|
Post by Admin on Oct 7, 2021 21:17:38 GMT
Note 2: Nuclear data capture analysis Stephan Schiffels, Alexander Peltzer, Anja Furtwängler, Verena J. Schuenemann
Sample selection for nuclear capture For the nuclear capture, we chose 40 samples from our mitochondrial results to prepare libraries for further nuclear capture. Our choice was based on obtained coverage of the respective sample, low mitochondrial contamination (<3%) and significant damage patterns to only include samples with patterns of ancient origin into our nuclear capture. Furthermore, we estimated potential yield based on endogenous content in our previously screened samples. The final choice of these samples can be seen in Supplementary Data 2.
Capture Samples that were further investigated for nuclear DNA content underwent Uracil-DNA-Glycosylase (UDG) treatment. Two aliquots of 20 µl of each DNA extract of the samples selected for the enrichment of nuclear DNA were used to generate two sequencing libraries per sample after Meyer and Kircher (22). To reduce the effect of damage-derived C to T and G to A misincorporations typical for ancient DNA, the extracts were treated with USER™ enzyme (New England Biolabs) containing Uracil-DNA-Glycosylase and Endonuclease VIII (27). To each library a library specific set of indexing sequences were added to each end of the library fragments by amplification with tailed primers in a 10 cycle PCR (23). For all indexed libraries an additional amplification was performed. For each library 20 µl were used for four reactions with a total volume of 100 µl respectively and the following concentrations: 1x AccuPrime™ Pfx Reaction Mix, 0,3 µm IS 5, 0.3 µm IS 6 and 0.02 Units/µl AccuPrime™ Pfx DNA Polymerase (Invitrogen). The amplification product was purified with the MinElute Purification Kit (Qiagen) after the manufactures instruction and an aliquot was quantified with the Agilent 2100 Bioanalyzer according to the manufactures instructions using the Agilent 2100 Bioanalyzer DNA1000 chip (Agilent Technologies) to determine the distribution of the fragment size and the concentration of the library. The non-UDG and UDG treated libraries were enriched by hybridization to probes targeting approximately 1.24 million genomic SNPs. The target SNPs consist of panel 1 and 2 as described in Mathieson et al. (41) and Fu et al. (12), a large proportion of which are also present on the Affymetrix Human Origins, the Illumina 610-Quad and the Affymetrix 50k array. The probes had a length of 52nt covering a region of 105nt flanking the target SNPs in the center. The enrichment was performed as described in Fu et al. (25). The two UDG treated libraries per sample were pooled for the capture while the non-UDG treated libraries were captured separately. After the last purification all enriched libraries were pooled for sequencing.
PCA We performed principal component analysis on the joined data set using the “smartpca” software from the Eigensoft package (63). For the plot shown in Supplementary Fig. 3, we used a selected set of European populations: Abkhasian, Adygei, Albanian, Armenian, Balkar, Basque, BedouinA, BedouinB, Belarusian, Bulgarian, Canary_Islanders, Chechen, Croatian, Cypriot, Czech, Druze, English, Estonian, Finnish, French, Georgian, Greek, Hungarian, Icelandic, Iranian, Italian_South, Ashkenazi_Jew, Georgian_Jew, Iranian_Jew, Iraqi_Jew, Libyan_Jew, Moroccan_Jew, Tunisian_Jew, Turkish_Jew, Yemenite_Jew, Jordanian, Kumyk, Lebanese, Lezgin, Lithuanian, Maltese, Mordovian, North_Ossetian, Norwegian, Orcadian, Palestinian, Russian, Sardinian, Saudi, Scottish, Sicilian, Spanish_North, Spanish, Syrian, Turkish, Ukrainian. For the plot shown in Figure 4a, we added the following African populations: Ethiopian_Jew, Dinka, Luhya, Algerian, Mozabite, Saharawi, Somali, Yoruba, Mota, Mandenka, Biaka.
Supplementary Table 3: Y-Chromosomal haplotype results.
Sample ID YHaplogroup Comment
JK2134 J This individual was assigned to haplogroup J based on mutations: CTS8938/PF4577: 18567169T→G, F2817/PF4579: 18695159C→T, F4299/PF4589: 21144431T→A, S22619/Z7820: 21144432C→A, F4300: 21144433T→A, YSC0000228: 22172960G→T, L778/PF4616/YSC0000236: 23088142T→C.
