Pigmentation of archaic humans: introduction

This week I will be blogging on research that represents a collaboration between my undergraduate and graduate students and myself. We have undertaken the investigation of pigmentation-related loci in archaic humans. So far, we have treated the project as a partially open notebook -- completely open within the team but not to the public. As we approach the end of the semester I am now prepared to open it further, to share what we have been doing and see if we can probe deeper into the question.

Why pigmentation?

Our primary goal is simple: to determine what we can about the physical appearances of Neandertals and Denisovans. In the case of the Denisova genome, we know nothing at all anatomically about the population aside from the morphology of a third molar. Anything we discover about pigmentation genetics from this genome will be our first knowledge about the physical appearance of an ancient human population.

For Neandertals, of course, the story is different. We know quite a lot about the skeletal anatomy of these ancient people. And we already know one piece of information about their pigmentation genetics. Carles Lalueza-Fox and colleagues Lalueza-Fox:2007 investigated the MC1R gene in Neandertal skeletal specimens from Monte Lessini, Italy and El Sidrón, Spain, finding a uniquely derived amino acid substitution that is not present in any known living Europeans. Analysis of a cell line engineered to carry this mutation showed that it reduced signaling activity compared to the normal human version of this gene. Even though nobody living has yet been found with this mutation, we can infer that it has a similar effect to other mutations that contribute to red hair color, so at least some Neandertals were likely to have had this pigmentation. In humans, coding mutations in MC1R are also correlated with slightly lighter skin and the incidence of freckling. I described this study when it came out four years ago ("The 'flame-haired' Neandertals"), and it has had a huge influence on the way artist reconstructors have envisioned Neandertal appearance. Knowing this one little detail has added tremendously to our ability to communicate the science of Neandertal genetics.

Investigating this one single locus has already shown variation among the archaic human genomes. Genetic data from several Neandertals Green:draft:2010 and from Denisova Reich:Denisova:2010 cover the MC1R gene. As I wrote this spring ("Denisovans did not have red hair"), the Denisova draft genome has only the normal human wild-type allele at the location where El Sidrón 1252 and Monte Lessini have the derived amino acid substitution. Of the Vindija Neandertals, only one (Vi33.26) has any information at this site, and it also carries the normal human wild-type. From this, we can infer (tentatively) that Neandertals were likely variable for this change, not fixed. Denisovans may have shared the Neandertal change, but the existing genome shows no sign of it. Archaic humans came in different colors.

These results encouraged us to think that there is much more to discover about the pigmentation of archaic people. In humans today, the variation in MC1R is really only a very small part of the overall story of pigmentation. The genes involved in normal pigmentation variation among human populations are known to number more than a dozen Sturm:pigmentation:2009. In addition to those that have common variations in human populations now, there are many other genes necessary for melanin synthesis and expression Ganelan:2008 that exhibit evolutionary changes in different animal species. Today, humans who have light pigmentation attain this phenotype by a diversity of different mutational pathways Myles:2007. Different mutations are common in different human populations, and most seem to have appeared recently. Today's light-skinned Europeans and northeast Asians seem to have evolved light pigmentation convergently, from a dark-pigmented ancestral population Norton:2007.

With such a range of variation in humans, and many different genes to examine, we hypothesized that archaic humans might also have had mechanisms for pigmentation variation that converged on those in recent people. The known Neandertal and Denisova genomes all represent people who lived at high latitudes, 40 degrees north or higher. This stretch of Eurasia has recently been inhabited by light-pigmented populations, which is at least circumstantial evidence that we should expect light pigmentation in these archaic people as well. Looking at the sequences of pigmentation-related loci might allow us to discover the mechanisms that affected pigmentation in these ancient people. Or, if we didn't find any changes to these genes in archaic genomes, we might infer that our expectation about their pigmentation is incorrect. In this case, they may have shared dark pigmentation with ancestral Africans and today's lower-latitude populations across Africa, south and southeast Asia, and Australia.

At the same time, we set out to test a second hypothesis: that today's light-pigmented populations may have inherited some of their pigment-altering alleles from archaic people. Today's people outside Africa got around 3 percent of their ancestry from Neandertals Green:draft:2010. It is logical to suppose that the fraction of Neandertal ancestry might be even higher for a subset of genes, if Neandertals had alleles that helped adapt them to their high-latitude environment two-s. Obviously, pigmentation genes are strong candidates for such adaptive introgression. We set out to invent a method that could test for such adaptive introgression of pigmentation alleles from Neandertals into later populations.

Steps toward an answer

As I've geared up for the analysis of whole genomes, I've started to generalize about some of the ways that this kind of analysis is different from more traditional genetics. Genomics is most effective when we can break genomes into small parts and perform one simple step at a time. Across a whole genome each step is potentially very time-consuming, and so it is better to keep each step very simple. If a comparison involves several analytical steps, we have to consider which of those steps may be precursors to other kinds of analyses. Those are worth saving so that we don't have to do the same steps twice.

Pigmentation is a natural thing for us to study because the network of genes involved in pigmentation variation gives us large unit for comparison. We can always compare a single gene to the whole genome, but we'll gain a lot more statistical power when we can compare a pattern across many genes. The bonus of working with a whole genome is that millions of data points provide a very powerful test of some hypotheses.

Our hypotheses led us along two lines of inquiry on the archaic genomes. First we took the coding regions of a set of genes related to pigmentation and compared them between archaic and recent humans. On the most basic level, this comparison is as simple as lining up the human reference sequence next to the archaic human sequences and taking note of where they are different. We can go substantially further than this using samples from the 1000 Genomes Project. We can check every place where an archaic human sequence is different from any living people to confirm whether that gene variant is unique to the archaic sample. In some cases, some living people share amino acid-coding variants with archaic genomes.

Second, we considered whether genes related to pigmentation have a higher or lower probability of introgression from archaic humans into recent populations. For this comparison we must separate the alleles that might be shared with Neanderthals because of incomplete lineage sorting from those that later populations inherited from Neandertal introgression. Alleles that have come in by incomplete lineage sorting must be relatively old in humans, while alleles that have introgressed will tend to be young. Working out the ages of alleles that humans share with Neanderthals is not a trivial problem, as you will see from our solution.

As I write this introductory post, some parts of our analysis are still ongoing. But the results are very interesting and it's time to share.