The "flame-haired" Neandertals

17 minute read


What better to match an Irish tongue, than red hair? In fact, with language and red hair Neandertals would seem to be excellent candidates for a little scheme I have copying the encyclopedia...

Botticelli Venus as Neandertal

Well, that's the story, anyway. How true is it? It's been two weeks since the headlines, I've had a chance to reflect, and there's more news coming. Time to get this off my desk -- so let's unleash another Hawks FAQ:

What is this study, anyway?

It is tempting to just say that Lalueza-Fox and colleagues have given us the MC1R version of the FoxP2 paper by Johannes Krause et al. The similarities are obvious -- they've identified an interesting variant by probing for a particular gene and they've confirmed the Neandertal state by finding the same variant in two different specimens.

Of course, there is also an obvious difference -- the FoxP2 study found that the Neandertals shared the derived human variant; this study found that Neandertals had at least one unique MC1R variant.

It is important that the FoxP2 research came out first, because Krause and colleagues (2007) included many more controls. They probed two El Sidrón specimens looking specifically for the two human-derived mutations of FoxP2. In both specimens they found both mutations. They also probed for a number of other gene variants, to try to assess how contaminated the bones might be. Their results showed that the two specimens are in fact two different individuals, and that they are both males. They probed for human-derived Y chromosome variants: sites for which most people have a derived allele, and only a few Africans today have the ancestral allele. Because they found none of those human-derived variants in the Neandertal samples, they could infer that the sequences almost certainly came from Neandertals. Together, the results suggest that the Neandertals really did have the human-derived FoxP2 mutations.

Because of that success, the current study by Lalueza-Fox and colleagues (2007), can rely on some of the same confirmatory details. It adds a few additional elements that tend to confirm its validity.

The team began by probing the Neandertal specimen from Monte Lessini, Italy, for the MC1R gene. If you've never heard of this specimen, it is a fragmentary skull with a couple of postcranial pieces, from the Riparo Mezzena cave in northern Italy. It is thought to be around 50,000 years old. David Caramelli and colleagues (2006) successfully extracted mtDNA from the specimen, discovering an mtDNA variant that is quite distinctive, although still on a single branch with other Neandertal skeletons.

In the current study, the group successfully amplified a 128-bp fragment of MC1R, but found that most of the clones generated from their probing were identical to the modern human consensus sequence. They interpreted this as contaminant (although I note that, with such a short fragment, endogenous sequences will often be identical to the modern human reference). Still, they found one clone with a unique mutation, not found in humans. A single change like this in a single clone (out of many) might easily be chalked up to DNA damage. However, this mutation was A -> G, which is not one of the main pathways of DNA damage. They hypothesized that the mutation was a genuine part of the Neandertal's MC1R sequence. So far, so good.

Now, they used the variant site as a target for amplifying much shorter fragments. Their logic: ancient sequence amplifies better at small fragment sizes, so by looking for very short fragments including this site, they might confirm the presence of the unique Neandertal-derived variant -- first in the Monte Lessini specimen, and later in other Neandertals. Lalueza-Fox and colleagues found the site repeatedly in the Monte Lessini sample, and also found it once in the El Sidrón sample, so they concluded it was a genuine Neandertal variant.

Despite this success, there is much they can't tell, even about this one variant. They can't genotype the specimens -- the Monte Lessini specimen clearly has the human variant within it, but that might be contamination. Nor do they know the sequence of the rest of the MC1R gene in these samples. So, this is simply an observation and replication of a Neandertal-derived mutation. It is important because, like the FoxP2 observation, it was replicated in more than a single specimen.

That kind of independent replication will be necessary to authenticate any Neandertal-derived changes -- again, this is a landmark feature of this paper. But as with the FoxP2 paper, I have to wonder whether it is a bit wasteful of sample to probe the remains one gene at a time.

Was Neandertal hair really red? I mean, how could you know, really?

To tell you the truth, I have no idea whether it was red, although it looks a lot more likely now than it did a couple of weeks ago.

The logic behind the "flame-haired" assertion: Lalueza and colleagues (2007) found the Neandertal-derived mutation at a site that changes the amino acid sequence of the MC1R protein. The mutation is at position 307 of the sequence, and would be expected to change the arginine in the usual amino acid sequence to a glycine (hence, the name Arg307Gly, which we can abbreviate to R307G). So it would potentially have induced a difference in phenotype in individuals carrying the mutation.

Some functional MC1R variants in living Europeans (and elsewhere) have high penetrance and cause red-headed phenotypes. That is to say, most people who carry such mutations have reddish hair. Where there's smoke, there's fire -- nonsynonymous mutations in MC1R in mammals often change pigmentation phenotypes, and in humans these are generally red. Blond hair (versus brown or black) is associated with variants at other genes.

