Genetic Variation and Adaptation in Africa: Implications for Human Evolution and Disease

Abstract

Because modern humans originated in Africa and have adapted to diverse environments, African populations have high levels of genetic and phenotypic diversity. Thus, genomic studies of diverse African ethnic groups are essential for understanding human evolutionary history and how this leads to differential disease risk in all humans. Comparative studies of genetic diversity within and between African ethnic groups creates an opportunity to reconstruct some of the earliest events in human population history and are useful for identifying patterns of genetic variation that have been influenced by recent natural selection. Here we describe what is currently known about genetic variation and evolutionary history of diverse African ethnic groups. We also describe examples of recent natural selection in African genomes and how these data are informative for understanding the frequency of many genetic traits, including those that cause disease susceptibility in African populations and populations of recent African descent.

African populations have the highest levels of genetic variation among all humans. Studies of these populations are used to understand demographic changes in Africa and to identify disease-related genes (e.g., APOL1).

Africa is where modern humans evolved and is the starting place for the global expansion of our species (Stringer and Andrews 1988; Stringer 1994; Templeton 2002). African populations also have the highest levels of genetic and phenotypic variation among all humans. This variation is informative for characterizing demographic history in Africa, including times when populations increased in size, contracted, migrated, or when admixture between them occurred. Genetic variation also provides data that are useful in identifying local adaptation to diverse environments (Campbell and Tishkoff 2008). Characterizing human genetic variation and examining phenotypic variation in extant African populations is fundamental to the identification of genes that play a role in function, adaptation, and complex disease susceptibility in Africans and populations of recent African descent.

LINGUISTIC, CULTURAL, AND SUBSISTENCE DIVERSITY IN AFRICA

There are over 2000 indigenous languages spoken in Africa. They are classified into four major linguistic families—Niger-Kordofanian (spoken primarily by agriculturalist populations located in large contiguous regions of sub-Saharan Africa from West Africa to eastern and southern Africa), Nilo-Saharan (spoken predominantly by pastoralist populations in central and eastern Africa), Afroasiatic (spoken predominantly by agro-pastoralists and pastoralist populations in northern and eastern Africa), and Khoesan (a language family that contains click consonants, spoken by hunter–gatherer San populations in southern Africa as well as the Hadza and Sandawe hunter–gatherers in Tanzania) (Fig. 1).

Figure 1.

The geographic distribution of the major linguistic groups in Africa. The map was drawn using information (geographic locations of ethnic speakers in Africa that are based on published sources) from Haspelmath et al. (2008) (wals.info/), www.ethnologue.com (International HapMap 2005), Greenberg (1963, 1972), Vansina (1995), and Ehret (1971, 1993, 1995, 2001b). Geographic range occupied by Bantu speakers, the major linguistic subfamily within the Niger-Kordofanian phylum mentioned in the text, is also shown. Putative centers of origin and estimated time of initial expansion based on linguistic studies for three of the four language families are also listed: AA, Afroasiatic (14 kya) (Ehret 1995); NS, Nilo-Saharan (Blench 1993; Ehret 1993; Blench 2006); NK, Niger-Kordofanian (5 kya) (Nurse 1997; Ehret 2001a). Afroasiatic-speaking pastoralists were the first food-producing populations to migrate into East Africa circa 5 kya (X) (Leakey 1931; Butzer 1969; Robbins 1972; Barthelme 1977); followed by Nilo-Saharan-speaking pastoralists circa 3 kya (Y) (Leakey 1931; Bower 1973; Ambrose 1982; Distefano 1990), and later Bantu-speaking agriculturalists after circa 2.5 kya (Z) (Posnansky 1961a,b). P and q represent initial expansion of pastoralists (2.5 kya) and later Bantu-speaking agriculturalists (after 2 kya) to southern Africa from East Africa, respectively.

African populations also live in a diverse range of environments (including deserts, savannahs, and tropical environments), have different exposures to infectious disease, and have diverse diets and subsistence patterns. Because African populations exist in such a broad spectrum of environments, they may have regional or population-specific genetic variants that play a role in local adaptation to different selection pressures. The study of African genomic variation is uniquely interesting because of the large proportion of human genetic variation observed in African populations and the length of human history in the continent.

PATTERNS OF GENETIC DIVERSITY AND DEMOGRAPHIC HISTORY IN AFRICA

The characterization of human genetic variation began with studies that described human blood groups and protein polymorphisms (Cavalli-Sforza 1994). Since then, it has grown to include numerous studies of mtDNA (mitochondrial DNA) that is inherited maternally, Y chromosome variation that is inherited paternally, and loci from the full nuclear genome that include sequences of repetitive DNA, single nucleotide polymorphisms (SNPs), and structural variation (SVs) (Campbell and Tishkoff 2008).

The pattern of genetic variation at these loci in modern African populations reflects their demographic and evolutionary histories (Campbell and Tishkoff 2008). Modern humans evolved in Africa ∼200 kya (thousand years ago) and migrated from Africa within the past 50–100 kya, successfully colonizing most of the terrestrial parts of the globe (Campbell and Tishkoff 2008). This is called the “Out-of-Africa” migration. Currently, the precise location of the origin of modern humans remains a contentious issue. Some argue that South Africa is the location where our species originated (Tishkoff et al. 2009; Compton 2011; Henn et al. 2011), whereas others argue that East Africa is where modern humans originated (Prugnolle et al. 2005; Ray et al. 2005). However, these inferences are biased by the lack of archeological and fossil data in tropical areas of Africa and the fact that the geographic location of populations in the present may have differed in the past.

