Neanderthal Man (39 page)

________________________________

We began the nuclear DNA sequencing from the libraries that Johannes had prepared from the bone as soon as we could. The results were stunning. When Udo mapped them to the human genome, he found matches for about 70 percent of all DNA fragments. Yet contamination with modern human DNA, as judged from the mtDNA results, was extremely low. This meant that more than two-thirds of the DNA in the bone had come from the dead individual! By comparison, only 4 percent of the DNA from our very best Neanderthal remains did so; more typically, the proportion was well below 1 percent. This bone was as well-preserved as the mammoth that Hendrik Poinar had sequenced and the Eskimo that Eske Willerslev in Copenhagen had sequenced. But both of those specimens had been deep frozen in the permafrost shortly after death. This explained why the majority of the DNA in those specimens was not bacterial, but I could not explain why the individual from Denisova Cave had produced so much DNA. Whatever the reason, it certainly made the analysis of the genome much easier. In fact, our biggest issue was how to weed out the microbial DNA fragments in the library rather than how to fish out the few endogenous DNA fragments, as we had done with the Neanderthals. Now the major question was a good one: Just how much of the nuclear genome could we get? As always, we didn’t want to use the outermost surface of the bone fragment. First, it seemed irresponsible to use it all up, as we didn’t know how much of the larger piece Eddy and his group had used up in Berkeley. Second, if any part of the bone was contaminated from people handling it, it would be the surface. So Johannes used the internal part of the bone to produce two extracts. From test runs of libraries prepared from these DNA extracts, Martin Kircher calculated that we would be able to get even more coverage of the genome than we had for the Neanderthals.

When Johannes made libraries from the extracts, he applied one of Adrian Briggs’s innovations to deal with the chemical damage that  changed C nucleotides in the DNA to U nucleotides. Adrian had shown that most of these U nucleotides were found close to the ends of the ancient DNA molecules, and how to remove the damaged ends. In doing so, he lost an average of one or two nucleotides at the ends of about half the ancient molecules but he also got rid of the vast majority of errors in the DNA sequences. Since it was no longer necessary to take frequent C to T errors into account, the mapping of the fragments to the human genome became easier. Johannes made two large libraries with this method. Not only were about 70 percent of the DNA fragments in those libraries from the Denisova individual, but those DNA fragments now carried many fewer errors than the Neanderthal DNA fragments. This was real progress. Yet, I was nervous, knowing that Eddy’s group might also be at work on the same project, or even polishing a nice manuscript presenting the genome. So I tried to get everything to move as fast as possible, asking the sequencing groups to set other projects aside and sequence these libraries as fast as they could.

I was also very curious about the strange-looking tooth that Anatoly had given us. Only DNA work would tell us if it came from the same type of person as the finger bone. Johannes, as careful as any dentist treating a live patient, drilled a small hole in the tooth and made extracts from the powder he retrieved and, in turn, libraries from the DNA in the extracts. From the libraries he then fished out mtDNA fragments. In addition, we immediately sequenced random DNA fragments from the libraries to see how much of the DNA was endogenous to the individual.

There was good news and bad news. The good news was that he was able to reconstruct the entire mtDNA genome. There were two differences between it and the finger bone, which meant both that it was from a different person and that they were the same type of humans. The bad news was that the fraction of endogenous DNA in the tooth was only 0.2 percent. We were now even more mystified about why the finger bone contained so much endogenous DNA. I speculated that the finger might have rapidly desiccated after death, which might have limited the degradation of the DNA by enzymes in the dying cells and stopped bacterial growth. I joked that perhaps this person died with her pinky pointing up into the air so that it mummified before bacteria had too much of a chance to multiply.

