Portal de Periódicos da CAPES

The discovery of resistance to malaria of sickle‐cell heterozygotes

2002; Wiley; Volume: 30; Issue: 5 Linguagem: Inglês

10.1002/bmb.2002.494030050108

1539-3429

A. C. Allison,

Genetic diversity and population structure

I was raised on a farm in the Kenya highlands overlooking the Great Rift Valley. Kenya is a beautiful country with diverse ecosystems; near the coast and Lake Victoria lie hot, moist regions, to the northwest of the capital (Nairobi) are fertile highlands, and large areas of the country are arid. There is a corresponding variety of plant and animal life and a diverse set of indigenous inhabitants speaking languages belonging to different groups. I often asked, "How are these people related to one another and to humans elsewhere in Africa and the rest of the world?" The excavations of Louis Leakey were unearthing fossils that were good candidates for being missing links between apes and men. His findings raised the possibility that East Africa might have an important place in the origin of man. Given all these wonders, and a high level of curiosity, I became fascinated by natural history and anthropology. I went on birding safaris with Kenya's authority, Leslie Brown, and visited Leakey at Olduvai and other sites. As a late teenager I was influenced strongly by Darwin's The Origin of Species and The Descent of Man and soon became a convinced Darwinian. At Oxford University I learned about the modern synthesis of evolution and genetics, according to which evolution results from changes in the frequencies of genes in populations. The mathematical analyses of R. A. Fisher and J. B. S. Haldane in the United Kingdom and of Sewall Wright in the United States had provided the theoretical basis of population genetics. Individuals of different genotypes vary in "fitness," a term that includes survival through reproductive age and relative fertility. Under some conditions genetic diversity (polymorphism) is stable, for example when a heterozygote has a fitness greater than that of either homozygote, whereas under other conditions polymorphism is unstable. Wright (and later Kimura) argued that much genetic diversity is because of random genetic drift, whereas British geneticists believed that natural selection plays a major role in maintaining genetic diversity. Selection was usually attributed to phenomena such as competition for resources or predation. There was at that time no example of natural selection operating on a common gene in humans, in contrast to selection against rare deleterious mutations. During my studies in biochemistry and genetics I longed for an opportunity to document as many genetic markers as possible in human populations. These would be more reliable than cultural or linguistic traits as indices of affinity between different human groups. Analyses of enough genetic markers would provide valuable information about the relatedness of human populations and their origins. Such information would be complementary to that obtained from the study of fossil hominid remains. Because the markers used for such analyses should not be subject to strong selection it was necessary to search for selective effects, which would be of interest in themselves. In 1949, during an interval of several months between completion of my basic science studies and entry into medical school, I participated in the Oxford University Expedition to Mount Kenya. Whereas my colleagues studied plants and insects, I collected blood samples from tribes all over Kenya for blood grouping and other genetic markers. Among these were assays for the sickle-cell trait. In 1910 a Chicago physician, James Herrick, observed sickle cells in the blood of an anemic dental student [1]. An account of the first sickle-cell patient has been written [2]. Herrick is also credited with the description of the clinical syndrome of coronary thrombosis. In 1917 Emmel [3] found that when blood from susceptible individuals is sealed under glass and allowed to stand at room temperature for several days red cells assumed the sickle shape. It was later demonstrated that sickling depends on a fall in oxygen tension [4]. In 1923 the sickling phenomenon was shown to be inherited as an autosomal dominant trait [5]. By 1949 it was clear that there are two distinct conditions, sickle-cell disease (also termed sickle-cell anemia), in which sickling occurs in venous blood, and the sickle-cell trait, in which more complete deoxygenation is required for sickling to occur. This can be achieved quickly and reproducibly by addition of a reducing agent such as sodium metabisulfite. By analogy with thalassemia major and minor, it was widely believed that carriers of the sickle-cell trait are heterozygous, and persons with sickle-cell disease are homozygous, for the gene concerned, an interpretation supported by the family studies