Figure 1.2. Reconstruction of a piece of mtDNA from the Neanderthal from Neander Valley. Above, a modern reference sequence is shown. Each line below represents one cloned molecule amplified from the Neanderthal type specimen. Where these sequences are identical to the reference sequence, I have placed a dot; where they differ from the nucleotide, I have written them out. In the bottom line is the reconstructed Neanderthal nucleotide sequence. At each position, we require that a change from the reference sequence is seen in a majority of clones and in at least two independent PCR experiments (either the ones shown or others). From Matthias Krings et al., “Neandertal DNA sequences and the origin of modern humans,” Cell 90, 19–30 (1997).

 

At this point, we compared our 379-nucleotide Neanderthal mtDNA sequence to the corresponding mtDNA sequences from 2,051 present-day humans from all around the world. On average, twenty-eight of the positions differed between the Neanderthal and a contemporary person, whereas people alive today carry an average of only seven differences from one another. The Neanderthal mtDNA was four times as different.

We then looked for any indication that the Neanderthal mtDNA was more like the mtDNA found in modern Europeans. One might well expect to find this, since Neanderthals evolved and lived in Europe and western Asia; indeed, some paleontologists believe that Neanderthals are among the ancestors of today’s Europeans. We compared the Neanderthal mtDNA with that of 510 Europeans and discovered that it carried, on average, twenty-eight differences. We then compared it with mtDNA from 478 Africans and 494 Asians. The average number of differences from the mtDNAs of these people was also twenty-eight. This meant that, on average, European mtDNAs were no more similar to the Neanderthal mtDNA than were mtDNAs from modern-day Africans and Asians. But maybe Neanderthal mtDNAs were similar to mtDNAs found in just some Europeans, as one would expect if Neanderthals had contributed some mtDNA to Europeans. We checked this and found that the Europeans in our sample whose mtDNAs were most like that of the Neanderthal showed twenty-­three differences; the Africans and Asians closest to the Neanderthal carried twenty-two and twenty-three, respectively. In short, we observed not only that the Neanderthal mtDNA seemed very different from the ­mtDNAs of modern humans worldwide but also that there was no indication of any special relationship between the Neanderthal mtDNA and any subset of European mtDNAs alive today.

However, just counting differences is not enough to reconstruct the evolutionary history of a piece of DNA. The differences found between DNA sequences represent mutations that occurred in the past. But some  types of mutations are more frequent than others, and some positions in DNA sequences are more prone to mutate than others. At such positions, more than one mutation—especially the types that happen more frequently—may have occurred in the history of a DNA sequence. Therefore, to estimate the history of this particular piece of mtDNA, we needed to apply models for how we believed it had mutated and evolved, bearing in mind that certain positions might have mutated more than once, thus obscuring previous mutations. The result of such a reconstruction is depicted as a tree, in which a DNA sequence on the tip of a branch links back to a common ancestral DNA sequence. These ancestral sequences are depicted as the points where branches join on the tree (see Figure 1.3). When we did such a tree reconstruction, we saw that the mtDNAs of all humans alive today trace their ancestry back to one common mtDNA ancestor.

Figure 1.3. A mtDNA tree, illustrating how the mtDNAs of people alive ­today trace their ancestry back to a common mtDNA ancestor (the so-called Mitochondrial Eve, indicated by a circle) who existed more recently than the mtDNA ancestor shared with the Neanderthal. Nucleotide differences are used to infer branching order, and the numbers refer to the statistical support for the branching order shown. Modified from Matthias Krings et al., “Neandertal DNA sequences and the origin of modern humans,” Cell 90, 19–30 (1997).

