Humans, Neanderthals, and Denisovans independently adapted to a wide range of geographic environments and their associated food odors. Using ancient DNA sequences, we explored the in vitro function of thirty odorant receptor genes in the genus Homo. Our extinct relatives had highly conserved olfactory receptor sequence, but humans did not. Variations in odorant receptor protein sequence and structure may have produced variation in odor detection and perception. Variants led to minimal changes in specificity but had more influence on functional sensitivity. The few Neanderthal variants disturbed function, whereas Denisovan variants increased sensitivity to sweet and sulfur odors. Geographic adaptations may have produced greater functional variation in our lineage, increasing our olfactory repertoire and expanding our adaptive capacity. Our survey of olfactory genes and odorant receptors suggests that our genus has a shared repertoire with possible local ecological adaptations.
Two prior studies examined genetic variation in contemporary humans, Altai Neanderthal, and Denisovan for OR7D4and OR2M3.Both studies identified novel Denisovan variants and subjected them to functional testing. Our work exponentially increases knowledge on functional differences in two novel OR variants to 29 additional open reading frames ( Table S1 ) for ORs with known human receptor-odor relationships and well-established agonist responses.We also more than double the number of ancient samples by including two additional Neanderthal genomes (Chagyrskaya, Vindija) and one ancient human (Ust’-Ishim) who lived in the same Altai montane locality ( Table S2 ). The contemporary human sample used is 1000 Genomes ( Table S3 ).There is more variation in the contemporary human ORs studied compared to extinct lineages. Functional testing indicates that novel variants found in extinct lineages alter sensitivity more than specificity. Our wide survey of the OR family suggests that the genetic and functional olfactory repertoires of extinct lineages were highly over-lapping and that increased genetic variation in the human repertoire occurred after our split with other migratory members of our genus.
The human genus Homo underwent the most radical ecological niche expansion of all primates when migrating out of Africa and adapting to diverse global environments.Denisovans and Neanderthals ancestors dispersed from Africa earlier than contemporary humans (∼750,000 versus 65,000 years ago)and separated from each other ∼300,000 years ago Figure S1 ). Neanderthals were geographically wide-ranging from Western Europe to the Middle East and Asia. Although Denisovans have been primarily found in Siberia,the Tibetan plateau,and possibly beyond Wallace’s Line,their genetic signature is found in Asian, Papua New Guinea, Australian, and indigenous American groups.Olfactory stimuli from divergent environments following independent dispersals from Africa may have left traces of variation in Homo OR genes. Whether there is a shared genetic and/or functional olfactory repertoire across the genus Homo or not may provide clues to evolutionary adaptive trends.
Terrestrial animals probe the environment for information about food, mates, and dangerby binding odor molecules to odorant receptors (ORs). Variation in mammalian ORs is linked to dietary niche, habitat, and sociability.In comparison, human ORs are strongly biased toward food odors,which reflects the evolutionary importance of dietary niche to our species. Although evolutionarily recent changes in human OR protein function have been linked to major dietary shifts, such as scavenging, hunting, animal milk consumption, cooking, domestication,what about deeper evolutionary changes?
OR10J5. There were three significant screening responses to lyral, helional and eugenol ( Figure S3 A). The Denisovan response to top odor lyral was lower than that of the human reference ( Figures 4 and 3 ). The G21Rvariant of OR10J5 found in the Denisovan is located at the very end of the N terminal end, just before the start of TM1. The role of this region in OR function is undetermined.
OR10G3. The screening assay revealed significant responses for all seven odors in the Denisovan OR and in the human reference ( Figure S3 A). Denisovan variants had significantly stronger dose responses to vanillin and ethyl vanillin compared to contemporary humans ( Figure 4 and 3 ). Neither of the Denisovan variants (S73G, E197 K) was located in conserved amino acid regions ( Figure 3 ). TM2 is not involved in odorant binding or receptor function, implying that S73Gprobably did not change the receptor response. TM5 (E197 K) forms part of the binding cavity but position 5.36 is located at the very limit of ECL2. K at this position is a rare residue in contemporary human ORs (3.6%), suggesting a functional, adaptive reason for this change. The location of the variant suggests it may be involved in ligand entry.
