https://orcid.org/0000-0002-4271-1527
Abstract
Sub-Saharan Africa is the most promising region of the world to conduct high-throughput studies to unveil adaptations to infectious diseases due to several reasons, namely, the longest evolving time-depth in the Homo sapiens phylogenetic tree (at least two-third older than any other worldwide region); the continuous burden of infectious diseases (still number one in health/life threat); and the coexistence of populations practising diverse subsistence modes (nomadic or seminomadic hunter-gatherers and agropastoralists, and sedentary agriculturalists, small urban and megacity groups). In this review, we will present the most up-to-date results that shed light on three main hypotheses related with this adaptation. One is the hypothesis of coevolution between host and pathogen, given enough time for the establishment of this highly dynamic relationship. The second hypothesis enunciates that the agricultural transition was responsible for the increase of the infectious disease burden, due to the huge expansion of the sedentary human population and the cohabitation with domesticates as main reservoirs of pathogens. The third hypothesis states that the boosting of our immune system against pathogens by past selection may have resulted in maladaptation of the developed hygienic societies, leading to an increase of allergic, inflammatory and autoimmune disorders. Further work will enlighten the biological mechanisms behind these main adaptations, which can be insightful for translation into diagnosis, prognosis and treatment interventions.
Introduction
The richness of sub-Saharan African (SSA) populations in terms of genetic, cultural, linguistic and environmental features incurs the risk of becoming a cliché (1–4), but it is still pretty much a ‘raw diamond’ waiting to be cut (5,6). SSA is inhabited by surviving nomadic hunter-gatherers living in tropical forests; nomadic/semi-nomadic agropastoralists following herds and sedentary farming/agriculture groups that have been transforming landscapes since 6 thousand years ago (kya); cosmopolitan groups in the transition from traditional ways of life; and highly admixed populations living in fully industrialized megacities. The 70° north-to-south axis of the SSA region encompasses several ecological zones (7): tropical rain forest; tropical moist deciduous forest; tropical dry forest; tropical shrubland; tropical mountain systems; subtropical humid forest; subtropical dry forest; subtropical steppe; and subtropical mountain systems. Numerous reservoirs, vectors and pathogens pullulate in these diverse ecological zones, especially so in locations closer to the equator (8).
The information contained in the SSA populations largely waits to be exposed via the high-throughput technologies such as arrays, exome/genome sequencing, transcriptomics, methylomics, microbiomics and proteomics (Fig. 1). The application of these approaches in SSA populations requires overcoming several challenges (6), namely, the active involvement of SSA researchers and communities, the large funding supported by global cooperation, the collection of diverse biological samples representative of the continental palimpsest and the development of laboratorial and bioinformatics tools (e.g. biobanks and cancer cell line repositories) dedicated to SSA big data. Here, we will review incipient but diversified high-throughput research conducted in SSA populations in order to demonstrate that the pay-off does justify addressing these challenges.
Diverse omics studies are beginning to be performed in various groups from sub-Saharan Africa, and sometimes in comparison with Europeans. Genomics. DNA can be extracted from a saliva sample and analysed by genotyping specific variants in an array or by sequencing the whole exome/genome. The genomic profiles from various individuals affiliated in a population group enable the detection of signs of selection that occurred along time in that group. Transcriptomics. RNA extracted from tissues can be sequenced in the form of cDNA and reveal differential gene expression profiles between individuals or between cells after exposure to an infectious stimulus. These differences in expression can be due to the influence of genetic variants, the so-called expression quantitative trait loci (eQTL). Methylomics. Screenings on methylated genes along the genome allow to detect differentially methylated genes between populations that may be due to genetic or environmental factors. Microbiomics. Microbiome characterization can be performed in various biosamples (as faeces and saliva) through sequencing of the 16S rRNA gene, allowing the phylogenetic classification and quantification of the microbes present in the samples to be compared.
