Dna: The Secret of Life

by Watson, James

  Matthias Krings, Pääbo's graduate student, took on the project. He was pessimistic at first, but favorable early analyses to assess the bones' state of preservation emboldened Krings to press ahead. His search for viable DNA was focused not in the cells' nuclei, as one might expect, but in the little bodies called mitochondria, which are scattered throughout the cell outside the nucleus and produce the cell's energy. Each mitochondrion contains a small loop of DNA, some 16,600 base pairs in length. And because there are from 500 to 1,000 mitochondria in every cell, but only two copies of the genome proper (in the nucleus), Krings knew that those decaying Neanderthal bones were much more likely to yield intact mitochondrial sequences than intact nuclear ones. Furthermore, since mitochondrial DNA (mtDNA) had long been a staple of studies of human evolution, Krings would have plenty of modern human sequences against which to make comparisons.

  A major worry for Krings and Pääbo was contamination. In the past a number of claimed successes at sequencing ancient DNA had proved to be erroneous when the sequence turned out to be from a modern source that had contaminated the sample. Every day each of us sloughs off a vast number of dead skin cells, showering our DNA into the environment to wind up we know not where. The polymerase chain reaction (PCR), with which Krings expected to amplify the stretch of mtDNA he hoped to find, is so sensitive that it can act upon a single molecule, amplifying any DNA it might encounter regardless of whether the source is ancient or still kicking. What if the Neanderthal DNA was too degraded for PCR to work, but the reaction proceeded nevertheless, amplifying a DNA sequence from an invisible contaminating particle that had flaked off Krings himself? Krings might then have to explain how he and the Neanderthal happened to have the same mtDNA sequence – a result unlikely to please the young man's boss, and even less his parents. To insure against this possibility, Krings and Pääbo arranged for a separate laboratory, Mark Stoneking's at Pennsylvania State University, to replicate the study. Contamination might occur there, too, but probably not with DNA from Krings, a continent away. And if both labs obtained the same result from the sample, it would be reasonable to suppose they had found a bona fide Neanderthal sequence.

  "I can't describe how exciting it was," says Krings of the moment he first glimpsed the sequencing results. "Something started to crawl up my spine." Although, as feared, some sequences showed evidence of contamination, in others he could see something wondrous: a collection of intriguing similarities to, and differences from, the modern human sequence. Piecing together segments, he was able to reconstruct a Neanderthal mtDNA stretch running 379 base pairs. But the results weren't yet in from Penn State. Those sequences, however, proved to be the same: the identical 379 base pairs. "That's when we opened the champagne," Krings recalls.

  The Neanderthal sequence had more in common with modern human mtDNA sequences than with those of chimpanzees, telling us that Neanderthals were unquestionably part of the human evolutionary lineage. At the same time, however, there were dramatic differences between the Neanderthal sequences and all 986 available sequences of modern human mtDNA to which Krings compared his sample. And even the most similar of those 986 sequences still differed from the Neanderthal one by at least 20 base pairs (or 5 percent). Subsequently, mtDNA has been sequenced from two other Neanderthals, one found in southwest Russia, the other in Croatia. The sequences, as expected, are not identical to the original one – we would expect to see variation among Neanderthal individuals just as we would among modern humans – but they are similar. The sum of the genetic evidence leads us to conclude that while Neanderthals do have their place on the evolutionary tree of humans and their relatives, the Neanderthal branch is a long way from the modern human limb. If, when they encountered each other in Europe 30,000 years ago, Neanderthals and moderns had indeed interbred, Neanderthal mtDNA sequences would have entered the modern human gene pool. That we see no evidence of such Neanderthal input implies that modern humans eliminated the Neanderthals rather than interbreeding with them. But whether they achieved the lethal result by direct confrontation or by more subtle means is something the DNA can't tell us.

  Studies of Neanderthal DNA have shown that we are genetically distinct from Neanderthals. But the overall lesson of molecular studies of human evolution has tended to run in the opposite direction, revealing just how astonishingly close we are genetically to the rest of the natural world. In fact, molecular data have often challenged (and overthrown) long-held assumptions about human origins.

  The great chemist Linus Pauling was the father of modern molecular approaches to evolution. During the early 1960s, he and Emile Zuckerkandl compared the amino acid sequences of corresponding proteins from several species. These were the early days of protein sequencing, and their data were inevitably limited. Nevertheless, the pair noticed a striking pattern: the more closely related two species are in evolutionary terms, the more similar are the sequences of their corresponding proteins. For example, comparing one of the protein chains of hemoglobin molecules, Pauling and Zuckerkandl noted that over its total length of 141 amino acids, there is only one difference between the human version and the chimpanzee, but the difference between humans and horses is 18 amino acids. The molecular sequence data reflect the fact that horses have been evolutionarily separated from humans longer than chimpanzees. Unearthing evolutionary history buried in biological molecules is now common practice; at the time, however, the idea was novel and controversial.

