A year ago, a team of American researchers, led by Edward Golenberg of the University of California at Riverside, extracted a fragment of DNA from chloroplasts of a fossil Magnolia leaf. They excavated the fossil from the clay beds of a Miocene lake (17 to 20 million years old) in Clarkia, Idaho, making the DNA easily the oldest in the world. Palaeontologists were impressed; genetic information from fossil DNA could clear up many unresolved questions of taxonomy and rates of evolution. But according to biochemical theory, DNA cannot survive for so long.
Golenberg’s discovery fired the imagination of chemists, biochemists and geologists, as well as palaeontologists. The search for ancient organic material such as amino acids and DNA is at best daunting and some might say futile. Researchers still do not know exactly what happens to these molecules when they become fossilised. They are just beginning to address the question of how such fragile molecules could have survived for millions of years, perhaps unchanged, when proteins, for example, can be denatured by heating.
They have some clues, however. To have the best chance of survival the molecules would have to have been fixed, probably compressed, almost instantly, in a closed system, without any biological input in the shape of bacteria, oxygen and water. There are unusual environments where this might conceivably happen: peat bogs, rocks formed from muds, permafrost and hypersaline pools have yielded extraordinarily well preserved human tissues. The idea of extracting DNA from such material intrigued researchers throughout the 1980s. What made it possible were revolutionary developments in another field entirely – molecular biology.
Advertisement
Allan Wilson of the University of California at Berkeley pioneered this novel application of molecular biology. In 1984, Wilson, Russell Higuchi and their colleagues became the first to recover useful DNA sequence information from the remains of an extinct species. Working on a museum specimen of the quagga, a zebra-like species that became extinct in 1883, they isolated DNA from the animal’s dried muscle tissue. They then attempted to produce enough copies to work with by inserting bits of the DNA into bacteria and extracting the DNA when the bacteria had copied it – a process called cloning.
Among the clones from the quagga DNA were two containing pieces of mitochondrial DNA. By comparing their sequences, made up of 229 base pairs each, with those of corresponding known sequences in the mountain zebra, the cow and humans, Wilson and his colleagues found that the zebra and quagga mitochondrial DNA differed by only 12 base pairs. They decided that a few of these 12 differences could be due to changes after death, including cloning artefacts, but they could still construct an evolutionary tree that showed that the zebra and the quagga had a common ancestor around 3 to 4 million years ago. This confirmed evidence from the fossil record. It seemed to support the notion that mitochondrial DNA changes at a constant rate over evolutionary time, and so could be used as a molecular evolutionary clock. This idea dates from the early 1960s, but is still attracting controversy (see ‘Molecular clocks run out of time’, ¿ìè¶ÌÊÓƵ, 10 February 1990).
Mitochondrial DNA comes from mitochondria, the ‘power houses’ of living cells. As there are as many as 2000 mitochondria in a cell it is much more common than the longer nuclear DNA, which carries more genetic information but is also less likely to have survived intact. If the survival of such ancient DNA was a general phenomenon, Wilson pointed out, researchers from fields as disparate as palaeontology, archaeology, forensic science and evolutionary biology all stood to benefit.
Svante Paabo, archaeologist turned molecular biologist, was working with Wilson, so he saw at first hand the potential of applying the techniques of molecular biology to samples of ancient tissue. In 1985 Paabo tested 23 different samples of ancient tissue for surviving genetic material. He succeeded in isolating DNA fragments from only one, an Egyptian mummy, dated by a conventional radiocarbon method to be 2400 years old. Most of the DNA was less than 500 base pairs long, but some had 5000 base pairs or more. Paabo used cloning to copy randomly the sequences he found, and produced one segment of DNA that was 3400 base pairs long. Researchers anticipated new opportunities for tracing relationships between ancient human populations. But there were problems. The surviving DNA was highly damaged and in small pieces, which made cloning difficult. And the cloning process itself could introduce errors into the sequences.