Genetic evidence also indicates that human populations underwent several major population expansions at different time periods over the last 100 ky (Rogers and Harpending 1992; Harpending et al. 1993; Excoffier and Schneider 1999; Atkinson et al. 2009; Cox et al. 2009): an initial major population expansion in Africa ∼110–70 kya (Rogers and Harpending 1992; Harpending et al. 1993; Excoffier and Schneider 1999; Atkinson et al. 2008, 2009) and subsequent migration and expansion events around the globe at ∼60–55 kya (Atkinson et al. 2009), ∼40–25 kya, and ∼12 kya (Harpending et al. 1993; Atkinson et al. 2009; Cox et al. 2009).

Prior to the domestication of plants and animals ∼10,000 years ago, all human populations practiced hunting-gathering/foraging for subsistence. Although we do not know their original homeland, ancestors of the extant African hunter–gatherer populations, including the current Khoesan speakers, migrated into central, eastern, and southern Africa probably >20 kya (Nurse 1997; Ehret 2002). It is speculated that speakers of Khoesan languages were widespread throughout a contiguous area that encompassed southern and eastern African regions (Sutton 1973; Smith 1992). Based on archaeological and linguistic data, it is hypothesized that recent migration events that occurred, due to changes in subsistence patterns, modified the genetic landscape in Africa through admixture of populations practicing pastoralism and/or agriculture with indigenous hunter–gatherer populations. For example, the expansion of agriculturalists from Central-West Africa and the migration of pastoralist populations from northeastern Africa have contributed to the current genetic landscape in Africa. One of the largest migration events in Africa that was mediated by a shift in subsistence patterns was the migration of Bantu-speaking populations, a subfamily of Niger-Kordofanian languages, from their proposed homeland in Nigeria/Cameroon across sub-Saharan Africa within the past 5000 years (Fig. 1) (Nurse 1997; Ehret 2001a). The result of this expansion is the current predominance of Bantu languages in many African countries. Other major migration events have included the migration of Nilo-Saharan pastoralists from their homeland in Chad/Sudan, both westward across the Sahel ∼7000 years ago and eastward into Kenya and Tanzania within the past ∼3000 years (Blench 1993; Ehret 1993). Afro-Asiatic-speaking agro-pastoralists migrated from their homeland, which encompassed the Nile Valley to the Ethiopian highlands, into western, northern, and eastern Africa within the past 8000–5000 years (Ehret 1995, 1998) (Fig. 1). Demographic history in Africa has also been influenced by the more recent migration of non-Africans into Africa. Most noteworthy is the migration of southwestern Asians into Northern and Eastern Africa within the past ∼3000 years (Van Beek 1967; Butzer 1981; Phillipson 2009). Europeans, southern and eastern Asians have also migrated into Southern Africa within the past several hundred years.

Consistent with the “Out of Africa” model of modern human origins, analyses of genetic data indicate that Africans have higher levels of genetic diversity than non-Africans (Cann et al. 1987, 2002; Tishkoff et al. 1998; Marth et al. 2004). Overall, the extant patterns of genetic variation in global populations suggests serial founder events, with increasing genetic distance correlated with geographic distance from East Africa, indicating the cumulative effect of genetic drift as humans expanded into the rest of the world (Prugnolle et al. 2005; Ramachandran et al. 2005; Hofer et al. 2009). Studies of autosomal genetic variation show that genetic distance between populations also increases with geographic distance within Africa (Prugnolle et al. 2005; Ramachandran et al. 2005; Hofer et al. 2009; Tishkoff et al. 2009). The greater genetic diversity in sub-Saharan Africans, as compared to other continental populations, may also be partially a result of ancient admixture with yet-to-be identified archaic population(s) (Zietkiewicz et al. 1998; Labuda et al. 2000; Plagnol and Wall 2006; Hammer et al. 2011; Lachance et al. 2012), analogous to the proposed admixture between Neanderthals and modern humans in Eurasia (Reich et al. 2010, 2011; Abi-Rached et al. 2011; Green et al. 2011; Yotova et al. 2011). Genomic introgression from Neanderthals has been detected in some East African populations but was inferred to be a signature of recent back-migrations from Eurasia (Wang et al. 2013).