JK2888 E1b1b1a1b2 This individual was assigned to haplogroup E1b1b1a1b2 based on mutation: V22: 6859957T→C, and to upstream E1b1b1a: CTS2661: 14410669C→T and E1b1b1: PF1619: 13848122T→C, CTS2620: 14393170A→C, M5360: 23618826C→T.
JK2911 J This individual was assigned to haplogroup J based on mutations: CTS687/PF4503:6953311A→T, CTS1250/PF4510/YSC0001255:7296343G→T, PF4513/NA:7759610C→T, PF4519/NA:8669451C→G, PF4524/NA:10009851G→A, PF4530/NA:13597365C→T, CTS2769/PF4538:14476551T→A, F1973/PF4546/YSC0001304:15581303G→A, F2114/PF4551:16262942G→A, CTS5628/PF4555:16401405C→G, CTS5678/PF4556:16427564A→T, F2502/PF4564:17495914G→A, CTS7738/PF4568:17637446T→C, CTS7832/PF4569:17693210A→G, F2769/PF4576:18552360G→C, F2973/PF4585/YSC0001312:19194316C→T, F4299/PF4589:21144431T→A, S22619/Z7820:21144432C→A, F4300/NA:21144433T→A, F3176/PF4592/YSC0001314:21329083T→C, PF4595/NA:21858778C→A, YSC0000228/NA:22172960G→T, M304/Page16/PF4609:22749853A→C, L778/PF4616/YSC0000236:23088142T→C, CTS11571/PF4617:23163701C→A, CTS11750/PF4618/YSC0001250:23250894C→T, CTS12047/YSC0001253:23443976A→G
|
|
|
Post by Admin on Oct 7, 2021 22:15:08 GMT
Note 3: Mitochondrial DNA sequence processing and alignment Alexander Peltzer, Kay Nieselt, Wolfgang Haak
Adapter sequences were trimmed from the 3’ ends of reads using Clip&Merge v1.7 (57), requiring an overlap of 10 bp between the adapter and the corresponding read for subsequent read merging. A minimum read length filtering of 25 nt has been used to ensure that reads, which were too short, were not incorporated into the further analysis. The resulting sequences have been aligned using the Burrows-Wheeler-Aligner (BWA) version 0.7.12 (73), with the parameters –n 0.01 and –l 1000 to disable seeding and allow more mismatches due to the age of the samples. All reads have been aligned against the GRCh37 build of the human genome with the Reconstructed Sapiens Reference Sequence (RSRS) (74) to account for NUMTs.
Note 4: Sequence based mitochondrial analysis Alexander Peltzer, Martyna Molak, Wolfgang Haak
Modern comparative data
To allow for further comparisons between the ancient samples from PrePtolemaic, Ptolemaic and Roman time periods in Egypt, we used an extensive private database of 37,368 HVR-I sequences (nucleotide positions 16,059- 16,400) from contemporary populations in Europe, the Near East and Africa (29). These have been pooled according to geographical information provided in the original publications and collected from literature. All sequence data used were updated from the original publication to the mtDNA phylogeny of phylotree.com v.17 (75). Our reference dataset consisted of these populations: I. Populations from central, eastern, western and southern Europe, representing the genetic context of Slovenia, Hungary, Italy, France, Serbia, England, Faroe-Islands, Finland, Norway, Sweden, Spain and Iceland. II. Populations from North-Africa, representing the genetic context of Egypt, Mauretania, Tunisia, West-Sahara and Morocco. III. Populations from Sub-Saharan Africa, represented by populations from Sudan, Ethiopia, Burkina-Faso, Cameroon and Guinea. IV. Populations from the Near East, represented by populations from Syria, Iran, Iraq, Turkey, Yemen, Kuwait, United Arab Emirates, Lebanon, Israel, Oman, Saudi-Arabia, Qatar and Jordan. V. Populations from the Caucasus and West Asia, represented by populations from Georgia, Armenia and Pakistan. Citations for the studies reporting the original HVR-1 sequences are listed in Supplementary Data 4.