published independently in 1949 by Neel [6] and Beet [7]. It was known that about 8% of African Americans carry the sickle-cell trait [8], but little information was available on the distribution of the trait in Africa. During the 1949 expedition I found that among tribes living close to the coast of Kenya or to Lake Victoria, the frequencies exceeded 20%, whereas among several tribes living in the Kenya highlands or in arid country, the frequencies were less than 1%. These differences cut across linguistic and cultural boundaries and were independent of blood group markers that we documented [9]. In hospitals close to the coast and near Lake Victoria I was shown many children with sickle-cell disease, which was frequently lethal. These observations raised questions; first, if there is strong selection against the sickle-cell homozygote, why is the frequency of the heterozygote high? Second, why is it high in some areas but not others? I formulated an exciting hypothesis; the heterozygotes have a selective advantage, because they are relatively resistant to malaria. This would operate only in areas of intense transmission of Plasmodium falciparum, and the selective advantage of the heterozygote would maintain a stable polymorphism. Testing the hypothesis had to wait until I had completed medical studies and received training in parasitology. Linus Pauling [10] has described how in 1945 he heard from William B. Castle, a Harvard physician, about sickle-cell anemia in which red cells assume this shape in the absence of oxygen. Pauling concluded that the hemoglobin (Hb)11 in these cells becomes aggregated into long, thin filaments when deoxygenated. In 1949 his group [11] found that hemoglobin from patients with sickle-cell anemia is indeed abnormal; at near physiological pH it has a lower negative charge than normal adult hemoglobin. Heterozygotes were shown to have a mixture of about equal quantities of sickle-cell (S) and normal adult (A) hemoglobin. Pauling designated sickle-cell anemia a "molecular disease," a term that provided a stimulus for research defining the molecular basis of other disorders. I spent most of 1953 undertaking postdoctoral research on the hypothesis that sickle-cell heterozygotes (AS) are relatively resistant to malaria. The first experiments were carried out in Nairobi, Kenya on adult volunteers. We studied members of the Luo tribe, who had come from a region close to Lake Victoria where malaria was hyperendemic. Many had migrated to Nairobi in search of work and had lived for years in a non-malarious environment, some longer than others. A laboratory had been established by a pharmaceutical company to test newly developed antimalarial drugs. Humans were infected with P. falciparum and treated with the new drug. If the parasitemia reached a predefined level, or if the subject had uncomfortable symptoms, he was treated with chloroquine. Resistance to the latter was then unknown in East Africa, and at that laboratory no patient had serious complications from the tests. Until the introduction of antibiotics after the 1939–1945 war, induced malaria infections were used routinely as a treatment for neurosyphilis [12]. We found that AS subjects were partially resistant to P. falciparum infections, as shown by lower parasite rates and counts [13]. These experiments were almost certainly biased by powerful effects of acquired immunity, which can persist for years when an adult moves from a malarious to a non-malarious environment. Because we had to use all the AS subjects available but could be more selective with the AA subjects, the two groups of Luo studied were not matched adequately to eliminate such effects. This experience, and discussions with experts on malaria epidemiology, convinced me that it was necessary to study naturally acquired infections in young children aged 6 months to 4 years. Parasite rates in younger infants are low, for several possible reasons, and after the age of 4 years children surviving repeated attacks of malaria show effects of acquired immunity, including decreased parasite counts. I also wanted to study an ethnically homogeneous population living in an area where malaria was highly endemic and where no chemoprophylaxis was used. The necessary blood samples were acquired in Buganda, the region around Kampala, where a medical school, Makerere, and well equipped laboratories were located. On each weekday there were farmers' markets to which women from the countryside, accompanied by small children, brought their produce. A government dispenser went to the market to provide healthcare. He knew his business and was liked and trusted by the local communities. I accompanied him and, as a medical doctor with drugs not available normally, acted as a consultant. In return I received a fingerprick or heelprick blood sample from children in the right age group. Children