variations that would account for the differences between them.
But as early as 1975, this prediction wasn’t holding up. In a seminal study, Mary-Claire King and Allan Wilson at the University of California, Berkeley, found that humans and chimps have an almost identical set of proteins (Science, vol 188, p 107). Later, direct comparisons showed that human coding sequences differ by only 1.2 per cent from those in chimps, and even mouse genes are 85 per cent the same as equivalent genes in humans.
“After King and Wilson, the prevailing view was that you wouldn’t see important differences in proteins,” says geneticist Andrew Clark of Cornell University in New York. And so many scientists began to suspect that species differ from each other mostly because of differences in gene expression, rather than in the genes themselves.
Central to this “regulatory theory” is the idea that alterations in a gene’s cis-regulatory sequences can change how it interacts with transcription factors – the proteins that switch genes on or off – and radically alter where, when and how much the gene is expressed. Think of cis-regulatory sequences as the coin slots on a bar-room pool table and transcription factors as the coins. Insert the correct combination of coins and you release the balls. But change the slots and the balls are stuck, or require a different set of coins to release them. In the same way, mutations in cis-regulatory sequences can block the transcription factors, or allow different ones to dock, permitting expression of the gene in tissues where it was never expressed before.
Developmental biologists were among the first to embrace this regulatory theory. Time and again they found that mutations altering where and when a gene was expressed could dramatically affect the way an embryo developed. They also found that a wide variety of species used near-identical developmental genes that differed only in the way they were expressed. The inescapable conclusion was that differences between species were largely down to gene expression. “By about 1995, the writing was on the wall,” recalls Sean Carroll of the University of Wisconsin, Madison.
For example, a subset of Hox genes, which are involved in mapping the body plan, help control the development of both fins in fish and limbs in four-legged animals. In mammals, Hox genes have cis-regulatory DNA sequences that are completely absent in fish. They promote gene expression at certain stages of development, triggering the formation of feet and toes.
Such evidence has convinced Carroll that cis-regulatory changes are more important than alterations in coding DNA. “This work is brutally hard so we don’t have a huge number of examples,” he admits. “But if we extrapolate from model organisms, most of the time the action is going to be in the non-coding DNA.”
Direct comparisons between species are hard to do, but there are ways to search for regulatory differences. If Carroll is right and cis-regulatory variations are the raw material of evolution, you would expect to find lots of variation within species as well as between them. This means that if you compare the genomes of different people, for example, you ought to find many cis-regulatory variations.
Until recently, no one had tried to test this prediction. But a little over a year ago, Matthew Rockman and Gregory Wray of Duke University in Durham, North Carolina, compiled data from more than 400 studies to get a sense of how prevalent cis-regulatory variations are in humans (Molecular Biology and Evolution, vol 19, p 1991). Many of the studies were done by medical geneticists studying genes implicated in common diseases, such as asthma. Most did not set out to look for cis-regulatory variations and only started looking at non-coding sequences when they found that coding DNA variations couldn’t account for all the differences they saw in disease susceptibility.
By compiling all the studies they could find, Rockman and Wray estimate that there are at least 16,000 cis-regulatory sites that vary from person to person. By comparison, a previous study suggested that the equivalent figure for coding sites is 13,000 (Nature Genetics, vol 22, p 231). In other words, there’s more cis-regulatory variation than coding variation in the human genome – and presumably this holds true in other species as well.
Rockman says these estimates should serve as a wake-up call for biologists who are trying to understand human evolution by focusing on coding sequence variation. But he admits they are just estimates. For a more accurate picture, systematic surveys are the way to go.
One approach might be to scan the genome for cis-regulatory variation. But this is a bit like looking for the proverbial needle in a haystack. “The information in non-coding DNA is much harder to get at than the information in coding DNA,” explains Carroll. While the information in the genetic code is well known, the information in non-coding DNA is cryptic. Most of the time it is impossible to identify cis-regulatory elements from their DNA sequence alone. Also, cis-regulatory elements aren’t necessarily near the genes they regulate. Some are tens or even hundreds of thousands of base pairs away. And even when researchers can identify cis-regulatory elements, it is impossible to predict how they will affect gene expression. Many seem to have no effect at all, or only act in one tissue or within a limited time window. The only way to know for certain is to test each one in every tissue over the lifetime of an organism – an impossible task.
So instead of sequence gazing, Christopher Cowles and Eric Lander from the Whitehead Institute in Cambridge, Massachusetts, tried a different approach. They reasoned that cis-regulatory variation could result in different versions (or alleles) of the same gene being expressed at different levels in the same individual. So they crossed two inbred strains of mouse and analysed the expression of 69 randomly chosen genes in the offspring (Nature Genetics, vol 32, p 432). For four of these genes, they found good evidence of expression differences caused by cis-regulatory variation. From their sample they estimate that 6 per cent of mouse genes contain cis-regulatory variation. Other groups have used similar methods on cultured human cells and fetal tissues, and come up with estimates ranging from 18 per cent to 54 per cent.