Enzymes copy the past
In the same year, researchers at the Cetus Corporation invented the polymerase chain reaction (PCR), a technique for accurately amplifying minute amounts of DNA using enzymes. In each cycle of this process the DNA double strand is first unwound by heating, and short sequences-primers-are added to form a double section, one at each end, of each single strand of the target sequence. Then the enzyme DNA polymerase is added; this catalyses the making of the complementary sequence, forming two complete double strands both identical to the original. Each cycle doubles the amount of DNA originally in the sample. The next five years saw some remarkable developments in molecular biology, and archaeologists were quick to take note. By 1989 Paabo had extracted mitochondrial DNA from 12 specimens of dry soft tissues of various ages from four species. Among these were four-year-old dried pork muscle, tissues from several Egyptian mummies taken from museums and collections and skin from two extinct animals, a marsupial wolf (Thylacinus cynocephalus) from Zurich and a giant ground sloth (Mylodon) found in a cave in Chile. This sloth DNA was the oldest ever reported, at the time, at 13,000 years.
Although overall he extracted between 1 and 200 micrograms of DNA per gram of dry tissue, the samples of ancient human tissue yielded less than 200 nanograms of DNA per gram. Why? Paabo expected the ancient DNA to be damaged in some way. The genetic information of the DNA double helix takes the form of various combinations of four bases: cytosine, thymine, adenine and guanine, arranged along a sugar-phosphate backbone. The first two bases belong to the group known as pyrimidines and the other two are purines (see Figure 1). They are paired by hydrogen bonds between the two strands, adenine to thymine and guanine to cytosine. When he looked more closely and compared the four-year-old pork DNA with that of the ancient human DNA by chromatography, Paabo found a striking difference. The ancient DNA showed hardly any cytosine and thymine – less than 5 per cent of what there should have been. Pyrimidines, especially thymine, are known to be susceptible to oxidative damage. Apart from missing and altered bases, such damage could take the form of modification to sugar residues and cross links between molecules. The DNA was also short, around 100 base pairs long, all of which made cloning difficult.
Paabo and Wilson now think that cloning, which involves the intermediate step of replication in bacteria, also leaves open the possibility that the bacteria’s own genetic repair mechanisms might alter the DNA sequence as they try to repair the damage, introducing so-called cloning artefacts. In PCR, by contrast, damage is more likely to stop or slow down the enzyme, so that only DNA is amplified.
Next, Paabo took a microgram of the same sample he used for cloning and tried to amplify three segments of mitochondrial DNA by PCR. After 40 cycles, he had only produced the two smallest segments (84 and 121 base pairs), which suggested that the DNA was shorter than the third segment (471 base pairs). Further work showed that the longest segment he could amplify was 140 base pairs. But the size of the DNA did not seem to correlate with the age of the samples: four-year-old pork DNA was just as degraded as that from a 13,000-year-old ground sloth. By analogy with the way nuclear DNA is rapidly degraded after death – by the action of hydroxyl (OH) radicals, reactive species derived from water-Paabo concluded that the size of the DNA reflects how fast the tissue had been dried. What determined the extent of oxidative damage to the DNA, which resembled that caused by gamma rays, remained unclear. Did it happen quickly, then level off? Was the environment in which individual specimens were preserved especially important?
Golenberg’s discovery a year later took everyone’s breath away. The idea of an apparently perfect leaf of the extinct species Magnolia latahensis, estimated to be at least 16 million years old, was itself remarkable. To have extracted and, in 30 cycles of PCR, amplified a 790-base-pair fragment of DNA from such a specimen was beyond anyone’s wildest dreams. The DNA was from chloroplasts, structures within plant cells that resemble mitochondria in their role as powerhouses. Golenberg amplified a segment of the chloroplast gene that codes for part of a carboxylase enzyme. To verify that the amplified DNA came from the fossil, not from contamination, he compared this sequence with other known sequences for this gene, and with that of present day Magnolia species thought to be related to the fossil species.
The Clarkia site where Golenberg’s colleague, Jack Smiley of the University of Idaho, excavated the leaf, is a haven for geologists. They are attracted by its fossil beds; layers of clay and ash, about 9 metres thick, that once formed the bottom of a lake. The lake formed rapidly between 17 and 20 million years ago, possibly when an ancient stream valley was dammed by lava. What makes these sediments so interesting is that they and the fossils they contain – so far researchers have excavated mainly leaves and a few fish – seem to have formed in an anoxic environment. Golenberg thinks the leaves fell or were blown off the trees directly into the lake, where they immediately sank to the bottom and lay there undisturbed, away from any oxygen. During the 1980s Karl Niklas and others working at the Clarkia site had found cell structures, such as bits of plant cell walls and chloroplasts, and intact pigment molecules – ring compounds called flavonoids. Golenberg thinks the environment of the leaves would have largely prevented the type of oxidative damage to pyrimidine bases that Paabo and others had seen in ancient DNA.