Many of the recent population genetic studies of African populations are based on analysis of genetic markers that were genotyped in a small number of populations that are part of the CEPH-HGDP collection (Centre d’Étude du Polymorphisme Humain–Human Genome Diversity Panel) or the International Haplotype Map (HapMap) Project (Batzer and Deininger 2002; Rosenberg et al. 2002; Li et al. 2008). Only six African populations are included in the CEPH-HGDP panel, whereas the International HapMap Project includes just three African populations (Niger-Kordofanian-speaking Yoruba from Nigeria, Bantu-speaking Luhya from Kenya, and Nilo-Saharan-speaking Maasai from Kenya). A more recent international collaborative effort that aims to catalogue human genetic variation through whole genome resequencing, the “The 1000 Genomes Project” (2010), originally had only two African populations represented in the project but has since expanded to include the Esan from Nigeria, individuals from The Gambia, the Luhya from Kenya, and the Mende from Sierra Leone. However, it should be noted that all of these populations speak Niger-Kordofanian languages and share recent genetic ancestry. The 1000 Genomes project has also expanded to include people of recent African descent including African Americans from the southwestern United States and African Caribbean individuals from Barbados. Although these projects are important in their description of overall human genetic diversity, they are limited in their coverage of African populations, despite the recent additions (Consortium 2010). Because these projects have sampled a small number of African populations it is likely that a substantial portion of human genetic variation will be missed. Thus, it is important to continue to add African populations that are underrepresented in human genomic studies, particularly those speaking languages belonging to the other three language families in Africa. Analysis of genomic variation across a broad range of African populations will be useful for elucidating fine-scale population structure and demographic patterns in African populations.

The current studies of global human genetic variation suggest that human populations may have been divided into distinct subpopulations in Africa prior to the migration of modern humans Out-of-Africa ∼100 kya (Labuda et al. 2000; Satta and Takahata 2004; Adeyemo et al. 2005; Plagnol and Wall 2006; Bryc et al. 2009; Tishkoff et al. 2009; Wall et al. 2009; de Wit 2010; Patterson 2010; Sikora 2010; Henn et al. 2012; Pagani et al. 2012; Schlebusch et al. 2012). For example, we (Tishkoff et al. 2009) identified 14 ancestral population clusters in Africa with four predominant clusters that broadly represent populations from major African geographical regions and those that speak the four African language families mentioned above. The remaining 10 are mainly restricted to specific geographic regions, languages, or in some cases, individual populations (e.g., Hadza hunter–gatherers). Other studies of African genetic diversity (Adeyemo et al. 2005; Bryc et al. 2009; de Wit 2010; Patterson 2010; Sikora 2010; Henn et al. 2012; Pagani et al. 2012; Schlebusch et al. 2012) have largely recapitulated the regional clusters (Tishkoff et al. 2009) while also adding some fine-scale genetic structure between populations from different ethnicities or populations that speak different languages within the regions studied.

Studies of genetic variation in Africa suggest that even though high levels of mixed ancestry are observed in most African populations, the genetic variation observed in Africa is broadly correlated with geography, language classification (Adeyemo et al. 2005; Bryc et al. 2009; Tishkoff et al. 2009) and subsistence classifications (Tishkoff et al. 2009). For example, genetic variation among Nilo-Saharan and Afroasiatic-speaking populations from both Central and East Africa (Tishkoff et al. 2009; Bryc et al. 2009) reflect the geographic region from which they originated, and generally shows a complex pattern of admixture between these populations and the Niger-Kordofanian speakers who migrated into the region more recently. Consistent with linguistic evidence regarding the origin of Nilo-Saharan languages in the Chad/Sudan border, the highest proportion of Nilo-Saharan ancestry is observed among southern Sudanese populations (Tishkoff et al. 2009). Additionally, North/Northwest African populations are genetically differentiated from sub-Saharan African populations, but have considerable shared ancestry with East African Afroasiatic speakers (Tishkoff et al. 2009; Henn et al. 2012) and southwest Asian populations (Rosenberg et al. 2002; Behar et al. 2008, 2010; Li et al. 2008; Kopelman et al. 2009; Tishkoff et al. 2009; Atzmon et al. 2010; Bray 2010; Hunter-Zinck 2010; Henn et al. 2012), likely reflecting historic migrations from those source populations into northern Africa. Finally, several studies of genetic variation in the “colored” population from South Africa (Tishkoff et al. 2009; de Wit et al. 2010; Patterson et al. 2010) are consistent with the documented history of this population, which indicates that they have South African Khoesan, Niger-Kordofanian, European, South Asian, and Indonesian ancestors (Tishkoff et al. 2009; de Wit et al. 2010; Patterson et al. 2010).

mtDNA AND Y CHROMOSOME VARIATION

A refined view of uniparental genetic diversity in Africa has been hampered by a lack of extensive sampling in Africa and a lack of consensus on the proper method to estimate mutation rates that are essential to make inferences about human demography. The time to the most recent common ancestor (TMRCA) for mtDNA lineages, based on large datasets, ranges from 121.5 to 221.5 kya, (Ingman et al. 2000; Behar et al. 2008), which is older than most estimates of the Y chromosome TMRCA, which are based on patchy sampling within Africa (60–140 kya) (Thomason 1988; Cruciani et al. 2011; Wei et al. 2013). The fact that there has been limited sampling in Africa is further exemplified by recent results that include more African samples and infer that the TMRCA of the human Y chromosome is similar (120–200 kya) (Mendez et al. 2013) or even older (237–581 kya) than the TMRCA of mtDNA (Ingman et al. 2000; Behar et al. 2008; Francalacci et al. 2013; Mendez et al. 2013; Poznik et al. 2013). Generally, there is a lack of precision in TMRCA estimates reported by different studies using the same marker systems (Francalacci et al. 2013; Mendez et al. 2013; Poznik et al. 2013). Nonetheless, the consensus based on numerous studies, is that human populations exhibit structure (e.g., genetic differences among populations from different regions) based on both mtDNA and Y chromosome variation (Wallace et al. 1999; Ingman et al. 2000; Maca-Meyer et al. 2001; Herrnstadt et al. 2002; Mishmar et al. 2003).