This finding, which was already known from Allan Wilson’s work in the 1980s,{1} is in fact expected for mtDNA, since each of us carries only a single type and cannot exchange pieces of it with other mtDNA molecules in the population. Since mtDNA is passed on only by mothers, the mtDNA lineage of a woman will die out if she has no female descendants—so in each generation some mtDNA lineages vanish. Therefore, there must once have existed a woman—the so-called Mitochondrial Eve—who carried an mtDNA lineage that would turn out to be the ancestor of all human ­mtDNAs today, simply because all other lineages since that time have been lost, purely by chance.

According to our models, however, the Neanderthal mtDNA did not trace back to this Mitochondrial Eve but went further back before it shared an ancestor with the mtDNAs of humans alive today. This finding was immensely exciting. It proved beyond any doubt that we had recovered a piece of Neanderthal DNA—and it showed, at least with respect to their mtDNA, that the Neanderthals were profoundly different from us.

My colleagues and I also used the models to estimate how long ago the Neanderthal mtDNA shared an ancestor with current human ­mtDNAs. The number of differences between the two types of mtDNA is an indication of how long they have been transmitted through generations independently of each other. The mutation rates of widely separated species—mice and monkeys, say—will differ, but among closely related species—such as humans, Neanderthals, and the great apes—they have been constant enough to allow scientists to estimate, based on the differences observed, when two DNA sequences last shared an ancestor. Using the models for how fast different types of mutations occur in mtDNA, we estimated that the mtDNA ancestor common to all humans alive today, the Mitochondrial Eve, lived between 100,000 and 200,000 years ago, as Allan Wilson and his team had found. However, the ancestor that the Neanderthal mtDNA shared with human mtDNAs lived about 500,000 years ago; that is, she was three or four times as ancient as the Mitochondrial Eve from whom all present-day human mtDNAs are descended.

This was amazing stuff. I was now fully convinced that we had recovered Neanderthal DNA and that it was very different from the DNA of modern humans. However, before publishing our findings, we needed to overcome the last hurdle: we needed to find an independent laboratory that could repeat what we had done. Such a lab would not need to determine the entire 379-nucleotide mtDNA sequence, but it would have to retrieve one of the regions that carried one or more substitutions setting Neanderthals apart from humans today. This would show that the DNA sequence we had determined really existed in the bone and wasn’t some strange and unknown  sequence perhaps floating around in our laboratory. But to whom could we turn? This was a delicate issue.

Although many labs would undoubtedly want to participate in such a potentially high-profile project, there was the risk that if we chose one that had not worked as intensely as we had on minimizing contamination and addressing all the other problems associated with ancient DNA, it might fail to successfully extract and amplify a relevant sequence. If that happened, our results would be deemed irreproducible and thus unpublishable. I knew that no one had spent as much time and effort on this sort of work as we had, but we eventually settled on the laboratory of Mark Stoneking, a population geneticist at Penn State University. Mark had been a graduate student and then a postdoc with Allan Wilson at Berkeley, and I knew him from my time spent as a postdoc there in the late 1980s. He was one of the people behind the discovery of the Mitochondrial Eve and one of the architects of the out-of-Africa hypothesis of modern human origins—the idea that modern humans originated in Africa some 100,000 to 200,000 years ago, then spread around the world and replaced all earlier forms of humans, such as the Neanderthals in Europe, without admixture. I respected his judgment and integrity and knew him to be an easygoing person. Moreover, one of his graduate students, Anne Stone, had spent the 1992–1993 academic year in our laboratory. Anne, a serious-minded and ambitious scientist, had worked with us on the retrieval of mtDNA from some Native American skeletal remains, so she was familiar with our techniques. I felt that if anyone could repeat what we had done, she could.

I contacted Mark. As expected, he and Anne were excited about trying this, so we parted with one of the last pieces of bone that Ralf had given us. We told Anne and Mark which part of the mtDNA they should try to amplify, so that they would have the best chance of hitting one of the positions in the mtDNA sequence that carried a mutation typical of our Neanderthal sequence. But we sent them no primers or other reagents, just a piece of bone that had been kept in a sealed tube since its trip from Bonn. This precaution minimized any chance that a contaminant might pass from our lab to theirs. We also did not tell them what positions were typical of Neanderthal mtDNA, not because I didn’t trust them but because I wanted to be able to say we had done everything we could to avoid even an unconscious bias. In short, Anne would have to synthesize the primers and do everything independently of us without knowing exactly what result we expected. Once we’d sent the bone off by FedEx, we just had to wait.

