MOST people, if they think of evolution at all, see it as a neat progression from spineless blobs in the primeval soup to that most sophisticated of creatures – ourselves. The notion that bacteria might in some way be more advanced than us seems absurd. Yet this is exactly what a small band of molecular biologists is proposing.
These scientists believe they have found evidence that single-celled organisms, such as bacteria, which are too primitive even to package their genetic material inside a nucleus, have the most advanced genes on Earth. By contrast, they say, human genes and those of all other so-called higher organisms have changed little since the first living things emerged around 3.6 billion years ago. Needless to say, not everyone agrees with this view.
At first glance, the debate seems quite esoteric. It hinges on apparently worthless stretches of DNA called introns and asks whether these genetic scraps evolved with the first living organisms or appeared less than a billion years ago. But the implications are huge. If introns were there in the beginning then human genes would reflect the primordial pattern. Simple organisms, conversely, would in a narrow sense be highly evolved, since their genes are devoid of introns. The formation of introns early in evolution could also explain how a few tens of thousands of useful proteins evolved from a huge number of possible combinations of amino acids.
Advertisement
The battle has been fought hard but decorously in the pages of prominent science journals. In private, though, the mud slinging is more acrimonious. Last year a paper published in Science seemed to stack the odds in favour of the “introns-late” view. But within months the “introns-early” people had retaliated with what they say is the strongest evidence yet to support their hypothesis. Their report has yet to be published but has been aired at scientific gatherings.
Has the debate been settled at last? First some history. In 1977, researchers made an embarrassing discovery. They began to realise that genes from their “model” organism, the bacterium Escherichia coli, were different from those throughout most of nature. Until then, all the evidence had pointed to a direct correspondence between the structure of genes and the proteins for which they code. In E. coli, the string of nucleotides of a gene is translated directly via a messenger RNA molecule, into a string of amino acids.
Surprising waste
Biologists expected to find this one-to-one correspondence everywhere, from orchids to elephants. No one was prepared for the discovery that most genes in higher organisms are not like that. It turns out that the information that codes for proteins in genes like our own is contained in small packets, which came to be called exons, interrupted by stretches of DNA that apparently code for nothing – the introns. On average each gene contains about half a dozen introns which are typically ten times as long as the exons.
This discovery solved a couple of questions that biologists had been puzzling over for some time. The first was the fact that many organisms seemed to have too much DNA in their cells for the number of proteins they produced. Second was the fleeting appearance of very long strands of RNA when these cells were actively manufacturing proteins. This RNA is now recognised as the initial step in the transcription of a gene and contains both exons and introns. The introns are then spliced out and the exons are joined end to end, forming an intact molecule of messenger RNA from which a protein is made. But this discovery of “split genes” – genes broken up by introns – was to pose more questions than it solved. Where, for example, did the introns come from? And how were they removed in the production of mature messenger RNA?
Evolutionary shuffle
One of the first people to tackle these questions was Harvard biologist Walter Gilbert. In 1978, he suggested that exons might code for some kind of functional unit within proteins, in which case, the intron/exon structure of genes could speed up evolution. Novel genes could be produced quickly by shuffling entire exons, rather than by relying on the slower process of incremental mutations at specific points in genes.
This would explain why natural selection let introns invade genes without evicting them en masse – but not necessarily when. Yet it wasn’t long before two other researchers, James Darnell of Rockefeller University, New York, and Ford Doolittle of Dalhousie University, Nova Scotia, pointed out that introns must have been present in the earliest, primordial genes because there could be little short-term benefit in splitting up previously intact genes.
Gilbert later melded these ideas into what he called the exon theory of genes. The earliest genes, he said, were assembled by exon shuffling. Primordial exons were short and coded for polypeptides only 15 to 20 amino acids long – a size that corresponds to what Gilbert calculated to be the smallest possible building block of proteins. “Over the sweep of evolutionary time, introns are lost and more complicated exons are formed,” he proposed in a classic paper in 1987. The average exon in modern genes, for instance, is two to three times longer than the proposed primordial exons. Bacteria simply take this process to the limit, losing all their introns in a bid to streamline their genetic material.
