MICROSOFT does it every few years. So does Apple. But how often does Mother Nature release a new operating system? The last one probably made its debut during the Metazoan radiation, when complex multicellular life appeared on the scene. And now, some 600 million years later, scientists think they鈥檙e beginning to discover how it works.
We鈥檙e talking about a genetic operating system where the lines of code are embedded in a genome rather than in silicon. And at the heart of this system, according to John Mattick and his colleagues at the University of Queensland in Brisbane, is a motley collection of RNA molecules widely derided as useless. They propose that these RNAs, called 鈥渘on-coding鈥� RNAs because they don鈥檛 carry instructions for making proteins, are part of a network linking genes that could be responsible for fast-tracking the evolution of humans and other complex organisms.
If Mattick is right, and it will be a few years before we know, biology textbooks will need to be rewritten to reflect this radical shift in thinking. And scientists will need to learn a whole new genetic language.
Advertisement
Among the central tenets of genetics set out 50 years ago is the belief that DNA is a gene鈥檚 鈥渂iological software鈥�, coding solely for a particular protein through the intermediary of messenger RNA (mRNA). This idea was modified when ribosomal and transfer RNAs were also found to be involved in protein synthesis. But RNA was, and still is, viewed as nothing but protein-producing machinery.
That鈥檚 fine for bacteria, whose genomes consist almost entirely of closely packed sequences of protein-coding DNA. They have a primitive genetic operating system that Mattick and his colleagues say lacks the power to catalyse the evolutionary 鈥渂ig bang鈥� that led to higher organisms.
In 鈥渆ukaryotes鈥� with nucleated cells, on the other hand, such as yeasts, plants and humans, protein-coding genes have a 鈥渕osaic鈥� structure. The mosaic consists of coding regions called exons and intervening, non-coding regions called introns. When a gene is transcribed to make an RNA copy, the raw transcript includes all the introns and exons. Then the introns are removed and the exons are 鈥渟pliced鈥� end-to-end to form the mRNA that directs protein synthesis.
But that鈥檚 only half of the tale, says Mattick. What about introns? Why are they there? The received wisdom is that introns and other non-coding RNAs that are not directly involved in protein synthesis鈥攆or example, RNAs transcribed from regions between genes鈥攁re 鈥渆volutionary hangovers鈥� of little or no function. Mattick refuses to go along with that. 鈥淪omeone once said, 鈥榯he best science is done at the point of greatest surprise鈥�,鈥� he says. 鈥淲hat鈥檚 amazing to me is that faced with the most surprising observation鈥攖he mosaic structure in the genes of higher organisms鈥攊t didn鈥檛 make anyone think.鈥�
Mattick has been puzzling over introns since they were discovered in 1977. After 25 years, he鈥檚 now convinced that much of this 鈥渏unk鈥� plays a crucial role in gene-to-gene communication. He first published the idea in rudimentary form back in 1994, and in September last year, he and Michael Gagen, a physicist at UQ, fleshed it out in a lengthy paper in Molecular Biology and Evolution. They suggested that introns and other non-coding RNAs in eukaryotes are 鈥渘etworked鈥� into a control system to turn genes on and off at the proper times鈥攁 network so powerful it enabled higher organisms to rapidly evolve and diversify. 鈥淚鈥檝e thought this through more than anybody else, I鈥檓 sure,鈥� says Mattick. 鈥淎nd I can see lots of reasons for this being the case. If non-protein-coding RNAs are functional, they must be networking.鈥�
Mattick鈥檚 logic is simple. Either introns and other non-coding RNAs are functional or they are not. If they are, the only way they can perform that function is by influencing other components in the gene-expression machinery. And unlike mRNA, which ferries information only between a gene and its protein, non-coding RNAs gossip with everyone, say Mattick and Gagen. They interact with one another, with mRNA, with protein and with DNA. In other words, they network.
A network is a set of interlinked nodes. Its structure and function depend on how the nodes interact. On the London Underground, for example, each tube station is a node, and the tracks link them. To the frustration of commuters, this is a static network, in which connections between nodes are fixed.
A molecular network in a cell is much more powerful than a static network. It is both complex and dynamic, with links between nodes forming and breaking all the time. Cognitive scientist Janet Wiles, another of Mattick鈥檚 UQ colleagues, explains that each gene in a genome is like a node in its genetic network. The interactions between genes are its links. As genes switch 鈥渙n鈥� or 鈥渙ff鈥�, links are made or broken in the network, determining gene expression.
