Another day, and another genome-wide association study. Genetic determinants of hair, eye and skin pigmentation in Europeans:
…We carried out a genome-wide association scan for variants associated with hair and eye pigmentation, skin sensitivity to sun and freckling among 2,986 Icelanders. We then tested the most closely associated SNPs from six regions–four not previously implicated in the normal variation of human pigmentation–and replicated their association in a second sample of 2,718 Icelanders and a sample of 1,214 Dutch. The SNPs from all six regions met the criteria for genome-wide significance. A variant in SLC24A4 is associated with eye and hair color, a variant near KITLG is associated with hair color, two coding variants in TYR are associated with eye color and freckles, and a variant on 6p25.3 is associated with freckles. The fifth region provided refinements to a previously reported association in OCA2, and the sixth encompasses previously described variants in MC1R.
A methodological point which they acknowledge: their assessments of phenotype were rather subjective and contingent upon self-reporting. When it comes to eye, and to a lesser extent hair, color think this is not that big of a deal, but I think that there are probably going to be serious issues in subjective self-reporting of skin color in an already fair complected population. I would bet that that is one reason they didn’t find an association with skin color on some of the SNPs.
They basically confirmed the OCA2 finding in regards to it being the main determinant of blue vs. brown eyes. Additionally, there are other loci of smaller effect, primarily TYR. I was surprised that they noted that several SNPs seem to distinguish more strongly between blue and green eye colors than between either and brown. I now think binning phenotypes into blue vs. non-blue or light vs. dark might miss some serious genetic action.
While the association of eye color with a few large loci of major effect seems supported by this study, the results for hair color seem are complicated and equivocal. Not when it comes to red hair of course, MC1R loss-of-function mutations which turn off brown pigment production & up-regulate red-yellow pigment have long been known. But I was more curious about the blonde vs. brown hair color variation. The signals they picked up in this study were numerous and I haven’t really teased them apart in my head yet, but, it seems to highlight why an equivalent of the OCA2 findings haven’t popped up for explaining the prevalence of blond hair. Just not that simple.
Finally, another story here is pleiotropy. The association of the red hair associated SNPs on MC1R with freckles, very fair skin and UV sensitivity has long been known, but this paper points to multiple other SNPs which have multi-pronged affects. Interestingly, one SNP, A allele of rs1540771, is reported to be associated with UV skin sensitivity and brown hair! This might be a spurious finding, but suggests that a SNP doesn’t have to have the same direction of effect globally in terms of pigmentation (i.e., the implication is that the SNP might increase or maintain hair pigmentation and decrease skin pigmentation). As some of you might know I’m fascinated by the phenomenon of very dark skinned blondes in Melanesia; so this sort of genetic process whereby a SNP can have countervailing tendencies in different regions of the body might explain some of that. In any case, I do suspect that a substantial, though not all, of the eye and hair color variation might have a link to selection for fair skin.
Addendum: So what’s up with all the moderate minor allele frequencies? If selection is at work how come it hasn’t fixed? A form of balancing selection? Or perhaps the selective sweep is still running through the population….