Encore! Why dream of recreating dinosaurs when we can already resurrect their genes for a repeat performance?

Bob Holmes
2474 words
23 April 2005
New Scientist
volume 186; issue 2496
(c) 2005, New Scientist, Reed Business Information UK, a division of Reed Elsevier, Inc. All Rights Reserved.

DID the first bacteria, billions of years ago, live in hot springs? Were dinosaurs nocturnal? Why did our ancestors evolve a susceptibility to heart disease? No matter how hard you look, you will never find fossils that can answer such questions. In the realm of science fiction, you could simply raise ancient creatures from the dead to find out what they were like. But why dream? There is already a way to resurrect extinct organisms – or at least their genes and the proteins they made.

"It's almost as if we have our own molecular time machine. We can recreate a little bit of the past in our test tube and see how it functions," says Belinda Chang, an evolutionary molecular biologist at the University of Toronto, Canada. Although only a handful of extinct genes have been resurrected so far, Chang and others working in this field believe their approach is set to boom. "I think people don't realise how doable it actually is," she says. "It's as if the time has come, and we'll start to see more and more of these types of studies."

Chang's enthusiasm is understandable, given the potential of gene resurrection. Not only can it provide answers to questions about extinct organisms that were once thought unanswerable, it also gives evolutionary biologists the chance to follow the twists and turns of evolution's path through time – and perhaps even discover what might have happened had it taken a different route. If that is not enough, there is a practical pay-off, too. Medical researchers and drug companies are interested in what the technique can tell them about treating present-day diseases .

Raising a gene from the dead may sound magical, but it is firmly grounded in science. In fact, the idea has been around since the early 1960s, when Nobel laureate Linus Pauling, then at Caltech in Pasadena, and his colleague Emile Zuckerkandl suggested that it ought to be possible to take the DNA sequences of a given gene from several species of living organisms and work backward to deduce the sequence of their common ancestor. By creating that ancestral gene and giving it the molecular machinery it needs to make its protein, the resurrection would be complete.

Pauling and Zuckerkandl were far ahead of their time. Back then, researchers had barely begun to understand the rudiments of the genetic code in which three-letter DNA "words", or codons, specify particular amino acids in the corresponding protein. Cheap, fast gene sequencing was still decades away, so there were none of today's gene databases to use as a starting point in deducing ancestral sequences. And it took a Herculean effort to synthesise even short stretches of DNA, let alone a whole gene.

Fast-forward to the present day. Data banks are bursting at the seams with gene sequences from all sorts of organisms, and it costs just a few thousand dollars to synthesise a gene once you know the DNA sequence – easily within the budget of any well-funded lab. What was wishful thinking just four decades ago is now straightforward molecular biology, though still a lot of hard work.

Take, for example, a gene that Chang and her colleagues resurrected when they set out to learn whether dinosaurs might have been nocturnal. The team chose to focus on the gene for rhodopsin, the pigment protein that all vertebrate eyes use to detect dim light. Since rhodopsin occurs in all living land vertebrates, the team reasoned that dinosaurs must have had it as well. They opted to recreate the rhodopsin gene of the 240-million-year-old progenitor of the archosaurs, a group that includes dinosaurs along with present-day birds and alligators.

With her target chosen, Chang's next step was to work out what the extinct rhodopsin gene looked like. The original was long gone: DNA does not survive long in fossils, breaking down within a few tens of thousands of years. Instead, her team rifled through genetic data banks to get 30 vertebrate rhodopsin sequences from living relatives of the ancestral archosaur. Using well-established evolutionary genealogies, they then sorted these sequences into a family tree and began working backward. Taking two closely related species – chicken and pigeon, say – they lined up the rhodopsin gene sequences and compared them, letter by letter. Where both species share the identical letter, they concluded that the common ancestor of chickens and pigeons also had that letter. Where letters differ, the next-nearest relative helped resolve the issue. Suppose, for example, chickens have an A where pigeons have a G, and the next most related species, the zebra finch, also has a G. The most likely explanation is that the common ancestor of all three had a G at that position, as did the later chicken-pigeon common ancestor, while the lineage leading to chickens underwent a mutation to A .

