Denisovites and people: new data

The ancestors of the Chinese and Japanese interbred with two different populations of Denisovans

Alexander Markov, "Elements"

Fig. 1. "Graphic summary" of the article under discussion in Cell. A new method for detecting foreign genetic impurities has allowed us to establish that the ancestors of modern East Asians interbred with representatives of two different populations of Denisov man. In the genomes of the inhabitants of South Asia and Oceania there are traces of crossing with only one of them.

Thanks to the achievements of paleogenomics, it became known that the ancestors of modern humans interbred with Neanderthals and Denisovans. Today, traces of hybridization with extinct populations can be detected without the use of paleogenetic data. Applying a new method of searching for foreign DNA fragments to the genomes of 5,639 modern inhabitants of Eurasia and Oceania, American geneticists showed that there were at least two episodes of hybridization of Sapiens with Denisovans. At the same time, there are traces of both episodes in the genomes of East Asians, and only one in South Asians and Papuans. The two Denisovan populations that interbred with sapiens differed in the degree of kinship with the Altai Denisovan, whose genome is so far the only read genome of a Denisovan human.

Late representatives of the genus Homo, such as Sapiens, Neanderthals and Denisovans, wandering through the expanses of the Old World, repeatedly interbred with each other. Specialists in paleogenetics and comparative genomics are gradually deciphering the sequence of hybridization episodes and the direction of genetic flows.

To date, five episodes of hybridization have been most convincingly confirmed (Fig. 2):

1) the influx of genes from unidentified archaic people (possibly late erectus) to Denisovans,
2) the influx of genes from ancient Sapiens (who came out of Africa earlier than 60,000 years ago) to the Altai Neanderthals,
3) from Altai Neanderthals to Denisovans,
4) from Neanderthals to sapiens who came out of Africa about 55,000 years ago (due to this episode, about 2% of Neanderthal DNA is present in the genomes of non–African sapiens),
5) from the Denisovans – to the ancestors of the modern inhabitants of Australia and Oceania, as well as, to a lesser extent, to the ancestors of the modern population of South and East Asia.

Fig. 2. Episodes of hybridization, more or less reliably established by 2016. Figure from the article by M. Kuhlwilm et al., 2016. Ancient gene flow from early modern humans into Eastern Neanderthals (see: The genes of archaic Sapiens were found in Altai Neanderthals, and the genes of Heidelberg people were found in Denisovans, "Elements, 02/25/2016).

Until now, there was no clarity on whether hybridization with Denisovans was an isolated event.

There must have been other episodes. In particular, a number of facts indicate repeated crosses of the ancestors of modern Europeans and Asians with Neanderthals (Q. Fu et al., 2015. An early modern human from Romania with a recent Neanderthal ancestor; B. Vernot et al., 2016. Excavating Neandertal and Denisovan DNA from the genomes of Melanesian individuals). In the genomes of some African peoples, such as Pygmies, Hadza and Sandawe, traces of hybridization with an unknown archaic population were found, which apparently separated from the Sapiens ancestors at about the same time when the Sapiens lineages and the common ancestors of Neanderthals and Denisovans separated (J. Lachance et al., 2012. Evolutionary History and Adaptation from High-Coverage Whole-Genome Sequences of Diverse African Hunter-Gatherers; P. Hsieh et al., 2016. Model-based analyses of whole-genome data reveal a complex evolutionary history involving archaic introgression in Central African Pygmies).

Foreign impurities in the genomes of modern humans can be searched in several ways. Firstly, it is possible to search for DNA fragments that match the known Neanderthal and Denisov sequences. But then we will not be able to detect impurities, the source of which were extinct populations for which there is no paleogenetic data. To find such impurities, statistical methods are used that do not take into account the sequences of ancient genomes. These methods are based on the fact that the introduced (introgressed) DNA fragments must have a certain length distribution (the more time has passed since hybridization, the shorter their average length) and contain sets of closely linked (located in the neighborhood) rare polymorphisms. If we are talking about hybridization, which occurred after the ancestors of different races and peoples began to disperse around the world, then introgressed areas should occur only in some, but not all modern populations (for example, Neanderthal admixtures are present in all non-African Sapiens, but are absent in Africans).

This approach, known as "S*-statistics", has been developed and improved for more than 10 years (see: V. Plagnol, J. D. Wall, 2006. Possible Ancestral Structure in Human Populations; B. Vernot, J. M. Akey, 2014. Resurrecting surviving Neandertal lineages from modern human genomes).

