The Contribution of Neanderthal Introgression to Modern Human Traits

Abstract

DNA retrieved from ancient specimens revealed that Neanderthals, our closest extinct relatives, admixed (mated) with modern human contemporaries. As a consequence, Neanderthal DNA survives scattered in the genome of present-day human populations. Patterns of Neanderthal introgressed genomic sequences suggest that Neanderthal alleles had distinct fates in the modern human genetic background. Some Neanderthal alleles facilitated human adaptation to new environments, while others had deleterious consequences. In this review, we discuss the body of work on Neanderthal introgression generated over the past decade. We describe how evolutionary forces shaped the genomic landscape of Neanderthal introgression and highlight the impact of introgressed alleles on human biology and phenotypic variation.

Introduction

Due to their close evolutionary relationship to modern humans, Neanderthals have been the subject of intense interest among both the general public and scientists. This fascination motivated the first studies of ancient DNA that investigated the relationship between Neanderthals and modern humans¹. Since the mid-19th century, numerous Neanderthal fossils displaying a distinctive set of morphological features [BOX 1] have been unearthed at several sites, from Portugal eastwards to the Siberian mountains and as far south as the Levant (Figure 1). Neanderthals inhabited this vast geographic region from ~300–430 thousand years ago (kya) until ~40 kya^2,3, overlapping in time and space with modern humans. In 2010, the sequencing of ancient DNA from ~40,000-year-old Neanderthal specimens - a milestone achievement - revealed that Neanderthal DNA survives today, scattered in the genomes of present-day people. This discovery motivated the development of analytical frameworks for detecting traces of archaic introgression (i.e. the acquisition of genetic material through mating/gene flow from archaic hominins into the modern human gene pool), allowing for Neanderthal ancestry to be mapped throughout the genomes of thousands of contemporary humans. The possibility of resurrecting Neanderthal DNA opened up a new era of research in the field of human evolutionary genetics and offered a number of fascinating insights about human evolution and population history. Here we review the most relevant scholarship from the past decade that illuminated the evolutionary forces underlying current patterns of Neanderthal ancestry, and the fitness and functional consequences of Neanderthal introgressed alleles. Finally, we discuss the major challenges and new horizons of this growing research field.

BOX 1. Morphological Differences Between Neanderthals and Modern Humans.

Craniofacial Morphology.

Neanderthals are large-brained hominins and their cranial capacity overlaps with modern humans. However, there are numerous differences in craniofacial morphology [BOX 1 Panel A]. In profile, Neanderthal crania are elongated and low in comparison to the rounded, globular shape characteristic of modern humans¹³¹. The occipital bone at the rear of the skull has a prominent protrusion known as the chignon or occipital bun¹³², together with a centrally located depression (suprainiac fossa)¹³³. The Neanderthal face is also distinctive, characterized by mid-facial prognathism, wherein the midface protrudes anteriorly as a result of enlarged maxillary sinuses¹³⁴. This contrasts with a flatter facial profile in modern humans. The Neanderthal face is also characterized by 1) a more pronounced and continuous supraorbital torus (brow ridge)¹³⁵, 2) long and thin zygomatic arches (cheek bones)¹³⁵, 3) lack of a distinct chin¹³⁴, 4) a broad mandible with a retromolar gap between the third molar and the mandibular ascending ramus¹³⁴, 5) unique morphology of middle ear ossicles¹³⁶, and 6) an oval shaped foramen magnum¹³⁷.

Dental Morphology.

Neanderthal incisors are labially convex, shoveled and characterized by lingual (tongue side) tubercles¹³⁸. Their molars are taurodont with fused roots and an enlarged pulp chamber¹³⁹ [BOX 1 Panel B]. Biomechanical analyses suggest taurodontism does not result in significant differences in withstanding dental stress and may have instead evolved as a way to reduce tooth wear over time or simply occurred as a byproduct of genetic drift¹³⁹. Additionally, Neanderthal molars have a larger surface area at the enamel-dentine junction that may affect mineral deposition and enamel formation during ontology, leading to faster dental maturation^140,141. The increased complexity of the Neanderthal enamel-dentine junction may also contribute to the presence of a ridge known as the mid-trigonid crest found in most Neanderthal lower molars, but rarely in modern humans¹³⁸.

Postcranial Morphology.

Neanderthals and modern humans display fundamentally different body proportions [BOX 1 Panel C]. Neanderthals are generally stockier than modern humans, with a less gracile skeleton, shorter limbs and broader joints¹³⁷. Additionally, the chest has a distinctive barrel-shape with a wider rib cage, though this does not seem to reflect significant differences in chest volume¹⁴². The Neanderthal pelvis is characterized by flaring ilia and long, thin pubic rami¹⁴³, which together with relatively shorter limbs may have resulted in differences in gait and posture^144,145. Neanderthal females may have also had wider pelvic inlets than modern human females to accommodate increased neonatal cranial volume¹⁴⁶.

As in modern humans, Neanderthal phenotypes exhibit variability in metric and non-metric (qualitative) traits. There is no simple reason for the morphological differences between Neanderthals and modern humans and most speculation has centered on the craniofacial anatomy. For example, some Neanderthal features, such as their unique nasal morphology, stockier bodies, and shorter limbs, may reflect adaptations to colder climates¹⁴⁷. Alternatively, some research suggests that Neanderthal cranial features reflect strong, prolonged masticatory stress¹⁴⁸, while other research suggests that Neanderthal cranial features may at least partially be attributed to genetic drift^149,150. Regardless, the differences observed in skeletal morphology between Neanderthals and modern humans reflect their underlying genetics and are representative of differences in their ecological, behavior, and life history trajectories.

