Update: John Hawks’ lab is working in the same area, and he disagrees with the specific results presented here. Always reminds you to be careful about sexy results presented at conference! (someone should do a study!)
So claimed Peter Parham at a Royal Society meeting last week, Human evolution, migration and history revealed by genetics, immunity and infection. You can actually listen to the talk by pulling down the mp3 file. To get the part about human evolution and introgression, jump to 24 minutes in.
Here is the general sketch: It looks like ~50 percent of the HLA Class I alleles in Europeans derive from Neandertals, ~70-80 percent of HLA Class I alleles in East Asians derive from Denisovans, and that and ~90-95 percent of HLA Class I alleles in Papuans derive from Denisovans. If you recall, ~2.5% of the total genome content of non-Africans seems to be Neandertal, while ~5% of the total genome content of Papuans seems to be Denisovan. The total genome content proportions are rough estimates, there may be some wiggle room in there. But you can see that the HLA allele admixture estimates from these ancient Eurasian lineages is greater by an order of magnitude. Why?
Parham is at pains to point out that there is a major distinction in the nature of the genealogies of alleles which have generally been buffeted by neutral dynamics, and those which have been subject to selection. The HLA region is among the most polymorphic in the human genome, and that is due to the fact that balancing selection maintains diversity (likely a great deal of this through negative frequency dependence over the long term). Presumably this is the target of selection when one conceives of the Red Queen’s Hypothesis in terms of pathogen-host immune system coevolution.
In the presentation it is clear that something seemed off in some of the HLA haplotypes which these researchers had analyzed. They “looked” as if they were introgressed. There has been evidence of this before on other genes. But, with the draft sequences of ancient Neandertals and the Denisovan, scholars could check to see if inferences of admixture between archaic and neo-African lineages were borne out by matching them against the actual sequenced ancient DNA. In some cases they did. In others instances the inferences were wrong (or, the archaic introgression was from a lineage which hasn’t been sequenced yet). In this case Parham reports that his researchers found that the alleles found at high frequency in eastern Eurasia and Oceania seem to derive from the same lineage as that of the Denisovan. Intriguingly, he also adds that the Europeans are about ~50 percent admixed at the HLA Class I locus. If I heard Parham correctly, there are two major points in relation to human evolutionary history:
- East Asians have the Denisovan allele, when they don’t have Denisovan ancestry
- They don’t have the Neandertal alleles, when they do have Neandertal ancestry
- The Papuans are nearly fixed for the Denisovan allele, and lack the Neandertal one
This is why the term “introgression” is key. We’re not talking simple admixture. Rather, admixture followed by selection, whether negative selection which purges introduced alleles, or positive selection which increases the allele’s frequency. We already saw a recent possible case of introgression with a dystrophin allele. This is much more exciting, as the HLA alleles have clear functional relevance, and are known to be targets of natural selection. If adaptation occurs via introgression, this would be one of the key candidate regions a priori. Additionally, the deviation from expectation as inferred by admixture estimates is so great that you have to wonder if selection is responsible for the difference at such a functionally relevant locus. In the East Asian case if these results hold (and I hear Parhman correctly that East Asians carry the Denisovan variant) you see a case where admixture was at such a low level that it’s not detectable, but natural selection preserved a signature of the admixture by amplifying the frequency of an introgressed allele.
In the summation of his presentation Parhman makes a lot of good points about how useful the variation of the Neandertals and Denisovans probably was for the neo-Africans. First, if they went through a bottleneck they may have lacked a large complement of HLA alleles. Because of the need for diversity at this locus to combat pathogens admixture may have been like an injection of mutations which were already preselected for a high degree of utility. Secondarily, Eurasian hominins were probably well adapted to local Eurasian pathogens. The fast that East Asians have much more detectable Neandertal ancestry (~2.5%) than Denisovan (~0.0%), but the Denisovan HLA Class I allele is much more prominent in these populations, indicates that Neandertal and Deninsovan variants were adaptive for different pathogens endemic to their regions of Eurasia!
Finally, a special shout out to Greg Cochran and John Hawks. They’ve been talking to me about introgression as a concept relevant to human evolution since 2005, so a lot of these findings are pretty unsurprising.