OR5P3. In the screening assay, Vindija Neanderthal did not have a significant response but the human reference responded to five of the seven odorants ( Figure S3 B). Vindija dose responses, which included higher concentrations of the top three responses for contemporary humans (coumarin and both enantiomers of carvone), did not exceed control ( Figure 3 ). The cell surface expression for Vindija indicated that the OR protein was present at the cell surface, albeit at a slightly lower level than that in contemporary humans. Vindija F159 Lis in the extra-cellular part of TM4 ( Figure 3 ), near 4.53, which is involved in mouse OR trafficking.We observed slightly lower trafficking of the Vindija protein. A similar mutation (S155 A) in human OR1A2 decreases in vitro responses to (S)-(−)-citronellal.If OR5P3 F159 Lis involved in odorant binding, the mutation of this position from phenylalanine to leucine might prevent the π-π stacking interaction between the aromatic residue and coumarin. We conclude that the Vindija protein is not functional and this might be attributable to many potential reasons (a few examples are lack of odorant binding, or fail in activation mechanism, or fail to bind the G protein).
OR2C1. Screening assay responses were strong but not statistically significant ( Figure S3 A). The dose response assay for octanethiol produced a statistically significant response in the Denisovan version of this OR ( Figures 4 and 3 ). The shared C149 Wcorresponds to the conserved W. The W residue is highly conserved in GPCRs but less so in ORs (58%). The location of the novel Denisovan I214 Tin TM5 is below residues involved in canonical ligand binding cavity and it points into the receptor rather than the surface. In addition, prior functional tests for C149 found a similar response.The C149W allele may stop protein function and may have produced octanethiol-specific anosmia.
OR1C1. The only significant response was from Chagyrskaya Neanderthal to androstadienone ( Figure S3 B) and it was weak ( Figure 4 and 3 ). Chagyrskaya Neanderthal OR1C1 variant Y120His part of the highly conserved MAYDRY motif involved in the activation of mammalian ORs ( Figure 3 and S3 B), which might explain why this variant alters function.
OR1A1. All but 2-heptanone induced a response in the screening assay ( Figure S3 A) and the highest responses were for allyl phenylacetate, a honey-like odor (particularly for Denisovan) ( Figures 4 and 3 ). Neither of the two Denisovan variants (V461I, T257 M) was in an amino acid region critical for mammalian OR function nor were they involved in the odorant binding cavity—perhaps explaining their minimal functional impact.
Despite using the same DNA sequence as previous studies of OR2B11and OR6P1,neither responded to any of the ∼350 odors (100 μM) against which they were tested ( Figure S4 )—neither did the extinct lineages. In previous studies, OR2B11and OR6P1responded strongly to coumarin and Anisaldehyde, respectively, at concentrations higher than 100 μM. We found that such concentrations often cause cell toxicity or OR non-specific cell responses.
There were only three Neanderthal genes containing novel variants across all three Neanderthal samples. Their dose responses were not correlated with those of contemporary humans ( Figures 2 A, 2C and 2D ). Only OR1C1 had a detectable response, but it was significantly lower than that of living humans ( Figures 3 A , 3C, 3D, 4 S2 , and S3 ). Despite the higher number of novel OR variants in Denisovans and higher dose responses compared to contemporary humans ( Figures 2 C and 2D), the OR responses for six genes and those for human reference were significantly correlated (R= 0.87) ( Figure 2 B). Sampled Denisovan ORs were less sensitive to odors that contemporary humans perceive as floral but much more sensitive to odors perceived as spicy, balsamic, or unpleasant (e.g., sulfur 4x greater and balsamic 3x greater than in contemporary humans) ( Figure 4 ). Higher dose responses in Denisovan ORs appear to be driven by two amino acid variations in two ORs containing novel variants ( Figures 3 and S3 ).
The transmembrane domains are represented in tubes and numbered from 1 to 7. The location of each of the variants is pointed on the human consensus OR homology model by a colored ring. For each variant, dotted lines connect the variant location to a panel showing the ORs dose-response to odorants that were significantly activating ORs in screening. The xaxis of panels represents the logarithmic transformation of the odorant concentration (M) and the yaxis the normalized luminescence (Luciferase (L) normalized by Renilla (R) luminescence, see STAR Methods ) generated by the OR activation in luciferase assay. The empty vector pCI is added as a negative control. Error bars are SEM (standard error of the mean) of n = 4 replicates. The Homo lineage origin of the OR variant is color coded in the OR homology model and the dose-responses as modern human = orange, Denisova = blue, Neanderthal Vindija = light purple, Neanderthal Chagyrskaya = dark purple. Statistical significance of OR response between of the human and Homo lineage versions is assessed by Extra sum-of-square F test on the dose-response and shown as ∗∗∗p<0.001, ∗∗0.001
(C) Boxplots of in vitro OR responses for all samples showing median, box boundaries (first and third quartiles), and two whiskers (upper whisker extends to the largest value no further than 1.5 inter-quartile range from the third quartile; lower whisker extends to the smallest value at most 1.5 inter-quartile range of first quartile, outliers identified with red asterisk), (D) OR response by lineage and gene. Each OR is represented by a different color and each point represents the natural log of the response to an odorant. Dotted lines correspond to the linear regression for the entire set of ORs responses for a given lineage. The corresponding equation and R2 values are shown on the regression plot. Raw and analyzed data are in DataS1.
