A Melanocortin 1 Receptor Allele Suggests Varying Pigmentation Among Neanderthals

See allHide authors and affiliations

Science  30 Nov 2007:
Vol. 318, Issue 5855, pp. 1453-1455
DOI: 10.1126/science.1147417


The melanocortin 1 receptor (MC1R) regulates pigmentation in humans and other vertebrates. Variants of MC1R with reduced function are associated with pale skin color and red hair in humans of primarily European origin. We amplified and sequenced a fragment of the MC1R gene (mc1r) from two Neanderthal remains. Both specimens have a mutation that was not found in ∼3700 modern humans analyzed. Functional analyses show that this variant reduces MC1R activity to a level that alters hair and/or skin pigmentation in humans. The impaired activity of this variant suggests that Neanderthals varied in pigmentation levels, potentially on the scale observed in modern humans. Our data suggest that inactive MC1R variants evolved independently in both modern humans and Neanderthals.

One gene responsible for skin and hair color variation in humans is melanocortin 1 receptor (mc1r), which encodes a seven-transmembrane heterotrimeric GTP-binding protein (G protein)–coupled receptor (GPCR) (Fig. 1) (1). Red hair and pale skin result from both complete and partial loss-of-function alleles in human MC1R (huMC1R) because they alter the balance between eumelanin and pheomelanin synthesis (2). Thus, we hypothesize that the retrieval of mc1r sequences from extinct species can potentially provide information on their phenotypic traits.

Fig. 1.

Structure and activity of MC1R. (A) Depicted are the seven transmembrane helices of MC1R. The R307G mutation, found in two Neanderthals, is positioned within helix 8, or the fourth intracellular loop. This amphipathic helix contains many conserved basic residues and is part of the intracellular receptor C terminus. (B) Partial loss of function of the R307G variant. For functional characterization, COS-7 cells were transiently transfected with constructs coding for the wild type (huMC1R) and the R307G variant. As a control, cells transfected with a plasmid encoding green fluorescent protein (GFP) were used. Transfected cells were tested for agonist-induced cAMP accumulation (11). Shown are the mean ± SEM of three independent representative assays, each performed in duplicate.

Neanderthals are an extinct hominid group that lived in Eurasia ∼400,000 to 28,000 years ago (3). Recently, metagenomic approaches recovered about 1 megabase of the Neanderthal genome (4, 5), implying that amplification of nuclear DNA by the polymerase chain reaction (PCR) may be feasible in well-preserved Neanderthal remains. We studied two Neanderthal fossils, Monti Lessini (Italy) (6) and El Sidrón 1252 (Spain) (7). Both samples have a low degree of amino acid racemization (<0.10) and high amino acid content (>20,000 parts per million), suggesting good DNA preservation (8). Amplifications of the hypervariable region 1 of the mitochondrial DNA showed that endogenous Neanderthal DNA was preserved in these individuals (6, 9).

We assumed that retrieval of Neanderthal nuclear DNA sequences from these samples was possible if the amplicon length was short and large numbers of clones were generated. Because of the low level of divergence between Neanderthals and modern humans, it is impossible to distinguish contamination if the sequences are identical or polymorphisms are shared between the species (10). Therefore, we focused on identifying Neanderthal-specific substitutions. We successfully amplified a 128–base pair (bp) fragment of mc1r (11) from the Monti Lessini sample. Most clones were identical to the modern human sequence, most likely representing contamination of the Neanderthal bone with modern human DNA. However, 1 of the 25 clones had an A-to-G substitution at nucleotide position 919, resulting in an Arg-to-Gly change at amino acid position 307, which was not previously observed in modern humans (12, 13). We hypothesized that this was not due to PCR error, because most errors are C-to-T (or G-to-A) changes, due to cytosine deaminations in the template DNA (14, 15).

Because the number of amplifiable ancient DNA molecules increases exponentially with decreasing amplification length (16), we designed primers to amplify a shorter DNA fragment and performed four different amplifications for El Sidrón 1252 and one for Monti Lessini. We sequenced at least 10 clones per reaction. The Arg307→Gly307 (R307G) substitution was present in all amplification products at frequencies ranging from 7 to 25% (fig. S1). Because the humanlike sequences probably at least partially represent contamination with modern human DNA, we cannot decide whether the two individuals are homozygous for R307G or heterozygous. Thus, we concentrated on authentication of the R307G variant. The likelihood of an incorrect nucleotide, due to postmortem damage or PCR errors, decreases as the number of independent amplifications increases (14). Therefore, we amplified a different Monti Lessini extract in two additional laboratories with different primer sets. The R307G substitution was found in all three laboratories for the Monti Lessini sample but only in Barcelona for the El Sidrón sample. Altogether, we observed this substitution in 9 of 12 amplifications, with frequencies ranging from 4 to 36% (table S1). Moreover, there is no known damage in ancient DNA resulting in A-to-G substitutions (15). Even if our results were due to a previously unknown kind of template damage that occurs at the same frequency as cytosine deamination (2%) (14), the probability of obtaining the same result in 9 out of 12 independent amplifications is ∼10–13 (11). These results suggest that the R307G substitution is a reproducible, albeit minority, sequence not attributable to damage.