BOX 1 Figure. Morphological differences between Neanderthals and modern humans.

(A) Comparison of craniofacial features in Neanderthals and modern humans. (B) Example of non-taurodont and taurodont molars exhibiting enlarged pulp chambers. (C) Comparison of Neanderthal and modern human skeletal morphology. Features that are typically attributed to Neanderthals are labeled.

Figure 1.

Neanderthal geographic range. Dating and location of some of the archaic hominin specimens for which nuclear ancient DNA is available are also shown. Asterisks (*) indicate genomes sequenced at high coverage. Created with BioRender.com

The Era of Neanderthal Genomics

For decades, a contentious and fascinating question in human evolution has been whether or not Neanderthals interacted with our ancestors, and whether these interactions involved mating. While initial attempts to answer this question faced some limitations^4,5, the sequencing of the first draft of the Neanderthal genome in 2010 conclusively revealed that Neanderthals did admix with modern human contemporaries - likely between 40 and 54 kya^6,7 - to an extent that ~1–4% of the genome of present-day human populations in Eurasia was inherited from Neanderthals⁸. That same year, DNA extracted from a fingerbone of a previously unknown hominin from Denisova cave in the Altai Mountains of Siberia⁹, revealed that this group, called the ‘Denisovans’, contributed to the genomes of present-day people in Oceania, East and Southeast Asia, providing further evidence for historical admixture with archaic hominins^9–13. Over the following decade, advances in ancient DNA extraction and sequencing (reviewed in Orlando et al.¹⁴) facilitated the generation of three Neanderthal and one Denisovan genome of quality approaching that of high-quality genomes obtained from living individuals: the "Altai Neanderthal" genome, from a female Neanderthal (“Denisova 5”) excavated at Denisova cave in the Altai Mountains of Russia, dated to ~120–130 kya and sequenced to 52x depth¹⁵; the "Vindija Neanderthal" genome, from a female Neanderthal (“Vindija 33.19”) from Vindija Cave in Croatia, dated to ~50 kya and sequenced to 30x depth¹⁶; the "Chagyrskaya Neanderthal" genome, from a female Neanderthal (“Chagyrskaya 8”) from Chagyrskaya Cave in Russia, dated to ~80 kya and sequenced to 27x depth17; the "Altai Denisovan" genome, from a female Denisovan (“Denisova 3”), from Denisova cave in the Altai Mountains of Russia, dated to between 74 and 82 kya and sequenced to 30x depth¹⁸. These genomes, along with moderate quality DNA retrieved from additional hominin specimens (such as El Sidron in Spain, Feldhofer in Germany, and Mezmaiskaya cave in Russia)^8,19–22, and archeological sediments^23,24, provided unprecedented insights into Neanderthal population history and genetic legacy.

Nuclear DNA sequences from 430,000 years old hominin remains from Sima de Los Huesos (Spain) indicate that Neanderthals diverged from the modern human lineage ~520–630 kya^15,16, and at least 430 kya from their Denisovan sister group²⁵ (Figure 2). After their separation, the Neanderthal and Denisovan groups continued to interbreed and exchange genes¹⁵. In 2018 the discovery and genome sequencing of “Denisova 11” (Figure 1), the direct offspring of a Neanderthal mother and a Denisovan father in the Denisova Cave, provided incontrovertible further evidence that mating between hominin groups was common in the Late Pleistocene²⁶.

Figure 2: Simplified demographic model illustrating inferred population relationships and admixture events among modern humans and archaic hominins.

Adapted from Nielsen et al. 2017 Nature¹⁵⁵. Events depicted in the schematic include (1) gene flow from early modern humans (MH) into Neanderthals^20,35,156, (2) from a super-archaic hominin lineage into Denisovans^15,35, and (3) gene flow from a super-archaic hominin lineage into early modern humans in Africa^36,38–40; (4) admixture events between Neanderthals and Denisovans in the Altai Mountains^15,157, also supported by the sequencing of the first generation offspring of a Neanderthal mother and a Denisovan father²⁶; (5) gene flow from Neanderthals into the ancestors of all non-Africans^{8,11,27,28,45}, (6) putative gene flow from Neanderthals into the ancestors of East Asians^16,31,158 (represented by a dashed arrow), (7) back migration from Eurasia to Africa³⁴; and multiple Denisovan introgressions into the ancestors of (8–9) Oceanians ^{9–11,13,18,45} and (10) East Asians¹². For simplicity we represent the multiple, distinct introgressing Denisovan-like lineages as arising from a single Denisovan source.

Genetic evidence indicates that Neanderthals lived in small, highly inbred, and geographically structured populations^15–17. Early Neanderthals were differentiated into at least two geographic groups: an eastern group typified by the Altai Neanderthal from Denisovan cave (Siberia), and a western group predominantly in Europe²². This western group appears to have persisted through time, showing genetic continuity with late European Neanderthals such as the Vindija Neanderthals²², and may have migrated and replaced subpopulations of the easternmost part of the Neanderthal distribution between 90 and 120 kya^17,21,26.

Gene Flow Between Neanderthals and Modern Humans

Following the discovery of archaic hominin ancestry in modern human genomes, genetic research focused on the genomic and geographic distribution of Neanderthal introgression. The release of the Altai Neanderthal genome¹⁵ catalyzed the rapid development of methods to detect Neanderthal introgressed regions in present-day human genomes. In 2014, two large-scale genomic studies^27,28 generated the first genomic maps of Neanderthal introgression in present-day Europeans and East Asians. These studies, which implemented two different computational approaches to identify introgressed genomic regions (S* and the Conditional Random Field, CRF), showed that each Eurasian genome harbors substantial amounts of Neanderthal sequence, allowing reconstruction of up to 35% of the Neanderthal genome from present-day modern human genomes. The sequencing of the Vindija Neanderthal genome, closer to the introgressing Neanderthal population, enabled identification of an additional ~10% of Neanderthal sequence from present-day Eurasians¹⁶.

