SLC24A5 is very important, but we don’t know why


The golden of pigmentation genetics started in 2005 with SLC24A5, a putative cation exchanger, affects pigmentation in zebrafish and humans. Prior to that pigmentation genetics was really to a great extent coat color genetics, done in mice and other organisms which have a lot of pelage variation.

Of course there was work on humans, mostly related to melanocortin 1. But more interesting were classical pedigree studies which indicated that the number of loci controlling variation in pigmentation was not that high. This, it was a mildly polygenic trait insofar as some large effect quantitative trait loci could be discerned in the inheritance patterns.

From The Genetics of Human Populations, written in the 1960s, but still useful today because of its comprehensive survey of the classical period:

Depending on what study samples you use variance on a locus of SLC24A5 explains less than 10% or more than 30% of the total variance. But it is probably the biggest effect locus on the whole in human populations when you pool them altogether (obviously it explains little variance in Africans or eastern non-Africans since it is homozygous ancestral by and large in both groups).

One aspect of the derived SNP in this locus is that it seems to be under strong selection. In a European 1000 Genomes sample there are 1003 SNPs of the derived variant, and 3 of the ancestral. Curiously this allele was absent in Western European Mesolithic European hunter-gatherers, though it was present in hunter-gatherers on the northern and eastern fringes of the continent. It was also present in Caucasian hunter-gatherers and farmers from the Middle East who migrated to Europe. It seems very likely that these sorts of high frequencies are due to selection in Europe.

The variant is also present in appreciably frequencies in many South Asian populations, and there seems to have been in situ selection there too, as well as the Near East. In Ethiopia it also seems to be under selection.

It could be something due to radiation…but the Near East and South Asia are quite high intensity in that regard. As are the highlands of Ethiopia. About seven years ago I suggested that rather that UV radiation as such the depigmentation that has occurred across the Holocene might be due to agriculture and changes in diet.

But a new result from southern Africa presented at the SMBE meeting this year suggests that this can not be a comprehensive answer. Meng Lin in Brenna Henn’s lab uses a broad panel of KhoeSan populations to find that the derived allele on SLC24A5 reaches ~40% frequency. Probably a high fraction of West Eurasian admixture in these groups is around ~10% being generous. Where did this allele come from? The results from Joe Pickrell a few years back are sufficient to explain: there was a movement of pastoralists with distant West Eurasian ancestry who brought cattle to southern Africa, and so resulted in the ethnogenesis of groups such as the Nama people (there is also Y chromosomal work by Henn on this).

Sad human with two derived alleles of SNP of interest

Lin reports that the haplotype around SLC24A5 is the same one as in Western Eurasia. Iain Mathieson (who is now at Penn if anyone is looking for something to do in grad school or a post-doc) has told me that the haplotype in the Motala Mesolithic hunter-gatherers and in the hunter-gatherers from the Caucasus are the same. It seems that this haplotype was widespread early in the Holocene. Curiously, the Motala hunter-gatherers also carry the East Asian haplotype around their derived EDAR variant.

I don’t know what to make of this. My intuition is that if a haplotype like this is so widespread nearly ~10,000 years ago recombination would have broken it apart into smaller pieces so that haplotype structure would be easier to discern. As it is that doesn’t seem to be the case.

And we also don’t know what’s going on withSLC24A5. Obviously it impacts skin color. It has been shown to do so in admixed populations. But it is hard to believe that that is the sole target of natural selection here.

13 thoughts on “SLC24A5 is very important, but we don’t know why

  1. “there was a movement of pastoralists with distant West Eurasian ancestry who brought cattle to southern Africa, and so resulted in the ethnogenesis of groups such as the Nama people (there is also Y chromosomal work by Henn on this).”

    In case you missed it: A paper last month showed that the source of this is a movement of East African pastoralists into South Africa <1500 years ago:
    http://biorxiv.org/content/early/2017/06/05/145409

    The EA Pastoralists themselves were ~50% Eurasian and EA admix in Khoi-San estimated ~15%; given the apparently strong natural selection for the allele it's likely that it spread very quickly among Khoi-San

  2. SLC24A5 and West Eurasian admixture most likely made its way to the Khoisan population from Cushites who migrated Southwards from the Horn into “Bantu” Africa and this introduced Cushitic/West Eurasian admixture into some population groups that far south in Africa (e.g. Tutsi) and from there it spread.

  3. This does seem to be yet another area where Greg Cochran got it right again–this light skin gene seems to have advantages outside of both agricultural diet and vitamin consumption.

    I do wonder if there was anywhere that *dark* skin has been selected for in the last 10,000 years. A pet theory of mine is that dark skin would have been favored in the extremely radiation-rich environment of the Pacific Islands, and that Melanesian introgression into the (probably lighter-skinned) China-derived Lapita people could have introduced such alleles.

  4. SLC24A5 and West Eurasian admixture most likely made its way to the Khoisan population from Cushites who migrated Southwards from the Horn into “Bantu” Africa and this introduced Cushitic/West Eurasian admixture into some population groups that far south in Africa (e.g. Tutsi) and from there it spread.

    nilotic speakers more likely IMO.

  5. Keep an eye out for a paper that will be published sometime next month in Science, concerning novel loci implicated in human skin pigmentation! We looked at 4 genes, aside from SLC24A5, that may contribute to skin pigmentation, especially in African populations.

