IMAGINE being given a 3000-page cookery book, subtitled 鈥渁 complete guide to the culinary arts鈥. As you leaf through it you are disappointed to find only 100 recipes in the entire volume. The rest of the pages seem to be filled with gibberish. You can spot a sentence here, a coherent paragraph there, but most of it is random words strung together with no apparent meaning. What sort of book is this?
Welcome to one of the enduring puzzles of genetics. Even though the DNA in a cell鈥檚 nucleus is often described as a recipe book for making the proteins that build and operate the cell, scientists have known for more than two decades that only a small fraction of the DNA 鈥 perhaps only three per cent in humans 鈥 actually encodes this sort of information. So what is the rest of this stuff cluttering up the book of life, the 97 per cent of the genome that doesn鈥檛 make proteins?
Some of it contains the control switches that direct cells to produce, for example, one set of proteins in the brain and another set in the liver. These gene regulators, which turn on the right genes at the right times in the right cells, are mostly scattered throughout the region immediately preceding each gene. Another 10 per cent or so of the genome consists of telomeres, which cap the ends of chromosomes to prevent erosion, and centromeres, which are attachment sites for the cellular cables that separate chromosomes during cell division. These two structures are composed of simple patterns of five or six bases 鈥 the letters of the genetic alphabet 鈥 repeated thousands of times. They play such vital roles in the cell that evolution has apparently provided plenty of spares just in case, says Robert Moyzis, director of the Center for Human Genome Studies at Los Alamos National Laboratory in New Mexico.
Advertisement
But the debate centres on the still vast, unexplained stretches that serve no known purpose. Many biologists would dismiss around 85 per cent of the genome as simply 鈥渏unk DNA鈥. How could it play a vital role, they argue, when organisms such as bacteria and puffer fish thrive with very little redundant genetic material? However other researchers believe that if only they could understand the language of junk DNA its function would be revealed. 鈥淚f I give you the Bible written in Mongolian, you wouldn鈥檛 say 鈥榯his is junk鈥 simply because you don鈥檛 know how to read it,鈥 says Edward Trifonov, a structural biologist at the Weizmann Institute of Science in Rehovot, Israel. Others who share this view point to recent research that appears to show language-like patterns in junk DNA which might suggest that it carries an important message. Controversy over these findings has only intensified the debate.
Not everyone agrees on what constitutes junk. Many common non-protein-coding sequences have no obvious function. In human chromosomes, for example, a 282-base sequence known as 鈥淎lu鈥 occurs in about a million separate locations, comprising about 10 per cent of the entire genome 鈥 more than twice as much as all the protein-coding genes put together. A second sequence, 鈥淟INE-1鈥, makes up another 5 per cent, and similar sequences occur in most other vertebrates. These recurrent sequences are transposable elements, or 鈥渏umping genes鈥, that have colonised the chromosomes by making copies of themselves and inserting these copies into new locations within the genome.
Selfish or selfless?
Populating the genome like fleas on a dog, Alus and their ilk provide no obvious benefit to the rest of the organism. Most geneticists therefore regard them as molecular parasites, mere bits of 鈥渟elfish鈥 DNA that serve no purpose other than their own continued survival. Researchers sequencing the human genome quickly grow tired of Alus and LINEs. 鈥淭hey certainly frustrate people in the genome project,鈥 says Moyzis. 鈥淢ost of the sequencers consider them nuisance regions to get through to get on to the interesting bits.鈥
A few researchers, however, are not so sure. 鈥淎lus were very handily tossed off as junk by several prominent scientists,鈥 says Ross Hardison, a biochemist at Pennsylvania State University. 鈥淚 think that鈥檚 made it difficult for people to appreciate the examples that are functional.鈥 Carl Schmid, an Alu expert at the University of California at Davis, has found that at least a few Alus behave more altruistically, producing RNA in response to stressful conditions, as part of the so-called 鈥渉eat shock response鈥. Other researchers also have evidence that some Alus might enhance the activity of nearby genes.
In cases like these, the sequences probably began as junk and were later pressed into service when they happened to fulfil some need, says Hardison. Such after-the-fact tinkering is a common feature of evolution. It may crop up in some other unlikely parts of the genome, in stretches known as trinucleotide repeats, which are sequences of the same three nucleotides, monotonously repeated over and over again. Geneticists know that misalignments during DNA replication 鈥 the equivalent of doing up the buttons of a shirt incorrectly 鈥 can easily add or delete repeats from these sequences. Indeed, they believe these repeats probably originated as a string of such errors and thus ought to be nothing but genetic junk. But, to their surprise, they have found that this sort of junk can have a function, although an undesirable one. Trinucleotide repeats that are too long can cause several human diseases, including Huntington鈥檚 disease and fragile X syndrome (see 鈥淲hen DNA turns traitor鈥, New 杏吧原创, 25 March 1995).