Several studies have shown that the proportion of Neanderthal ancestry varies across present-day populations (Figure 3A). For instance, East Asians have on average 20% more Neanderthal introgressed sequence compared to Europeans^27–29. Whether such a signal might be attributable to multiple admixture events with Neanderthals after their divergence from Europeans (Figure 2), negative selection against deleterious Neanderthal alleles²⁷, or dilution of Neanderthal ancestry in Europeans by unadmixed populations ³⁰ has been debated^{27,28,31–33}. However, using a novel probabilistic method, called IBDmix, for identifying archaic introgressed sequences that does not rely on an unadmixed modern human reference population, a study found that levels of Neanderthal ancestry among Eurasian populations appear more uniform than previously reported³⁴.

Figure 3. The geographic and genomic landscape of Neanderthal introgression.

(A) Proportion of Neanderthal ancestry in geographically diverse populations. Reproduced with permission from Prüfer et al. (2017) Science¹⁶. (B) Neanderthal introgressed genomic regions are depicted by colored ticks throughout the genomes of East Asians (red) and Europeans (blue). Introgressed deserts are large genomic regions (≥10 Mbp) depleted of Neanderthal introgression. Grey regions denote genome assembly gaps, and black ovals indicate the approximate position of each centromere. Reproduced with permission from Vernot & Akey (2014) Science²⁸. (C) Frequencies of Neanderthal introgressed haplotypes in East Asians (top), Europeans (middle), and South Asians (bottom). Positive selection after admixture with Neanderthals likely drove some of these haplotypes to high frequency. Gray dashed lines mark the 99th percentile. Haplotypes above the line are considered strong candidates for adaptive introgression. Large red dots indicate haplotypes with a significant phenotypic association. Reproduced with permission from Gittelman et al. (2016) Current Biology⁵².

Other approaches^35,36 focused on inferring the ancestral recombination graph (ARG) (the full sequence of gene genealogies along modern human and archaic hominin genomes), allowed an even more extensive inspection of the history of introgression events. For example, analyses using ARGweaver-D³⁵ (an extension of ARGweaver³⁷) found that ~3% of the Neanderthal genome introgressed from early modern humans between 200 and 300 kya, while ~1% of the Denisovan genome introgressed from an unknown “super-archaic” lineage more than 225 kya. Additionally, 15% of these super-archaic regions were subsequently passed from Denisovans into present-day humans in Oceania³⁵. Both ARG inference^36,38 and other recently developed methods ^39,40 also support introgression from a (possibly different) super-archaic lineage into the ancestors of sub-Saharan Africans. By mapping Neanderthal introgressed haplotypes throughout the genomes of modern humans, these and other studies (reviewed in detail in Racimo et al. ⁴¹ and Ahlquist et al.⁴²) contributed to our understanding of Neanderthal introgression across geographically diverse human populations and to refining models of human population history at both recent and deep time scales. A schematic of inferred introgression events between modern humans and archaic hominin groups is shown in Figure 2.

The Fate of Neanderthal Introgressed DNA

In the past few years, researchers gravitated toward studying the evolutionary forces that shaped the genomic distribution of Neanderthal introgressed DNA, and how introgression might have contributed to our biology by conferring increased risk of certain diseases or facilitating human adaptation to certain environments.

While most Neanderthal alleles are thought to have neither benefit nor cost (i.e., neutral variants), there is growing evidence supporting the effects of both negative and positive natural selection on Neanderthal introgressed DNA. One of the first lines of evidence comes from the first two maps of Neanderthal introgression into modern human genomes^27,28, where certain large regions – termed introgression deserts – were identified as depleted of Neanderthal ancestry in any modern human genome (i.e., with Neanderthal ancestry less than 0.1%²⁷, though the largest deserts are robust to the choice of this threshold¹¹) (Figure 3B). In the absence of natural selection, introgressed tracts are expected to be distributed at random across the genome, only broken up and replaced through recombination and genetic drift. Thus, the presence of a 17 Mbp long Neanderthal desert on chromosome 7²⁸, encompassing FOXP2, a gene involved in speech and language⁴³, and the 5-fold depletion of Neanderthal ancestry observed on the X chromosome²⁷ suggest that negative selection might have removed Neanderthal introgressed alleles from these regions. In theory, the probability of loss of a low-frequency haplotype increases with decreasing population size, thus autosomal introgression deserts could be partly attributed to intense bottlenecks in modern humans soon after admixture with Neanderthals²⁷. Interestingly, Denisovan introgression deserts in the genomes of present-day Oceanic individuals overlap with Neanderthal introgression deserts identified in the genomes of Eurasians, providing further evidence for selection acting on these genomic regions in geographically diverse populations¹¹.