Yes, yes, MC1R mutations, red hair. What are the gory biochemical details?

I'm glad you asked. There are two reasons why you want to know the pigmentation pathway: First, many of the genes involved in pigmentation have been under strong positive selection in recent humans, so this is an active target of recent evolutionary change. That helps to explain why recent humans aren't like Neandertals -- we have our own, new, strange alleles.

Second, understanding the mechanism of pigmentation helps to reveal why we don't necessarily know that Neandertal hair was red. The pathways involving MC1R tend to influence the ratio of red versus black pigment, but they may be affected by other genetic changes.

The reddish color is produced by pheomelanin. This is the same pigment that makes orangutans red -- in fact, pretty much every reddish mammal is made so by pheomelanin. In contrast, eumelanin is brown or black in color. These pigments are generated in melanosomes, which are specialized organelles that function within skin cells called melanocytes. Melanosomes are "donated" to other cells of the epidermis (keratinocytes), spreading pigment across the skin. The concentration of melanosomes in the cells of the epidermis generates skin pigmentation -- low melanin production makes light skin, more pheomelanin than melanin makes reddish skin. Hair is pigmented by melanocytes within the hair follicle; signaling to these follicular melanocytes occurs in concert with hair growth. Melanosomes concentrate within the cortex of the growing hair, lending it their color. Gray and white hair occur as melanocytes become inactive with age.

To mobilize these pigment-producing melanosomes, the melanocytes need signals from outside the cell. The Mc1r protein studs the surfaces of the melanocytes and receives these signals (reviewed by Rees 2004). When Mc1r is highly receptive, the melanocytes produce mostly eumelanin. This means that hair and skin will be in a range from light to brown-black, depending on the activity of other genes.

Some derived MC1R variants in Europeans are loss-of-function mutations, which tend to make the Mc1r protein less capable of receiving extracellular signals. There are basically two ways to impede Mc1r: make it less able to bind with the signaling molecules, or make it less capable of embedding in the cell membrane, so there will be less of it to respond to signals. Each of these approaches is taken by some functional MC1R alleles in Europeans. These loss-of-function mutations tend to increase the ratio of pheomelanin to eumelanin produced by the melanocytes. That makes the overall color (of hair or skin) redder. A handful of the functional mutations yield red hair phenotypes, particularly in homozygotes -- including R151C, R160W, and D294H. There are a few other mutations that have slighter effects on hair pigmentation, sometimes called "weak" or r variants, in contrast to the three "strong" or R variants. It's a confusing terminology -- the R and r are considered together with wild-type (wt) alleles at those sites, but there are other functional variants in addition to these. For the most part, functional changes to MC1R have all been independently derived from the ancestral sequence (Harding et al. 2000), yielding a big multiallelic system where genotype is very important and linkage less so.

The effects of the strong and weak alleles are known from phenotypic association-testing -- take a large sample of people from one population, take down their hair and skin color information, and then genotype them. Now, the Neandertal-derived R307G allele is not present in any living humans, at least not that anybody has yet sampled. So it is impossible to directly test its phenotypic associations in people. Considering that there are many functional mutations to human MC1R in Europeans, and only a few of them correlate with red hair phenotypes, if we stopped here we would have no reason to think that Neandertals expressed a red hair phenotype.

But nowadays we have cleverer ways of predicting phenotypes. For one thing, you can engineer cell lines to express the mutation:

To investigate whether the Arg307Gly substitution affects the function of human MC1R, both wild-type and the Arg307Gly variant were expressed in COS-7 cells, and basal and agonist-induced intracellular cAMP levels were determined. Human MC1R responded to the natural agonist α-MSH [that is, melanocyte stimulating hormone] with a ~4-fold increase in intracellular cAMP levels. In contrast, cells expressing the Arg307Gly variant had intracellular cAMP levels reduced to 40% of the wild-type levels and 50% less than wild-type when activated by an agonist.
...[R]adioligand binding and ligand-binding-independent measurements revealed a significantly reduced cell surface expression of the Gly207 variant. However, we observed no difference in the ability of both receptor variants to bind α-MSH. Altogether, our data support that the Arg307Gly allele has a partial loss of function caused by a reduced cell surface expression of receptor protien and an altered G protein coupling efficacy (Lalueza-Fox et al. 2007:2, citations omitted).