Generally, these sorts of experiments take longer than expected: a company fails to deliver primers in the promised time, a reagent you test for contamination turns out to have human DNA in it, a technician falls sick the day he is to run the sequencing machine with the crucial sample. We waited for what seemed like an eternity for Anne to call from Pennsylvania. And then one night she did call. The tone of her voice immediately told me she was not happy. She had cloned fifteen amplified DNA molecules from the region of interest, and they all looked like any person’s today—in fact, like my own or Anne’s mtDNA. This was a crushing defeat. What did it mean? Had we amplified some freak mtDNA? I could not believe that was the case. If it was from some unknown animal, it would not be as close to human mtDNA as it was, yet it could hardly be mtDNA from some unusual human if it was roughly four times as different as all the human mtDNAs that had been studied. There was always the possibility that the sequence we had come up with was created by some chemical modification of the ancient DNA that consistently attacked the same positions in the sequence; however, such a modified mtDNA sequence would be expected to look like a human sequence with extra changes added by this unknown chemical process, rather than like a sequence that had branched off the human lineage in the past. And even then, why wouldn’t Anne find the same sequence we did? The only plausible explanation seemed to be that Anne had more contamination in her experiments than we did—so much so that it outnumbered the rare Neanderthal molecules. What could we do? We could hardly go back to Ralf and ask him for another piece of the valuable fossil on the chance that the next experiment would be more successful than the first.

Perhaps, even if Anne’s experiments had more contamination than ours did, she could sequence thousands of mtDNA molecules from her piece of bone and thereby find some rare ones that looked like ours. But in the meantime we had done experiments to estimate the number of Neanderthal mtDNA molecules in the Neanderthal bone extracts we had used to start our PCRs. As it turned out, there were about fifty. By comparison, a source of contamination such as a dust particle might contain tens of thousands, or hundreds of thousands, of mtDNA molecules. So any such fishing expedition was very likely to fail.

I discussed this conundrum at length, not only with Matthias but in our weekly lab meetings with the subgroup of my lab working on ancient DNA. Throughout my career I have found these extensive discussions with scientists working in my lab to be very useful; indeed, I think they  have been crucial to whatever successes we have had. In such discussions, ideas that would never occur to people focusing solely on their own work are often hatched. Moreover, scientists without a personal stake in a project’s outcome provide a reality check, since they are free of the wishful thinking all too common among those who are working on a project they love and on which their scientific future may depend. Often my role in these discussions is to moderate and select the ideas that seem promising enough to pursue.

Once again, our meeting bore fruit, and we came up with a plan. Anne would be asked to make primers that would not be a perfect match to modern DNA. Instead, the final nucleotide on their tips would match a nucleotide uniquely seen in our putative Neanderthal sequence. Such primers would not (or only very weakly) initiate amplification from modern human mtDNA and thus would favor the amplification of Neanderthal-like mtDNAs. We discussed this plan thoroughly—especially the crucial point of whether Anne’s effort could be considered an independent replication of our finding if she used information from our sequence to make the primers. Obviously, it would have been more aesthetically pleasing if Anne had been able to come up with the same sequence that we did without any prior knowledge of the sequence. However, we could tell her to synthesize Neanderthal-­specific primers that would bracket two other positions that also carried unique nucleotides. And we would not tell her where or how many of those positions there were. If she found the same unique nucleotide changes that we had, then we would all be convinced that such molecules were indeed native to the bone itself. After much further discussion, we agreed that this was a legitimate way forward.