The depletion of Neanderthal (and Denisovan) ancestry on the X chromosome garnered attention particularly in light of well-established theoretical and empirical results on speciation and hybridization. Prior work demonstrated that hybrid incompatibilities known as Dobzhansky-Muller incompatibilities (DMIs) preferentially accumulate on the X chromosome, and these incompatible alleles tend to have mild, recessive effects that are exposed as hemizygous in the heterogametic sex (i.e. XY males in humans), hence reducing the frequency of introgression on the X relative to the autosomes (see Masly and Presgraves⁴⁴). In addition, it is thought that in male heterogametic species "incompatibilities causing hybrid male sterility accumulate more quickly than those causing hybrid inviability or hybrid female sterility" ⁴⁴. Both brain-¹¹ and testis-expressed genes, especially those involved in meiosis^27,45,46, are depleted for Neanderthal ancestry, concordant with this expectation of DMIs accumulating on the X. However, it is unlikely that DMIs alone could have shaped most of the landscape of depletions in Neanderthal ancestry, as far too many DMIs would be required to sufficiently explain this landscape^32,47,48. Alternatively, the increased efficacy of negative selection on the X chromosome due to exposure of partially recessive deleterious alleles in males could have contributed to this same depletion³². Others have hypothesized that the substantial depletion of Neanderthal ancestry on the X chromosome could be due to sex bias (i.e. differences in the proportion of females and males) in gene flow between Neanderthals and modern human contemporaries, perhaps as much as a three to one bias towards Neanderthal males mating with modern human females³². These explanations are not mutually exclusive, though, as it is likely that several mechanisms might have played a role in shaping the genomic distribution of Neanderthal introgression seen today.

One consequence of negative selection acting against Neanderthal introgressed alleles is that Neanderthal ancestry proportion should decrease over time. In line with this expectation, one early study ⁴⁹ found that the proportion of Neanderthal ancestry in ancient modern humans sampled across Eurasia between 7 and 45 kya decreased linearly from 3–6% to around 2% through time, implying long-term, continuous selection acting against deleterious Neanderthal introgressed variants. However, simulation studies^32,47 and a re-analysis of ancient DNA data⁵⁰ indicate that Neanderthal ancestry decreased rapidly during the first ~10 generations after admixture, then stabilized to levels close to those observed in present-day humans, rather than steadily declining.

Two studies^32,47 used models of background selection (the loss of neutral genetic variation caused by genetic linkage to loci subject to negative selection) to explain these widespread signals of deleterious Neanderthal introgressed variation in modern humans. Neanderthals had a substantially smaller effective population size than any modern human population^15,16; as a consequence, weakly deleterious "nearly-neutral" mutations occurring in Neanderthals would have a decreased probability of loss relative to modern humans, as this type of evolutionary dynamic would be governed by genetic drift rather than negative selection⁵¹. However, once exposed to the higher effective population size of modern humans (following admixture events), the evolution of these mutations would then be governed by negative selection. Though there may only be thousands of these cryptically deleterious Neanderthal introgressed alleles in each modern human individual, the genetic linkage driving background selection extends the effects of selection against these Neanderthal alleles to a sizable proportion of the genome⁴⁷. Accordingly, both studies^32,47 found that cryptically deleterious Neanderthal introgressed alleles were key to explaining observed genome-wide patterns of depletion of Neanderthal introgression.

Further work focused on understanding the impact of negative selection on Neanderthal introgressed variation across different genomic regions. Neanderthal ancestry appears to be depleted in evolutionarily constrained regions of the genome, such as regions of high gene density, high evolutionary conservation, and low values of the "B" statistic^{11,27,45,48,52,53}. Over the past 8,000 years, non-synonymous Neanderthal alleles decreased in frequency in modern humans, although Neanderthal alleles altering gene expression levels do not show as strong a depletion, suggesting that the Neanderthal alleles that remain in our genomes tend to alter gene regulation rather than protein sequence⁵³. Among regulatory elements, untranslated regions of mRNAs (UTRs) and promoters are more depleted for Neanderthal ancestry than enhancers⁵⁰. Furthermore, tissue-specific enhancers – especially those specific to fat cells, mesenchymal cells, and T cells – show the highest levels of Neanderthal ancestry, whereas enhancers active in multiple tissues (pleiotropic enhancers) and those specific to muscle and brain are the most depleted of Neanderthal ancestry^54,55. Together, these results suggest that Neanderthal enhancers with targeted effects, especially those involving immunity and metabolism, were the best-tolerated regulatory change, whereas changes to both widespread expression and brain- or muscle-specific expression were least-tolerated.

Neanderthal Introgression and Gene Expression

Several studies have investigated the effects of Neanderthal introgressed variation on gene expression^52,53,56 using recent, massive studies of gene expression variation in humans like the Geuvadis consortium ⁵⁷ and the Genotype-Tissue Expression (GTEx) project⁵⁸. An allele-specific expression analysis of the GTEx dataset ⁵⁶ found that transcripts containing Neanderthal alleles show widespread differences in expression relative to their modern human counterparts, and an approximately equal proportion of genes show Neanderthal allele upregulation versus downregulation. However, brain- and testis-expressed transcripts containing Neanderthal alleles appear to be consistently downregulated⁵⁶, in line with previous studies showing the depletion of Neanderthal ancestry in brain ¹¹ and testis genes^27,45,46. This pattern could be explained by a type of "functional epistasis" between Neanderthal cis-regulatory elements (regulatory genomic features nearby the regulated gene) and the modern human trans-regulatory background (e.g. transcription factors or regulatory RNAs encoded elsewhere in the genome, that interact with the cis-regulatory element). Under the functional epistasis model, these cis- and trans-elements diverged between Neanderthals and modern humans, but remained co-adapted within each lineage, leading to functional incompatibilities when modern human trans-elements are mixed with Neanderthal cis-elements via introgression – for instance, a Neanderthal enhancer whose transcription factor binding site has diverged from that of modern humans would lead to downregulation of the Neanderthal transcript⁵⁶. Although functional epistasis does not necessarily affect fitness, the depletion of Neanderthal ancestry in brain- and testis-expressed genes^27,46, along with Neanderthal allele-specific downregulation in the same tissues⁵⁶, suggest that some of the purported DMIs between Neanderthals and modern humans might have taken this form.