COS-7 cells are transformed vervet monkey cells, originally taken from the kidney of a monkey in the 1960's. Today, it is possible to transfer genes into these cells, to determine the function of alternate alleles when expressed. This is a fairly quick way to check gene activity in culture; organismal expression may be examined by making knock-in mice expressing the human allele. Mouse models have been used to examine the activity of other MC1R variants, and it wouldn't be surprising to see them developed to investigate this allele.

So, the Neandertal-derived MC1R allele, R307G, doesn't respond to extracellular signaling in the same way as the wild-type human allele, and we can presume that would increase the ratio of pheomelanin-to-eumelanin production in melanocytes.

Does this mean red hair? Here's a reason to suspect otherwise: Neandertals existed long enough for hair and skin pigments to come to an equilibrium, while recent Europeans have not.

Selection on pigment-altering alleles has been powerful in Europe during the recent past -- and although some alleles have approached fixation, most have not. It is possible that the alleles that confer red hair in Europeans are at their stable equilibrium point, but even if they were, there is no way of knowing whether their phenotypic effect would remain stable in the long run as alleles at other loci continued to change in frequency. The red hair phenotype depends not only on significantly reduced MC1R activity, but also on the continued functioning of the rest of the melanocyte signaling pathway. Other loss-of-function alleles may interfere with the phenotype -- and the frequency of red hair would ebb over time as other loss-of-function mutations increased.

This is, of course, speculative. Still, if Neandertals were strongly selected for pigmentation variants, we ought to expect many more loss-of-function mutations with ample time to reach fixation. Perhaps even though they possessed an MC1R variant that causes red hair in Europeans, most Neandertals were nonetheless blond. Dark hair seems unlikely -- otherwise, why select for MC1R loss of function?

Our problem is that we really don't know what the optimum pigmentation phenotype would be in Pleistocene Europe. We have no reason to think that today's phenotypic variation in Europe is anywhere near the optimum -- and to confirm this, we need only look at another population where pigmentation evolution has taken a parallel course: East Asia. Asians and Europeans have each been selected for alleles that lighten skin pigmentation, but each population has a different set of them. The MC1R variation in Europe is greater, with more effects on hair pigmentation, but other genes exhibit greater adaptive variation in Asia (Norton et al. 2006). The important insight is that neither Europe nor Asia has reached an equilibrium -- if the different alleles from these populations increased their overlap, the Asian variants might introgress into Europe, or vice-versa. We have no way of knowing which alleles are "better" in the sense of fitness, or which might "win out" given enough time.

So, I hesitate to say that Neandertal hair was red. They had a potentially functional mutation to the pigmentation pathway, analogous to changes that make red hair in humans, but their genetic background may have been different enough that such a change would not have had the same effect.

Could it just be relaxed selection and genetic drift?

Neandertals were a relatively small population. Purifying selection against functional changes in MC1R may have been relaxed because dark pigment is less important at high latitudes. Neandertals certainly existed over long enough time for a neutral allele to drift to high frequency -- and in this case we can't exclude the hypothesis that the allele was at 10% or even lower. So, considered in isolation, there would be no reason to think that the allele was selected. No other explanation necessary.

Still, although genetic drift would be consistent with the presence of the allele and what we know about its frequency, we can test this hypothesis by comparison with recent humans. There is clear evidence for strong selection on pigmentation in recent Europeans. Today's MC1R variants haven't existed for very long -- they all postdate Neandertals -- and they could not have reached their current frequencies in the large population of recent humans by drift alone. I suspect by analogy with recent Europeans that Neandertals would have been under selection for pigmentation variants as well.

But the credibility of the selection hypothesis for Neandertals depends on exactly why MC1R functional variants have been selected in recent Europeans. Recent Europeans can be a relevant comparison only if their adaptive environment was similar to the Neandertals. If latitude is the ruling factor in pigmentation selection, then recent Europeans may be a good model for the selection pressures on Neandertals. If, on the other hand, diet is an important environmental factor linking pigmentation phenotypes to fitness (for instance, because a diet based on cereal grains provides low vitamin D availability), then recent Europeans would be a poor model for Neandertal selection pressures -- after all, Neandertals weren't agriculturalists. An anti-rachitic (i.e., rickets-preventing) diet has long been suggested as a reason why arctic groups like the Inuit lack very light skin and hair (Loomis 1967). Neandertals did not eat very much seafood, but their dietary differences from later Europeans may have had an effect on the selection pressures related to vitamin D.

Still, I have trouble believing it.

Carles Lalueza-Fox suggested that Neandertals had "the whole range of hair colour we see today in modern Europeans." Any comment?

Well, there's nothing impossible about this idea of polychrome Neandertals, but I doubt it.