Geographical mapping Similar to our approach in the MDS analysis, we evaluated the geographic distribution of mtDNA variance using a geographical mapping of our calculated FST values on our HVR-1 dataset and our corresponding ancient populations. Genetic distances were computed in Arlequin v.3.5.2.2 and afterwards combined with longitudes and latitudes and then subsequently mapped to a geographic map. We applied the R packages ggmap and ggplot2 in a custom script to plot FST values to our geographic setting (76), creating a map of FST shown in Supplementary Fig. 1. For each of the populations used in this study, we defined a point of reference, which described best the available geographic information. In cases where we did not have exact geographic information, we used the innermost longitudinal / latitudinal value defined for the respective country.
Note 5: Frequency based mitochondrial analysis Alexander Peltzer, Wolfgang Haak, Kay Nieselt
Test of population continuity We followed an approach first used and defined by Brandt et al. (29) by first generating counts of 22 haplogroups determined manually to be most descriptive for our three ancient populations. Our priors on c (see Supplementary Data 5) were set based on the number of generations between two selected populations to test, with an assumed generation length of 25 years for our ancient populations. To determine potential influences of chosen effective population size Ne on our computations, we iteratively evaluated a range of population sizes for the specific region (see Supplementary Data 5) and investigated whether these changed our analysis significantly. The basic assumption of the method described by Brandt et al. (29) is to test a model of population continuity, e.g. compare allele frequencies of our ancient population with modern typed data from the same location. As Brandt et al. (29) already state, the idea is based on the assumption that mtDNA can only obtain new alleles by migration from an outside source and not by genetic drift only. The idea is therefore to test whether a model of genetic drift fits to our given dataset and then test whether this model fits the data or has to be rejected. Internally this is done using a hypergeometric distribution, which is then evaluated using a MCMC package implemented in the scope of Brandt et al (29). We used the dataset of 120 Ethiopian and 100 Egyptian modern mtDNA genomes from Pagani et al. (17) to determine whether we can detect genetic discontinuity between ancient and modern Egypt. However, the method was only able to detect strong signs of discontinuity between our ancient populations and modern Ethiopians. For modern Egyptians, neither a significant value supporting discontinuity nor continuity was observed (see Supplementary Data 5). To ensure that the prior mean on the drift parameter t/N (where t is the number of generations separating the populations and N is the effective population size) did not influence our results, we chose our N based on historic written data (see Supplementary Data 5) and furthermore evaluated a range of N=(20K, 50K, 70K, 100K) effective population to account for possible wrong choices on N in our basic assumptions.
|
|
|
Post by Admin on Oct 17, 2021 20:41:57 GMT
Shades of complexity: New perspectives on the evolution and genetic architecture of human skin
Abstract Like many highly variable human traits, more than a dozen genes are known to contribute to the full range of skin color. However, the historical bias in favor of genetic studies in European and European-derived populations has blinded us to the magnitude of pigmentation's complexity. As deliberate efforts are being made to better characterize diverse global populations and new sequencing technologies, better measurement tools, functional assessments, predictive modeling, and ancient DNA analyses become more widely accessible, we are beginning to appreciate how limited our understanding of the genetic bases of human skin color have been. Novel variants in genes not previously linked to pigmentation have been identified and evidence is mounting that there are hundreds more variants yet to be found. Even for genes that have been exhaustively characterized in European populations like MC1R, OCA2, and SLC24A5, research in previously understudied groups is leading to a new appreciation of the degree to which genetic diversity, epistatic interactions, pleiotropy, admixture, global and local adaptation, and cultural practices operate in population-specific ways to shape the genetic architecture of skin color. Furthermore, we are coming to terms with how factors like tanning response and barrier function may also have influenced selection on skin throughout human history. By examining how our knowledge of pigmentation genetics has shifted in the last decade, we can better appreciate how far we have come in understanding human diversity and the still long road ahead for understanding many complex human traits.