The complete genetic operating system Mattick and Gagen propose regulates both gene expression and the proteins that carry out structural and functional roles in the cells. Introns provide the additional connections, with each RNA molecule doing different tasks at the same time鈥攁n example of parallel processing similar to what happens in your brain or a supercomputer.
This extra level of organisation may have been key in the evolution of complex, multicellular eukaryotes. Gagen likens it to the evolution of the modern computer. 鈥淭he first computers were hard-wired,鈥� he says. 鈥淲hen you wanted to change the computation, you had to redesign the network. This was difficult and took forever.鈥� So early computers were like simple organisms鈥攙ery cleverly designed, but programmed for one task at a time.
Enter the von Neumann computer, which can be electronically 鈥渞ewired鈥� by flicking a large number of switches. A closed switch is like connecting a wire between nodes, while opening a switch is like taking a wire away. Suddenly, instead of performing just one task, computers had the power to multitask. They could be reprogrammed in an instant and that enabled them to leap ahead in what Gagen calls 鈥渆xploration of program space鈥�.
Mattick and Gagen suggest that evolution, too, may have found a way to exploit multitasking, by rewiring its molecular networks on the fly. It did this by changing the control codes鈥攊ntrons and other non-coding RNAs鈥攖hus altering which genes are turned on at which times. 鈥淐hanging the sequence of codes changes the stored program. An infinity of different programs is now instantly available,鈥� says Gagen. And all without changing the underlying hardware: the basic suite of genes. Effectively, evolution got an upgrade that allowed it to race ahead.
The operating systems of modern desktop computers have millions of lines of code. Likewise, a genetic operating system capable of multitasking and parallel processing must have stacks of control molecules. And these couldn鈥檛 just appear out of the blue. Gagen says they are most likely to come from the part of the eukaryotic genome that has grown in size with increasing complexity鈥攊n other words, introns.
While the exact number of protein-coding genes in the human genome is hotly debated, current estimates hover between 30,000 and 40,000. This is only about twice as many as the worm and fruit fly, and five or six times as many as some bacteria. The difference in the number of proteins is somewhat higher, because cells can use 鈥渁lternative splicing鈥� to recombine the same bits of mRNA in different ways to produce more than one protein. But even that doesn鈥檛 seem enough to explain the huge difference in complexity between humans and worms or bacteria.
Silent majority
In contrast, higher organisms have vastly more non-protein-coding RNA sequences than simpler organisms. No one knows exactly how many more, because so few researchers have paid attention to non-coding RNAs, but Mattick estimates that introns and other non-coding RNAs make up 97 to 98 per cent of the entire transcriptional output of the human genome.
That estimate seems plausible. The human genome project reported last year that coding sequences comprise on average only about 5 per cent of each protein-coding gene. The other 95 per cent are introns. But there are also non-coding RNAs that aren鈥檛 derived from introns in genes. They could make up another 2 or 3 per cent of the genome鈥檚 transcriptional output, says Mattick.
Whether or not you agree with this figure, non-coding RNAs are clearly more common than we once thought鈥攁nd geneticists are finding new ones all the time. One well-documented example is the human X (inactive)-specific transcript, or XIST. The XIST gene doesn鈥檛 make a protein. Instead, its end product is a 17-kilobase RNA molecule that shuts down one of a female鈥檚 two X chromosomes before birth. Another, the H19 gene, produces a large non-coding RNA that promotes tumour development if it mutates. Others include His-1 RNAs, involved in viral defences in humans and mice, and spinocerebellar ataxia 8 RNA, or SCA8, a mutation found in patients with spinocerebellar ataxia type 8.
Then, last October, three research papers in Science announced a slew of tiny non-coding RNAs just 21 to 25 nucleotides long. They showed that these so-called microRNAs (miRNAs) perform unique functions during an organism鈥檚 development. Thomas Tuschl and his colleagues at the Max Planck Institute for Biophysical Chemistry in G枚ttingen, Germany, isolated 33 new miRNAs鈥�14 from fruit fly embryos and 19 from human cells. David Bartel and his team at MIT鈥檚 Whitehead Institute for Biomedical Research found 55 in the nematode Caenorhabditis elegans. And Victor Ambros and Rosalind Lee of the Dartmouth Medical School in Hanover, New Hampshire, reported 15 C. elegans miRNAs, 10 of which were also found by the Bartel group. All three groups proved that at least some of these miRNAs regulate gene activity, in part by turning other RNAs on or off, making Mattick a happy man.
