Genesis of Eden

Genesis Home           

Top Translator New Scientist 18 April 1998

THE genetic code is the product of early natural selection, not simply random, say scientists in Britain. Their analysis has shown it to be among the best of more than a billion billion possible codes. Nearly every living organism uses the same DNA language to describe the proteins it makes. A gene is essentially a series of three-letter words which is spelt out using four possible chemical letters, known as bases. Cells use the genetic code to translate this series of three-letter words, or codons, into a chain of amino acids-a protein. Of the 64 possible words, 61 specify one of the 20 amino acids and 3 give the order "stop" to the proteinmaking machinery. Roughly 1020 genetic codes are possible, but the one nature actually uses was adopted as the standard more than 3.5 billion years ago. Now Steven Freeland, a postgraduate student at the University of Cambridge, and evolutionary biologist Laurence Hurst at the University of Bath, have shown that strong selective pressures must have acted on the code during its evolution. "The code has evolved to minimise errors during translation," Hurst concludes. First, Hurst showed that the natural code is far better than the vast majority of randomly generated codes at minimising the errors caused by genetic mutations. He found that single-letter changes to a codon, which meant that the wrong amino acid was inserted into a protein, tended to specify amino acids that were very similar chemically to the correct ones, thereby minimising the impact on the protein. Freeland then built on that model by taking into account errors that occur during the decoding, or translation, of the gene. He reasoned that if the code had evolved to minimise translation errors, it should minimise chemical differences most between the correct and incorrect amino acid at the third base in the codon. The translation machinery misreads this base 10 times as often as the second. In an analysis that gave extra mathematical weight to the vulnerable sites most likely to be mistranslated, Freeland showed that no more than one in a million random codes was better at reducing the impact of errors than the natural code. Hurst and Freeland will say in a forthcoming igsue of the Journal of Molecular Evolution that it is extremely unlikely that such an efficient code arose by chance-natural selection must have played a role. The natural genetic code became universal very early in evolution. Although unconventional codes exist in a handful of places-such as mitochondria-Hurst says there has always been strong pressure for the code to be universal. That's because most organisms exchange DNA. For instance, a virus that infects two species can transfer DNA from one to the other. "Imagine if you got DNA coding for an excellent protein, but when you read the DNA you got a different protein," Hurst explains. "That's not going to do you any good." Eors Szathmary, who is an evolutionary biologist at the Institute for Advanced Study in Budapest, Hungary, says Freeland's analysis supports the theory that evolutionary forces shaped the code. But he points out that another factor probably played a role: the modern genetic code descended from a simpler form with fewer codons. If that's true, similar amino acids might be specified by similar codons simply because they share a common lineage. Jonathan Knight