1 INTRODUCTION Skin pigmentation is often held out among complex traits as a relatively simple phenotype to study as the melanogenesis pathway is well characterized and the phenotype comparatively straightforward to measure. This is a mischaracterization of a trait, which is, like many widely variable traits, highly polygenic with population-specific alleles, complex epistatic interactions, pleiotropic effects, and gene-specific evolutionary trajectories shaped by global and local selective pressures and drift. Much of our current knowledge of the genetic architecture of human pigmentation variation has been centered on European populations. These studies fall into two primary groups: those identifying genetic loci that explain differences in pigmentation between European and non-European populations (Beleza et al., 2013; Bonilla et al., 2005; Lamason et al., 2005; Norton, Koki, & Friedlaender, 2007; Shriver et al., 2003) and those identifying loci that explain pigmentation variation within European populations (Candille et al., 2012; Eiberg et al., 2008; Flanagan et al., 2000; Kayser et al., 2008; Liu et al., 2015; Praetorius et al., 2013; Sulem et al., 2007; Valverde, Healy, Jackson, Rees, & Thody, 1995). While these studies built a strong foundation for our understanding of the genetics of human pigmentation variation, this Eurocentric bias has also limited our understanding of the genetic architecture of pigmentation globally. We now know that loci important in European pigmentation variation may not be relevant in non-European populations and may be small in number compared toloci influencing pigmentation outside of Europe (Crawford et al., 2017; Martin et al., 2017).
In the following pages, we will review some of the major changes in our understanding of the genetic bases of skin color as more detailed studies are carried out in more diverse populations. In doing so, we aim not to produce a comprehensive list of the genes contributing to skin color variation in humans, but to highlight the complex effects, distributions, and interactions among a few genes of particular note. Then, we will delve into how population history and selective pressures have shaped the genetic architecture of pigmentation and consider two alternative functions of skin—tanning response and barrier function—which recent evidence suggests may also exert selective pressures on skin. Finally, we will consider important technological advances in how skin color is measured and predicted that both inform and are informed by the field's efforts to capture a more accurate global picture of skin color and genetic diversity.
CHEDDAR MAN observed alleles Eye, Hair & Skin Pigmentation
SNP GENE rs16891982 SLC45A2 C rs28777 SLC45A2 C rs12203592 IRF4 T rs683 TYRP1 C rs1042602 TYR C rs1393350 TYR G rs12821256 KITLG T rs12896399 SLC24A4 G rs2402130 SLC24A4 A rs17128291 SLC24A4 A rs1545397 OCA2 A rs1800414 OCA2 T rs1800407 OCA2 C rs12441727 OCA2 G rs1470608 OCA2 G rs1129038 HERC2 T rs12913832 HERC2 G rs2238289 HERC2 A rs6497292 HERC2 A rs1667394 HERC2 T rs1426654 MYEF2 SLC24A5 G rs3114908 ANKRD11 CT rs1805006 MC1R C rs2228479 MC1R G rs11547464 MC1R AG rs1805007 MC1R C rs1110400 MC1R T rs1805008 MC1R C rs885479 MC1R G rs8051733 DEF8 A rs6059655 RALY G rs6119471 ASIP C rs2378249 PIGU A
Additional for Eye & Hair pigmentation rs4959270 EXOC2 AC rs796296176 MC1R C rs1805005 MC1R G rs201326893 MC1R C rs1805009 MC1R G Hair Texture rs17646946 TCHHL1 G rs7349332 WNT10A C
|
|
|
Post by Admin on Oct 17, 2021 22:17:01 GMT
2 OLD GENES, NEW UNDERSTANDING Much in the field has changed since the topic of pigmentation genetics was last reviewed in the Yearbook of Physical Anthropology in 2007 (Parra, 2007). At that point, the vast majority of genetic variation had only been deeply characterized in a limited number of individuals, let alone populations. The largest catalogue of common variants was the International HapMap Project which had released its data just a few years prior and focused exclusively on a small sample of white Americans with northern and western European ancestry, Yoruban individuals from Ibadan, Nigeria, Han Chinese individuals from Beijing, and Japanese individuals from Tokyo (The International HapMap Consortium, 2003). The geographic distribution of these individuals artificially created an impression of discontinuous genetic variation along (U.S.) racial lines. The Human Genome Diversity Project (HGDP) attempted to rectify this limited geographic coverage, but had less genetic data and was still hardly representative of the full geographic or genetic distribution of the human species (Cavalli-Sforza, 2005). Both projects lacked any associated phenotype data.