Elsewhere in the genome, however, researchers have known for several years that there are numerous examples of what Moyzis calls 鈥渢he rusting hulks of accidents that occurred in evolution鈥. Some of these broken genes, or pseudogenes, form in much the same way as Alus, when an accidental copy of the gene inserts into the genome randomly. Stripped of its all important context 鈥 the local DNA that would normally control its expression 鈥 such a gene is a piece of useless junk as surely as a radio without control knobs.
Other pseudogenes form when errors in cell division create duplicate copies of whole genes adjacent to one another on the chromosome in a molecular stutter. The duplicate gene, no longer crucial for the normal functioning of the cell, can accumulate mutations freely. Such duplicate genes thus become grist for the evolutionary mill and can evolve new functions or merely degenerate into pseudogenes.
Genes in the human beta-globin cluster, which codes for one of the protein components of haemoglobin, have suffered both these fates. Around 200 million years ago a single gene coded for this protein, but successive duplications have given rise to six related genes in the cluster. Five of the six genes still function, but they have specialised: one produces a globin protein only during early embryonic development, two code for proteins during fetal life, and two make globins used after birth. The sixth gene has fallen into disrepair and no longer makes a protein. The coding portions of the six beta-globin genes occupy barely 3 per cent of the 100 000 bases in the cluster. Noncoding regions within each gene, known as introns, take up another 5 per cent, and sequences that control the genes a few per cent more. Most of the globin cluster, therefore, appears to be nothing more than spacers filling the gaps between genes, says Morris Goodman, a molecular evolutionist at Wayne State University in Detroit.
Finding function
But even in these apparently redundant regions, Goodman sees subtle signs of what could be called function. For example, in the genomes of humans, apes and Old World Monkeys a 6000-base 鈥渏umping gene鈥 has been inserted into the spacer between the embryonic and fetal globin genes. Goodman thinks the expansion of this spacer may explain why a formerly inactive fetal gene now produces the lion鈥檚 share of fetal globins in these organisms. Though the idea is still unproven, he suggests that this change in spacing may alter the way the DNA molecule folds up, bringing the other fetal gene under the influence of DNA sequences that regulate its expression.
Indeed, researchers are only just beginning to explore how loops, twists and folds in the long DNA chain 鈥 its so-called 鈥渉igher-order structure鈥 鈥 affect the expression of genes contained within them. And this work could reveal a purpose for noncoding DNA. Cells package their DNA by spooling it on protein drums known as nucleosomes and folding strings of nucleosomes into tight pleats, then pleats of pleats, and so on. Such close bundling keeps the genes dormant. To activate them, the cell must unwind the relevant portion of the chromosome so that enzymes can get at the genes to begin transcribing them. 鈥淲hat controls that is anybody鈥檚 guess right now,鈥 says Moyzis. 鈥淭he working idea is that other regions of the DNA that are not part of the traditional coding regions are responsible for that signal. That will probably be a major area of exploration in the next five or ten years.鈥
One of the few visionaries who suggest that noncoding DNA may be teeming with information of this sort is Emile Zuckerkandl, a molecular evolutionist who directs the Institute of Molecular Medical Sciences in Palo Alto, California. According to his theory, cells need a lot of noncoding DNA in and around the coding regions, in the form of introns and intergenic spacers, to enfold the coding portions and wrap them into properly labelled bundles that open at the right moment. But Zuckerkandl admits that his theory will be impossible to prove until geneticists know more about the higher-order structure of DNA. In the meantime, they must draw on other evidence to determine whether introns and spacers have any real function.
It was just such evidence that caused a stir among genome-watchers last winter when Rosario Mantegna, Eugene Stanley and their colleagues from Boston University reported finding language-like patterns in noncoding DNA. The team had divided DNA sequences into short 鈥渨ords鈥 of arbitrary length (three to seven bases) and counted the number of times each word appeared in the complete sequence. They found a pattern of word frequencies that closely matched that seen in human languages, a pattern known as Zipf鈥檚 Law. Moreover, they reported that noncoding DNA showed a stronger Zipf鈥檚 Law pattern than coding DNA. Stanley admits that finding an abstract statistical pattern and understanding its biological significance are two completely different things. 鈥淭his work doesn鈥檛 prove anything,鈥 he says. 鈥淏ut I鈥檇 like to think it opens doors for richer investigation.鈥
The study has, however, proven controversial. Critics contend that word-frequency patterns are virtually meaningless. 鈥淭he fact that you see a Zipf pattern isn鈥檛 interesting,鈥 says Richard Voss, a physicist at Yale University, who studies statistical patterns. 鈥淚t has nothing to do with linguistic features. It occurs if you look at the number of islands on the Earth鈥檚 surface and rank them according to size.鈥 Voss also points out that Mantegna and Stanley鈥檚 data set, which was chosen to include all known sequences longer than 50 000 bases, contained only 40 sequences. When the same analysis is applied to other sets of sequences, the difference between the word-frequency patterns of coding and noncoding DNA is less pronounced or absent, says Voss.