American geneticists have developed a new, improved version of S*-statistics, which they called Sprime. The Sprime method allows you to analyze large sections of the genome at once (instead of sequentially counting short sections) and take into account data on multiple genomes at once (instead of pairwise comparisons). In addition, unlike previous versions, Sprime successfully copes with such a complicating circumstance as the limited migration of carriers of introgressed DNA fragments to areas inhabited by people without such fragments. For example, Sprime would make it possible to find Neanderthal impurities in non-Africans, even if some (not too large) number of carriers of these impurities returned to Africa in the distant past and interbred with the local population. Of course, without the involvement of paleogenetic data, Sprime would not have determined that these were Neanderthal impurities, but it would have shown that these were introduced DNA fragments.

Having tested the new method on virtual (computer-generated) genomes with a known history, the researchers were convinced that it has increased reliability and sensitivity compared to previous versions. So, if two populations diverged hundreds of thousands of years ago, the period of hybridization was tens of thousands of years ago, and at the same time about 3% of foreign genes got into the gene pool of the studied population, then the Sprime method will detect about half of the introgressed fragments (the rest will be too short and/or contain too few polymorphisms unique to the second population), and the share of false positive results will be no more than 7%.

After making sure that the method works, the authors applied it to 5639 complete genomes of modern humans from all over Eurasia. Papuans and Yoruba Africans were also included in the sample (the latter were used as a control, since it is believed that they have practically no Neanderthal and Denisovan admixtures).

The number and total length of the identified foreign fragments in the genomes of people from different populations coincided with the previously obtained estimates of the proportion of Neanderthal and Denisovan DNA in these populations (taking into account the fact that the Sprime method, as mentioned above, detects about half of the introgressed sites). In particular, it was confirmed that East Asians have more impurities than Europeans and South Asians.

Recall that all these alien fragments were identified using the Sprime method without using paleogenetic data. Now, of course, it was necessary to compare them with the known ancient genomes. The authors did so, using for comparison two of the most qualitatively read genomes – the Altai Neanderthal and Denisovan. We analyzed only those introgressed areas within which both Neanderthals and Denisovans have at least 10 single-nucleotide differences from modern humans (more precisely, from Africans who do not have Neanderthal and Denisovian impurities). This made it possible for each introgressed site to determine the degree of its similarity, on the one hand, with the Altai Neanderthal, and on the other – with the Denisovan. When assessing the similarity, only those positions were taken into account in which Neanderthals and (or) Denisovans differ from Africans.

Fig. 3. Diagrams showing the number of introgressed DNA fragments with different levels of similarity to the genome of the Altai Neanderthal (horizontal axis) and Denisovan (vertical axis). The populations shown in figures 16-19 (from Mexicans to Peruvians) are not Indians, but people of mixed, mainly European origin. Other explanations in the text. A drawing from the discussed article in Cell.

The results obtained are shown in Fig. 3. The diagrams constructed for 20 non-African populations show the frequency of occurrence of introgressed DNA fragments with different levels of similarity with the genomes of the Altai Neanderthal and Denisovan. These diagrams deserve careful examination.

In all the diagrams we see a small compact area with high values in the lower left corner. These are fragments that were identified by the Sprime program as introgressed, but which at the same time have nothing in common with either Neanderthal or Denisov DNA. Most likely, this is a false positive signal, that is, fragments that are not actually introgressed and were identified by the program incorrectly. Judging by the simulation results, the program should give 5-7% false positive results, which is confirmed.

In all the charts, we also see an impressive rise in the lower right corner. This is our Neanderthal heritage. The sequences inherited from Neanderthals are similar to the Altai Neanderthal genome by an average of 80%. At the same time, their similarity with Denisovans is low – about 10-20%. It should be clarified here that Denisovans are a little closer to Neanderthals than to Sapiens, which means that the common ancestors of Denisovans and Neanderthals managed to accumulate some amount of common differences from Sapiens even before they split up. That is why the similarity of Neanderthal fragments with the Denisov genome is significantly different from zero.

The Papuans have the most Denisovan admixtures among the studied populations (the last diagram). The average level of similarity of Denisovan inclusions in Papuan genomes with the known Denisovan genome is about 50% (and the same 10-20% with the Altai Neanderthal genome). This means that those Denisovans who left their mark on the genomes of Papuans were quite distant relatives of the girl from Denisova Cave, whose genome was sequenced. They represented another population of Denisovans, which lived, perhaps, far from Altai (maybe somewhere in the south or southeast Asia).