heterozygous for HbS were found to have lower parasite counts than children with HbA. In particular, the high parasite counts shown previously by Field [14] to be correlated with mortality were significantly less frequent in the AS group [13]. I postulated that AS children are more likely to survive through early childhood in a highly malarious environment than AA children. From this hypothesis it would be expected that high frequencies of AS would be confined to areas where malaria transmission was intense. Kenya, Uganda, and Tanganyika, now Tanzania, included areas where malaria was hyperendemic, separated by high or arid areas where the vector mosquitoes (Anopheles sp.) cannot survive. I surveyed AS frequencies in nearly 5000 East Africans and found that high frequencies were confined to populations living in malarious areas whereas low frequencies were found in tribes living in areas where there was no malaria transmission [15]. Among the Baamba living in the Semliki forest of western Uganda and populations living in Tanzania south of Lake Victoria AS frequencies as high as 40% were found. These differences cut across tribal and linguistic affinities. Clearly an environmental factor, malaria, had a greater effect on the distribution of the sickle-cell trait than affinities defined by linguistic groupings and blood groups [16]. It may be thought that my conclusion was obvious. If so, it is more obvious in retrospect than it was at the time. Published observations on Ugandan populations had attributed the large differences in sickle-cell frequencies observed to the degree of contact of East Bantu-speaking tribes with those speaking Hamitic languages [17]. Observations on the variability of sickle-cell frequencies in Kenyan and Sudanese populations were postulated to result from the introduction of the gene into Africa from the east, probably by sea [18]. Because the HbS mutation in Indians and Arabs is different from those in Africa, as discussed below, the invader hypothesis is untenable. With the publication in 1954 of three papers, one describing the partial resistance of AS subjects to malaria [13], one showing that the distribution of the sickle-cell gene in East African populations corresponded to that of malaria [15], and a third on the implications of the findings for population genetics [19], my position was staked out clearly. Attacks on this claim soon followed. A report that African Americans with AS could be infected artificially with malaria [20] actually demonstrated that in non-immune subjects malaria infections have to be terminated at levels lower than those required to show differences between AA and AS subjects. It was also published that there were no differences in the low parasite densities observed in AA and AS East African adults [21], showing that in adults living in an area of high malaria endemicity the effects of acquired immunity overshadow those attributable to abnormal hemoglobins. Fortunately, researchers in East, Central, and West Africa read my comments on the necessity for studying children between 6 months and 4 years of age and repeated my work in several populations. The results confirmed my findings in detail, as reviewed in Ref. 22 and Table I. Consequently it became generally accepted that AA children living in areas where malaria is endemic are more likely than AS children to have high parasite counts that had been correlated with mortality [14], usually because of cerebral malaria or severe anemia. Several investigators found a relative deficiency of AS children among patients whose deaths were attributable to malaria (Table II). In a well studied West African population [23], AS children were found subsequently to have more than 90% protection from severe malaria (p < 1 × 10−11). Similar findings have been reported from Kenya [24]. In 1954 I postulated that high frequencies of the HbS gene would be found only in areas where P. falciparum was formerly endemic [15]. Many thousands of assays during the second half of the last century have supported that proposal, without exception. In Africa high frequencies of the HbS gene are confined to the malaria belt north of South Africa and south of the Sahara (see Fig. 1 and Fig. 2). Low frequencies of HbS occur in the non-malarious highlands of East Africa and parts of West Africa where HbC is frequent. It was known that HbS occurs sporadically in Mediterranean countries, and two foci of high frequency were defined in Greece; 16% of persons living around Lake Copais in central Greece were found to be AS heterozygotes [25], and even higher frequencies (up to 32%) were observed in populations living in the Chalkhidiki peninsula in northern Greece [26]. Both of these areas were notoriously malarious before control measures were introduced. Tribal populations in the Nilgiri region of southern India were found to have frequencies of AS