A virtue of this theory was that it could explain how relatively few proteins evolved from a virtually infinite range of possibilities. Given that the average protein is some 200 amino acids long, and that the pool of amino acids from which proteins are manufactured is 20, the total number of possible such proteins is 20200. There would be “neither the time nor carbon in the Universe to explore all these possible structures and find the relevant one,” says Gilbert. If, however, natural selection worked with mini-exons, then the number of possibilities to be explored is dramatically reduced, perhaps to a few million. This neat argument was enough to convert many molecular biologists to the theory.
But not all. One critic was Periannan Senapathy, then at the National Institutes of Health, Division of Computer Research and Technology in Bethesda, Maryland. Senapathy likes the idea that introns were present in primordial genes but claims that Gilbert’s theory doesn’t explain certain features of modern genes, such as why the length of exons in modern genes is random and why almost all exons end the same way, with three nucleotides representing a “stop” signal.
As a result, Senapathy has come up with his own split-gene theory. The first DNA, he says, formed when nucleotides came together in random order to produce small strands. These grew in length until, by chance the three nucleotides that code for “stop” were tacked onto the end. Some of the resulting fragments – the exons – coded for parts of protein, the rest including the introns was genetic junk.
By the time Senapathy published his findings, in 1986, there was already growing opposition to any notion that introns appeared early in evolution. Prominent among the critics were the British biologists Tom Cavalier-Smith, now at the University of British Columbia, and John Rogers from the University of Cambridge. The way they see it, the irregular pattern of introns in modern genes means they were inserted within the last billion years; as does the conspicuous absence of introns in today’s simple organisms. Had introns really been present in the earliest organisms, one would have to argue that tens of thousands subsequently vanished in many different species. And that, say Jeffrey Palmer and John Logsdon of Indiana University (see Diagram), is pretty unlikely.
Such arguments persuaded Doolittle to switch from the introns-early to the introns-late camp. Not so Gilbert, who maintains that losing introns is much easier than proponents of the introns-late hypothesis admit – given that streamlined genes would be more efficient at storing information.
So who is right? In principle, comparing the positions of introns in genes in different species should help to resolve the debate. In practice, it doesn’t. In some genes, introns occupy the same places in plants and animals – a coincidence that is hard to explain unless introns appeared early in evolution, before animals and plants diverged. But in other genes, intron positions vary from species to species, more in keeping with the introns-late theory.
Now all hopes for settling the matter are pinned on attempts to predict the positions of as yet undiscovered introns. According to Gilbert’s theory, this is made possible by the fact that each exon in a gene codes for a well-defined segment of a protein. In other words, introns should always be found between stretches of DNA with specific biological functions – a demand not made by Senapathy’s split-gene theory or the intron-late hypothesis. True to this picture, Gilbert and his colleagues have managed to predict and then find some introns sitting between stretches of DNA that code for specific protein segments. But lately the tide has been turning against them.
Arlin Stoltzfus, a researcher in Doolittle’s laboratory, used a battery of statistical tests to look for correspondence between intron position and protein structure in four genes of ancient origin. He found correspondence in only one of these genes, which could, he says be explained by pure chance.
The analysis looks powerful. “[It] may represent a turning point in the debate,” wrote John Mattick, a biologist at the University of Queensland, in a recent review. Gilbert, however, remains unimpressed. Introns are slippery customers, he says: they can vanish, slide from place to place, invade new genes. “Their argument fails to take into account the problem posed by [this] biological reality.”
At a meeting in Japan last December, Gilbert presented his latest counter-attack. His work has still to be published but looks in great detail at the structure of DNA, and particularly at how the positions of introns affect the “sense” of DNA code. Each amino acid is coded for by a triplet of nucleotides. Coding regions of genes therefore consist of a series of such triplets, which can be caricatured as 3-123-123-123-etc. Molecular biologists refer to this as the reading frame. Shift the phase of the frame by one nucleotide and you get 31-231 231-23-etc, which will produce a different protein, or, more likely, no protein at all. Gilbert argues that introns would interfere with such reading frames only if they were inserted randomly, and by implication late in evolution. And based on an analysis of some 431 genes of ancient origin, he claims this simply doesn’t happen.
But nobody is yet claiming victory. Maybe Gilbert is wrong and late introns weren’t inserted randomly. Maybe his opponents are wrong about the positions of introns. Indeed, maybe we’ll never know how our genes came to acquire the chemical structures they have today, or whether bacteria have the most sophisticated, streamlined genes of all. Gilbert, for one, seems resigned to the inscrutability of biology’s deep past. “That’s the nature of evolution,” he shrugs.