The genetics of Scottish hair color variation

In the aughts the elucidation of human pigmentation genetics was of one the major successes of 'omic' techniques. The fact that humans exhibit some continuous variation in complexion was strongly suggestive that more than one gene was at work to generate the range of the phenotype. On the other hand pedigree based studies going back to the 1960s suggested that only a modest number of large effect genetic variants were producing the variance. Today we can say with reasonable certainty that about half a dozen genes account for almost all the between population variation in pigmentation. For example, on the order of 1/3 of the difference between Africans and Europeans in regards to skin pigmentation can be accounted for a genetic difference on the gene SLC24A5. Unlike height the genetic architecture of human pigmentation variance was ideally suited to being explored by the power of contemporary GWA; moderate frequency and moderate effect variants are numerous.

Of course pigmentation is more than skin. The original work in mice paid attention to coat color as well as flesh, and humans vary in hair and eye color in addition to skin. While skin color variance is controlled by genetic differences across half a dozen genes, most of the variance in human eye color seems to be localized around the OCA2/HERC2 locus (we're talking blue vs. brown mostly). Additionally, there is naturally an overlap between the genes which have been implicated in the variation of skin, hair and eye pigmentation, though the details of exact relationship between the genetic architecture of these traits must be kept in mind so as to explain relatively dark-skinned individuals with light eye color (such as the actress Stacey Dash, who is of Native American, African American and European American ancestry).

Nevertheless, one concern in modeling hair and eye color as opposed to skin color is that in the case of the latter there has been significant quantitatization in the form of skin reflectance values (those with dark skin naturally absorb more radiation from a source). With eye and hair color categorical labeling discards useful information in the form of phenotypic variance which is artificially collapsed into a few classes. A new paper attempts to take a step forward and correct this defect, at least in relation to hair color. Genetic determinants of hair and eye colour in the Scottish and Danish populations:

We assayed the hair of a population of individuals of Scottish origin using tristimulus colorimetry, in order to produce a quantitative measure of hair colour. Cluster analysis of this data defined two groups, with overlapping borders, which corresponded to visually assessed dark versus red/light hair colour. The Danish population was assigned into categorical hair colour groups. Both cohorts were also assessed for eye colour. DNA from the Scottish group was genotyped at SNPs in 33 candidate genes, using 384 SNPs identified by HapMap as representatives of each gene. Associations found between SNPs and colorimetric hair data and eye colour categories were replicated in a cohort of the Danish population. The Danish population was also genotyped with SNPs in 4 previously described pigmentation genes. We found replicable associations of hair colour with the KITLG and OCA2 genes. MC1R variation correlated, as expected, with the red dimension of colorimetric hair colour in Scots. The Danish analysis excluded those with red hair, and no associations were found with MC1R in this group, emphasising that MC1R regulates the colour rather than the intensity of pigmentation. A previously unreported association with the HPS3 gene was seen in the Scottish population. However, although this replicated in the smaller cohort of the Danish population, no association was seen when the whole study population was analysed.

In this paper they looked at the genes which have been implicated previously in pigmentation variation. The main value-add seems to be mapping the genetic variants of interest onto three quantitative dimensions of hair color variance, L*(dark/light), a*(red) b*(yellow). Remember that hair has two primary pigments, eumelanin, the brown one which we are familiar with, and pheomelanin, which results in a reddish tint. Many red-haired individuals have relatively low levels of eumelanin combined with normal or high levels of pheomelanin, resulting in their distinctive complexion. When very dark-haired people bleach their hair initial treatments may result in a copper-color as the eumelanin no long masks phenomelanin (I know this from personal experience). It seems that red-hair can be conceived as a serial loss of function across many genes; in other words, a wide range of pigment producing genes have to have reduced function.*

Here's a table which shows the relationship between hair and eye color:

You can see the associations through visual inspection, but it is also statistically significant. Those with dark hair tend to have dark eyes, and those with light eyes tend to have light hair. But the associations are not exact, and that is informative as well.

Below I've reformatted table 3. What you see are genes, the quantitative trait of red shading, and the categorical value of the trait associated with a particular genetic marker. Remember that R² represents the proportion of the variance of the trait, in this case the value of a*, which can be explained by the variance in genes.

Note that MC1R is validated as a major predictor of red hair color. Quantitative measurement should confirm what we already know, making our comprehension more precise, and possibly smoke out relationships which were masked previously because of the manner in which one classified traits. The KITLG locus had previous been associated with increased odds of blonde hair, and here we see its quantitative relationship to red hair. A similar table with eye color confirms the importance of OCA2.

Here's another way to visualize the effect an SNP on KITLG has on hair color (red and yellow shading respectively) by genotype:

Many of these findings were replicated in a Danish sample, but since they didn't use a quantitative measure there wasn't as much value-add.

Citation: BMC Genetics 2009, 10:88doi:10.1186/1471-2156-10-88

* These rules are not hard & fast, there are daywalkers, individuals with normal pigmentation excepting their red hair.