We will focus on hypotheses related with human adaptation to infectious diseases (Fig. 2). Adaptation, in which an advantageous phenotype is enriched in response to an environmental challenge (9), takes place through various modes: positive selection, which increases the frequency of an advantageous allele either via classic sweep (strong selection on a new mutation), standing variation (selection on a pre-existing mutation) or polygenic adaptation (selection acting at once on multiple loci); balancing selection (preserves deleterious alleles in populations where the driving agent is acting due to heterozygote advantage or frequency-dependent selection); or adaptation through admixture (a neutral variant of group A is selected for in the recipient group B). Genome-wide selection scans in global human populations have shown that immune-related genes display strong selection signatures, supporting the hypothesis that pathogens have been key players in driving human evolution (10). However, it is still premature to attribute an estimate to the proportion of the human genome selection events due to pathogen pressure. Several identified selected genomic regions contain several genes, rendering it difficult to attribute the most probable selection cause (11). Population samples from SSA and other tropical regions, where infectious diseases abound, are still largely absent from genome-wide association tests that are an important source of evidence for adaptations (12). Continuing efforts in conducting genome-wide selection scans in populations representative of the global human diversity will enlighten the contribution of pathogen-driven selection in human evolution.
Recent and old host–pathogen interactions. Coevolution. The long period of human evolution restricted to sub-Saharan Africa (from 300–70 kya; thousand years ago) favoured coevolution of old pathogens, which adapted to low host population densities (allowing repletion of the reservoir of susceptible individuals) and slow progression to disease (passing a latent/asymptomatic infection period). Crow diseases. The agricultural development (in Africa, at around 6 kya) led to the appearance of crowd diseases, which are generally caused by zoonotic pathogens and highly virulent. Hygiene hypothesis. This hypothesis states that the 20th century-packed human urban populations have low exposure to pathogens, and now the selected boosted immune system attacks the self, increasing allergic reactions, chronic inflammatory and autoimmune disorders.
Coevolution of host and pathogens
Ehrlich and Raven (13) coined the term coevolution to describe relationships between two entities exerting reciprocal selective pressures. Coevolution is a highly dynamic process (14) as selection continually challenges the attained adaptive peaks, rendering coevolving species maladapted most of the time. A very detailed introduction to coevolutionary concepts, first genomic evidence and appropriate algorithms is provided in a recent review (15). Here, we will provide some examples of coevolution in the field of human infectious diseases (16,17), focusing on the impact in the human genomic diversity. As coevolution implies time, it is expectable to affect pathogens that have been infecting the human species since before/close to its African origin around 300 kya. Kodaman et al., 2014, (17) indicated Helicobacter pylori (Hp), Mycobacterium tuberculosis (Mt) and human papillomavirus as examples of agents of old human infections, preceding the out-of-Africa migration at around 70 kya. Concordantly (Fig. 2), these pathogens are adapted to low host population densities, that would be the norm before the out-of-Africa migration (18), and lead to slow progression to disease after an extended period of latent/asymptomatic infection. This scenario allows predictions to be made in terms of differential adaptation of worldwide population groups (Fig. 3). Deeper SSA lineages would be best adapted, while maladaptation would be expected following the out-of-Africa migration, due to the selective pressures from the new European and Asian environments. The European and Asian groups would then have to re-adapt locally. Disruptive coevolution would be expected of the historical migrations bringing into contact mismatched human and pathogen ancestries.
Differential adaptation of worldwide population groups for old pathogens. Coevolution. The more time for coevolution in sub-Saharan Africa led to a naturally better protection of these individuals with a lower virulence of African strains. Maladaptation. Following the out-of-Africa migration, maladaptation due to the new environments potentially increased the virulence in the Asian and European groups, who would then have to re-adapt locally. Disrupted coevolution. Encounter of mismatched host–pathogen groups, motivated by historical migrations, led to higher virulence being displayed in mismatched cases.