(A) Comparison of human (xaxis) and Neanderthal (yaxis) OR responses to odorants in vitro, (B) Comparison of human (xaxis) and Denisovan (yaxis) OR responses to odorants in vitro.
Of the 29 new genes examined and additional genomes examined for 7D4, we identified 11 genes containing 14 novel variants that were subjected to functional testing. Because gene function in not reliably predictable for ORs from sequence data,we directly measured the functional responses of ORs containing novel variants. Each OR protein, expressed in a cell line, was screened against seven odorants previously identified in the literature as evoking responses: OR1A1,OR1C1,OR2C1,OR10J5,OR5P3,and OR10G3.Dose response assays for the top screening responses included seven concentrations of the odors delivered separately. To compare dose responses, we computed an Activity Index based on the potency and efficacy of an OR/odorant pair ( Figure 4 ). We distinguish between genes and proteins by prefixing proteins with OR and eliminating the prefix when referring to genes (e.g., 1A1 for the gene, OR1A1 for the protein). Functional testing included 784 OR/odorant pairs with at least 1 concentration in total—42 odorants on 2 ORs (living and extinct) for a total of 84 OR/odorant pairs ( Figure S3 ), 2 ORs against 350 odorants for a total of 700 OR/odorant pairs ( Figure S4 ).
Amino acid sequences for the sampled ORs were the basis of a cladogram, which visualizes relationships across samples ( Figure S2 ). Extinct lineages formed a clade with Vindija Neanderthal being the most distinct, which was unexpected because genome-wide studies have indicated that Vindija is most genetically similar to Chagyrskaya Neanderthal.The clade of extinct lineages (Neanderthal and Denisova) was closest to the ancient human Ust’-Ishim. Ancient samples were closest to living East Asians and then living South Asians—both groups harbor genetic signatures of introgression with Neanderthals and Denisovans.
Extinct lineages and Ust’-Ishim had fewer DNA and protein variants compared to 1000 Genomes ( Figure 1 and Table S6 ). Nucleotide variants in extinct lineages averaged 0.19% across the 17 genes containing variants—0.11% across all 30 genes, including ones with no differences from the human reference. In comparison, nucleotide variation in 1000 Genomes averaged 0.82% across 30 genes (all genes contained variants). The fixation index (Fst) measures genetic variance because of population structure (typically weighted by population size) and ranges from 0 (no differentiation) to 1 (complete differentiation).The Fst values by gene for 1000 Genomes populations are lower than other large-bodied mammals with wide geographic dispersal ( Table S7 ).The 1000 Genomes populations are not highly differentiated across the sampled 30 ORs—an Fst mean of 4%. In contrast, the genus Homo (1000 Genomes and ancient samples) has an Fst mean of 11%, the lower end of possible significant differentiation by population structure. Of the 13 genes with Fst >12%, only two had novel variants in extinct populations, which suggests that the small number of novel variants are not highly influential in differentiating populations ( Table S7 ).
The percentage of variation was calculated by taking the total variant count per gene divided by total basepairs per gene for each population. We used raw counts for ancient populations and raw counts for the consensus sequence of each gene for each of 26 groups in 1000 Genomes. We used the consensus sequence for each of the 1000 Genomes groups rather than total raw counts because the sample sizes are divergent. By using the consensus raw count, we compare percentage variation of one gene to another gene.
Of the 30 genes examined (29 new genes and 7D4 for the newly added genomes), two ancient genes were identical to the human reference (4Q3, 8B3), twenty contained variants already observed in 1000 Genomes ( Table S4 : 2016 VCF Shared Variants), and only 11 genes contained a total of 14 variants not found in 1000 Genomes ( Table S5 ). Denisovan had nine novel variants (of which two were synonymous) compared to the Neanderthal five (of which 2 were synonymous). No novel variants were observed in the ancient human Ust’-Ishim. The primary pattern across sampled genes suggests a pattern of shared variation (i.e., variants were introduced before global divergence of Homo).
Discussion
4 Saraiva L.R.
Riveros-McKay F.
Mezzavilla M.
Abou-Moussa E.H.
Arayata C.J.
Makhlouf M.
Trimmer C.
Ibarra-Soria X.
Khan M.