While the consistent downregulation of certain Neanderthal transcripts points to negative selection against Neanderthal introgression, more than a quarter of the putative adaptive Neanderthal introgressed haplotypes tested harbor expression quantitative trait loci (eQTLs)⁵², some of which result in a population-specific response to viral stimuli⁵⁹, thus supporting a role for gene expression changes in positive selection for Neanderthal introgression. Further work also suggests that archaic trans-eQTLs may modulate expression of genes within introgression deserts⁶⁰, greatly extending the potential regulatory reach of archaic introgressed variation. More broadly, gene flow with archaic hominins not only influences transcriptional variation among individuals by introducing derived archaic variants, but also by reintroducing ancestral variants that were lost in non-African populations⁶¹. Interestingly, these reintroduced ancestral alleles are more likely to occur on introgressed haplotypes with regulatory effects than derived archaic alleles, suggesting less negative selection against or possibly some positive selection for reintroduced ancestral alleles⁶¹.

Neanderthal Introgression and Human Adaptation

The legacy of Neanderthals in modern humans is not strictly confined to negative effects: introgression introducing variants that spread in the recipient population due to positive selection, a phenomenon known as adaptive introgression (reviewed in Martin and Jiggins⁶²), is thought to have occurred between Neanderthals and modern humans. In general, strong positive selection increases allele frequency over time and may lead to fixation of the allele, simultaneously pulling alleles linked on the same haplotype to high frequency. Thus, positive selection acting on an adaptive introgressed tract of DNA will lead to a high-frequency haplotype of introgressed alleles (Figure 3C). In the past few years, several studies identified signatures of positive selection on Neanderthal introgressed haplotypes overlapping genes involved in brain development, neuronal function, both adaptive and innate immunity, lipid metabolism, skin and hair pigmentation, and the musculoskeletal system. Here we focus our discussion on strongly supported examples of Neanderthal adaptive introgression belonging to the broad functional categories of skin and hair pigmentation, metabolism, and immunity.

Skin and Hair Pigmentation

One of the most well-supported candidates of adaptive introgression is a large Neanderthal introgressed haplotype (~50 kbp) found at high frequency (70%) in Europeans encompassing BNC2, a gene on chromosome 9^{12,27,28,45,52,63–65}. BNC2 encodes a zinc finger protein that is expressed in keratinocytes and has been associated with skin pigmentation⁶⁶ and freckling⁶⁷ in Europeans. Another strong candidate of adaptive introgression spans the POU2F3 gene on chromosome 11. This gene encodes a transcription factor expressed in the epidermis and is responsible for the mediation of keratinocyte proliferation and differentiation⁶⁸. While nearly absent in Europeans, the Neanderthal haplotype encompassing this gene is present at high frequency in East Asians (~60%)^{12,28,45,52,63}.

Another example of adaptive introgression is a 200 kbp long haplotype on chromosome 3 encompassing the HYAL2 gene involved in the cellular response to UV radiation and changes in skin pigmentation⁶⁹. HYAL2 expression decreases significantly after exposure to UVB irradiation^70,71, disrupting tissue repair processes and ultimately leading to sunburn⁷⁰. The Neanderthal introgressed haplotype is found at high frequency in East Asians (>50%), but absent elsewhere^27,63,72.

A 29.7 kbp Neanderthal introgressed haplotype encompassing the OCA2 gene on chromosome 15 is found at high frequency (~60%) in East Asians, and intermediate frequency (~20–30%) in Europeans, South Asians, and Melanesians^52,63,73. The OCA2 gene encodes a transmembrane protein that impacts skin, hair, and iris pigmentation⁷⁴. Specifically, a Neanderthal introgressed allele⁷⁵ associated with blue iris pigmentation⁷⁶ and blonde and red hair color⁷⁷ was under selection in modern human populations.

Other examples of Neanderthal adaptive introgression include KRT71 and KRT80, two genes in the type II keratin gene cluster on chromosome 12. A 100 kbp archaic haplotype encompassing the gene KRT71 has a frequency of 65% in Europeans, 52% East Asians, and 38% South Asians¹². KRT71 encodes an epithelial keratin protein that is expressed in hair follicle inner root sheaths and contributes to hair pigmentation^78,79. An additional ~18 kbp Neanderthal haplotype overlapping KRT80 is found at high (20–60%) frequency in Near and Remote Oceanians, intermediate frequencies in Europeans and South Asians and low frequencies in East Asians⁷³. KRT80 encodes an epithelial keratin that contributes to the structural integrity of epithelial cells.

Metabolism

A genome-wide association study (GWAS) performed on a large cohort of over 8,000 Mexicans and Latin Americans identified a novel locus on chromosome 17 associated with a ~20% increase in risk for developing type 2 diabetes⁸⁰. The risk haplotype carries five single nucleotide variants (SNPs), including 4 missense SNPs, in the gene SLC16A11, thought to play a role in hepatic lipid metabolism⁸⁰. This five-SNP haplotype is present at ~50% frequency in Mexican populations, at lower frequency in East Asians and nearly absent elsewhere. Comparison with the Altai Neanderthal genome showed that all 5 SNPs are homozygous in the archaic sequence. The inferred age of this haplotype (~799 kya), combined with its length and geographical distribution, suggest that it introgressed into modern humans from Neanderthals.

Two other major metabolic candidates have recently emerged as well. One study identified a Neanderthal adaptive introgressed tract in Europeans spanning the TSHR gene on chromosome 14⁸¹. TSHR encodes the thyroid stimulating hormone receptor which binds a signaling molecule known as thyrotropin involved in a number of physiological pathways, including the growth of the thyroid gland and thyroid-related metabolic processes⁸². Mutations in TSHR result in disease phenotypes such as Graves’ disease and congenital hyper- and hypo-thyroidism⁸³. Several studies also demonstrated the important role of TSHR in adipocyte differentiation and lipolysis^84,85.