For one thing, there's no evidence of polymorphism. Two Neandertal specimens have the same Neandertal-derived allele, with no clear evidence (i.e., clear non-contamination evidence) for any other alleles. Of course, this is a very small sample -- really we only know about two copies of the gene in the entire population. Lalueza-Fox and colleagues (2007) did a little calculating and figured that the most they could say is that the Neandertal-derived allele probably has a frequency over 10 percent. That's not saying much about polymorphism.

Now, Europeans are unusual in having many functional variants. But Europeans today are not a stable entity -- alleles that influence pigmentation phenotypes have been strongly selected, and so have been increasing in frequency rapidly. We don't know what will happen to these in the future -- maybe they'll continue to increase to fixation; maybe they'll stall out and Europeans will remain polymorphic forever. But we do know that some pigment-related alleles have increased to near fixation, including the near-fixed European variant of SLC24A5. If selection on MC1R was induced by UV levels, selection ought to have favored one of these over the others, in the long term. Even if two (or more) alleles were equal in fitness, over the long term genetic drift would fix one or the other of them. With Neandertals, we are definitely talking long term evolution -- they had plenty of time to come to an equilibrium. So they were likely to have been fixed for some MC1R allele.

That all assumes, though, that selection on pigment phenotype was directional. If instead selection was frequency-dependent, then Neandertals might have harbored substantial polymorphism. Today's Europeans are polymorphic for hair and skin phenotypes influenced by MC1R alleles. In addition to this, Europeans have polymorphisms in at least a half dozen other genes that result in hair, eye, or skin color polymorphisms. Maybe there's something unique about Europe selecting for pigmentation polymorphisms recently, and maybe that selection also was present in Neandertal times.

The most common suggestion is sexual selection. But there's no reason for sexual selection in one region to remain the same over a long period of time.

And it's rather unlikely that red hair was the target of this selection. The red hair phenotype is mostly recessive in recent people, and recessive alleles rise exceedingly slowly in frequency and are much less likely to survive genetic drift. Maybe a reddish-brown hair color, in heterozygotes, was enough to promote selection on the allele. But more likely, the target of selection has been some other phenotypic effect of the recent alleles. That means the red-hair-is-sexy sexual selection hypothesis isn't very likely.

Another possibility is that Neandertal pigmentation varied across geographic space. These two specimens are from Spain and Italy -- not exactly the world headquarters of light hair today (though they have their share). If a single loss-of-function allele of MC1R was moderately common in both these populations, we may suggest that genetic exchanges across Europe were likely fairly substantial. The selection supporting this variant may have been present in southern Europe, but it may have been more effective to the north, and spread southward by gene flow.

But everyone knows that it was Ayla who was blonde, not Creb! Please don't take away my Daryl Hannah fantasies!

Yeah, well, Jean Auel really did her research for that series of books, and has been a great supporter of the field as well as perhaps its most outstanding popularizer.

But that particular element of the story, we can now say, pretty much got it backward. It looks like early modern Europeans may have had none of the current variety of MC1R alleles in Europe. And far from being dark-skinned, dark-haired schlubs, Neandertals had their own MC1R variant. Maybe more than one.


Gerstenblith MR, Goldstein AM, Fargnoli MC, Peris K, Landi MT. 2007. Comprehensive evaluation of allele frequency differences of MC1R variants across populations. Hum Mutation 28:495-505. doi:10.1002/humu.20476

Healy E, Jordon SA, Budd PS, Suffolk R, Rees JL, Jackson IJ. 2001. Functional variation of MC1R alleles from red-haired individuals.

Krause J, Lalueza-Fox C, Orlando L, Enard W, Green RE, Burbano HA, Hublin J-J, Bertranpetit J, Hänni C, Fortea J, de la Rasilla M, Rosas A, Pääbo S. 2007. The derived FoxP2 variant of modern humans was shared with Neandertals. Curr Biol 17:1-5. doi:10.1016/j.cub.2007.10.008

Lalueza-Fox C and 16 others. A melanocortin 1 receptor allele suggests varying pigmentation among Neanderthals. Science (early online) doi:10.1126/science.1147417

Loomis WF. 1967. Skin-pigment regulation of vitamin-D biosynthesis in man. Science 157:501-506.

Norton HL, Kittles RA, Parra E, McKeigue P, Mao X, Cheng K, Canfield VA, Bradley DG, McEvoy B, Shriver MD. 2006. Genetic evidence for the convergent evolution of light skin in Europeans and East Asians. Mol Biol Evol 24:710-722. doi:10.1093/molbev/msl203

Sulem P and 24 others. 2007. Genetic determinants of hair, eye, and skin pigmentation in Europeans. Nat Genet (early online) doi:10.1038/ng.2007.13