heterozygotes up to 30% [27]. Since then, HbS has been reported to be polymorphic in about 50 tribal populations, i.e. isolated, dispersed, and endogamic populations in southern and central India [28]. These are regarded by some as descendants of early inhabitants of the Indian subcontinent. These regions were highly malarious until control was introduced. The haplotypes of tribal Indians provide evidence for a unicentric origin of their βS mutation and its difference from African mutations [28]. The authors suggest that the Indian mutation must have arisen and spread before tribal dispersion. These findings indicate that in Africa, Greece, and India malarial selection independently increased the frequency of HbS genes, up to heterozygote frequencies of 30–40%. When a method for cultivating malaria parasites in vitro was developed in 1977, an independent assay for the resistance of red blood cells containing HbS to parasite multiplication became available. It was soon reported that P. falciparum infection increased the rate of sickling of erythrocytes containing HbS and that the parasites were killed under these conditions [29]. Moreover, parasite growth within red cells containing HbS was restricted under conditions of low partial pressures of O2 [30]. During the later stages of P. falciparum maturation, parasite-induced changes in the red cell membrane lead to trapping within venules. Under these conditions hypoxia-induced polymerization of HbS, even in AS cells, could limit parasite multiplication. Trapping in the venous system allows sufficient time for HbS polymerization to occur, as discussed below. My 1954 papers [13, 15, 19] certainly generated interest, and they were followed up by reviews [22, 31]. The general message was that humans are subject to natural selection, just as other animals are, and disease is a potent agent of such selection. The concept of heterosis, heterozygous advantage leading to stable polymorphism, could be illustrated simply by the abnormal hemoglobins. An article for Scientific American [32] was also read widely, and my published maps showing the correspondence between the distribution of the sickle-cell gene and that of malaria [32, 33] have been reproduced frequently. The story is found in many textbooks of biochemistry and genetics. However, the "malaria hypothesis" is often attributed to Haldane [34], on the basis of a paper that is widely quoted but seldom read. Following a presentation of Haldane at an international conference, Montalenti [35] pointed out that the distribution of thalassemia in Italy corresponded to that of malaria. Haldane agreed with Montalenti's suggestion that malarial selection might have influenced the distribution of thalassemia. Haldane then repeated this proposal [34] without acknowledgment to Montalenti. At this time I was doing field work in Africa unaware of these publications. In fact 1949 was an annus mirabilis when the modern phase of research on hemoglobinopathies was initiated. Sickle-cell hemoglobin was discovered, and the inheritance of thalassemia and sickle-cell disease was documented. Montalenti observed high frequencies of thalassemia heterozygotes in formerly malarious parts of Italy, and I observed high frequencies of sickle-cell heterozygotes in formerly malarious parts of Kenya. These high frequencies posed a genetic problem because of selection against both homozygotes. A selective advantage of the heterozygotes was an obvious explanation. The fact that we were dealing with erythrocytic abnormalities, coupled with the distribution of the traits, suggested to Montalenti and me, independently, that malaria might produce such an advantage. Haldane added the prestige of his imprimatur, but he never published a paper on malaria or an abnormal hemoglobin and never mentioned sickle cells. Haldane is rightly acknowledged for his contributions to population genetics. In addition to formulating hypotheses I actually demonstrated that sickle-cell heterozygotes are resistant to malaria and thereby showed the action of natural selection through disease in human populations. I spent most of 1954 in the laboratory of Linus Pauling at the California Institute of Technology. My allotted task was to study the polymerization of HbS. Max Perutz et al. [36] had shown that deoxygenated HbS is relatively insoluble and proposed that sickling involved actual crystallization of hemoglobin in red cells. Pauling believed otherwise, and my experiments were intended to resolve the difference of opinion between two Nobel prizewinners. I found that when HbS was deoxygenated the viscosity of the solution was increased dramatically [37]. Microscopic examination showed the existence of two phases, lens-shaped, birefringent aggregates suspended in a homogeneous, non-birefringent phase. The phases were separable by centrifugation, which allowed quantification of the