A ~100 kbp Neanderthal tract has been identified spanning the TBC1D1 gene on chromosome 4, and segregates at intermediate frequency (20–40%) in South and East Asians, and high frequency (~30–90%) in Oceanic populations⁷³. TBC1D1, and its paralog TBC1D4, are proteins downstream of the insulin-stimulated Akt kinase cascade that modulate glucose uptake by regulating vesicular trafficking of the glucose transporter GLUT4⁸⁶. Mutations in TBC1D1 in both humans and mice are associated with obesity^87–89. Taken together, these findings suggest that Neanderthal introgression played a key role in both energy regulation and lipid-related pathways.

Immunity

One of the earliest and most replicated signals of Neanderthal adaptive introgression spans the cluster of three OAS (OAS1, 2, and 3) genes on chromosome 12^{27,52,63,65,90,91}. The OAS genes are a core component of the innate immune system, particularly with respect to antiviral response. Interferons induce these OAS genes and activate RNase L, leading to degradation of both viral and cellular RNAs, thereby inhibiting viral protein synthesis⁹². An adaptive Neanderthal OAS haplotype is observed at ~30% frequency in European and South Asian populations, and at 20% frequency in populations in East Asian and the Americas^52,91.

Signals of adaptive introgression have also been found spanning the TLR1/6/10 immune gene cluster on chromosome 4^{27,52,63,93,94}. The TLR genes (Toll-like receptors) are also involved in innate immunity, though are part of the antibacterial and antifungal response. These receptors recognize peptides or sugars on the membrane of bacteria or cell wall of fungi and trigger an intracellular cascade leading to a pro-inflammatory response⁹⁵. Dannemann et al.⁹³ identified one Neanderthal haplotype shared by Native Americans, Europeans, and Asians, as well as one Neanderthal and one Denisovan haplotype unique to Asians. Beyond these two major gene clusters, adaptive introgression signals have been detected overlapping genes involved in autoimmune disorders (GMEB2^27,96), antibacterial immune response (GBP4/7¹¹), adaptive immunity (CCR9/CXCR6^12,52; and IGHA1/IGHG1/IGHG3/IGHG412), and immune response to the influenza A virus (PNMA1/MIDEAS^59,96).

On a much larger scale, proteins interacting with RNA viruses (virus-interacting proteins, VIPs) are enriched for Neanderthal ancestry in Europeans⁹⁷, suggesting that Neanderthal variants might have played a role in antiviral immune response in these populations. A subset of VIPs that interact with coronaviruses show signals of positive selection within the last 25,000 years in East Asians, suggesting that an ancient coronavirus epidemic in the ancestors of East Asians might have driven the evolution of these proteins⁹⁸. Additionally, genetic factors underlying both severity of and protection against COVID-19 have been mapped near the CCR9⁹⁹ and OAS1 genes^100,101, respectively, both of which are Neanderthal adaptive introgression candidates, highlighting past and current impacts of Neanderthal introgression on human survival against life-threatening pathogens.

It is important to note that current methods to detect adaptive introgression operate solely on genetic data, and as such cannot provide direct insights into the phenotypic effects of these archaic variants. Inferences about the effects of these variants rely on knowledge of the functions of overlapping genes and/or statistical associations with known phenotypes, meaning that we rarely know the strength or even direction of the effect of archaic variants, let alone which tissues or developmental stages are affected. Consequently, the future of research into archaic introgressed variation will rely heavily on functional studies of these candidates of adaptive introgression.

The Contribution of Neanderthal Introgression to Human Phenotypic Variation

While signatures of natural selection on our genomes can provide insights into how archaic DNA affected human fitness over time, the massive scale of modern genomics and the depth and diversity of association studies contributed to revealing connections between archaic DNA and a variety of human traits. GWAS compare the genotypic state at millions of sites throughout the genome to a phenotype of interest, looking for sites where the genotype matches the phenotype more often than expected at random. In a complementary manner, phenome-wide association studies (PheWAS) scan many phenotypes against the genotypes at individual genomic loci, looking for phenotypes that match genotypes more often than expected at random. Applying both of these frameworks, one study¹⁰² compared 46 groups of phenotypes from electronic medical records with genotypes for over 11,000 patients of European ancestry in the Electronic Medical Records and Genomics (eMERGE) network and found that Neanderthal variants are associated with skin lesion disorders, obesity, blood disorders, tobacco use, mood disorders, and depression. Another study¹⁰³ extended these same approaches to 136 phenotypes across over 112,000 individuals of European ancestry included in the UK Biobank and found associations between Neanderthal variants and hair and skin pigmentation, tanning, sunburn incidence, mood, smoking, height, chronotype, and heart rate. Interestingly, several of these skin-related phenotypes are driven by Neanderthal haplotypes overlapping the BNC2 gene, a strong candidate for adaptive introgression (see subsection “Skin and Hair Pigmentation”).

Much as different populations of modern humans inherited different regions of the genome from Neanderthals, there is little overlap between Neanderthal-associated traits and disease risks in Europeans and East Asians¹⁰⁴. Despite this, common phenotypic themes emerge across these two populations just as they did for adaptive introgression: Neanderthal introgressed variants have effects on skin and hair pigmentation, metabolism, and the immune system. In this context, recent research into the host genomics of COVID-19 identified Neanderthal variants contributing to both susceptibility to⁹⁹ and protection against COVID-19 in Eurasian populations^100,101. Incidentally, the COVID-19 susceptibility haplotype⁹⁹ and the protection haplotype¹⁰¹ are near the chemokine receptor genes CCR9/CXCR6 and the OAS gene cluster, respectively, both of which are strong candidates for adaptive Neanderthal introgression (see subsection “Immunity”).