The R307G substitution has not been described in more than 2800 modern humans that have been fully sequenced for mc1r (12, 13). We genotyped this polymorphism in the CEPH Human Genome Diversity Panel (17) (fig. S2), as well as in all people involved in the excavation and genetic analysis, to determine whether this variant exists in extant humans (11). No individual had the R307G allele. If the R307G allele occurs in modern humans, it must be at a very low frequency, and it is unlikely that such a rare variant would appear as contamination in three separate laboratories. We additionally investigated whether nonhuman contamination could explain this result. BLASTN was used to compare the longest sequence with the R307G variant and showed that it was most like human, with 98% sequence similarity, followed by primate sequences with progressively decreasing identity (table S4). None of the sequences in GenBank matched perfectly to the R307G variant, and nonprimate mammalian sequences differ considerably, excluding other common sources of contamination as possible origins of the sequence. Therefore, we concluded that the R307G substitution is an endogenous Neanderthal sequence.

The R307G mutation is positioned at the cyctoplasmatic surface of MC1R within the so-called helix 8 or fourth intracellular loop (18, 19) (Fig. 1A). Mutations of conserved basic residues within this amphipathic helix alter receptor function in numerous GPCRs (19, 20). Although Arg307 is replaced by Lys in fox, cow, and sheep and by Met in mouse (21), the position has been shown to be intolerant of most mutations (22) (fig. S5). To investigate whether the R307G substitution affects the function of huMC1R, both wild-type MC1R and the R307G variant were expressed in COS-7 cells, and basal and agonist-induced intracellular cyclic adenosine monophosphate (cAMP) levels were determined. huMC1R responded to the natural agonist α–melanocyte-stimulating hormone (α-MSH) with an ∼fourfold increase in intracellular cAMP levels (Fig. 1B and table S2). In contrast, cells expressing the R307G variant had intracellular cAMP levels that were 40% of the wild-type levels and were 50% less than wild-type levels when activated by an agonist. Whereas basal cAMP levels of the wild type strongly correlated with the amount of transfected plasmid DNA, this was not the case for the basal activity of R307G (fig. S3). Analysis of the stimulation curves (Fig. 1B) showed that the median effective concentration values did not differ between the human and R307G variants. To exclude the possibility that the observed differences resulted from overexpression in COS-7 cells, we established stable cell lines with a single expression cassette of either MC1R variant at a predefined locus (11). These stably transfected CHOK1 cells also showed reduced agonist-induced cAMP levels for the R307G variant (table S2).

In order to determine whether these reduced basal and agonist-induced cAMP levels observed in R307G were caused by either lower cell-surface expression levels (fig. S4) or reduced G-protein coupling properties, we determined MC1R protein expression levels with an enzyme-linked immunosorbent assay and performed binding assays on intact cells (11). Total expression of the full-length receptor protein did not differ between the two variants (table S3). In contrast, radioligand binding and ligand-binding–independent measurements revealed substantially reduced cell-surface expression of the R307G variant (table S3). However, we observed no difference in the ability of both receptor variants to bind α-MSH (table S3). Altogether, our data support the idea that the R307G allele has a partial loss of function caused by reduced cell-surface expression of receptor protein (23, 24) and altered G protein–coupling efficacy.

Alleles conferring partial loss of function of MC1R have been associated with pale skin color and red hair in humans (1, 12). We tested a functional cAMP assay on extant partially functional huMC1R alleles associated with pale skin color and red hair. We confirmed partial activity of these variants, indistinguishable from that of R307G (table S2). Pale skin color and lighter hair are more likely when MC1R alleles are in a homozygote or compound heterozygote stage. Although we cannot decide currently whether the Neanderthal individuals analyzed were homozygous for R307G or heterozygous, we can obtain a minimum frequency for this variant in Neanderthals. If we assume that both individuals were heterozygous, we obtain a minimum allele frequency for R307G of 50% for the two individuals investigated. From this figure, the minimum frequency of the mutation in all Neanderthals that is compatible with observing two mutant alleles when four alleles are sampled at random is 0.1 for P >0.05. This translates into at least 1% of homozygous Neanderthal individuals that may have had reduced pigmentation levels, possibly even similar to the pale skin color and/or red hair observed in modern humans. These results once more raise the question of whether reduced pigmentation may have been advantageous in Europe, for example via ultraviolet-light–mediated vitamin D synthesis, or whether it just reflects a loss of constraint for the mc1r gene in regions of reduced solar irradiation (25). Our data do not support the hypothesis that phenotypic similarities between these two human groups are explained by gene flow (26) and do support the hypothesis of convergent evolution of reduced-function MC1R alleles, as suggested between modern European and Asian populations (27, 28).

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Tables S1 to S4


References and Notes

View Abstract

Navigate This Article