Over the past decade, rapidly decreasing sequencing costs alongside deliberate efforts to better explore the rich genetic and phenotypic variation within continental groups and the role that geography, history, and selection have played in shaping modern pigmentation diversity has led to a deeper understanding of the interactions among genes (Veeramah & Hammer, 2014). While the geographic distribution and ethnic diversity of populations studied has improved (Table 1), the historical bias in favor of relatively wealthy countries persists and limits our understanding of this culturally and biologically important phenotype. Figure 1 highlights the greater number of studies with large sample sizes performed in European and European-derived populations, meaning that much human genetic diversity is missed. As pigmentation genetics research is undertaken in understudied populations in Africa, South Asia, and Southeast Asia, we increasingly recognize that our understanding of the function of genetic variants is highly bound by the context in which they were first identified. The evolutionary history of the human species, which inhabits wildly diverse environments, adds an additional layer of complexity where the contribution of the same gene—or even the same allele—in determining pigmentation variation differs across populations due to interactions across many genes, local adaptation, and patterns of genetic drift and gene flow across our broadly geographically dispersed species. The comparatively limited amount of research done in non-European populations suggests that even genes like MC1R, SLC24A5, and OCA2 that have been nearly exhaustively characterized in European populations, are incompletely understood and have more to teach us about the complexity of pigmentation genetics.
Table 1. Summary of skin color genetics association studies discussed in this manuscript
Study Study typea Trait N Continent Population Nb Associated genes Abe, Tamiya, Nakamura, Hozumi, & Suzuki, 2013 Candidate gene (4) Melanin Index 456 Asia Japanese 1 OCA2 Ang et al., 2012 Candidate gene (2) Melanin Index 492 Asia Peninsular Malaysia 2 SLC24A5, SLC45A2 Beleza, Johnson, et al., 2013 Admixture mapping Melanin Index 699 Africa Cape Verdean 4 SLC24A5, TYR, OCA2, SLC45A2 Candille et al., 2012 GWAS Melanin Index 469 Europe Ireland (146), Poland (72), Italy (109), Portugal (142) 0 Crawford et al., 2017 GWAS Melanin Index 1,570 Africa Ethiopia, Tanzania, Botswana 6 SLC24A5, MFSD12, DDB1, TMEM138, OCA2, HERC2 Eaton et al., 2015 Candidate gene (9) Melanin Index 419 Asia East Asian Ancestry (Canada) 1 OCA2 Han et al., 2008 GWAS Melanin Index 10,755 Europe European Ancestry (U.S. and Australia) 3 SLC24A4, IRF4, SLC45A2 Hernandez-Pacheco et al., 2017 GWAS Melanin Index 658 Americas Hispanic/Latinos from Puerto Rico (285) and African Americans (373) 4 SLC24A5, SLC45A2, BEND7, PRPF18 Jonnalagadda, Norton, Ozarkar, Kulkarni, & Ashma, 2016 Candidate gene (5) Melanin Index 533 India Western India 1 SLC24A5 Liu et al., 2015 GWAS Self-Report 17,262 Europe Dutch (5,857), Australian (4,296), UK (5,278) 5 SLC45A2, IRF4, HERC2/OCA2, MC1R, ASIP Marcheco-Teruel et al., 2014 Candidate gene (15) Melanin Index 1,019 Americas Cuba 2 SLC24A5, SLC45A2 Martin, Lin, et al., 2017 GWAS Melanin Index 456 Africa KhoeSan 4 SLC24A5, TYRP1, SMARCA2/VLDLR, SNX13 Nan et al., 2009 GWAS Tanning response 15,155 Europe European Ancestry (U.S.) 5 SLC45A2, IRF4, TYR, OCA2, MC1r Norton et al., 2007 Candidate gene (6) Melanin Index Asia Island Melanesia 2 OCA2, ASIP Norton, Werren, & Friedlaender, 2015 Candidate gene (9) Melanin Index 583 Americas European Ancestry (U.S. and Canada) 3 SLC45A2, IRF4, HERC2 Paik et al., 2011 Linkage analysis Melanin Index 345 Asia Mongolia 4 GRM6, ATF1, WNT1, SILV/Pmel17 Quillen et al., 2011 Candidate (14) Melanin Index 515 Americas Mexico (95), U.S. Hispanics (247), Colombian (173) 4 SLC24A5, SLC45A2, OPRM1, EGFR Rawofi et al., 2017 GWAS Melanin Index 305 Asia East Asian Ancestry (Canada) & Chinese 1 ZNF804B Shriver et al., 2003 Admixture Mapping Candidate (3) Melanin Index 592 African/European African American (232), European American (187), and African Caribbean (173 U.K.) 2 TYR, OCA2 Stokowski et al., 2007 GWAS Melanin Index 968 South Asian Ancestry South Asian Ancestry 3 SLC24A5, SLC45A2, TYR Sulem et al., 2007 GWAS Skin sensitivity 6,918 European Iceland (7,246), Dutch (1,214) 1 ASIP Visconti et al., 2018 GWAS Tanning ability 17,6,678 European European & European Ancestry 20 PDE4B, RIPK5, PA2G4P4, SLC45A2, PPARGC1B, IRF4, AHR/AGR3, TRPS1, TYRP1, BNC2, EMX2, TPCN2, TYR, DCT, ATP11A, SLC24A4, HERC2/OCA2, MC1R, ASIP, KIAA0930 Zhang et al., 2013 GWAS Tanning ability 9,678 European European Americans 5 ASIP, HERC2, IRF4, MC1R, TYR a All genome-wide association studies (GWAS) considered more than 200,000 SNPs. The number of genes (many with more than one SNP) is indicated for candidate gene studies in parentheses. b The number of genes significantly associated is listed. For GWAS, 10−8 is used as cutoff, for candidate gene studies, associations are listed as reported in manuscript.