We transmitted the necessary information to Anne; she ordered the new primers; we waited. By now it was mid-December, and Anne had told us earlier that she planned to fly to North Carolina to visit her parents over Christmas. I obviously could not tell her to cancel, much as I wished she would. Finally, after almost two weeks, the phone rang. Anne had sequenced five molecules from her new PCR products. All of them contained the two substitutions we had seen in our Neanderthal sequence—substitutions that are rare or absent in modern humans. This was an enormous relief. I felt we all deserved a Christmas break. We called Ralf in Bonn to relay the good news. As I’d often done during my years in Munich, I celebrated New Year by taking a skiing trip with some wildlife biologists to remote valleys in the Alps on the Austrian border. This time, while skiing up the spectacular valleys, I could not refrain from formulating the paper  that would describe the first DNA sequence from a Neanderthal. To me, what we were about to describe was even more spectacular than the steep and snowy landscape surrounding me.

Matthias and I met up again in the lab after Christmas vacations and sat down to write our paper. One major question was where to send it. Nature, the British journal, and its American counterpart Science, enjoy the most prestige and visibility in the scientific community and in the general media, and either would have been an obvious choice. But they both impose strict length limits on manuscripts, and I wanted to explain all the details of what we had done—not only to convince the world that we had the real thing but also to promote our painstaking methods of extracting and analyzing ancient DNA. In addition, I had become disenchanted with both journals because of their tendency to publish flashy ancient DNA results that did not meet the scientific criteria our group considered necessary. They often seemed more interested in publishing papers that would give them coverage in the New York Times and other major media outlets than in making sure the results were sound and likely to hold up.

I discussed all this with Tomas Lindahl, a Swedish-born scientist at the Imperial Cancer Research Fund Laboratory in London. Tomas, a preeminent expert in DNA damage, is soft-spoken, yet not one to shy away from controversy when he knows he’s right. He has been something of a mentor to me since 1985, when I spent six weeks in his laboratory studying chemical damage in ancient DNA. Tomas suggested we send the paper to Cell, a highly respected and influential journal that specializes in molecular and cell biology. Publication there would send a signal to the community that the sequencing of ancient DNA was solid molecular biology and not just about the production of sexy but questionable results; moreover, Cell allowed long articles. Tomas called its celebrated editor, Benjamin Lewin, to gauge his interest, since such a manuscript was somewhat outside Cell’s usual scope. Lewin told us to submit it and he would send it out for the usual peer review. This was great news. We now had sufficient space in which to describe all our experiments and present all the arguments for why we were convinced we had genuine Neanderthal DNA.

Today, I still consider this paper to be one of my best. In addition to describing the painstaking way in which we had reconstructed the mtDNA sequence and why we considered it genuine, it laid out the evidence that our mtDNA sequence fell outside the range of variation seen today and the  implication that Neanderthals had not contributed mtDNA to modern humans. These conclusions were compatible with the out-of-Africa model of human evolution that Allan Wilson, Mark Stoneking, and others had proposed. As my colleagues and I said in our paper: “The Neandertal mtDNA sequence thus supports a scenario in which modern humans arose recently in Africa as a distinct species and replaced Neandertals with little or no interbreeding.”

We also tried to describe all the caveats we could think of. In particular, we pointed out that mtDNA offers only a limited view of the genetic history of a species. Because it is transmitted only from mothers to offspring, it reflects exclusively the female side of history. Therefore, if Neanderthals interbred with modern humans, we would detect it only if females crossed over between the two groups. This need not have been the case. In more recent human history, when human groups that differ in social status have met and interacted, they almost always had sex with one another and produced offspring. But this generally occurred in a biased way with respect to what males and females do: in other words, the partner from the socially dominant group was most often male, and the offspring of these unions tended to remain in their mother’s group. Of course, we do not know if such a pattern was typical of modern humans when they came to Europe and met Neanderthals some 35,000 years ago. And we do not even know if modern humans were socially dominant in any sense that would be comparable to what we see among human groups today. But it is clear that looking only at the female side of inheritance tells us only half the story of what happened.