But this still leaves some questions unanswered. Earlier this year, having attempted their own PCR analysis of fossil leaves from Clarkia, Paabo and Wilson warned that ‘there are reasons to be only cautiously enthusiastic at this point’. The biggest problem, as they see it, is water. Unlike Paabo’s specimens of ancient dried tissue, the Clarkia leaves were found in water-soaked sediments. The effect of water on DNA is well studied. In aqueous solution, DNA suffers spontaneous loss of purine residues – the bases adenine and guanine. Once some bases are missing, the DNA strand is more likely to break.
As long ago as 1972, Tomas Lindahl, a biochemist, calculated how fast this depurination process takes place. Paabo and Wilson repeated the calculation for molecules of DNA each containing 800 base pairs. They estimated that a gram of leaf tissue would contain about 1012 such segments from its chloroplasts. Starting from this number, and assuming a temperature of 15 °C (not unreasonable, as there is geological evidence that the Miocene climate was rather warm and humid) and a pH of 7, the last 800 base pair fragment would have lost its purines after only 5000 years. A more acid or alkaline solution would have speeded up the process, while at only 5 °C the process would have been two to four times slower. If this theory is correct, any Miocene DNA would have disappeared long ago.
‘I don’t say they haven’t got a Magnolia sequence. The sequence they published is definitely Magnolia,’ Paabo says. But could this DNA really be 16 million years old? So far, Paabo has been unable to repeat Golenberg’s results. When he repeated the extraction of DNA from other leaves excavated from Clarkia last autumn, also working with Michael Clegg and Jack Smiley, he found, like Golenberg, that about 1 extract in 10 contained high molecular weight DNA. But this was not Magnolia DNA: ‘The high molecular weight DNA – it looks like bacterial DNA,’ Paabo says. This raises the tricky question of how bacteria could be present without affecting the DNA, and indeed how they got there in the first place. Were they entombed in the sediments along with the leaves? If not, does this mean the fossil leaves were not, after all, compressed within a closed system?
Golenberg dismisses these criticisms: ‘I have reproduced these results. The question of bacterial contamination is irrelevant. To say this (Miocene Magnolia sequence) is not what we have, you would have to stand on your head. It would be like the boy who is taken to a zoo and shown a giraffe, which he has never seen before, and saying it can’t possibly be a giraffe, such a thing does not exist, because he has never seen one.’ Golenberg cites several other researchers who, he says, will soon publish findings that back up his original results.
The key to the debate seems to lie in what happened to the DNA in the leaf between its initial deposition and the point at which it became compressed in the sediments. Did it degrade? One possibility is that the effect of oxidation on plant DNA is different from that on animal DNA – there are huge differences between animal and plant cells. The leaf cells would have been alive when they were deposited; did repair mechanisms continue to function for a while, even after the leaf was blown off the tree?
New links for old
More is known about repair to animal and bacterial DNA than to plant DNA. DNA in living cells is continually being damaged and repaired. It is susceptible to attack by toxic substances and ionising radiation that cause various types of damage including missing or altered bases, breaks on one or both strands, or cross-links between DNA and proteins. Single strand breaks are repaired within minutes, double strand breaks are a little more difficult. Enzymes play a key role: missing bases, for example, which can result from the reaction between water and the bond linking them to the sugar backbone, which in a single human cell happens to as many as 10,000 purine bases every day, are restored with the help of the enzymes DNA polymerase and ligase. Even simple bacteria repair their DNA. No one knows whether these repair mechanisms help to preserve fossil DNA.
Geochemical clues
Meanwhile, other researchers are trying to find some answers by studying diagenesis-the physical, chemical and biological changes sediments undergo before they become rocks. Geoffrey Eglinton, James Maxwell and others at the University of Bristol are one of several groups studying the chemistry of organic molecules over periods of geological time. They would like to know not only how ancient biomolecules become damaged in the early stages of diagenesis, but also how to obtain more information from them. They are trying to find out how mobile the molecules are-whether they have stayed put, or moved away from their original place-and how the chemistry of sediments interacts them.