Dino eyes

Through a series of inferences like this, plus refinements that took account of some kinds of mutation being more common than others, Chang eventually reconstructed a complete sequence for the ancestral archosaur's rhodopsin gene – more than 1000 letters long. It was then a simple matter to make the gene by piecing together short DNA fragments with the right letters, insert it into cultured cells, and coax them to produce a protein that had not existed on Earth for 240 million years (Molecular Biology and Evolution , vol 19, p 1483). Then came the moment of truth – how well would the ancient protein work?

Chang compared the resurrected rhodopsin with a modern one from a cow's eye by applying standard tests that are used to evaluate the performance of rhodopsins. It proved just as capable of detecting light and triggering the next biochemical step in the visual pathway. In other words, the ancestral archosaur had one key part of the machinery to see in dim light just as well as modern mammals, which generally have good night vision, and thus could have had what it takes to be nocturnal. "It's often difficult to infer anything about behaviour and physiology just from fossil bones," says Chang. "This offers the possibility of saying something about the biology of animals that no longer exist, which I think is absolutely fascinating."

Gene resurrection does have its uncertainties, though. For one thing, few proteins work in isolation. Instead, once formed, they usually need to interact with others to function properly. "We're taking an ancestral protein and expressing it in present-day cells," says Joe Thornton, an evolutionary biologist at the University of Oregon in Eugene. "The concern would be that perhaps there's some difference in the way the protein functions in its ancestral context and how it functions in the cells of today." But a protein missing its crucial context would most likely just not work, rather than take on some other, novel role, he notes – so resurrectors who produce a working protein are probably safe in assuming they are looking at its real function.

A second problem is that reconstructed sequences almost always end up with at least a few ambiguities, where researchers cannot be certain which letter belongs in a certain position. "We're always dealing with the best guess, rather than the actual reconstruction of the ancestral sequence," says Mikhail Matz, a molecular evolutionist at the University of Florida's Whitney Laboratory for Marine Bioscience in St Augustine. One solution to this uncertainty is to recreate and test each one of the possible variations of genes and proteins. If all behave similarly, then you can safely ignore the ambiguity. These complications mean that some resurrections will work better than others, says Steve Benner, a biological chemist at the University of Florida in Gainesville. All else being equal, it should be easier to resurrect a small gene than a large one, because it offers fewer opportunities for critical errors to creep in. Similarly, a gene that has already been sequenced in many different living organisms gives researchers more information to work with, leading to less ambiguity. And more recent ancestors should be easier to reconstruct than more distant ones, because evolution will have had less opportunity to obscure the ancestral sequence.

Nevertheless, even long-extinct genes can be studied with confidence if they have not changed much. For instance, Benner's team has reached more than a billion years into the past to study the lifestyle of the ancestor of all bacteria. Here, one of the most keenly debated questions is whether early life evolved in hot springs or at more moderate temperatures. Various lines of circumstantial evidence have not resolved the issue, so Benner decided to try a more direct approach. He would resurrect a gene central to the successful functioning of the first bacteria and find out the temperature at which the protein it produced worked best.

The gene he chose is called EF-Tu . It codes for a protein that helps transport amino acids for protein synthesis. With such a fundamental role to play, EF-Tu varies very little even among distantly related organisms. So, even allowing for all the ambiguity in retracing the genetic sequence over a billion years, Benner's team came up with only three likely ancestral sequences. And all the proteins they produced from these sequences proved most stable at between 55 and 65 °C (Nature , vol 425, p 285). In other words, at least this one crucial protein from the ancestral bacterium seems best adapted to moderately hot springs – not temperate seas or puddles, but not super-hot springs either.

From the nightlife of dinosaurs to the temperature preferences of the first bacteria, resurrecting DNA can provide unexpected snapshots of the past. But it can also reveal the process of evolution itself and so explain some enduring mysteries. Take the evolution of steroid hormones and the specific receptors to which each binds. Modern mammals have six such hormone-receptor pairs. But how did these six hormone-plus-receptor pairs evolve from the original one?