The distribution pattern of uniparental lineages in Africa suggests ancient geographical and cultural subdivision in African. African populations exhibit greater diversity in the Y chromosome than those of other continents (Hammer 2002). Based on the latest classification of Y chromosome haplogroups (Karafet et al. 2008), A, B, E, and J haplogroups are found in Africans, whereas C, D, and the lineages descending from haplogroup F are almost exclusively observed outside of sub-Saharan Africa (Fig. 2). The oldest lineages, A and B, with a TMRCA over ∼75 kya, are present only in Africans. The younger lineages, E and J haplogroups, with a TMRCA less than 75 kya, are present in both African and non-African populations (Fig. 2) (Seielstad et al. 1998; Scozzari et al. 1999). Haplogroup E represents the great majority of Y chromosome haplotypes across Africa (Cruciani et al. 2002). Furthermore, the sublineages within these haplogroups exhibit unique regional distribution patterns within Africa (Cruciani 2002, 2010, 2011; Hammer 2002; Luis et al. 2004; Wood et al. 2005; Batini 2007, 2011; Berniell-Lee et al. 2009), with some restricted to specific geographical regions and/or language families (Fig. 3), consistent with substructuring of diverse African populations.

Figure 2.

Evolutionary tree of Y chromosome and mtDNA haplogroups. (A) Nomenclature for major lineages of Y chromosome haplotypes. The M, P, and YAP labels leading to haplogroup/s are SNPs and indels that are used to define these haplogroups. Haplogroup F encompasses haplogroups F to T. Haplogroups A, B, and E are mainly found in Africa, whereas the rest are found mainly outside Africa. (B) Overview of mtDNA haplogroup phylogeny. In the mtDNA haplogroup nomenclature, the letter names of the haplogroups run from A to Z, with further subdivisions using numbers (from 0) and lowercase letters (from a). The naming was done in the order of their discovery and does not reflect the actual genetic relationships. Haplogroup M and N encompasses all of the haplogroups lettered A to Z excluding haplogroup L. Haplogroups L (L0–L6) are mainly found in Africa, whereas the rest are found mainly outside Africa (Richards et al. 1998; Macaulay et al. 1999; Quintana-Murci et al. 1999; Salas et al. 2002; Kivisild et al. 2004; Behar et al. 2008).

Figure 3.

Distribution of mtDNA and Y chromosome haplogroups across Africa. The black lines represent the approximate geographical boundaries of the distribution of each haplogroup cluster, with each of the clusters predominantly observed in the demarcated regions.

Analogous to variation observed on the Y chromosome, mtDNA haplotypes also show geographic structuring. African populations are characterized by having the oldest haplogroup lineages (L0–L5), with TMRCA over ∼80 kya (Bandelt et al. 1995, 2001; Chen et al. 1995, 2000; Graven et al. 1995; Soodyall et al. 1996; Watson et al. 1996, 1997; Alves-Silva et al. 2000; Torroni et al. 2001; Salas et al. 2002), as well as the M1 haplogroup, an L3 sublineage. Although lineages M, N, and all of their descending lineages with inferred TMRCA less than 80 kya are found in most global populations, they are observed almost exclusively outside of sub-Saharan Africa. There is also a unique distribution pattern of mtDNA haplotypes across Africa with some lineages restricted to specific geographical regions and/or language families (Fig. 3) (Newman 1980; Vigilant 1990, 1991; Chen et al. 1995, 2000; Ehret 1995, 2006; Watson et al. 1996, 1997; Rando et al. 1998; Krings et al. 1999; Richards et al. 2000; Pereira et al. 2001; Salas et al. 2002; Thomas et al. 2002; Destro-Bisol et al. 2004; Kivisild et al. 2004; Beleza et al. 2005; Coia et al. 2005; Jackson et al. 2005; Cerny et al. 2006, 2008, 2009; Gonzalez et al. 2006; Batini et al. 2007; Gonder et al. 2007; Behar et al. 2008; Coudray et al. 2008; Quintana-Murci et al. 2008, 2010; Castri et al. 2009; Coelho et al. 2009; Poloni et al. 2009; Saunier et al. 2009; Stefflova et al. 2009; Veeramah et al. 2010). Additionally, there is also a predominance of some Y chromosome and mtDNA lineages in specific regions of Africa. There is also sharing of many of these lineages between regions (Watson et al. 1996, 1997; Pereira et al. 2001; Salas et al. 2002; Beleza et al. 2005; Behar et al. 2008; Quintana-Murci et al. 2008), suggesting that there have been both recent and ancient migration events between these regions over the last 20 ky (Ehret 1967).