|
|
|
Post by Admin on Oct 18, 2021 3:52:20 GMT
Figure 1 Global distribution of skin pigmentation studies. For each study listed in Table 1, a circle is placed over the approximate location where the samples were collected and scaled to the number of individuals included. Samples of European ancestry collected in the U.S., Canada, or Australia were placed over Europe; samples collected in Canada from predominantly Chinese participants were placed in China 2.1 MC1R: lessons in genetic diversity One of the most well-studied pigmentation loci in or outside of Europe is the melanocortin-1 receptor, or MC1R gene. MC1R is a seven-pass transmembrane G-protein coupled receptor found on the surface of melanocyte cells. When the α-melanoctye stimulating hormone (α-MSH) binds to the Mc1r protein, eumelanin production occurs. However, mutations that interfere with this binding process can result in the production of red-yellow pheomelanin rather than brown or black eumelanin. Although small in size (951 bp), MC1R exhibits high levels of variation in populations across Eurasia (Harding et al., 2000; Rana et al., 1999) and pleiotropic influence on several pigmentation phenotypes. MC1R is perhaps most well-known for its role in influencing hair color, with multiple nonsynonymous variants strongly associated with red hair and fair skin color within European and European-derived populations (Box, Wyeth, O'Gorman, Martin, & Sturm, 1997; Duffy et al., 2004; Flanagan et al., 2000; Valverde et al., 1995). In addition to red hair and lighter skin pigmentation, derived alleles at MC1R are also associated with freckles, increased nevus count, and an increased risk of melanoma (Bishop et al., 2009; Flanagan et al., 2000). Two decades ago, MC1R was the only gene known to have a substantial effect on normal human pigmentation variation. The ratio of synonymous to nonsynonymous substitutions in MC1R in the genomes of humans compared to chimpanzees indicates a shared selective sweep in all modern humans which may have been associated with the loss of fur approximately 1.2 million years ago (Rogers, Iltis, & Wooding, 2004). When lightly pigmented skin under dark fur was exposed to the intense ultraviolet radiation (UVR) of equatorial Africa, a strong selective pressure would have favored darker skin in our hominin ancestors. Work on nucleotide substitutions at this locus in modern populations revealed a lack of variation in African populations and an abundance of non-synonymous substitutions in European populations. Two views were presented on the modern distribution of MC1R variation: it was argued to be under diversifying selection by some (Rana et al., 1999) and simply representing neutral drift by others (Harding et al., 2000). However, all sides agreed that the apparent lack of diversity of MC1R was evidence of functional constraint in the African continent. While functional constraint in the MC1R locus is a logical conclusion based on the evidence presented, the lack of variation was taken to mean that skin pigmentation itself was functionally constrained in Africa. This view was compounded by these studies' reliance on African Americans and immigrants from West Africa to represent the full spectrum of the continent's genetic and phenotypic diversity. Afro-descendant individuals have variable amounts of African ancestry that, particularly in the Americas, stems predominantly from West and Central Africa as a result of the trans-Atlantic slave trade (Bryc et al., 2010; Campbell & Tishkoff, 2010; Tishkoff et al., 2009; Zakharia et al., 2009). The use of African diaspora populations as representative of African genetic diversity is further problematic because of common adherence among scientists and study populations to the folk “one-drop rule” which ascribes Black racial identity to individuals with any amount of African ancestry regardless of multifactorial ancestry. Over the past decade, advances in our