This hints at several ways in which fossil DNA could have survived. Formation of the damaging hydroxyl radical from water, for example, is catalysed by copper ions. So reactant inhibition is one possibility: perhaps humic acids in the sediments chelate metals, or hydrogen sulphide from bacteria precipitates them as sulphides. Or perhaps radicals are trapped by molecular mops such as carotenoids, which chemists know survive in ancient sediments. Or has compression physically prevented reactions such as the breakup of carbon-nitrogen bonds? Paabo has suggested that pressure could have squeezed out the water from the chloroplasts of Golenberg’s Clarkia leaf, leaving the DNA surrounded by lipid molecules. Acidity and temperature will almost certainly have contributed. Or perhaps the chemical composition of the clay that surrounded the Clarkia leaf, and the way this restricted access to the DNA by water molecules, may hold the key. If so, a new generation of molecular palaeontologists may be looking with new eyes at fossil-rich sites. Molecular biology has enabled researchers to ask some intriguing questions. Now they are looking to a combination of biochemistry, geology and chemistry to provide at least some of the answers.
* * *
1: DNA, PCR and the body in the carpet
PCR amplification of DNA received official approval as a forensic tool in February this year. It was accepted as evidence for the first time in a British court in a murder case in which the better established DNA fingerprinting and profiling techniques proved useless. The story began in 1989 when builders discovered a skeleton-the remains of a body wrapped in a carpet-at a house in Cardiff. Facial reconstruction in clay by a medical artist brought the skull to life, and witnesses identified it as that of Karen Price, a teenager who had disappeared eight years previously from a children’s home. Dental records were inconclusive, and in the search for further evidence to back up the provisional identification, the police sent a sample of femur to Erika Hagelberg at the John Radcliffe Hospital in Oxford. In 1989 Hagelberg, working with Bryan Sykes and with Robert Hedges from the University of Oxford’s archaeology unit, had used PCR to amplify mitochondrial DNA from human bones up to 750 years old, including samples taken from the English Civil War Cemetery in Abingdon, by PCR. At first the possibility of contamination-from people, traces of DNA in the research laboratory, in reagents and equipment and from previous PCR reactions-made some archaeologists sceptical about this new method of dating bones.
From a gram of the unidentified femur, Hagelberg was able to extract 1.5 micrograms of DNA. She sent it to Alec Jeffreys, inventor of DNA fingerprinting, at the University of Leicester. He identified the DNA as about 1 per cent human nuclear DNA: the rest was contamination by DNA of microbiological origin, either bacterial or fungal. Even after burial for such a relatively short time the human DNA had become degraded, the average length of the segments being 150 base pairs. This meant that DNA fingerprinting and profiling, which require segments up to thousands of base pairs long, could not be used. Hagelberg and Jeffreys decided to try PCR to amplify six variable marker regions of DNA, each one a two-base pair repeating sequence, about 100 base pairs long in total. The advantage of such a short marker is that it should be intact, even within the damaged DNA researchers had to work on. The disadvantage is that it is less variable, and therefore less specific to an individual.
Next, they compared the six sequences with DNA taken from the blood of Karen Price’s parents. In each case they obtained a reproducible match with two gene variants, one from each parent. A biostatistical test established that the likelihood of a chance match were 200,000 to 1 in favour of identification, which Cardiff crown court was prepared to accept as proof beyond reasonable doubt that the body was that of Karen Price.
What it does not prove is that PCR can be applied to identify any old bones. The amplified DNA must be compared with a known reference; in the absence of a national DNA database, this effectively means a DNA sample from another member of the family. But if there is a match, it is highly unlikely to be anything but genuine. Jeffreys has already extracted DNA from the bones claimed to belong to Joseph Mengele, and is trying to obtain blood samples from the man’s family. Such comparative DNA analysis opens up the possibility of identifying the remains of bodies buried in mass graves. A Home Office scientist is now mastering Hagelberg’s extraction technique.
Last month Hagelberg published further evidence for the accuracy of her method-perfect matching of an amplified gene sequence that codes for cytochrome b in pig, from a sample taken from a leg bone of salted pig meat carried on the Mary Rose. Contamination from human DNA was impossible because the particular sequence is in a region of DNA 23 per cent different from the human one.