In the genomes of South Asians (the third row of diagrams: Hindus, Bengalis, etc.), there is also a Denisovan admixture, although not so large. At the same time, the level of similarity of the introduced sites with the genome of the Denisov girl is the same as that of the Papuans: about 50%. This means that the source of the impurity was probably the same population of Denisovans, not too close to the Altai, as the Papuans.

There are practically no Denisovan impurities in the genomes of Europeans (first row of diagrams), with the exception of Finns (Diagram No. 2), who have a little bit of Denisovan DNA. Most likely, this admixture got to the Finns along with a portion of Asian genes (M. Sikora et al., 2014. Population Genomic Analysis of Ancient and Modern Genomes Yields New Insights into the Genetic Ancestry of the Tyrolean Iceman and the Genetic Structure of Europe).

The most interesting and unexpected result is associated with East Asians (three Chinese samples and one Japanese, diagrams 6-9 in the second row). Their Denisovan admixture consists of two heterogeneous parts: some Denisovan fragments are 50% similar to the genome of the Altai girl (as in Papuans and South Asians), and others are about 80%. The two-hump distribution is statistically significant. This means that the ancestors of the Chinese and Japanese most likely had two episodes of hybridization with Denisovans. Once they interbred with the same population that left a trace in the genomes of Papuans and South Asians, and the second time with other Denisovans closer to the Altai population.

The episode with the participation of "50 percent" Denisovans could be the same as that of the ancestors of Papuans and South Asians. In other words, the common ancestors of Papuans and residents of south and East Asia could also participate in this episode. Or, maybe, only the ancestors of the Papuans actually interbred with the Denisovans, and only then, as a result of some migrations, the ancestors of the Hindus and Chinese received a Denisov admixture from them. As for the episode with the participation of "80 percent" Denisovans, only the ancestors of East Asians were involved in it.

In principle, it is possible to determine the time sequence of hybridization episodes by the length of introgressed fragments. In the genomes of East Asians, Neanderthal fragments are on average slightly shorter than the "50 percent" Denisov fragments, and those, in turn, are shorter than the "80 percent" ones. This indicates a possible sequence of events: first, crossing with Neanderthals, then with "50 percent" and, finally, with "80 percent" Denisovans. However, the authors honestly note that the differences in the length of the fragments are not statistically significant, and therefore it is premature to draw any conclusions based on them.

The new data did not confirm the point of view about multiple episodes of hybridization with Neanderthals. Neanderthal DNA fragments do not break up into any clusters, which means that either there was only one episode, or repeated crosses occurred with Neanderthal populations very close to the one with which the ancestors of non-African Sapiens hybridized initially about 55,000 years ago. The second option is more likely, because Asians have significantly more Neanderthal DNA than Europeans, and this cannot be explained by the action of selection (that is, by the fact that selection more effectively cleaned out the Neanderthal admixture from the ancestors of Europeans). Most likely, the ancestors of Asians were re-crossed by Neanderthals, who were close relatives of those with whom the common ancestors of all non-African Sapiens had previously interbred.

The authors also looked at which of the identified foreign DNA fragments were subjected to positive selection, that is, they benefited our ancestors and therefore reached a high frequency in modern gene pools. The previous conclusions were confirmed, according to which the selection supported a number of Neanderthal alleles associated with the structure of the skin, hair and, most importantly, the immune system. This is logical, because the sapiens who came out of Africa encountered new pathogens in Eurasia, so the genes of the immune defense of the local population could be very useful. Our ancestors received fewer useful genetic variants from Denisovans than from Neanderthals (see: Our ancestors borrowed important genes from Neanderthals and Denisovans to protect against viruses, "Elements", 07.10.2011; Tibetans inherited a gene from Denisov people that saves from hypoxia, "Elements", 10.07.2014).

Thus, thanks to the development of methods of paleogenetics and bioinformatics (analysis of genomic sequences), our knowledge about the history of hybridization of Sapiens with "archaic Homo" is rapidly being refined and detailed. Progress cannot fail to impress, especially if we recall that less than eight years have passed since the very fact of such hybridization was first firmly established (see: Neanderthal genome read: Neanderthals left a trace in the genes of modern humans, "Elements", 10.05.2010).

Source: Browning et al., Analysis of Human Sequence Data Reveals Two Pulses of Archaic Denisovan Admixture // Cell. 2018.

Portal "Eternal youth" http://vechnayamolodost.ru