Another, even more important limitation of mtDNA stems from the way it is inherited. As noted, an individual’s mtDNA does not exchange bits and pieces with another individual’s mtDNA. Furthermore, if a woman has only sons, her mtDNA will become extinct. Because chance plays such a strong role in the history of mtDNA, even if some had passed from Neanderthals to early modern humans in Europe at some point between 35,000 and 30,000 years ago, it may well have disappeared. This limitation does not exist for the chromosomes in the cell nucleus: recall that they exist in pairs in every individual, with one chromosome in the pair coming from the mother and the other coming from the father. When sperm or egg cells are formed in an individual, the chromosomes break and rejoin in an intricate dance that results in pieces being exchanged between them. Therefore, if we are able to study several parts of an individual’s nuclear genome, we would get several different versions of the genetic history of a group. 

For example, even if, in some parts, the variants perhaps contributed by Neanderthals were lost, this would probably not be the case for all parts. Therefore, by looking at many parts of the nuclear genome, one arrives at a picture of human history that is much less influenced by chance. For this reason, we concluded in our paper that our results “do not rule out the possibility that Neandertals contributed other genes to modern humans.” However, given the evidence at hand, we clearly favored the out-of-Africa hypothesis.

Our paper was peer-reviewed and accepted for publication by Cell after only minor revisions. As is typical for all top journals, the editors at Cell insisted that we not talk about our results before publication in the July 11 issue.{2} They prepared a press release, and I flew to the press conference they organized in London for the day of the publication. It was my first press conference and the first time I’d ever found myself at the center of such intense media attention. Much to my surprise, I enjoyed trying to get across the essence of our work, doing my best to describe both our conclusions and the caveats involved. It was not that easy, because our data had direct implications for a pitched battle that had been raging in the field of anthropology for over ten years.

This battle had been initiated by the out-of-Africa hypothesis, which Allan Wilson and his colleagues had proposed based largely on the patterns of present-day human mtDNA variation. Initially, the idea had been met with ridicule and hostility by the paleontological community. Almost all paleontologists at the time subscribed to the so-called multiregional model for the origin of modern humans—holding that modern humans evolved on several continents, more or less independently, from Homo erectus. They saw a deep history dividing current groups of humans: the ancestors of current Europeans, for instance, were thought to be the Neanderthals and perhaps earlier European hominins; the ancestors of current Asians were thought to be other archaic forms in Asia, going back to Peking Man. However, a growing number of respected paleontologists, foremost among them Chris Stringer at the Natural History Museum in London, now viewed the out-of-Africa model of modern human origins as the best fit to both the fossil record and the archaeological evidence. Chris had been invited by Cell to the press conference, where he announced that our retrieval of Neanderthal DNA was to paleontology what the lunar landing had been to space exploration. I was of course pleased, though not surprised, by his praise. I was even more pleased when the “other side,” the multiregionalists, had good things to say at least about the technical  aspects of our work—particularly when the most vociferous and pugnacious of them, Milford Wolpoff, of the University of Michigan, declared in a commentary in Science that “if anyone would be able to do this, it would be Svante.”

All in all, I was stunned by the attention our paper received. It was reported on the first page of many major newspapers and on radio and TV news shows worldwide. In the week after the paper appeared, I spent most of my time on the phone with journalists. I had worked on ancient DNA since 1984 and had gradually realized that it must in principle be possible to recover Neanderthal DNA. And nine months had now passed since Matthias called and awakened me to say he saw a DNA sequence that did not look human come out of one of our sequencing machines. So I’d had time to get used to the idea and, unlike most of the rest of the world, was not shaken by our achievement. Once the media frenzy had died down, though, I felt the need for some perspective. I wanted to reflect on the years that had led up to this discovery and to think about where I would go next.