No one knows how the age of bone is related to the preservation of its DNA. Hagelberg has worked extensively with medieval bones, and routinely obtains 800 base pair fragments. In contrast, she could only amplify DNA from 1 out of 4 samples of femural bone taken from much younger skeletons from a grave dating from the English Civil War.
Lynne Bell of University College London, who studies sections of bone with scanning electron microscopy, says that appearances can be deceptive. Even very poor bones with the outer and most of the inner sections decayed can still have a ring of well-preserved tissue between that yields amplifiable DNA. So archaeologists attempting to survey sites systematically by this method would have to take several samples from each.
* * *
2: Rats have a legacy of fossil proteins
DNA is not the only type of molecular fossil. Proteins, or fragments of proteins, can also sometimes survive for thousands of years and although they give less information than DNA they can be useful because they are often species-specific. These differences enable researchers to establish the position of branches in evolutionary trees, for example. The idea that similarities among large biological molecules in different species indicates a genetic connection is now well established. Researchers have been sequencing modern proteins for about 30 years to look for these similarities. Could they do the same for ancient proteins? At first the idea seems highly unlikely. Proteins can deteriorate by denaturing, hydrolysis of peptide bonds, and by modification of amino acid side chains by, for example, loss of amide groups, cross linking and oxidation. They also turn into their mirror image at what, in geological terms, is a fast rate. But bones and teeth, common relics of the past, contain collagen, already tightly wound into a triple helix, fixed in a hydroxyapatite matrix. Shells form a calcite matrix where proteins can escape attack. Even the haem group at the heart of a haemoglobin molecule is fairly inaccessible.
In 1983 Thomas Loy of the British Columbia Provincial Museum caused quite a stir among paleontologists when he succeeded in washing bloodstains from prehistoric weapons, crystallising the blood protein haemoglobin from them, and identifying what species the blood came from. The way haemoglobins from different species precipitated depended partly on variations in sequences of amino acids between species. These differences give each haemoglobin a slightly different shape and surface electrostatic charge.
During the 1980s Jerold Lowenstein of the University of California and others used the biological method of immunoassay to detect this and other proteins. Lowenstein found the common protein albumin in a number of preserved tissues from extinct animals, including a frozen mammoth, a Tasmanian wolf and the more recently extinct (17th century) Steller’s sea cow, as well as in blood from ancient weapons. More recently, he applied immunoassay to a more unusual source of evolutionary information: fossil urine.
The urine of North American pack rats, their African and South West Asian relatives the hyrax, and North American porcupines crystallises – mainly as calcium oxalate and calcite – into a solid mass called a midden. The rats polish it to a smooth finish with their bodies until it resembles amber. Ancient middens from pack rats and other mammals are already well known as a source of fossil plants. Lowenstein decided that since liquid urine contains traces of the water-soluble protein serum albumin, perhaps some of it could have survived within the urine crystals. From the albumin it would be possible to find out what species of pack rat created the midden, and, together with evidence from the plant fossils, whether the rats’ eating habits changed over time.
Immunoassay exploits the specificity of antibodies-their ability to detect small regions, between six and ten individual amino acids, of a protein. Such small segments may survive as fossils but they are too small to sequence: immunoassay is sensitive enough to detect them. In a three-step process any molecules of albumin – the antigen – in a solution of the urine are immobilised on the inside of a small plastic cup. Then antisera antibodies, made in rabbits to albumins of deer mouse, a species related to the pack rat, are added. The antisera will bind to albumin molecules from the rat urine if these molecules still retain their immunological function. Excess antisera are washed out, then goat (anti-rabbit) antibodies, labelled with radioactive iodine, are added. These bind to the rabbit antibodies. The amount of radioactivity in the cup depends on the amount of albumin and the specificity of the rabbit antibodies.
The urine, from five different middens in the same cave, ranged between 19,700 and 2440 years old. From these samples, Lowenstein found that the albumin had lost most of its immunological activity in the first thousand years, but after that loss was more gradual, so they retained enough information to determine their species. Next, Lowenstein hopes to improve the resolution of the technique – at present about the same as for radiocarbon dating, 40,000 years – by using monoclonal antibodies, which are even more specific and so could detect even smaller protein fragments.