Whatever the outcome of this debate, some researchers remain convinced that the noncoding DNA found within genes, at least, must be doing something valuable. Trifonov argues that cells have put up with these introns for hundreds of millions of years, patiently cutting them out every time a gene is transcribed and splicing the coding parts together. 鈥淲hy have a whole headache of such precise splicing and throwing away to begin with?鈥 he asks. 鈥淚t would be stupid to introduce such an awkward system for nothing. That means [introns] are doing something very important.鈥
This argument gains force from the fact that cells have a mechanism available to them to eliminate introns, but don鈥檛 use it. When a gene makes a protein it must first produce a transcript of itself and cut out the noncoding introns. By making a DNA copy of this spliced transcript and then reinserting it into the genome introns can be eliminated for good 鈥 exactly the mechanism by which some pseudogenes arise. 鈥淚f introns are really useless, I don鈥檛 understand why we haven鈥檛 gotten rid of them,鈥 says Moyzis. 鈥淚 have a hard time believing they鈥檙e not there for some reason.鈥
Slippery term
But 鈥渇unction鈥 is a slippery term. Everyone knows that human bureaucracies tend to accumulate layer upon layer of workers who perform tasks of no real importance, says Ford Doolittle of Dalhousie University in Halifax, Nova Scotia. Even so, these paper-pushers come to depend on one another, so that the absence of one rubber-stamping drone can bring the system to a grinding halt. In the same way, he concludes, 鈥渋t might also be the case that if you had an intron in gene X for a hundred million years and then removed that intron, the gene may no longer function. The cell is used to having it there, so you can鈥檛 get rid of it. But it would be naive to conclude that it performs a necessary function.鈥
Besides, critics ask, if introns and intergenic spacers really are so critical to gene expression, shouldn鈥檛 evolution tend them like a meticulous gardener, weeding out stray mutations? But the opposite is true, says Phillip Sharp, a biologist at the Massachusetts Institute of Technology, who shared a Nobel prize in 1993 for the discovery of introns. Except for a few regions here and there, introns and intergenic spacers freely accumulate mutations 鈥 even large deletions or insertions. For example, while the coding parts of globin genes in mice and humans are very similar, the noncoding regions bear almost no resemblance, Sharp notes. This suggests that they have become riddled with mutations.
The strongest evidence that much of the genome is useless, however, comes from species that simply do without. For example, the puffer fish, Fugu rubripes 鈥 the same fish whose tasty flesh tempts thousands of Japanese gourmands to court death from one of the world鈥檚 most potent toxins 鈥 has a genome only one eighth the size of most vertebrates. The puffer fish carries no fewer genes than other fish, says Sidney Brenner, a geneticist at the University of Cambridge who is studying its genome. Instead, it achieves its genomic brevity by having radically shorter introns and intergenic spacers. Closer examination shows that the genome contains very few pseudogenes. Repetitive sequences such as transposable elements and satellite sequences are almost absent as well 鈥 perhaps because they never chanced to accumulate, says Brenner. 鈥淚 like to call it the discount genome. It gives you a 90 per cent discount on sequencing. You only have to do a tenth the work to get the information.鈥 Yet this compact genome, says Brenner, contains all the essentials for understanding genome function. 鈥淎ll the rest is just rubbish that鈥檚 got into our genomes.鈥
Over at the University of Oxford, evolutionary theorist Mark Pagel goes one step further. He believes he has evidence that evolution sometimes puts a lid on all this genomic rubbish. Up to a point, accumulations of genetic junk may make little difference to the cell, Pagel and his colleague Rufus Johnstone have hypothesised. But if too much junk builds up, it becomes too expensive and time-consuming for the cell to replicate all of it.
Pagel and Johnstone tested their idea by comparing the genome sizes of 24 species of salamanders, a group of creatures notorious both for their gigantic genomes 鈥 some of them are more than 40 times larger than the human genome 鈥 and for the large variation in genome size between species. The researchers predicted that species with rapid cell division and earlier hatching, in other words those with the highest rate of DNA transcription and replication, would tolerate less junk DNA than species with more leisurely development. The data bear out their prediction. Junk DNA accumulates in the genome until the cost of replicating it becomes too great, concludes Pagel.
Even so, most vertebrates have genomes much smaller than those of salamanders and much larger than those of puffer fish. Some molecular biologists maintain that as long as DNA鈥檚 higher-order folding remains largely a mystery, it鈥檚 too early to call off the search for the hidden language of 鈥渏unk DNA鈥. 鈥淪urely we cannot answer all the questions about the Mongolian Bible,鈥 says Trifonov. 鈥淲e don鈥檛 know the Mongolian language yet. But calling it junk DNA is a capitulant鈥檚 point of view. You don鈥檛 study it any more.鈥