Latitudinal Clines of the Human Vitamin D Receptor and Skin Color Genes

Abstract

The well-documented latitudinal clines of genes affecting human skin color presumably arise from the need for protection from intense ultraviolet radiation (UVR) vs. the need to use UVR for vitamin D synthesis. Sampling 751 subjects from a broad range of latitudes and skin colors, we investigated possible multilocus correlated adaptation of skin color genes with the vitamin D receptor gene (VDR), using a vector correlation metric and network method called BlocBuster. We discovered two multilocus networks involving VDR promoter and skin color genes that display strong latitudinal clines as multilocus networks, even though many of their single gene components do not. Considered one by one, the VDR components of these networks show diverse patterns: no cline, a weak declining latitudinal cline outside of Africa, and a strong in- vs. out-of-Africa frequency pattern. We confirmed these results with independent data from HapMap. Standard linkage disequilibrium analyses did not detect these networks. We applied BlocBuster across the entire genome, showing that our networks are significant outliers for interchromosomal disequilibrium that overlap with environmental variation relevant to the genes' functions. These results suggest that these multilocus correlations most likely arose from a combination of parallel selective responses to a common environmental variable and coadaptation, given the known Mendelian epistasis among VDR and the skin color genes.

Keywords: adaptation; epistasis; linkage disequilibrium; network analysis; skin color; vitamin D.

Figure 1

Allelic networks identified by BlocBuster. Each dot represents an allelic node, identified by its SNP and nucleotide state. Out of the total of 128 nodes in the data set, only the nodes that had an edge connecting them to another node are depicted. The 52 edges connecting nodes represent Custom Correlation Coefficient (CCC) values ≥ 0.65. These 52 edges defined seven discrete networks of alleles, labeled 65_1 through 65_7.

Figure 2

Significant latitudinal clines for haplotype 65_3 identified by BlocBuster in the skin color gene MC1R. Haplotype 65_3 shows a significant nonlinear regression with latitude due to the low frequency of this haplotype in sub-Saharan Africa, but exclusion of the sub-Saharan population still results in a significant linear regression with latitude, as shown by the thick straight line.

Figure 3

A plot of the frequencies of network 65_1 and its single-gene components vs. latitude. The components are the extended haplotype 70_1 in TYRP1, the extended promoter haplotype 70_5 in VDR, and a single nucleotide polymorphism in TYR. The line shows only the regression for the multilocus network 65_1.

Figure 4

A plot of the frequencies of network 65_2 and its single-gene components vs. latitude. The line shows only the regression for the multilocus network 65_2. SNP, single nucleotide polymorphism.

Figure 5

Frequencies of 65_2R4, 65_2, and 84_2 vs. latitude. (A) Frequency of individuals with 65_2R4 vs. latitude in 10 populations surveyed here and in eight HapMap populations. The line indicates the linear regression when the Han Chinese, Japanese, and sub-Saharan populations are excluded. (B) Frequency of individuals with 65_2 vs. latitude in 10 populations surveyed here and with 84_2 in eight HapMap populations. The line indicates the linear regression when the Han Chinese and Japanese populations are excluded.