Now that we had shown that the tooth came from the same type of human as the finger, Bence devoted himself to the analysis of its morphology  with renewed energy. Although I am no tooth expert, even I found it to be startlingly large. It was almost 50 percent larger than my molars. Bence pointed out that besides being very big, it was different from most Neanderthal molars with respect to both the absence and presence of certain traits in its crown. Also, its roots were unusual. Unlike Neanderthal molar roots, which tend to be closely spaced or even fused, it had strongly diverging roots. Bence concluded that the tooth morphology suggested that the Denisova population was distinct from both Neanderthals and modern humans. In fact, since the Denisova tooth lacked Neanderthal features that evolved about 300,000 years ago, he surmised that the ancestors of Denisova individuals had gone their separate way from Neanderthals before that. This was in line with what the mtDNA told us. But I was always cautious, some might even say overly skeptical, about the interpretation of morphological traits. Perhaps the Denisova people had reverted to having ancient-looking teeth after separating from either modern humans or Neanderthals. Only the nuclear genome would tell the complete story.

Our sequencing machines began churning out Denisova nuclear DNA sequences at around the same time that we were dealing with the reviewers’ comments and finalizing the Neanderthal paper. Thus, we didn’t have much time to look at the Denisova sequences immediately, but I imagined that we could analyze them quickly once we got to it. During the last four years we had developed computer programs to analyze the Neanderthal genome that could now be directly applied to the genome from the Denisova individual. Still, I remained afraid that Eddy might be far ahead of us, so I decided to scale down the Neanderthal Genome Analysis Consortium to a core and, I hoped, faster group, asking them to devote their full attention to the Denisova genome. Most crucially, we needed David Reich, Nick Patterson, and Monty Slatkin and his crew (see Figure 23.1). We initially called ourselves the “X-Man” group because we didn’t know what the Denisova individual was. Bence had by then told us that the finger was from a young individual, perhaps just three to five years old, and we had sequenced the maternally inherited mtDNA so it seemed inappropriate to use a designation that made everyone think of a macho comic figure. I considered “X-Girl” but thought that sounded too much like a Japanese manga character. Finally, I settled on “X-Woman”—and the name stuck. Right away, the X-Woman Consortium began having weekly phone meetings.

Udo mapped the DNA fragments to the human and chimpanzee genomes. It was comparatively easy given that we had used Adrian’s approach to remove the majority of the errors, but Udo warned me that the  mapping was preliminary. In spite of this, we distributed the data to the X-Woman Consortium. Not long after we had submitted the final version of the revised mtDNA paper to
Nature,
Nick Patterson sent me a report on its preliminary analysis of Udo’s preliminary mappings. When I read it, I felt grateful to the reviewer who had convinced us not to name a new species. Nick had found two things.

First, he found that the nuclear genome of the Denisova finger bone was more closely related to the Neanderthal genome than to the genomes of people living today. In fact, it seemed to be only slightly more different from the Neanderthal genome than the deepest differences one could find among humans living today—for example, between the Papua New Guinean individual we had sequenced and the African San individual. This was quite a different picture than the one painted by the mtDNA results alone, and my immediate suspicion was that gene flow from some other more ancient hominin in Asia was responsible for introducing the mtDNA into the Denisova individuals. After all, we had just shown that modern humans had interbred with Neanderthals, so gene flow seemed a reasonable guess. But it was something we needed to think carefully about.

Figure 23.1. Monty Slatkin, Anatoly Derevianko, and David Reich, at a meeting at Denisova Cave in 2011. Photo: B. Viola, MPI-EVA.

The second thing Nick had found was even more unexpected. Among the five humans we had sequenced for the Neanderthal analysis, the Denisova individual shared more derived SNP alleles with the Papuan individual than with the Chinese, European, or two African individuals. One possible explanation was that relatives of the Denisova individual had mixed with the ancestors of the Papuan individual, although given the distance from Siberia to Papua New Guinea I felt we might be jumping to conclusions. There could be some systematic error in what we did, and Udo again warned me that his mappings of the DNA fragments to the genome were preliminary. Perhaps there was something in the complex computer analyses that created extra similarity both between the Denisova and Neanderthal genomes and between the Denisova and Papuan genomes. Then both of Nick’s findings could be wrong.