Despite the large amount of ethnic and genetic diversity observed in African relative to non-African populations, there has been limited sampling in Africa. Additional sampling and characterization of genetic diversity, particularly from underrepresented central, southern and eastern African countries will be crucial to deciphering fine-scale genetic history of genetically, culturally, and linguistically diverse African populations and for reconstructing both ancient and recent human evolutionary history. Comprehensive analysis of African genetic variation will help us address some of the outstanding questions about the origin of our species such as: (1) precisely where and when modern humans originated in Africa; (2) the number and age of ancestral structured populations in Africa; (3) the age and direction of ancient migration events both within and out of Africa (e.g., we do not yet have a consensus from genetic studies on the timing and route/s modern humans took out of Africa to populate the rest of the globe); and, (4) determination of the archaic source population(s) from which genetic introgression occurred into African populations and when these events took place.

DARWINIAN SELECTION AND GENETIC VARIATION

Demographic events such as those described above, that is changes in population size, short- and long-range migrations, and admixture or gene flow from one population to another, influence the levels and patterns of genetic variation across the genome (Campbell and Tishkoff 2008). In addition to demography, natural selection, recombination and mutation, which take place in specific parts of the genome, can also influence genetic variability.

Natural selection occurs when the fitness effects of genotypes are unequal (Nielsen 2005). For example, a beneficial genetic variant is likely to increase in frequency in a population over time because of positive selection. Likewise, if a genetic variant is deleterious, it may be selected against and its frequency in a population will decrease over time. The occurrence of natural selection can thus create substantial differences in the patterns of genetic variation between human populations (Aquadro et al. 2001). Identifying loci that exhibit patterns of genetic variation that are indicative of natural selection is, therefore, important because these loci are likely to be regions of the genome that are, or have been, functionally important and play a role in adaptation to local environments. Furthermore, understanding which regions of the genome have been subject to natural selection will aid in the identification of genetic variation that contributes to phenotypic variation among human populations, including disease susceptibility (Ronald and Akey 2005). Here, we discuss results of several studies of local adaptation to distinct environments and diets within Africa.

Lactase Persistence

Adult mammals lose the ability to effectively digest lactose, the main carbohydrate in milk, after weaning. The inability to digest lactose is a result of decreased production of the enzyme lactase-phlorizin hydrolase, or lactase (Swallow 2003; Ingram et al. 2009). Individuals who are unable to digest lactose have what is called the “lactase nonpersistent” (LNP) trait. Although the LNP trait was once considered to be pathologic, it is now well understood that this trait is “ancestral” (i.e., all ancestors of modern humans had this condition) and widespread among human populations (Auricchio et al. 1963; Newcomer et al. 1983; Segal et al. 1983). Individuals with the “lactase persistent” (LP) trait continue to have high levels of lactase production into adulthood. The LP trait is common only in populations whose ancestors practiced cattle domestication or pastoralism (Ingram et al. 2009). For example, LP occurs at high frequencies in northern European dairying populations (e.g., Finns, Swedes, and Danes), and decreases in frequency in southern Europe, the Middle East, and in most of Asia (Durham 1991). LP is also found to be common in some pastoral populations in Africa (Tishkoff et al. 2007). Thus, it has been hypothesized that LP is an adaptive trait in human populations that practice cattle domestication and dairying (Swallow 2003).

In Europe, a common regulatory variant outside of the gene that encodes lactase (LCT) has been identified as being strongly associated with the LP trait (Enattah et al. 2002; Ingram et al. 2007; Tishkoff et al. 2007; Enattah et al. 2008). Several studies also suggest that this variant is causal and is a target of recent natural selection (Enattah et al. 2002; Poulter et al. 2003; Bersaglieri et al. 2004; Coelho et al. 2005). This suggestion is based on several lines of evidence. First, experiments in small intestine cell lines suggest that the European mutation increases the expression of the lactase gene and causes an increase in the production of the lactase enzyme (Enattah et al. 2002; Olds and Sibley 2003; Troelsen et al. 2003; Lewinsky et al. 2005). Additionally, several different tests indicate that the LCT locus has experienced recent strong natural selection (Poulter et al. 2003; Bersaglieri et al. 2004; Coelho et al. 2005). These studies have identified unexpectedly long genomic regions of genetic similarity flanking the LCT locus, which is a signature of recent positive selection. When positive selection causes a variant or variants to increase in frequency, the selected variant will cause neighboring genetic variation to “hitchhike” as the selected variant increases in frequency, thereby causing large tracks of identical sequences surrounding the variant under selection. This genetic feature has been identified in many European individuals with the regulatory variant who have the LP trait (Poulter et al. 2003). Finally, the estimated age of the European LP-associated variant is between ∼20,000 and ∼2000 years ago (Bersaglieri et al. 2004; Coelho et al. 2005; Tishkoff et al. 2007). These dates are consistent with the origins of dairy farming in southwest Asia that have been estimated from the archaeological record at ∼8–9 kya (Evershed et al. 2008).

Although this is a convincing example of gene/culture co-evolution in Europe, the genetic basis of the LP phenotype was largely unknown in Africa for many years. Previous studies (Mulcare et al. 2004) indicated that the European regulatory variant was noticeably absent in many African populations, including populations that have a history of pastoralism and drinking milk. This observation called into question whether the European variant is truly the causal variant for lactase persistence (Ingram et al. 2007). However, genotype/phenotype association studies of LP in Africa (Swallow 2003; Tishkoff et al. 2007) have identified three novel variants upstream of LCT, near the European variant, that are significantly associated with LP in African populations. All three of the African sites enhance the expression of LCT (Tishkoff et al. 2007; Enattah et al. 2008), and the most common of the three African mutations also shows strong evidence of recent strong positive selection. The estimated age of this African variant is ∼3000–7000 years, consistent with the introduction of cattle domestication south of the Saharan Desert within the past ∼5500 years (Tishkoff et al. 2007). Together, these examples from Europe and Africa show us that strong selective pressure can increase the frequency of numerous rare mutations that arise independently and regulate lactase gene expression where people practice dairying. The occurrence of multiple mutations that evolved under similar selection pressures is a good example of convergent evolution in several different human populations.