knowledge of African skin pigmentation genetics have been driven by a shift in populations used to represent diversity on this vast continent. The earliest studies in skin pigmentation genetics used the contrast in phenotype between northern Europeans and western Africans to uncover some of the genes playing a role in human skin pigmentation. This contrast between light and dark, derived and ancestral, was useful to some extent, but limited in what it could tell us about skin pigmentation genetics within the continent of Africa, which represents the most diverse gene pool in the world—both in terms of overall genetic diversity and skin pigmentation. Although studies of admixed populations can be very useful for identifying the alleles influencing traits that vary between the parental populations, it is important to recognize that these are association studies and, as such, favor the discovery of common alleles of large effect. Given the broader variation in skin color and greater genetic diversity in Africa, relative to Europe, admixture mapping studies favor the discovery of variants that have lightened the skin of Europeans (which is explained by fewer genes) over alleles that affect skin color variation within Africa (Beleza, Johnson, et al., 2013; Bonilla et al., 2005). Nevertheless, the greater diversity in African skin color and genetics is also evident in analyses of the relationship between individual admixture and skin reflectance in admixed populations as variance in skin color is greater in individuals with higher proportions of west African ancestry (Martin, Lin, et al., 2017; Parra, Kittles, & Shriver, 2004; Shriver et al., 2003). When studies are undertaken in other regions of Africa, a more complex picture of genetic and pigmentation variation emerges. In the last few years, the focus on geographically dispersed African populations within the continent, as opposed to African-derived populations in Europe and the U.S., has identified novel genes and complex interactions contributing to African skin pigmentation. These data, though relatively new, should not be surprising given what we have long known about the deep and complex evolutionary history of African populations and the varied effects of natural selection on populations living in vastly different UVR-regimes. Even for MC1R, thought to be tightly constrained across Africa, studies of individuals from southern Africa suggest that some relatively low levels of nonsynonymous MC1R variation may be tolerated or compensated for in lower UVR regions of the continent (John, Makova, Li, Jenkins, & Ramsay, 2003). Despite the overarching narrative of Africans as representatives of a homogenously darkly pigmented ancestral group, some have emphasized the diversity of pigmentation across the continent (Jablonski & Chaplin, 2000; Relethford, 1997). South African KhoeSan populations, in particular, have long been highlighted as an example of convergent evolution of lighter pigmentation in Africa. However, it was not until recently that large-scale research on the genetic architecture of skin color in Southern Africa was reported (Martin, Lin, et al., 2017). These authors describe associations with loci that had been previously discovered (SLC24A5, SLC45A2, and KITLG) and a novel association at SNX13. Interestingly, they show that a large portion of the phenotypic variance in these populations remains unexplained, suggesting that skin pigmentation genomics may be far more polygenic and complex—with 50 or more genes of small to moderate effect required to explain variation within the KhoeSan—than was presumed on the basis of large-effect single nucleotide polymorphisms (SNPs), like those in MC1R, discovered in Eurasian and African European admixed populations. Assumptions about darkly pigmented skin and functional constraints at MC1R also influenced our early understanding of pigmentation variation in populations from Southeast Asia and Oceania. Few studies have specifically focused on understanding the genetic architecture of pigmentation in these regions, so much of what we know is based on reports of allele frequencies from studies examining global variation at a particular pigmentation locus but without measured phenotypes. These studies indicate that MC1R variation in Oceanian populations may be constrained by purifying selection (Harding et al., 2000; Rana et al., 1999), while less constraint is evident in populations from Southeast Asia (Harding et al., 2000; Nakayama et al., 2006; Rana et al., 