proportion of aggregated HbS under different conditions. These observations provided compelling evidence that sickling is because of the aggregation of HbS into long rods, which are arranged in parallel, forming liquid crystals or tactoids. From these observations and the theories of Pauling I predicted that the rods must be helical aggregates [37]. I also found that agents forming covalent bonds with sulfhydryl groups prevent sickling. It had been reported that HbS has more available sulfhydryl groups than HbA, but we found two such groups in both molecules and published the first correct descriptions of the available and total sulfhydryl groups in HbS, HbA, and HbF [38, 39]. From studies with hemoglobin mixtures, it was clear that HbC facilitated the aggregation of HbS whereas HbF did not. The conditions for aggregation [37] closely paralleled those under which various erythrocytes (SS, SC, and AS) become sickled [40]. Many investigations of HbS polymerization followed, initially using procedures similar to mine, but more recently supplemented by contemporary physicochemical methods [41]. The polymerization of HbS is among the best understood of all protein self-assembly systems. Only hemoglobins with the β6 substitution of valine for glutamic acid (S and Harlem) aggregate, so the resulting hydrophobic site is required. When hemoglobin is deoxygenated, it changes from a "relaxed" to a "tense" conformation, which is required for HbS aggregation [41]. Aggregation does indeed form helical structures, as shown by electron microscopy. Fourteen strands of the fiber are organized into pairs, producing a fiber 21 nm in diameter. The process is strongly concentration- and time-dependent. The initial polymerization requires aggregation of a few molecules and is rate-limiting; on this "seed" further aggregation rapidly proceeds. At least 80% of AS cells become reoxygenated in the circulation before they can become sickled [41]. However, if the AS cells are trapped in the venous part of the circulation, the great majority become sickled. This is thought to occur during the late stages of P. falciparum multiplication. In 1956 Vernon Ingram [42] was working in the Medical Research Council Laboratory of Molecular Biology in Cambridge, United Kingdom, where Frederick Sanger had developed a "fingerprinting" method for sequencing the amino acids in proteins by using enzymes to cleave the proteins into separable peptides in which sequencing is easier. I provided Ingram with specimens of sickle-cell hemoglobin [43], in which it was demonstrated that there was a single amino acid substitution in the β-chain [43]. In HbA position 6 of the β-chain has an acidic glutamic acid residue, whereas in HbS it is valine. This was the first demonstration in any species that the effect of a mutation is a single amino acid substitution in the encoded protein. When the triplet codes for amino acids were determined and DNA could be sequenced, the nature of the mutation itself could be established. Sickle-cell anemia occurs because of a substitution of thymine for adenine in the DNA codon for glutamic acid (GAG → GTG); in consequence the β6 Glu in HbA becomes β6 Val in HbS. The existence of different mutations leading to HbS is discussed below. The frequencies of abnormal hemoglobins in different populations vary greatly, but some are unquestionably polymorphic. Thus hemoglobin E (β26 Glu → Lys), which is widely distributed in Southeast Asia [44], attains heterozygote frequencies of 55% among Khmers around Angkor Wat and decreases elsewhere in Thailand. The frequency is high in Cambodia and Laos. The gene is still polymorphic in Malaysia, Bangladesh, Assam, and parts of South China, Indonesia, and the Philippines. HbE is absent from the hill tribes of Thailand. This distribution is consistent with malarial selection. Growth of P. falciparum is restricted in cultured EE cells but normal in AE cells [45]. P. falciparum malaria is less severe in AE heterozygotes than in AA persons in Thailand [46]. There is genetic evidence for multiple origins of the βE-globin gene in Southeast Asia [47], and malarial selection may have operated independently on the different E mutants. Hemoglobin C (β6 Glu → Lys) is polymorphic in West Africa, attaining heterozygote frequencies approaching 20% in northern Ghana and Burkina Faso [48]. The frequencies decline in all directions, although the polymorphism persists through much of sub-Saharan West Africa. The distribution is again consistent with malarial selection. In culture P. falciparum growth is normal in AC cells but retarded in CC cells [45]. Epidemiological studies in Burkina Faso have shown some protection of AC heterozygotes and stronger protection of CC homozygotes against P. falciparum [48]. Although the evidence is less compelling than in the case of the sickle-cell gene, it strongly suggests a role for malaria