Malaria

Examination of population-level genetic variation and recent natural selection caused by infectious diseases can help us understand differences in disease susceptibility and incidence, and why some genetic variants, especially those that are potentially deleterious, are common in specific populations. Malaria is one such example that has had an important impact on patterns of human genetic variation and the geographic distribution of a number of genetic disorders (Fortin et al. 2002).

Malaria is a parasitic disease caused by species in the genus Plasmodium. There are several Plasmodium species that infect humans, including P. falciparum, P. ovale, P. malariae, and P. vivax. A large number of genetic variants that are associated with malaria-protective phenotypes have been identified and many of them cause serious disease (Ko et al. 2012; Gomez et al. 2013a). For example, variants associated with hemoglobinopathies, α and β thalassemias, and the structural hemoglobin variants S, C, and E, are classic examples of deleterious genetic variants that nevertheless confer protection from malaria (Weatherall 2001; Gomez et al. 2013b). Early studies of thalassemias (Haldane 1949; Flint et al. 1986) noted that these diseases have a geographic distribution that coincides with P. falciparum endemicity. In what is now called the “malaria hypothesis,” JBS Haldane suggested that the genetic variants responsible for the thalassemias may be maintained in some human populations because of an advantage associated with malaria resistance (Haldane 1949). In a recent analysis, Piel et al. (2010) revisited the global distribution of sickle cell anemia, a disease caused by a mutation in the β-globin gene (HbS). They showed strong geographical correlation between HbS allele frequency and malaria endemicity in Africa (Fig. 4), consistent with the malaria hypothesis.

Figure 4.

Global distribution of the HbS sickle cell anemia allele compared to the historic geographic distribution of malaria. (Top) Global distribution of HbS allele frequency predicted from Bayesian geostatistical modeling. (Bottom) Historical map of malaria endemicity. The classes are defined by PfPr₂₋₁₀ (PfPr₂₋₁₀ = proportion of 2- to 10-year olds with confirmed blood stage asexual parasites): malaria free, PfPr₂₋₁₀ = 0; epidemic, PfPr₂₋₁₀ ≈ 0; hypoendemic, PfPr₂₋₁₀ < 0.10; mesoendemic, PfPr₂₋₁₀ ≥ 0.10 and < 0.50; hyperendemic, PfPr₂₋₁₀ ≥ 0.50 and < 0.75; holoendemic, PfPr₀₋₁ − 1 ≥ 0.75 (this class was measured in children ages 0–1) (From Piel et al. (2010); reproduced, with express permission, from Nature Publishing Group © 2010.)

Additional studies of other genes have shed further light on our understanding of the co-evolution of humans and P. falciparum, and provide evidence of recent natural selection in the human genome. For example, studies of genetic variation at G6PD (glucose-6-phosphate dehyrogenase), a gene that is known to harbor mutations associated with G6PD enzyme deficiency and protection from P. falciparum infection, have revealed signatures of genetic variation consistent with recent natural selection (Tishkoff et al. 2001; Saunders et al. 2002; Verrelli et al. 2002; Sabeti et al. 2006). Among the G6PD variants that are associated with protection from malaria, the A-variant is most common in sub-Saharan Africa. Several studies, in African and non-African populations (Tishkoff et al. 2001; Sabeti et al. 2002; Saunders et al. 2002; Verrelli et al. 2002), have identified signatures of recent natural selection acting on the A-variant and inferred the TMRCA for this variant to range from ∼1200 to 12,000 ya, which is consistent with the idea that selective pressures caused by malaria increased in African populations as population densities increased after the introduction of domesticated plants and animals (Tishkoff et al. 2001).

Kidney Disease

In the United States, diseases like type 2 diabetes mellitus, hypertension, and prostate cancer are known to be more common in African Americans than European Americans (Landis et al. 1999; Brancati et al. 2000; Ong et al. 2007). This disparity in disease occurrence suggests that genetic factors mediating disease risk, together with environmental factors, may differ among European and African descent populations (Hunter 2005; Yang et al. 2005; Cooper 2013). It has been proposed that genetic risk factors could be more common in African populations because they may have been adaptive in past African environments but increase risk for disease in today’s Western environment (Campbell and Tishkoff 2008).

One example of such a disease at high prevalence in African Americans is chronic kidney disease (CKD). In 2009, the mortality rate of African Americans with CKD was significantly higher than the mortality rate in European Americans and patients of other ethnicities (United States Renal Data System 2011). Additionally, in 2009, end-stage renal disease (ESRD), a serious complication of CKD in which a patient has complete or near complete loss of renal function and requires either frequent dialysis or a kidney transplant, occurred at a rate that is 3.5 times greater in African Americans than European Americans (United States Renal Data System 2011). There are many different causes of CKD; however, the primary proximal causes include diabetes and hypertension. There are also inherited forms of CKD; some of these include polycystic kidney disease (PKD), focal segmental glomerulosclerosis (FSGS), and pediatric nephrotic syndrome (Friedman and Pollak 2011).

