Belinda Chang and her colleagues have used molecular biology to recreate history in their laboratory

How do you research the vision of a creature that's been extinct for more than 200 million years? Adventurous types might be inclined to try hunting down some fossil remains in the hope of extracting any intact genetic material, DNA. But this approach is still largely within the realm of movies and science fiction. Researchers are hard pressed to show that any genes they have garnered from samples older than a few hundred thousand years really are ancient DNA and not some modern-day contaminant.

Working with Thomas Sakmar and others at Rockefeller University, and with Michael Donoghue at Yale, I have taken a completely different approach. The aim has been to recreate history in the laboratory: to use molecular biology and statistical inference to synthesise a protein that was last made by living creatures almost a quarter of a billion years ago.

We focused on rhodopsin, a protein important in vision at low light levels. The photoreceptor cells lining the back of the eyes of vertebrates use it as a light-activated molecular switch. Light causes rhodopsin to change its shape, triggering a series of biochemical events within the photoreceptor that generates a nerve impulse to the brain. Thus rhodopsin is the first critical step of a complex signalling pathway in vision. Without it, the animal would be unable to perceive light, particularly at night. As with other proteins, the information that cells require to make rhodopsin - the recipe - is stored in coded form within a gene.

ÖйúAƬ

The creatures we wanted to study are called ancestral archosaurs. These ancient reptiles gave rise to some of the most spectacular animals ever to have walked the earth. Many of their descendants, including the dinosaurs of the late Cretaceous period, are extinct. Others, such as birds and crocodiles, live on.

By inference from existing vertebrates, it is likely that the eyes of archosaurs would also have used rhodopsin; so they too would have possessed a gene specifying its make-up. Different animal species make slightly different forms of the rhodopsin molecule. In recent years, researchers have analysed the make-up of the rhodopsins of a number of vertebrates, and the exact sequence of the genes that code for them. Our aim was to construct a rhodopsin family tree comparable to the family tree of the various species of the animals themselves. We used information on rhodopsin from about 30 vertebrates, ranging from mice and men to toads and eels. Along with statistical models describing the way that gene sequences can change over time, this information allowed us to work out the rhodopsin structure most likely to have existed in the archosaurs.

ÖйúAƬ

But is this inferred rhodopsin the same as the one that was found in the archosaurs? All we can say is that, based on understanding of the evolution of molecules, it is our best guess.

Using this structure as a guide, our next step was to assemble a series of nucleotides, the building blocks of DNA, and so synthesise an artificial rhodopsin gene. We put this gene into living tissue culture cells in our laboratory, where it acted as the instruction for making our putative archosaur rhodopsin.

Would our laboratory-recreated archosaur protein have the property fundamental to all rhodopsins from living animals? In other words, in the appropriate circumstances, would it respond to light? We showed not only that it did so, but also that it was able to activate the biochemical reactions leading to vision. Our guess about what the archosaur ancestor would have had in its eyes was plausible.

This is the point at which the work becomes interesting to palaeontologists, because knowing something of the chemical properties of this protein allows you to make inferences about the archosaur eyes in which it might have been found. Rhodopsin's chemistry determines things such as an eye's sensitivity to different colours and light intensities. The archosaur rhodopsin, as measured in our assays, is more sensitive to light at slightly redder wavelengths than those of existing vertebrates. Among present-day vertebrates, this most resembles the rhodopsin in birds: a hint that the vision of today's birds may have retained some of the characteristics of the ancestral archosaurs.

ÖйúAƬ

Our analyses suggest that the archosaurs might also have been able to see in dim light - perhaps as well as existing mammals. This would be consistent with the controversial possibility that the ancestors of birds, reptiles and mammals might have been nocturnal. But more evidence would be needed to assess this intriguing, largely unexplored, idea.

The evolutionary and molecular genetic approach that we have used to recreate history in the laboratory is not restricted to studies of vision. In theory it could be applied to any protein for which sufficient gene-sequence information is held in genetic databanks. The film Jurassic Park notwithstanding, these studies could even be extended to the creation of transgenic animals. By inserting the ancestral gene for rhodopsin, or any other protein of interest into an animal's genome, it might be possible to make better guesses about the lives and behaviours of long-extinct animals.

Belinda Chang is a researcher in Thomas Sakmar's laboratory at Rockefeller University. She spoke to Geoff Watts.