These are very rough estimates based on small samples, but new technologies are coming on the scene that should help firm them up. In what Cowles calls a “ground-breaking study” published last March (Nature, vol 422, p 297), Eric Schadt at Rosetta Inpharmatics in Kirkland, Washington, used DNA chips to look at how different strains of mice expressed nearly 24,000 genes. He found around 1000 genes that varied in their expression level and also contained variation in nearby non-coding sequences – probably cis-regulatory sequences. This led to a conservative estimate that 3 per cent of mouse genes have cis-regulatory variation. Leonid Kruglyak of the Fred Hutchinson Cancer Research Center in Seattle did the same kind of study in yeast and came up with a similar estimate (Science, vol 296, p 752).
Putting all these studies together, it is impossible not to conclude that cis-regulatory variation deserves more attention than it has been given in the past. “It underscores the fact that this is a prevalent form of variation,” says Cowles. Kruglyak adds: “Focusing just on coding changes misses a lot of variation that’s out there.”
But what does all this cis-regulatory variation mean for human evolution? There certainly appear to be lots of cis-regulatory differences between species, but as Carroll points out, it is impossible to say how many of these differences are important. Just because a variation in cis-regulation changes gene expression does not mean it makes the difference between, say, having language and not. “A lot of this genetic variation is going to be morphologically invisible,” he claims. That is to say, most of the time, it doesn’t matter.
But the same is true for variations in protein-coding sequences. Most changes don’t affect the protein at all. Even if they do, the protein often functions normally. And even if the protein does behave slightly differently, it may have no effect on how the organism looks or functions.
Cowles admits that most cis-regulatory variation will turn out to be background noise. But he believes studies like his are an important starting point. “This might be a fishing expedition,” he says, “but fishing expeditions often generate hypotheses.”
In fact, the results of another fishing expedition suggest that some cis-regulatory variations were important in the evolution of the human brain. Using DNA chips, Svante Pääbo of the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, and Ajit Varki of the University of California, San Diego, compared expression patterns for about 18,000 genes shared by humans, chimps, macaques and orang-utans (Science, vol 296, p 340). They found that in blood and liver tissue, humans and chimps expressed the genes in a very similar way. But when they looked at brain tissue, they found that chimps’ expression patterns were closer to those of other primates than those of humans.
The group concluded that a major acceleration in the evolution of regulatory factors in humans allowed them to develop unique gene expression patterns in the brain. “Some of those changes could have been involved in what makes the human brain different,” says Varki. The team is now trying to identify which changes make humans unique.
Meanwhile, Rockman has already found one candidate gene whose regulation may help explain what makes us human. The gene, whose identity he is keeping secret pending publication, codes for a neuropeptide and is regulated by a cis-regulatory region that affects the gene’s expression in the brain. The chimp’s cis-regulator is similar to the corresponding region in other primates, but differs markedly from that in humans. Rockman suspects this allows humans to express the gene differently to chimps, perhaps leading to important alterations in brain function.
But despite all the evidence in favour of the regulatory theory, the old idea that protein-coding sequences are what really matter simply won’t go away. In a recent study, Clark and colleagues compared 7645 genes shared by humans and chimps and found that nearly 9 per cent of the human versions had evolved after humans and chimps diverged (91av, 20 December 2003, p 9) “I was astonished at how many we got,” says Clark. Though he still thinks the regulatory theory is sound, Clark believes his findings challenge the notion that there are no important differences between human genes and their chimp counterparts.
With the completion of the chimp genome, more gene-for-gene comparisons will inevitably follow, and Carroll worries that researchers will turn to the low-hanging fruit of coding-sequence variation to explain how humans differ from chimps, which would be a huge mistake, he contends. Convinced that cis-regulatory variations will turn out to be the smoking guns of evolution, he insists: “The traits we most care about in humans, such as language acquisition and cognition, will be due to tinkering in gene regulation. It’s inescapable from what we know from developmental biology.”
Varki, though, is not quite as adamant. “Cis-regulatory elements are probably important,” he says, “but it doesn’t mean that other mechanisms aren’t also important. Given the many steps in human evolution, I can’t imagine that it doesn’t involve every trick in the book.” And even if we knew which sequences were responsible for which unique human attributes, it would be hard to pinpoint the most crucial changes, he argues, asking, “Which is more important: walking upright, having an opposable thumb or developing a big brain?”