Neanderthals and humans interbred in Europe for 5,400 years

Discussion
GWAS of facial features have identified dozens of associated genome regions1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17. In some cases, these regions overlap genes for which experimental evidence demonstrates their involvement in craniofacial development1,2,47. Furthermore, most of the SNPs associated with facial features are in non-coding regions, and enrichment analyses indicate that these SNPs are preferentially located in regulatory elements of the genome, active during craniofacial development1,2,4,17. Consistently, the novel loci (and associated SNPs) we identify here share features with findings from previous GWAS of facial morphology.

Considering the five chromosomal regions highlighted in Table 2: (i) 1q32.3 overlaps the Activating Transcription Factor 3 ATF3 gene (Fig. 3). ATF3 is an evolutionarily highly conserved transcription factor known to be involved in nervous tissue regeneration after trauma48. Although currently there is no evidence for a direct involvement of ATF3 in craniofacial development, it has been reported that ATF3 expression is regulated by FOXL2, a transcription factor whose mutations are known to lead to alterations of the midface49. Furthermore, strongest association was observed for SNPs intronic to ATF3 around an enhancer which has been shown to be active during craniofacial development50 (Fig. 3). (ii) Associated SNPs in 3q21.1 overlap the MYLK (Myosin light chain kinase) gene, which studies in mice have implicated in palate fusion during development51. (iii) The newly associated 8p11.21 region includes a cluster of disintegrin and metalloproteinase (ADAM) domain genes (Fig. 4). This is a family of surface proteins with adhesion and protease activity, members of which have been shown to be involved in craniofacial development52. Furthermore, one of the ADAM genes in the cluster on 8p11.21 (ADAM3A), has been suggestively associated with non-syndromic cleft lip/palate43. (iv) Associated SNP on 10p11.1 overlap a cluster of Zinc Finger proteins genes (ZNF, Fig. 4). This cluster includes ZNF25, which has been shown to be involved in osteoblast differentiation of human skeletal stem cells53, this is a process in which RUNX2 (a well-established craniofacial morphology gene, Table 1) also plays a major role54,55,56. (v) The mouse analyses performed here are consistent with the novel association we detect on human 22q12.1. In humans, maximum association is seen for SNPs intronic to the SEZ6L gene (Fig. 2). In mice, SNPs in Sez6l are also significantly associated, although association is strongest around the Ttc28 and Mn1 gene regions (Fig. 2). There is currently no evidence implicating SEZ6L directly in craniofacial phenotypes, but there is abundant evidence that Ttc28, Mn1 and other genes in this region are involved in mouse craniofacial development (Fig. 2)34,57,58. Interestingly, of the candidates highlighted here, three (ATF3, MYLK and SEZ6L) are the genes closest to the index SNPs and, in our RNAseq data analysis, we observe that two of these genes (MYLK and SEZ6L) are preferentially transcribed in CNCC cells (Supplementary Fig. 7B).

Genetic determinants of variation in facial features in contemporary human populations are also likely to have played a role during the evolution of facial morphology. We previously identified a region in 1p12 in which a tract introgressed from archaic humans (Denisovans) impacts on lip thickness. That chromosomal region had previously been shown to be associated with body fat distribution59 and bears a strong signature of natural selection, raising the possibility that Denisovan introgression could have facilitated adaptation to a cold environment. The evidence we observe of Neanderthal introgression in 1q32.3 impacting on mid-face height represents the second instance of archaic human introgression affecting facial morphology in modern humans. In this case, the possibility of examining similar skull traits in contemporary human and Neanderthal skulls allowed us to determine that the increase in mid-face height associated with archaic introgression in 1q32.3 is consistent with the modern human-Neanderthal morphological differentiation. Evaluating the consistency of phenotypic effects had not been possible in the case of Denisovan introgression in 1p12 as that case concerned only soft tissues (the lips). Analysis of skulls has long shown that facial morphology differs markedly between Neanderthals and modern humans with the mid-face, particularly the nasal cavity, showing major differences60. This includes markedly taller noses in Neanderthals than in modern humans. Furthermore, it has long been speculated that nose morphology (in Neanderthals as well as in modern humans) has been the subject of natural selection, particularly as an adaptation to environmental temperature and humidity61,62,63. Further genetic work, including future analyses of additional ancient DNA samples, could help shed light on this question.