Most African Americans have mixed ancestry originating from Africa and Europe; studies have inferred the average amount of European ancestry in African Americans to be ∼20% and the remainder to be predominantly of western and central African origin (Smith et al. 2004; Bryc et al. 2009; Tishkoff et al. 2009). It should be noted, however, that there is substantial variation in the level of African ancestry in individual African Americans. The proportion of African ancestry in a given individual can range from 99% to 1%. Additionally, the level of African or European ancestry in African Americans not only varies by individual but also by genomic region (Bryc et al. 2009). To identify genetic factors mediating disease predisposition in populations of recent African descent, like African Americans, a number of different approaches have been employed (Dong et al. 1999; Smith et al. 2004; Malhotra et al. 2005; Leak et al. 2007; Adeyemo et al. 2009; Sale et al. 2009; McDonough et al. 2011). One of these, called admixture mapping, is based on the premise that if a trait or disease of interest has a different prevalence in the parental populations prior to admixture, then it is likely that the genetic variation causing the disease or trait of interest in the admixed population is associated with chromosomal segment(s) derived from that parental population where the disease or trait is more prevalent (Winkler et al. 2010).

The identification of genetic risk factors for CKD in African Americans is an important example of the success of admixture mapping (Kao et al. 2008; Kopp et al. 2008; Genovese et al. 2010; Tzur et al. 2010). In 2008, two studies independently identified MYH9 as a candidate locus associated with CKD or ESRD in African Americans. Kopp et al. (2008) analyzed FSGS African-American cases and controls and showed a significant association on chromosome 22 near the MYH9 locus. They also found a marked increase of African ancestry in FSGS cases. Because MYH9 is expressed in the kidney, the authors considered this to be a good causal candidate gene. In the second study, Kao et al. (2008) also used admixture mapping to search for genetic variants associated with CKD and identified a single location significantly associated with disease on chromosome 22 in a population of nondiabetic ESRD patients. They also showed that the amount of European ancestry in the ESRD cases at that region was significantly lower than the average across the genome. Similar to Kopp et al. (2008), Kao et al. (2008) identified MYH9 as the likely candidate gene responsible for the strong association signal, given the expression patterns of MYH9 in the kidney.

Following publication of these two papers, numerous studies were conducted to find the functional variants responsible for these admixture signals (Kao et al. 2008; Kopp et al. 2008; Freedman et al. 2009a,b; Reeves-Daniel et al. 2010; Rosset et al. 2011; Freedman and Murea 2012). In 2010, two groups (Genovese et al. 2010; Tzur et al. 2010) used the data from the 1000 Genomes Project to identify potential functional variants that differ in frequency between Africans and Europeans near the MYH9 locus. Genovese et al. (2010) performed association tests of genome-wide variants with FSGS using African-American cases and controls (a genome wide association study or GWAS). They showed that the strongest signal with FSGS was not at MYH9 but at APOL1, which is a gene that is a short distance from MYH9. The strongest association signal was observed at a genetic variant termed G₁. When the authors controlled for the association of G₁ with FSGS they found a second associated allele that they termed G₂. The combined signal at G₁ and G₂ was 35 orders of magnitude greater than the signal found at MYH9. Tzur et al. (2010) also used the 1000 Genomes data to search for new variants near MYH9 that might be functional. They found four genetic variants that were potential candidates; two variants in APOL1 (also identified by Genovese et al. 2010), one in APOL3, located further upstream, and one in FOXRED2, located downstream from MYH9. They found that the APOL1 variants are more strongly associated with CKD than the MYH9 risk alleles in African-American and Hispanic-American ESRD cases.

In addition to the association study, Genovese et al. (2010) tested whether either the G₁ or G₂ gene variants exhibited a signature of recent positive selection. The results of their analyses suggest that at G₁ there is a strong signal of recent positive selection and a weak, but noteworthy, signal of recent positive selection at G₂. Because it is unlikely that G₁ and G₂ were positively selected for their deleterious role in the pathogenesis of kidney disease, Genovese et al. (2010), Tzur et al. (2010), and Oleksyk et al. (2010) speculated that these alleles must serve a separate beneficial role in an African environment. In Africa, APOL1 is an important factor in resistance to infection from Trypanosoma brucei brucei, one of the trypanosomes that cause African sleeping sickness. Genovese et al. hypothesized that this function may be the underlying reason for the apparent recent positive selection at APOL1 (also see Oleksyk et al. 2010; Tzur et al. 2010). To test this hypothesis, Genovese et al. (2010), examined whether the G₁ or G₂ alleles influence the resistance phenotypes conferred by APOL1 during infection caused by three subspecies of Trypanosoma brucei (Trypanosoma brucei brucei, Trypanosoma brucei rhodesiense, and Trypanosoma brucei gambiens). They showed that the G₁ and G₂ alleles effectively killed >60% of the T.b.rhodesiense samples and suggested that the signature of selection observed at these variants may be the result of the selective advantage conferred against Trypanosoma infection. However, it should be noted that T.b.rhodesiense is presently found in East Africa and the signature of selection identified by Genovese et al. was found in West African populations. Additionally, Tzur et al. (2010) did not find the APOL1 variants in people from Ethiopia, which implies that the G₁ and G₂ alleles may not be common in East African populations. The absence of the G₁ and G₂ variants in Ethiopia is difficult to reconcile with the observation that that G₁ and G₂ are most effective at killing Tryanosoma species that are common in East Africa. However, Genovese et al. (2010) suggest that changes in Trypanosoma biology and distribution and/or human migration could explain the lack of correlation between parasite range and the distribution of the G₁ and G₂ alleles today. They also propose that the G₁/G₂ variants at APOL1 could play a role in immunity to other pathogens common in western Africa.