There can be no doubt that many more non-coding RNAs are waiting to be found. 鈥淚 think Mattick鈥檚 case has strengthened through gradual accumulation of more and more cases of functional RNAs,鈥� says Laurence Hurst, an evolutionary geneticist at the University of Bath. 鈥淎s the issue is not whether some [non-coding] RNAs are functional but what proportion might be, the answer will only come by a very slow case-by-case analysis.鈥�
Mattick is already convinced. 鈥淚t鈥檚 all out on the table now,鈥� he says. 鈥淚 think there鈥檚 enough evidence to say this is not only probably, but almost certainly, right.鈥� He鈥檚 so sure that he鈥檚 winding down his other research to commit full-time to this work. It was not a decision Mattick took lightly鈥攁s a very senior researcher in bacterial pathogenesis, he has hundreds of refereed journal articles and conference presentations under his belt.
Mattick鈥檚 ideas have certainly piqued the interest of his peers. His third paper on this concept, published in EMBO Reports in November 2001, has been rated highly on BioMed Central鈥檚 鈥淔aculty of 1000鈥� website (), where biology research papers are ranked according to the recommendations of over a thousand experts. Until recently, Mattick鈥檚 paper was number two in the 鈥淢ost Viewed鈥� top 10 for all of biology, and number one in the 鈥淗idden Jewels鈥� top 10, a category for interesting articles from less widely read publications. He has also been invited to submit an article on non-coding RNAs for Nature鈥檚 Encyclopaedia of the Human Genome.
The interest in Mattick鈥檚 ideas comes as no surprise to Frank Gannon, executive director of the European Molecular Biology Organization (EMBO) in Heidelberg, Germany. 鈥淚 think he has highlighted an aspect of molecular biology which has been overlooked and which requires serious consideration,鈥� says Gannon. 鈥淚 am sure it will stimulate thought and action from laboratories around the world for many years to come.鈥�
It has certainly stimulated Jean-Michel Claverie of CNRS, France鈥檚 national research agency in Marseilles, to speak out. 鈥淚 don鈥檛 think much of this work,鈥� says Claverie, who is not a man to mince his words. 鈥淚n general, all these very global ideas don鈥檛 travel very far because they fail to take into account the most basic principle of biology: things arose by the additive evolution of tiny subsystems, not by global design. It is perfectly possible that one intron in one given gene might have evolved鈥攂y chance鈥攕ome regulatory property. It is utterly improbable that all genes might have acquired introns for the future property of regulating expression.鈥�
Stephen Holbrook of the Lawrence Berkeley National Laboratory in California finds Mattick鈥檚 concept more appealing. 鈥淭he idea of RNA forming a global communication network or genetic operating system may be overstated, but current evidence suggests that it is a component of such a network, together with proteins and small molecules such as cyclic AMP,鈥� he says. 鈥淭he contribution of each partner has yet to be determined and we are in the very early stages of modelling such complex biological systems.鈥�
Holbrook also sees an important practical application of Mattick鈥檚 ideas. If non-coding RNAs really play such a crucial regulatory role in cells, then bugs in the programming could show up as diseases or genetic abnormalities. In that case, he says, 鈥渨e can expect to find RNAs related to human diseases and as potential drug targets.鈥�
That prospect has enticed several biotechnology companies to begin sniffing around the subject. In the 1990s, Genetic Technologies Limited in Melbourne, Australia, was granted several world patents on intron sequence analysis in humans, animals and plants, and it aims to extend that to 鈥淸almost] all genes in all species鈥�. The company is now licensing these techniques to other biotech companies (快猫短视频, 18 May, p 5). Another biotech company, the Ibis Therapeutics division of Isis Pharmaceuticals in California, is developing drugs that bind to RNAs. Company president David Ecker says Ibis is interested in targeting miRNAs, but it鈥檚 waiting for the picture to become clearer first.
As for Mattick鈥檚 mission, the next step could be the decider. Using powerful computers, his colleagues at UQ have just started simulating a genetic operating system to work out how the introns might fit in. The modelling team will begin with a basic, intron-free network, and gradually add introns, checking at each step that the biology still makes sense. What they need is an operating system powerful enough鈥攚ith enough nodes linked in an efficient way鈥攖o generate complex life from its simple beginnings. 鈥淚t鈥檚 going to take a lot of data to get to the next point of saying this is the critical thing that underpins complexity,鈥� says Mattick. 鈥淏ut within 3 to 5 years, this idea will either be accepted or in the dustbin.鈥