Although the earliest (and largest) studies on the genetics of facial variation have been carried out in people of Europeans ancestry4,9,10, recent efforts have increasingly sought to examine non-Europeans1,2,5,32. Populations with admixed continental ancestry, such as Latin Americans, offer challenges and opportunities for such studies. In these populations, optimal correction for population stratification, considering both global and local genomic ancestry, is a challenging analytical problem for which an all-round solution is yet to be developed64,65,66. Use of genetic PCs and local ancestry correction approaches to deal with population stratification should therefore be undertaken with caution. Nevertheless, the extensive genetic and phenotypic diversity of Latin Americans is enabling GWASs that have led to important insights into the genetics of physical appearance1,2,22,23,24. This is illustrated here by the novel instance of archaic introgression detected in 1q32.3: the introgressed tract has a high frequency in Native Americans but is essentially absent in Europeans (Table 2, Supplementary Fig. 6). Given the wide-spread availability of 2D photographs, the automated landmarking approach we applied here could facilitate a more comprehensive world-wide sampling of human facial variation than hitherto attempted. The study of larger and more diverse study samples should enable a fuller assessment of the genetic architecture of facial variation in the global human population and of the evolutionary forces that have shaped this variation across the world.

The authors took advantage of open access to drill down into the high-quality genomes of three Neanderthals and a Denisovan – for some reason all of them female. The Neanderthals included two Siberians: the 100,000 year old Altai female who lived in Denisova Cave, and a 48,000 year female who resided in Chagyrskaya Cave. A third female aged about 57,000 years came from Vindija Cave in Croatia. The Denisovan genome came from a female who lived in Denisova Cave about 64,000 years ago.

The first surprise was discovering that the full variability of the ABO system seen in modern humans was present in the Neanderthals. “We thought for years that H. sapiens was the only one to have the full set,” says Condemi. That was because our closest relatives have only a partial set. Chimpanzees are all type A; gorillas are all type B. Until this study the only Neanderthal to be checked was blood type O. “So the assumption was they were all O,” says Condemi.

In fact the Vindija lady was a B (genotype BO), the Chagyrskaya lady an A (genotype AA), and the Altai lady A (genotype AA). The most parsimonious explanation, says Condemi, is that the African ancestor of Neanderthals and modern humans, perhaps Homo heidelbergensis, already possessed the full range of the ABO system.


The next surprise was that all three Neanderthals carried a rare Rhesus type which Condemi refers to as “Rhesus plus incomplete”. This variant had only ever been seen once before. In 2019, researchers analysing the DNA of 72 Western Desert Aboriginal people found that one of them carried the same novel Rhesus type.

onlinelibrary.wiley.com/doi/10.1111/trf.15047

“At the time, it was assumed to be a new Rhesus type that had arisen in Australia,” says Condemi. “Now we know that it had existed in the past and was lost.” The authors also found that in a population of 80 people from Papua New Guinea, only one carried this rare Rh type.

science.sciencemag.org/content/367/6484/eaay5012?rss=1

The finding confirms the evidence from DNA (all non-Africans carry about 2% Neanderthal DNA) that modern humans interbred with Neanderthals in the Middle East before heading to south Asia and Australia.

The blood-type findings also gives clues as to the Neanderthals’ disappearance.

First, there’s the finding that three Neanderthals separated by 50,000 years in time and 5,000 km of space all shared the same Rhesus type. This adds to the evidence from genome studies of their low genetic diversity, a factor that can put a species at risk of being wiped out by disease.

On the other hand, interbreeding with modern humans could have put them at risk of another kind.

Condemi says that if an Rh ‘partial complete’ Neanderthal mated with an Rh complete H. sapiens, there would be an 18% chance of the infant developing the condition known as “haemolytic disease of the newborn” and dying.

Bottom line: if you need to give a Neanderthal a blood transfusion, don’t use human blood unless it’s from a rare Western Desert Australian or Papuan.