GWAS also allows for estimation of the proportion of a traiťs variation attributable to a given genetic locus (usually termed ‘SNP heritability’, or SNP-h²). Recent work¹⁰⁵ used heritability estimates to ask the question of which traits have disproportionate contributions by DNA inherited from Neanderthals. While most traits showed a depletion of Neanderthal-associated heritability, several trait groups such as autoimmune diseases, white blood cell count, balding, sunburn, age at menopause, respiratory and bone traits, and chronotype, showed an enrichment of heritability at older Neanderthal variants. The older a Neanderthal variant is, the more likely it was common among Neanderthal populations and thus tolerated rather than eliminated by selection and therefore no longer contributing to trait heritability. Neanderthal variants involved in each of these traits – except for autoimmune diseases and white blood cell count –have the same direction of effect¹⁰⁵, suggesting the potential for polygenic selection on Neanderthal introgressed variants.

Polygenic Adaptation and Neanderthal Introgression

GWAS and PheWAS are very powerful tools for mapping the genetic basis of many traits, but suffer from multiple important limitations¹⁰⁶: First, individual genetic loci contributing to the phenotype need to have a fairly large effect in order to be detected; second,alleles contributing to the phenotype must be sufficiently common in the population for their effect to be detected (or the number of samples in the GWAS must be large enough to detect the allele enough times to reliably assess its effect); and third, the effects of different genetic loci are often assumed to be independent. By combining the ever-increasing knowledge of the complex networks interconnecting genes and their products with the genotypes and phenotypes used in GWAS and PheWAS, recent methods focusing on polygenic trait evolution promise to exceed these limitations on detectability. These approaches were recently applied to study the contribution of archaic introgression to polygenic traits¹⁰⁷, finding an enrichment of Neanderthal introgression in multiple immune and possibly brain-related pathways in both Europeans and East Asians, as well as metabolic and olfactory pathways in Europeans, and a curious (possibly Denisovan) signal of malaria resistance in Melanesians. The immune pathways identified also include some of the most replicated signals of adaptive introgression, such as the TLR and OAS gene clusters and STAT2^27,108.

Assaying the Functional Consequences of Neanderthal Introgression

Although many phenotypic associations have been identified for Neanderthal genetic variation, functional studies are needed to understand how they contribute to human biology. Early work^59,109 focused on the role of population-specific variation in immune response to bacterial and viral stimuli, respectively. Immune cells from African American and European American volunteers in these studies were exposed to Listeria or Salmonella¹⁰⁹ or a variety of viral ligands (proteins that bind immune cell receptors like the TLRs to trigger an immune response) and even influenza A virus⁵⁹, and gene expression was quantified after exposure. Both studies showed thousands of differences in gene expression between ancestry groups in response to immune stimulus, and a couple hundred of these differences were attributable to Neanderthal introgressed variation. Of these Neanderthal-associated response expression quantitative trait loci (reQTL, variants that change the expression of a gene in response to a stimulus), variants affecting expression of the TLR1 and PNMA1 genes also showed signatures of positive selection, corroborating their previous identification as subjects of adaptive introgression.

While approaches like GWAS/PheWAS and eQTL mapping identify statistical associations between allelic states and phenotypes, linkage disequilibrium (LD) between variants on a haplotype make it difficult (if not impossible) to identify which variant is causal. Archaic introgressed tracts intrinsically have greater LD than background (having been exposed to fewer generations of recombination, thus exacerbating this limitation. In vitro functional assays that test the effects of individual archaic alleles, such as Massively Parallel Reporter Assays (MPRAs) and lower throughput CRISPR/Cas9-based genome engineering, serve as a natural complement to association studies in identifying causal variants. Following the MPRA approach, two recent studies^110,111 identified a total of 25 Neanderthal alleles with experimentally validated regulatory effects on genes involved in immunity, skin and hair pigmentation, and insulin secretion. In addition, Jagoda et al.⁹⁶ looked at the regulatory effects of over 5,000 high-frequency Neanderthal introgressed variants using a combination of MPRAs, CRISPR knockouts, and chromatin conformation sequencing (Hi-C) in an immune cell line. They identified ~300 expression-modulating variants (emVars) which show an enrichment for chromatin marks indicating active promoters and strong enhancers, consistent with previous findings^50,54,55. They found that these emVars altered the binding sequences of transcription factors involved in innate immunity and thereby changed expression of downstream genes in innate immunity pathways. By combining reporter assays, transcription factor binding analysis, eQTL overlaps, and Hi-C interaction signals, they identified specific variants at the TLR cluster, GMEB2, the OAS cluster, and the IL23A/STAT2/PAN2 locus as putative targets of positive selection, all of which have been previously identified as candidates for adaptive introgression. Using CRISPR knockouts, they confirmed that three Neanderthal emVars regulate transcription factors with known roles in immune response. This represents the closest we have come to directly connecting Neanderthal variants with modern human phenotypic effects, and establishes a strong precedent for future work as functional assays and CRISPR become commonplace.