  Molecular approaches to studying evolution depend on the correlation of two variables: the length of time two species (or populations) have been separated and the extent of molecular divergence between them. The logic of this "molecular clock" is simple. To illustrate it, let us imagine some matchmaking between two pairs of identical twins, one of genetically identical females and one of identical males. Each female is wed to one of the males, and each couple is then placed on its own otherwise uninhabited island. From a genetic perspective, the populations of the two islands are at the outset indistinguishable. Now leave each couple and its descendants alone for a few million years. At the end of this period, mutations will have occurred in the population on one island that will not have occurred in the population on the other. And vice versa. Because mutations occur at a low rate and because individual genomes, being large, offer huge numbers of possible sites where mutations might occur, it is inconceivable that both populations will have acquired the same set of mutations. So when we sequence DNA from the descendants of each couple, we will find that many differences between the once-identical genomes have accumulated. We say that the populations have "diverged" genetically. The longer they have been separated, the more divergent they will be.

  But how do we tell time, so to speak, by looking at this "molecular clock"? Put another way, how can we measure the genetic divergence between ourselves, say, and the rest of the natural world? In the late sixties, long before the advent of DNA sequencing, Allan Wilson, a whimsical New Zealander at UC Berkeley, together with his colleague Vince Sarich, set about applying the Pauling-Zuckerkandl logic to humans and their closest relatives. But at a time when protein sequencing was still a dauntingly cumbersome and laborious affair, Wilson and Sarich found an ingenious shortcut.

  The strength of an immune reaction to a foreign protein reflects how foreign the protein is: if it is relatively similar to the body's own protein, then the immune reaction is relatively weak, but if it is very different the reaction is proportionately stronger. Wilson and Sarich compared reaction strengths by taking a protein from one species and measuring the immune responses it triggered in others. This gave them an index of the molecular divergence between two species, but to introduce a time dimension to this "molecular clock" they needed to calibrate it. Fossil evidence implied that New and Old World monkeys (the two major groups of monkeys) separated from their common ancestor around 30 million years ago – and so Wilson and Sarich set the immunological "distance" between New and Old World monkeys as equivalent to 30 million years' separation. Where di
d this put humans in relation to their closest evolutionary kin, chimpanzees and gorillas? In 1967 Wilson and Sarich published their estimate that the human lineage had separated from that of the great apes about 5 million years ago. Their claim provoked an uproar: in paleoanthropological circles conventional wisdom held that the divergence had occurred around 25 million years ago. Between humans and apes, the establishment insisted, there is clearly much more than 5 million years' worth of difference. It was, for many, cause enough to dismiss the Berkeley team's newfangled genetic method as untrustworthy, and, to declare that, anyway, geneticists should stick to their fruit flies, and leave humans to the anthropologists! Wilson and Sarich, however, weathered the storm. And subsequent research has shown that their dating of the human/great ape split was remarkably accurate.

  When the time came to extend his analysis of the human/ape divide from proteins to DNA, Wilson entrusted the effort to his graduate student Mary-Claire King (see Plate 48). The product, in 1975, was one of the outstanding scientific papers of the twentieth century. For a long time, though, such a triumphant outcome seemed unlikely, especially from King's perspective. Her work had not been going well, owing in part to the enormous distraction created at Berkeley by the anti-Vietnam War movement in the early 1970s. King considered going off to Washington, D.C., to work for Ralph Nader, but fortunately she sought Wilson's advice. "If everyone whose experiments failed stopped doing science," he wisely counseled, "there wouldn't be any science." King stuck with it.

  King and Wilson's comparison of the chimpanzee and human genomes combined a number of methods, including a clever technique called "DNA hybridization." When two complementary strands of DNA come together to form a double helix, they can be separated by heating the sample to 95ºC – a phenomenon called "melting" in the molecular geneticist's jargon. What happens when the two strands are not perfectly complementary – when there are mutations in one of them? It turns out that two such strands will melt apart at a temperature lower than 95ºC. How much lower will depend upon the degree of difference between the two strands: the greater the difference, the less the heat required to pry them apart. King and Wilson used this principle to compare human and chimpanzee DNAs. The closer the two were in sequence, the closer the double helix's melting point would be to the perfect-match standard of 95ºC. The closeness observed was surprising indeed: King was able to infer that human and chimpanzee DNA differ in sequence by a mere 1 percent. In fact humans have more in common with chimpanzees than chimpanzees do with gorillas, the genomes of the latter two differing by about 3 percent.

  So striking was the result that King and Wilson felt obliged to put forward an explanation for the apparent discrepancy between the rates of genetic evolution – slow – and of anatomical and behavioral evolution – fast. How could so little genetic change account for the substantial difference we see between the chimpanzee at the zoo and the species on the other side of the glass? They suggested that most of the important evolutionary changes had occurred in the pieces of DNA that control the switching on and off of genes. This way, a small genetic change could have a major effect by changing, say, the timing of the expression of a gene. In other words, nature can create two very different-looking creatures by orchestrating the same genes to work in different ways.