A week later Ed finished his own careful analysis of the new data. He found that there were very few Y chromosomal fragments among the DNA we had sequenced, so X-Woman really was a woman, or rather, given the tiny bone, a girl. The general lack of Y chromosomal fragments also indicated that male nuclear DNA contamination was low. When he looked at divergence of the Denisova DNA sequences from the human and Neanderthal genomes, he, like Nick, found that the Denisova genome shared more derived SNP alleles with the Neanderthal genome than with modern humans. So this suggested that the common ancestor of the Denisova girl and Neanderthals first diverged from the lineage that includes modern humans, and only then did the ancestor of the Denisova girl and Neanderthals go different ways. In other words, the Denisova girl and Neanderthals were more closely related to each other than they were to modern humans. Several questions arose as we discussed these data during our Friday meetings in Leipzig and during long phone meetings with Nick, David, Monty, and the others. How could the Denisova mtDNA be so different when the Denisova nuclear genome was closer to Neanderthals than to modern humans? Could the Denisova girl perhaps have had recent ancestors who included Neanderthals and some more archaic human form, perhaps late
Homo erectus?
Or could she be a mixture of modern humans and such an archaic hominin? We looked at each of these possibilities and none seemed to fit.

It took Udo a few months to refine the mapping of all the fragments to each of the comparison genomes. The final mappings didn’t change the picture, and I became convinced that the Denisova girl was a member of a population that shared a common origin with Neanderthals, but that had  lived separately from the Neanderthals for at least as long as Finns today have been separated from, say, the San in southern Africa. Denisova DNA sequences tended to be a bit closer to those of Eurasians than to Africans, but less so than were the Neanderthal DNA sequences. This was best explained by a common ancestry for the Denisova girl and Neanderthals so that when Neanderthals mixed with modern humans, Eurasian ancestors inherited DNA sequences that were somewhat similar to Denisova DNA sequences just because the Neanderthals were related to the Denisova girl.

So it was clear that the population to which the Denisova girl belonged had separated from Neanderthals before they met modern humans. What would we call this population? We certainly didn’t want to give them a Latin name that would force us to label them a subspecies or a species. Since they were only about as different from the Neanderthals as I am from a San, this would be ridiculous. But we needed to call them something. We needed what taxonomists would call a trivial name, such as “Finn,” “San,” “German,” or “Chinese.” “Neanderthal” was such a trivial name, named for Neander Valley in Germany,
Thal
being an old spelling of the German word for “valley.” Following this example, I suggested that we call them “Denisovans.” Anatoly agreed, so we unceremoniously announced our decision in a phone meeting and, from then on, we referred to the population that included X-Woman and the individual with the unusually large molar as the Denisovans.

One exciting issue remained: whether Nick’s finding, that the Denisovan girl shared more derived sequence variants (SNPs) with the Papuan individual than with the other four individuals we had sequenced, was a real discovery or instead due to some bug in a computer program or a quirk in the data. Over the next several weeks, we discussed different technical problems that could cause that data to look this way. But things remained ambiguous. There could, perhaps, be something special about the Papuan DNA sequences that made them appear to be slightly more similar to the Denisovan DNA sequences. To me, it seemed suspicious that we had seen no trace of this putative admixture in China since it would mean that Papuan ancestors could have met the Denisovans, who we knew existed in Siberia, without meeting the ancestors of the Chinese. Of course, maybe the Denisovans lived in other places beyond Siberia. We decided that the best way to address this was to sequence more present-day people. This slowed down our progress toward publication, but we didn’t want to make fools of ourselves by claiming something that would then turn out to be due to some technical oversight on our part. So we decided to sequence seven  more people from around the world. We chose an African Mbuti and a European from Sardinia, two people we wouldn’t expect to have anything to do with the Denisovans. We also included a person from Mongolia in central Asia as a person who lived not too far from the Altai area; a Cambodian as someone on mainland Asia not too far from Papua; and a Karitiana from South America as a representative of Native Americans, whose ancestors had come from Asia and could perhaps have met Denisovans in the past. Finally, we decided to sequence two people from Melanesia and chose a second Papuan and a person from the island of Bougainville.