Van Gerven L.
et al. A transcriptomic atlas of mammalian olfactory mucosae reveals an evolutionary influence on food odor detection in humans. 13 Jacobs Z.
Li B.
Shunkov M.V.
Kozlikin M.B.
Bolikhovskaya N.S.
Agadjanian A.K.
Uliyanov V.A.
Vasiliev S.K.
O’Gorman K.
Derevianko A.P.
Roberts R.G. Timing of archaic hominin occupation of Denisova Cave in southern Siberia. 36 Bastir M.
Rosas A.
Gunz P.
Peña-Melian A.
Manzi G.
Harvati K.
Kruszynski R.
Stringer C.
Hublin J.-J. Evolution of the base of the brain in highly encephalized human species. , 37 Weiss T.
Soroka T.
Gorodisky L.
Shushan S.
Snitz K.
Weissgross R.
Furman-Haran E.
Dhollander T.
Sobel N. Human olfaction without apparent olfactory bulbs. We tested 784 OR/odorant pairs on ten novel missense variants, which were located in 8 genes out of 30 (29 new genes and OR7D4 for the new Neanderthal samples). Five were functionally different than contemporary humans (OR1C1, OR2C1, OR5P3, OR10G3, OR10J5), one the same (OR1A1) and two without identifiable ligands (OR2B11, OR6P1). Given the small percentage of genes with variants altering OR function, members of the genus Homo likely shared an olfactory repertoire based on our sample, with Neanderthals and Denisovans smelling the same range of odors we do but having different dose responses to those odors. When OR function was altered by a novel OR, the difference was in sensitivity rather than specificity. Novel Denisovan OR variants (OR1A1, OR2C1, OR10G3, OR10J5) were twice as responsive as human equivalents to odors contemporary humans perceive as spicy, balsamic, and unpleasant ( Figure 4 ), but not to odors perceived as floral. Given the human specialization of olfactory sensory neurons to food-based odors,the variation in sensitivity to food odors in an extinct lineage in our sample is indirect evidence of a similar emphasis. Novel Neanderthal variants were three times less responsive than human ORs, including reduced responses to odors perceived as green, floral, and spicy ( Figure 4 ). There is some correlation in Neanderthal skull morphology that suggests their olfactory bulbs were smaller than contemporary humans,but the link between bulb size and olfactory acuity is unclear.
38 Marlowe F.W.
Berbesque J.C.
Wood B.
Crittenden A.
Porter C.
Mabulla A. Honey, Hadza, hunter-gatherers, and human evolution. 39 Selig K.R.
Silcox M.T. The largest and earliest known sample of dental caries in an extinct mammal (Mammalia, Euarchonta, Microsyops latidens) and its ecological implications. 40 Dominy N.J. Ferment in the family tree. 41 Fellows Yates J.A.
Velsko I.M.
Aron F.
Posth C.
Hofman C.A.
Austin R.M.
Parker C.E.
Mann A.E.
Nägele K.
Arthur K.W.
et al. The evolution and changing ecology of the African hominid oral microbiome. 42 Wang G.
Hayes J.
Ziegler G.
Roberts R.
Hopfer H. Dose-response relationships for vanilla flavor and sucrose in skim milk: evidence of synergy. Given that most variants are shared by the genus Homo, the small number of novel variants in extinct lineages may reflect adaptations following dispersals from Africa and after the split with humans. For example, the Denisovan OR strong response to honey-like odors (allyl phenyl acetate) may reflect the evolutionary olfactory ecology of the lineage. Honey is the most energy-dense natural food and is a prized component of extant hunter-gatherer diets (except where bees are rare or absent)—even great apes have a ‘honey tooth’.The hunt for sugar is as old as the earliest stem primates 54 mya who notably had a high prevalence of caries after a shift to a diet high in fruit—such an emphasis on sweet is allied with the increased sensitivity to vanilla odors in our study. Denisovan’s increased sensitivity to sweet odors suggests an evolutionary tuning to sources of natural sugar. Energy-dense foods like sugars are sought by larger-brained primatesand, based on oral microbiome data, Neanderthals and contemporary humans share functional adaptations in nutrient metabolism, including starch digestion, that are not found in our closest ape relatives.The high response to vanilla odors further suggests a tuning to sweet—an odor-taste pairing common in contemporary humans.
28 Mafessoni F.
Grote S.
de Filippo C.
Slon V.
Kolobova K.A.
Viola B.
Markin S.V.
Chintalapati M.
Peyrégne S.
Skov L.
et al. A high-coverage neandertal genome from Chagyrskaya Cave. 43 Gilad Y.
Man O.