selection in the distribution of the HbE and HbC genes. Deficiencies of this enzyme are polymorphic in many parts of the world. I collaborated in work showing that P. falciparum parasite densities were lower in G6PD-deficient Tanzanian children than others living in an area where chemoprophylaxis was not used [49]. This observation has been confirmed repeatedly [50]. An interesting example is a study of African females of the GdB/GdA− phenotype [51]. Because of the mosaic expression of X-linked genes, some of the erythrocytes of these women had normal G6PD whereas others were deficient. Malaria parasites were rarely found in GdA− cells, in contrast to GdB+ cells. G6PD-deficient cells do not efficiently support malaria parasite growth in culture [45]. These observations, together with the finding that high frequencies of G6PD deficiency are confined to formerly malarious parts of the world, are consistent with a role for malaria selection in the distribution of G6PD genes. After the field observations I analyzed the population genetics of the sickle-cell gene [19]. Given that heterozygote frequencies rise to 40% (4% of the population are SS homozygotes), a mutation rate of about 10−1 per gene per generation would be required to replace the loss of sickle-cell genes. Even if the fitness of SS is one quarter of the fitness of AA, the fitness of AS must be about 1.26 to maintain a stable polymorphism. Observations on mutation rates in Congolese subjects showed that they were far too low to explain the frequency of the sickle-cell gene [52]. I also presented calculations of the rates of change in sickle-cell gene frequency toward a stable polymorphism under malarial selection (Fig. 3) and the decline in the frequency of the gene when this selection is removed (Fig. 4). Later publications on this subject [53, 54] have not substantially changed those estimates. An interesting complication of this straightforward picture is introduced when the genes for more than one hemoglobinopathy are common in a population. For example, in Greece the HbS and β-thalassemia genes are polymorphic, and in West Africa this is true of the HbS and HbC genes. Heterozygotes for both hemoglobinopathies (HbS/β-thalassemia and HbS/HbC) suffer from variants of sickle-cell disease. Although these are less severe than the disease in HbS homozygotes, there is little doubt that before modern medical practice was introduced the fitness of the abnormal heterozygotes was less than the mean fitness of the population. It was calculated that in such tri-allelic systems a stable equilibrium might be attained; however, the stability is sensitive to changes in the relative fitness values assumed for the different genotypes and might easily tend to an unstable state [55]. Before this paper [31] and later [22] I postulated that these genes (HbS and β-thalassemia in one situation and HbS and HbC in another) would tend to be mutually exclusive in populations. Our own observations [56], as well as those of Greek investigators [67], showed that in areas of Greece where HbS is frequent β-thalassemia is relatively rare and vice versa (Fig. 5). Likewise, where HbC is common in West Africa HbS is relatively rare (Fig. 6). A recent report on the resistance to malaria of CC and, to a lesser extent, AC West Africans [48], develops this theme further. Because there is little selection against CC homozygotes but selection against SC and SS persons, the authors postulate that the HbC gene is replacing the HbS gene in West Africa. What happens when malarial selection operates on two independent genes, such as those for abnormal hemoglobins and G6PD deficiency? In this case the fitness of persons with both the abnormal hemoglobin and enzyme deficiency is not less than the mean fitness of the population, but heterozygotes for either trait have a fitness above the mean because of resistance to malaria. Hence it would be expected that the frequencies of the two independent genes in populations should be correlated positively, which is observed (Fig. 7). Has the frequency of HbS in the United States black population decreased as a result of eliminating malarial selection? This frequency (about 8% heterozygotes) is certainly less than that in many West African populations, even when dilution by genes from non-blacks is taken into account [22]. However, HbS frequencies vary widely in different West African populations, and their relative contribution to the American black population is not known with certainty. Hence the question cannot be answered with confidence. Calculations using the many genetic polymorphisms now available should throw light on the problem. Did the sickle-cell mutation originate only once and become widely distributed, or did it arise independently in different parts of the world? Two mutations found in non-transcribed sequences of DNA adjacent to the β-globin