Unlikely duo

"It's a real puzzle," says Thornton. "If you imagine a single ancient hormone with one receptor, there is no selection pressure to drive the evolution of a new hormone if there's not already a receptor to receive it. And conversely, there is no pressure to drive the evolution of a new receptor if there's not already a hormone for it to receive." The only obvious solution – that both evolved simultaneously – seems too improbable to be true, he says.

To tackle the conundrum, Thornton and his colleagues reconstructed the original steroid receptor gene by comparing 73 sequences from living organisms. By synthesising the ancient protein and testing to see which steroids activate it, they learned that the ancestral steroid receptor functioned specifically as an oestrogen receptor (Science , vol 301, p 1714). This suggests a solution to Thornton's chicken-and-egg problem, because two other modern steroid hormones, testosterone and progesterone, are produced as intermediates in the biochemical pathway to oestrogen. So these chemicals must have been present in the body before they were hormones. When the receptor gene was accidentally duplicated, the duplicate copies apparently evolved to recognise these intermediates, co-opting them for use as true hormones.

The remaining steroid hormones may have arisen in a similar way. Thornton has recently resurrected ancestral genes for their receptors and, although the work is not yet published, he says the picture is much the same.

But Thornton has even bigger ideas. Along with a few other advocates of gene resurrection, he dreams of not just understanding the path that evolution has taken, but of investigating alternative paths. Recreating a gene from the distant past, they say, effectively gives us a rewind button – a chance to go back and replay evolution in the lab. Thornton, for example, plans to start with the ancestral oestrogen receptor and push it to evolve in a different way by imposing selection pressures that force it to recognise a different steroid, such as testosterone. If the effort succeeds, he can compare his mark 2 testosterone receptor to the mark 1 version – that is, the real product of hundreds of millions of years of evolution – to see how their amino acid sequence differs.

If the two differ greatly, we will know that there is more than one way to skin that particular evolutionary cat. On the other hand, if the new receptor is almost identical to the existing one, then evolution probably had only one channel to follow. Either way, we will be a lot closer to making the history of life on Earth an experimental science.

SIDEBAR On the origin of illnesses

By Bob Holmes

MEDICAL researchers could reap huge rewards by resurrecting extinct genes. "A lot of the biomedical community has not appreciated how important historical views are to our understanding of disease," says Steve Benner from the University of Florida in Gainesville. "If you understand the past and how it emerged into the present, you understand the present better."

Researchers in Benner's lab are reconstructing ancient versions of enzymes called tyrosine kinases, which are responsible for switching on and off many processes involved in cell growth and division. He believes that understanding how these enzymes evolved among our ancestors will help pinpoint how and why they sometimes fail, leading to cancer and other diseases.

Cardiovascular disease has also caught Benner's eye. One cause of high blood pressure, which is a powerful risk factor for all sorts of cardiovascular problems, is a defect in a gene called uricase, which controls the breakdown of uric acid. The defect, which arose 10 or 20 million years ago, means that humans can have unusually high uric acid levels in their blood compared with other mammals, contributing to gout and salt-sensitive hypertension. By recreating the ancient uricases around the time of their loss, Benner hopes to learn more about why the loss happened. Did it benefit ancient humans in some other way? If so, that might suggest ways to treat hypertension, or warn of side effects from tampering with uric acid.

Quite apart from what ancient proteins tell us about how the human body and its medical problems evolved, they have another valuable feature: they worked. It is this that interests Janos Kodra, a chemist at Danish drug company Novo Nordisk. Biochemists who tweak existing proteins to change their behaviour in subtle ways must contend with the problem that almost every change they make will harm, and probably totally destroy, their function. But looking at ancestral proteins avoids the problem. "They are evolutionarily optimised," says Kodra. As a result, they give biochemists a pre-tested range of "new" variants to explore.

Document NEWSCI0020050422e14n0000b