Glusman G. A comparison of the human and chimpanzee olfactory receptor gene repertoires. 16 Jacobs G.S.
Hudjashov G.
Saag L.
Kusuma P.
Darusallam C.C.
Lawson D.J.
Mondal M.
Pagani L.
Ricaut F.X.
Stoneking M.
et al. Multiple deeply divergent denisovan ancestries in papuans. , 17 Browning S.R.
Browning B.L.
Zhou Y.
Tucci S.
Akey J.M. Analysis of human sequence data reveals two pulses of archaic denisovan admixture. Although local ecological adaptive pressures may have acted on ORs in extinct lineages to produce the few novel variants observed, extinct lineages were less variable in OR genes and proteins compared to 1000 Genomes. Differences in sample sizes might account for some of the striking differences, but the reduced variation is probably because of genetic drift effect or gene conservation.Purifying selection has been observed in chimpanzee ORs compared to a mix of relaxed and positive selection in human ORsand extinct lineages may also have been subject to this as well (as suggested by the presence of fewer variants, most of which code for synonymous proteins). The mean of Fst values across genes comparing 1000 Genomes to extinct lineages (11%) is higher than those for 1000 Genomes population comparisons (4%), which suggests that there are structural differences between them and us that reflect both explanations—drift and conservation. Denisovans have been previously noted to exhibit geographical and temporal population structure.Based on our data the last common ancestor shared by Homo (contemporary humans, Neanderthals, Denisova, and others) and Pan (chimpanzees, bonobos) had a conserved set of ORs. Contemporary humans derived away from the pattern of conservation more recently, with evolutionary pressure toward increased missense variation.
44 Trimmer C.
Keller A.
Murphy N.R.
Snyder L.L.
Willer J.R.
Nagai M.H.
Katsanis N.
Vosshall L.B.
Matsunami H.
Mainland J.D. Genetic variation across the human olfactory receptor repertoire alters odor perception. 45 Somel M.
Wilson Sayres M.A.
Jordan G.
Huerta-Sanchez E.
Fumagalli M.
Ferrer-Admetlla A.
Nielsen R. A scan for human-specific relaxation of negative selection reveals unexpected polymorphism in proteasome genes. 46 Lachance J.
Vernot B.
Elbers C.C.
Ferwerda B.
Froment A.
Bodo J.M.
Lema G.
Fu W.
Nyambo T.B.
Rebbeck T.R.
et al. Evolutionary history and adaptation from high-coverage whole-genome sequences of diverse African hunter-gatherers. 46 Lachance J.
Vernot B.
Elbers C.C.
Ferwerda B.
Froment A.
Bodo J.M.
Lema G.
Fu W.
Nyambo T.B.
Rebbeck T.R.
et al. Evolutionary history and adaptation from high-coverage whole-genome sequences of diverse African hunter-gatherers. 44 Trimmer C.
Keller A.
Murphy N.R.
Snyder L.L.
Willer J.R.
Nagai M.H.
Katsanis N.
Vosshall L.B.
Matsunami H.
Mainland J.D. Genetic variation across the human olfactory receptor repertoire alters odor perception. Despite sharing a genetic and functional olfactory repertoire with Neanderthals and Denisovans (shared variants suggest a shared olfactory world), contemporary humans have greater genetic variability and diversity (across populations) than extinct lineages. Such variation and diversity indicates a broad olfactory repertoireand may reflect cultural adaptations following our migrations from Africa. Relaxed selection on OR genes for groups no longer engaging in traditional lifestyles is possible.In our sample, 1000 Genomes groups outside Africa were less variable in most OR genes than those in Africa in our sample and other research has found that greater OR gene enrichment in African hunter-gatherer groups (Hadza and Pygmies) but not African agricultural (Yoruba) and pastoral (Maasai) groups.Tanzanian Sandawe hunter-gatherers show no OR allelic enrichment, however, which undermines the case for relaxed selection.High allelic diversity and OR generalization on a broad scale may have functional implications, such as increasing the effective size of the olfactory repertoire.
Our data provide insights into how the dispersal of human lineages outside of Africa (Denisova, Neanderthal, ancient human) may have affected olfactory gene repertoire and function. Understanding the evolutionary genetics and functional significance of observed OR variability in and among human populations and extinct relatives sheds light on the role of olfaction in key aspects of human culture, and perhaps our current success as a global species. The applications of these methods to a large dataset addressing the function of genetic variation in extinct hominins is an advance in the study of human evolution and allows future work with odorant receptors (and other proteins that can be created in the lab) and their adaptive function in human evolutionary timescales. Understanding our unique OR allelic diversity and its evolution is an important challenge to olfactory science.