  6. Keep an eye out for a paper that will be published sometime next month in Science, concerning novel loci implicated in human skin pigmentation! We looked at 4 genes, aside from SLC24A5, that may contribute to skin pigmentation, especially in African populations.

    Fascinating. I was just thinking when I read Razib’s post about the mysteriously light skin of some of the Igbo, who can range all the way from standard West African dark brown to skin tones within the light range for African Americans (even among men, who presumably aren’t beaching their skin). Unlike any other lighter-skinned group in Africa, however, no test I am aware of has shown Eurasian admixture.

  7. Unlike any other lighter-skinned group in Africa, however, no test I am aware of has shown Eurasian admixture.

    there are probably low levels of admixture. yoruba have it for sure. i have the word of david reich on this 🙂

  8. @Razib That 3,000 year old Tanzanian that you described on Twitter, one of the ancient genomes to be featured in Skoglund et al. 2017, was likely a early South Cushitic speaker. This would explain why his genetic profile is almost identical to Somalis and other lowland Cushitic speakers like the Afar and Oromo, and fits well with the archeological record, e.g. Savanna Pastoral Neolithic. Nilotic languages were only introduced to the SE Africa ~500-1000 years ago, and speakers like the Maasai and Datog most certainly once spoke South Cushitic languages but shifted to their current tongues, this would’ve involved some admixture. The Dinka are a more adequate proxy for Nilotic admixture, not the Maasai/Datog who owe up to ~50-70% of their ancestry to a Somali-like population. Groups like the Iraqw in Tanzania also still speak South Cushitic languages. The East African admixture in the Khoisan was likely introduced to the region with the arrival of Khoe-Kwadi (xSandawe) speaking agro-pastoralists from SE Africa who had exchanged gene-flow with an early South Cushitic speaking population, represented by that 3,000 year old Tanzanian and groups like the Iraqw.

    2. I highly doubt that the Yoruba or Igbo possess any meaningful Eurasian ancestry. That was suggested some time ago, but it’s more likely imho – based on the findings of Schlebusch and soon Skoglund – that West-Central Africans derived from an admixture event between “ancient East Africans” (a sister group to OOA) and a “Basal African” population slightly more basal than even Ballito Boy/Khoisan speakers. At a ratio of approximately 70:30.

  9. i spoke extensively with both skoglund and reich at this conference (i had lunch with david).

    This would explain why his genetic profile is almost identical to Somalis and other lowland Cushitic speakers like the Afar and Oromo

    skoglund specifically contrasted this individual with somalis. it possesses no “eastern farmer” ancestry. they do.

    . I highly doubt that the Yoruba or Igbo possess any meaningful Eurasian ancestry.

    david reich was adamant that there is eurasian admixture, because they detect neanderthal admixture. i asked him about his earlier assertions on this issue and he reiterated it quite emphatically. for whatever reason it is not a publisheable unit, but it is a finding they see consistently. (their assumptions may be wrong, who knows)

  10. @Razib

    1. What did he mean by “eastern farmer”? As in Iranian or are you talking about something further east? All of the analyses I’ve seen have modeled Somalis as approximately 65% East African and 35% Levantine Neolithic, nearly identical to this ancient Tanzanian individual. Somalis are a coastal population with ancient links to the Indian Ocean, so that might explain the “eastern farmer” ancestry he’s referring to, but I’d rather not speculate.

    2. I personally don’t share Reich’s confidence, but maybe he knows something I don’t. This has been suggested before (detection of Neanderthal signals), but I think it’s more likely that we’re actually detecting minor gene-flow from an archaic human population into the “Basal African” ancestors of the Yoruba?

  11. 1. What did he mean by “eastern farmer”? As in Iranian or are you talking about something further east? All of the analyses I’ve seen have modeled Somalis as approximately 65% East African and 35% Levantine Neolithic, nearly identical to this ancient Tanzanian individual. Somalis are a coastal population with ancient links to the Indian Ocean, so that might explain the “eastern farmer” ancestry he’s referring to, but I’d rather not speculate.

    yes somalis share drift with the ancient iranian samples from the zagros. the ancient tanzanian individual does not share drift with the ancient iranian samples. it shares drift exclusively with the pre-pottery neolithic levant samples.

    and he mentioned somalis only as an example. it was clear in context that iranian drift is found throughout the horn of africa. he was very clear in conversation and the talk and that tanzanian individual shares drift only with levant neolithic.

    re: similarities in models, you know that % that output from models vary depending on the parameters you’ve fixed those models with. in the context of ss-african pops eurasians as a whole will look rather similar % wise.

    but I think it’s more likely that we’re actually detecting minor gene-flow from an archaic human population into the “Basal African” ancestors of the Yoruba?

    that strikes me as plausible. i brought it up in conversation that there was no follow up so it didn’t pan out. david was clear that no, it was there (admixture from neanderthals and eurasians). of course he sees the basal african cline from cameroon to senegambia.

    the issue with conflating archaic african with neanderthal is that it would be strange if this was a clade against modern humans. possible, but unlike. order of possibility i’d think

    archaic african outgroup to (neanderthal – modern) > neanderthal outgroup to (archaic african – modern) > modern outgroup to (neanderthal – archaic)

    in the last scenario you’d conflate gene flow from either into moderns.

Comments are closed.