Outlook

More than a decade ago, the sequencing of the first Neanderthal genomes laid the foundation for some of the biggest revelations in human evolutionary genomics. The desire to understand our evolutionary relationship with our closest extinct relatives has driven major developments in the study of population history and archaic introgression. Though substantial progress has been made in recent years, much remains to be learned about the genetic and phenotypic legacy of Neanderthals in present-day humans. Over the next decade, research will likely shift into several different directions:

Investigating introgression from other archaic hominins

Although much research has examined the impacts of Neanderthal introgression, the study of Denisovan introgression remains in its infancy. One of the strongest and most fascinating cases of adaptive introgression, the EPAS1 haplotype conferring high-altitude adaptation in Tibetans, comes from Denisovans^112,113. Major genomics and molecular phenotype datasets like the 1000 Genomes Project and GTEx are largely focused on populations of European ancestry Eurasians to the exclusion of populations in Island Southeast Asia and Oceania where Denisovan introgression is prevalent, limiting our understanding of the fitness and functional consequences of Denisovan ancestry. Large-scale studies have only begun to address these gaps in terms of genomic representation^11,45,73, gene expression variation^114,115, and phenotypic associations¹¹⁶. A more comprehensive characterization of the full spectrum of human genetic and phenotypic variation in Southeast Asia and Oceania is necessary to address these major gaps in representation and to broaden our understanding of the complex history of human populations.

Genetic evidence suggests that admixture events with archaic hominins beyond Neanderthals and Denisovans did occur^{35,36,39,40,117,118}, though reliably identifying these introgressed haplotypes has proven quite difficult^73,119. To study these “ghost” super-archaic sequences and their impacts on modern humans, the field will need more sophisticated computational frameworks and more comprehensive genetic sampling of both archaic and modern human populations.

Identifying the phenotypic effects of individual introgressed variants

Omics methods developed over the past decade have provided a trove of candidate loci. However, identifying causal variants underlying phenotypic changes from these candidate loci has proven a difficult task. As we are entering the heyday of MPRAs and genome engineering with tools like the CRISPR/Cas9 system, causal variant validation is now within reach. Genome editing will likely be combined with repositories of induced pluripotent stem cell (iPSC) lines that harbor a diversity of Neanderthal introgressed sequence in its natural context¹²⁰, serving as an experimental template for various functional assays. These same iPSCs can also serve as a starting point for the growth of organoids, providing insights into previously unascertainable developmental and tissue-level phenotypes¹²¹. Notably, recent work¹²¹ used CRISPR on iPSCs to generate cortical organoids harboring a non-synonymous substitution in the NOVA1 gene nearly fixed between Neanderthals and modern humans (though not a product of introgression) showing marked phenotypic differences compared to the modern human derived organoids. However, concerns have been raised that the phenotypic consequences observed in such organoids might be caused by unintended off-target effects of CRISPR genome editing technology¹²². In the future, carefully designed validation studies combining in vitro and in vivo model systems¹²³, MPRAs, CRISPR-based genome engineering of 3D organoid models will be critical to winnowing the long lists of candidates of adaptive introgression and provide us with more detailed insights into the mechanisms and effects of archaic introgressed variation.

As Neanderthals are extinct, our knowledge of their phenotypic variation comes either from the fossil record or from genomic predictions^124,125. Studies integrating datasets across different disciplines will open new avenues to investigate the evolution of human traits (see BOX 2 for an example of such approaches).

BOX 2. Neanderthal Introgression and Brain Shape.

Comparisons of living primates and modern humans informed us on patterns of ontogeny and endocranial development that are shared among hominoids or unique to humans^151,152. To better discern differences between modern humans and Neanderthals, however, we must turn to analyses of fossilized endocranial remains. Though humans and Neanderthals exhibit nearly identical brain size, a study by Gunz and colleagues¹³¹ identified a stage of globularization in postnatal brain development that appears to be absent in the Neanderthal lineage. By using reconstructions of endocasts at progressing stages of development, this study found that while humans and Neanderthals exhibited similar brain volumes at birth, only Neanderthal endocrania maintained an elongated shape in later postnatal stages and into adulthood. The addition of a neonate skull found at Mesmaiskaya cave and virtual reconstructive methods using geographic morphometrics corroborated these results¹⁵³.

The integration of paleoanthropological, genetic and modern neurological approaches further illuminated our understanding of hominin endocranial evolution. In an expansion of earlier work, Gunz and colleagues¹⁵⁴ used an interdisciplinary approach to explore the molecular basis underlying cranial globularity and understand how Neanderthal introgression may contribute to endocranial shape in modern humans. Computed tomography (CT) scans from modern human and Neanderthal crania were used to develop a summary metric for endocranial globularity, which was then applied to magnetic resonance imaging (MRI) scans of thousands of human adults. Association analyses using these MRI scans, genotype data and genome-wide maps of Neanderthal introgression, identified Neanderthal alleles on chromosomes 1 and 18 associated with reduced endocranial globularity in modern humans. These Neanderthal alleles include eQTLs that affect the expression of two genes, UBR4, which encodes a ligase involved in neuronal development in the neocortex, and PHLPP1 encoding a regulator for a signaling pathway involved in brain growth and myelination, providing new insights into the neuroanatomical changes that might underlie modern human endocranial shape.

BOX 2 Figure.

CT scans of Neanderthal (left) and modern human (right) crania exhibiting the elongated cranial shape characteristic of Neanderthals and the globular shape characteristic of modern humans.

Moving beyond the realm of single nucleotide variation

Structural variants, large genetic differences typically >50–100 bp¹²⁶, comprise the majority of varying nucleotides among human genomes¹²⁷. Yet studies on archaic introgression have focused on single nucleotide variation and short indels¹²⁸. Recent work has unveiled three large adaptive introgressed structural variants, two of Neanderthal origin^129,130 and one of Denisovan origin¹²⁹. The advent of third-generation ("long-read") sequencing technology promises to massively expand the catalog of human structural variants, and to spur a new wave of methods development to address burgeoning questions of the contribution of introgressed structural variation to modern human phenotypes.