gene are so close to each other that the likelihood of crossover is very small. Thus correlations will persist through many generations, providing a marker for population affinities and movements. Restriction endonuclease digests of the β-globin gene cluster show five distinct patterns associated with the sickle-cell mutation, four being observed in Africa, the Bantu, Benin, Senegal, and Cameroon types [57], and the fifth in the Indian subcontinent and Arabia [28]. This is evidence that the HbS mutation occurred independently at least five times. The high levels of AS in parts of Africa and India resulted presumably from independent selection in different populations living in malarious environments. With the introduction of gene sequencing it has become apparent that the human genome is highly polymorphic. To what extent are these polymorphisms subject to selection, and is the strong selection acting on abnormal hemoglobins an isolated and unrepresentative situation? The answer to the second question is clearly no. Selection also operates on blood group genes. Relevant to malaria is the resistance to Plasmodium vivax of persons of West African origin lacking the Duffy blood group on their red cells [58]. The Duffy blood group is a receptor for this malaria parasite, and malarial selection probably explains the high frequency in West Africa of a promoter gene mutation altering the expression of the Duffy protein on red cells but not on other cell types [59]. A second example of selection operating on blood groups is the relative susceptibility of persons carrying the O group to cholera [60]. Polymorphisms that have attracted a great deal of attention include those involving major histocompatibility complex proteins and their associations with diseases. In the case of malaria Hill et al. [23] found that in West African children a major histocompatibility complex class I haplotype (HLA-Bw53) and an HLA class II haplotype (DRW13, subtype 02) are common whereas they are rare in non-African populations. These haplotypes are associated independently with protection against severe malaria [23]. The observations support the hypothesis that the extraordinary polymorphism of major histocompatibility complex genes has evolved primarily through natural selection by infectious pathogens. Other examples of human polymorphic genes influenced by selection are known. However, it is clear that many polymorphisms do not detectably influence the quantities or functions of the proteins encoded so that they are unlikely to have appreciable selective effects. At the one extreme are genes for abnormal hemoglobins, which are subject to strong selection, and at the other extreme are polymorphisms that are selectively neutral. The geneticists who postulated that polymorphism is maintained by selection and those who postulated random genetic drift were both right. It will be recalled that my genetic researches on East African populations had two objectives, to document genetic markers that might be used in assessing relationships to other human populations and their origins and to analyze selective effects. Regarding the first objective our studies of blood groups [9, 16] were a modest beginning of what has grown to be a large enterprise. Analyses of mitochondrial DNA sequences [61] and linkage disequilibrium [62] have provided compelling evidence that all modern humans originated in Africa, probably in Kenya or Ethiopia [63]. Fossils believed to represent early members of the hominid lineage (dating from 5 to 6 million years ago) have been found in Kenya [64] and Ethiopia [65]. Thus East Africa seems to be at least one of the places in which early humans developed, as Leakey predicted when I was a schoolboy. The distribution of P. falciparum in Africa before malaria control was introduced (modified from M. F. Boyd's Malariology) [66]. Frequencies of sickle-cell heterozygotes in different parts of Africa. The change in sickle-cell heterozygote + homozygote frequency in populations that would occur from a high or a low level when the fitness of the normal homozygote, sickle-cell heterozygote, and sickle-cell homozygote are 0.95, 1.19, and 0.30, respectively. The interrupted horizontal line represents an equilibrium frequency of 40% heterozygotes, the highest observed (from Ref. 19). The change in sickle-cell heterozygote + homozygote frequency that would occur in the absence of malaria, assuming that the fitness of the sickle-cell homozygote is 0.25 and that of sickle-cell heterozygotes and normal homozygotes is 1.0 (from Ref. 19). Frequencies of the sickle-cell gene plotted against frequencies of the β-thalassemia gene in different parts of Greece [22]. Frequencies of the sickle-cell and hemoglobin C genes in West African populations [22]. Frequencies of the sickle-cell and glucose-6-phosphate dehydrogenase traits in African populations [22].