1999). Researchers also identified the presence of some intermediate frequency non-synonymous MC1R polymorphisms in both regions, including Val92Met (rs2228479, a valine to methionine substitution at amino acid 92 in the gene) and Arg163Gln (rs885479). Both derived variants are also present in Europe, but at lower frequencies. The relatively higher frequencies of the derived alleles in East Asia (23% and 64% respectively in HapMap Han Chinese) compared to Western Europe (6% and 10% in HapMap Europeans) may be due either to drift or selection following the divergence of these populations. In the case of Val92Met, this selection may have followed introgression of the variant from Neandertals (Ding et al., 2014). In the absence of pigmentation data collected in Southeast Asia and Oceania, it was impossible to know if these nonsynonymous changes were associated with pigmentation variation as assumed. Yamaguchi et al. (2012) reported a nominal association of the derived allele 163Gln with lighter skin pigmentation in a Japanese sample. However, this marker was not associated with skin pigmentation in a recent genome-wide association study (GWAS) in individuals of East Asian ancestry living in Toronto (Rawofi et al., 2017), so further investigation is needed to confirm the effect of this polymorphism across East Asia. The studies described above have generally included relatively limited numbers of Melanesians, making it difficult to assess if MC1R variation is constrained by purifying selection in this high-UVR region as it is in similarly high-UVR regions of Africa. This is an important question because there are many instances where recurrent exposure to either high-UVR or low-UVR climates over evolutionary history could lead to convergent depigmentation or repigmentation in human populations. Studying the genetic architecture of populations with similar levels of melanin in similar climates is important to establish the flexibility of the melanogenesis pathway. A recent investigation of MC1R sequence variation in a sample of 188 Melanesians failed to support a model of purifying selection using a McDonald-Kreitman test (Norton, Werren, & Friedlaender, 2015). This same study genotyped the Val92Met polymorphism in 635 Melanesians from 20 populations on three different islands and reported that the frequency of the derived allele ranged from 0.04 to 0.33. However, the polymorphism showed no association with skin or hair pigmentation. Taken together, these results suggest that MC1R variation, though relatively common, does not play a significant role in the genetic architecture of Melanesian pigmentation, possibly due to interactions with other unidentified loci. An earlier study directly investigating the genetics of pigmentation variation in Island Melanesia focused on genotyping a limited set of 10 SNPs previously associated with pigmentation variation (Norton, Kittles, et al., 2007; Norton, Koki, & Friedlaender, 2007). However, these SNPs were identified primarily because they influenced variation in European populations or explained substantial differences in skin pigmentation between Europeans and other populations. It is not surprising, given this strong European ascertainment bias, that six of these polymorphisms were monomorphic or exhibited very low levels of heterozygosity across the region. Of the remaining loci (rs6058017 in the gene ASIP, rs1800404 in the gene OCA2, and Val92Met and rs2228478 in the gene MC1R), only the variants in ASIP and OCA2 showed significant associations with skin pigmentation across the region (Norton, Kittles, et al., 2007; Norton, Koki, & Friedlaender, 2007). In both cases, the ancestral allele, more common in African populations, was associated with darker skin color. However, the interpretation of these results is complicated by the strong population structure that characterizes Island Melanesia (Friedlaender et al., 2008). Both alleles exhibit significant frequency differences among islands, as does pigmentation. Without controlling for background levels of population substructure it is difficult to determine if the observed association reflects a true influence on phenotype or is instead simply consistent with background levels of population stratification. Studies exploring the genetic architecture of skin pigmentation in Island Melanesia, or any geographically dispersed populations, should control for this structure and utilize sequencing to identify novel alleles that are not subject to a strong European-ascertainment bias.
|
|