More recently, Ko et al. (2013) further examined genetic variation at APOL1 and showed that the G₂ variant has a similar frequency across diverse African populations (3%–8%) and that the G₁ variant is only common in the Yoruba population (39%). These authors identified another variant, termed G₃, which has a more widespread distribution than the G₁ variant, and showed a signature of recent natural selection in a West African Fulani population. Further studies of G₁/G₂ allele frequencies in diverse African populations and additional tests of recent natural selection are necessary to better understand the evolutionary history of this locus and how selection has influenced allele frequencies at APOL1. It is also important to note that the functional mechanism(s) underlying the association of G₁/G₂ with CKD is currently unknown. In addition to further studies of kidney disease in African populations, additional studies are also needed to demonstrate the functional roles that the G₁, G₂, and G₃ variants play in infectious disease and kidney disease.

The identification of regions in African genomes that are recent targets of natural selection has resulted in important insights into the genetic basis of many pathological and nonpathologic phenotypes common in African populations. However, despite the growth of evolutionary studies to understand the prevalence of disease phenotypes, there is a general lack of epidemiological studies that help us understand the prevalence of noncommunicable diseases in African countries and how genetic variation plays a role in presence and prevalence of these diseases. Dalal et al. (2011) reported that in 2004 about one-quarter of all deaths in sub-Saharan Africa were caused by noncommunicable diseases, and they estimate that by 2030 noncommunicable diseases will increase to 46% of all deaths in sub-Saharan Africa. Unfortunately, their review of the literature suggests that community-based epidemiologic studies of noninfectious diseases, such as diabetes, cardiovascular disease, and obesity, are lacking in many African nations compared to the focus on maternal–child health and infectious diseases. Thus, despite the current shift in the prevalence of noninfectious diseases, we are currently ill-equipped to describe who is being affected and what the genetic and environmental risk factors that contribute to noncommunicable disease susceptibility in Africa are.

In the coming decades, it will be crucial to focus on the prevalence of noninfectious diseases in African countries and to understand what genetic variants predispose African people to conditions like obesity and cardiovascular disease. The effort to understand the genetic basis of complex noninfectious diseases in Africa should be twofold—our efforts should be placed on large-scale studies of both families and unrelated people. These two approaches will allow us to better characterize how disease-causing variants are distributed among diverse populations, and family-based studies will also help us to identify rare genetic variants that are difficult to identify in studies of unrelated people. Comparison of African populations and those within the African diaspora across diverse environments may also help to disentangle genetic and environmental effects on variable phenotypes and disease risk. These efforts will help to elucidate factors contributing to complex diseases in Africa and may also elucidate the genetic basis of these conditions in populations that have recent African admixture, like African Americans, Hispanics, and peoples of the Caribbean.

CONCLUSIONS

The pattern of genetic variation observed at mtDNA, Y chromosome, and autosomal loci in African populations reflects the demographic history of these populations. These data indicate high levels of genetic diversity within and between African populations, and also show that the current pattern of genetic variation in Africa is a result of both ancient and recent migration and admixture events.

The genomes of Africans have also been impacted by natural selection, sometimes resulting in prevalence of genetic variants that cause disease. Studies of natural selection in African populations show that functionally important genetic variants may be common but geographically restricted within Africa because of local adaptation to a particular lifestyle or environment. Also, many common variants that are adaptive because of protection from an infectious disease may also result in susceptibility to a different, possibly noninfectious, disease in populations of recent African origin. This observation points to the importance of including ethnically diverse Africans in human genetic studies because these populations represent an important component of human genomic variation, and data from African populations help us to understand the context in which some genetic variants were selected.

Going forward, as the cost of whole genome sequencing decreases, it will become feasible to conduct large-scale genomic sequencing studies across ethnically diverse Africans. These data are necessary to understand the genetic basis of complex disease in Africa and also the genetic basis of complex disease in populations like African Americans. Currently, most large-scale studies that seek to identify genetic variants that are associated with complex disease risk are heavily biased toward the identification of genetic variants initially discovered in European populations. Large-scale sequencing studies in African populations will identify new non-European variants that can be added to the repertoire of variation that is included in large association studies. The inclusion of a more diverse panel of variants will increase what we know about the genetic basis of complex diseases because new variants will allow us to search for associations among sites that are common in people from many different ethnic backgrounds. These studies will shed light on modern human origins, African and African-American population history, and the genetic basis of nonpathologic traits as well as traits that affect disease susceptibility.