  The next, and biggest, bombshell from Wilson's Berkeley lab came in 1987. Using patterns of DNA sequence variation, he and his colleague Rebecca Cann figured out the family tree for our entire species (see Plate 49). It was one of the very few pieces of science ever to make the cover of Newsweek.

  As Krings would in his analysis of Neanderthals a decade later, Cann and Wilson relied on mitochondrial DNA. There were several reasons for using mtDNA, but as usual the practical ones were most important. In the days before PCR technology had entered the research mainstream, getting enough DNA to probe a particular gene or region could be quite a headache. And Cann and Wilson's study called for analyzing not one but 147 samples. They therefore needed as much DNA as they could get their hands on. A human tissue sample is massively rich in mtDNA compared with the chromosomal DNA found in cell nuclei. Still, Cann and Wilson would need plenty of tissue if they were to have any hope of extracting even mtDNA in sufficient quantities. Their solution: placentas. Usually discarded by hospitals after babies are delivered, these are a rich source of mtDNA. All Cann and Wilson had to do was persuade 147 pregnant women to donate their babies' placentas to science – 146, actually, because Mary-Claire King was more than willing to contribute her daughter's placenta. And they knew that to reconstruct the human family as completely as possible they would need tissue from the most genetically diverse range of donors they could assemble. Here America's melting-pot population offered a distinct advantage: they would not have to travel to Africa to get hold of African DNA – the slave trade had brought African genes to our shores. But Cann and Wilson would have to depend on collaborators in New Guinea and Australia to find Aboriginal women (not much represented in the U.S. gene pool) who were willing to participate.

  Your mtDNA is inherited from your mother. Your father's genetic contribution, contained in the head of a single sperm, did not include mitochondrial material. The sperm's DNA is injected into an egg cell that already contains mitochondria derived from the mother. Cann and Wilson would therefore be tracing the history of the human female line. Inherited from just one parent, mtDNA never gets an opportunity to undergo recombination, the process by which segments of chromosome arms are exchanged so that mutations are shuffled from one chromosome to another. The absence of recombination in mtDNA is a major advantage when we come to reconstruct the family tree based on similarity of DNA sequences. If two sequences have the same mutation, we know that they must be descended from a common ancestor (in whom that mutation originally arose). Were recombination occurring, however, one of the lineages could have acquired the mutation just recently through a recombinational shuffling event, so having a mutation in common would not necessarily indicate common ancestry. Now the logic for using mtDNA to make the family tree is simple. Similar sequences – those with plenty of mutations in common – indicate close relationship; sequences with many differences indicate a more distant relationship. In visual terms, close relatives – those that derive from a relatively recent common ancestor – will cluster close together on the family tree; distant relatives are more spread out, because their common ancestor is relatively far back.

  Cann and Wilson found that the human family tree has two major branches, one comprising only various groups within Africa and the other consisting of some African groups plus everyone else. This implies that modern humans arose in Africa – that is where the ancestors common to all of us lived. This idea was hardly new. Noting that both our closest relatives, chimpanzees and gorillas, are native to Africa, Charles Darwin himself inferred that humans had evolved there too. The most striking, and controversial, aspect of Cann and Wilson's family tree is how far back it goes in time. By making a number of simple assumptions about the rate at which mutations accumulate through evolution, it is possible to calculate the age of the family tree – the time back to the great-great-great-great- . . . -grandmother of us all. Cann and Wilson came up with an estimate of about 150,000 years. Even the most distantly related currently living humans shared a common ancestor as recently as 150,000 years ago.

  Like Sarich and Wilson's result two decades earlier, Cann and Wilson's was greeted by many in the anthropological community with outraged disbelief. One widely accepted view of human evolution held that our species was descended from individuals who left Africa about 2 million years ago before settling throughout the Old World. Such a model implied that the family tree should be about thirteen times deeper. Cann and Wilson's alternative, dubbed by the media "The Eve Hypothesis" or, less misleadingly, "Out of Africa," did not deny the more ancient migration, but rather implied that when modern humans arrived in Europe they displaced those populations of early hominids derived from the original exodus nearly 2 million years befo
re. Homo erectus, the species that spread out from Africa 2 million years ago, migrated through the Old World and gave rise, about 700,000 years ago, to Neanderthals, who were thus in effect their European descendants. Then, no more than about 150,000 years ago, another group, Homo sapiens or modern humans – also descendants of Homo erectus but a group that had evolved without ever having left the mother continent – now chose to repeat the odyssey out of Africa made eons before by their H. erectus ancestors. We have seen how the Neanderthals failed to interbreed with the new arrivals in Europe, and the same seems to have been true whenever H. sapiens encountered H. erectus. Wherever they met, the former displaced the latter. And the disappearance of the last Neanderthal, around 29,000 years ago, represents the extinction of the last of the nonmodern descendants of H. erectus.