We tried to uncover functional evidence of coevolution by conducting the first in vitro coinfection assays with human and Hp matched and mismatched ancestries, in African and European backgrounds, followed by human transcriptomic analysis (19). We showed that the host response to Hp infection was greatly shaped by the human ancestry, with major variability affecting the innate immune system and the metabolism. This variability was concordant with a decreased pro-inflammatory response in the SSA human epithelial cells and increased SSA lipid metabolism (already shown to be involved in the natural protection of SSA ancestry against viral dengue disease; (20)). The African human ancestry showed clear signs of coevolution with Hp, while the European ancestry was maladapted, in agreement with the fact that SSA populations have a lower incidence of gastric cancer than non-SSA populations, despite the almost ubiquitous African prevalence of Hp infection (21). Our model only mimics the initial stages of infection, and at this level the mismatched ancestry did not seem to be an important differentiator of the host gene expression pattern. But observations conducted in admixed Colombians (22) indicated that the severity of the gastric disease correlated positively with the proportion of African Hp ancestry in patients from Amerindian ancestry (mismatched ancestries), while patients with a large proportion of African ancestry infected by African Hp strains (matched African ancestries) had the best prognosis.
Population genetic evidence of coevolution was also reported by McHenry et al. (23), who evaluated the genetic variation in two tuberculosis cohorts (screened for variants on 29 innate and adaptive immune genes or a 4-million array) from Uganda and Mt (lineage inference) as a unit. The authors found that the most recent Mt lineage in Uganda (L4.6/Uganda, derived from the worldwide widespread European-L4 lineage) led to higher disease severity when the host presented the ancestral rs17235409 allele in SLC11A1 gene. This gene is an important regulator of macrophage responses to this mycobacterium, indicating that pathogen and humans have coevolved to modulate tuberculosis severity.
Phylogenetic analyses of the gastrointestinal parasites helminths elucidated that these eukaryotes were already infecting early hominids, more than 1 million years ago (24). These ‘old friends’ could also have coevolved with us (24), their antigens being recognized by toll-like receptors on dendritic cells (DC) and eliciting T regulatory (Treg) cell-mediated response. This continuous activation of the immune regulatory mechanisms is hypothesized to modulate the immune system homeostasis (25), including an ‘healthy’ gastrointestinal flora. Unfortunately, most of the published microbiome studies overlook the presence/absence of gastrointestinal parasites, as the less expensive 16S rRNA-based sequencing does not allow to reliably identify eukaryotic organisms. In one of the few exceptions, the authors (26) surveyed both bacteria (16S rRNA sequencing) and eukaryotic (faecal microscopy, quantitative PCR and limited shotgun metagenomic sequencing) gut communities in 575 pastoralists, agropastoralists and hunter-gatherers from Cameroon. They observed that hunter-gatherers have higher frequency of parasites (Ascaris lumbricoides, Necator americanus, Trichuris trichiura and Strongyloides stercoralis) versus non-hunter-gatherers, which was positively associated with gut microbial diversity, thus a healthier microbiome. In fact, a poorer gut microbial diversity has been associated with disbyosis and gut diseases, namely gastric cancer (27), despite still retaining information for geographical affiliation of the host in European or Asian cohorts (28). Additionally, the measurement of 21 plasma cytokines through a multiplex magnetic bead panel allowed the former authors (26) to detect that the parasite colonization was associated with multiple immune mechanisms, such as elevated levels of T helper cells 1 and 2 (TH1 and TH2) and proinflammatory cytokines—an activated immune system, although still unclear if contributing to homeostasis. This evidence of higher parasite infection and higher microbiome diversity replicates findings of previous work conducted also in Cameroon (29), but here the authors did not conclude for higher parasite (inferred only from microscope observation) infection of hunter-gatherers, calling for further research in other hunter-gatherer groups.
The infectious disease burden of the agricultural transition—emergence of crowd diseases
The establishment of the Holocene around 12 kya, characterized by a stable and agreeable climate pattern, allowed the agricultural transition in independent centres around the globe, forever changing human demography and ecology (30). It was now convenient to settle permanently nearby cultivated fields, increase the population density in these areas leading, consequently, to the development of the crowd diseases (30). These diseases depend on high host population densities, so that there are enough individuals to perpetuate the cycles of transmission of highly virulent pathogens, which are usually zoonotic transfers from wild or domesticated animals to humans, such as smallpox, measles and rubella (Fig. 2). So, expectations would be for higher pathogen selection pressures on agriculturalists than on hunter-gatherers.
Several studies have already been focused on generating genome-wide datasets in extended SSA regions and searching for selection signals (11,31–33). Results have been concordant in finding that the strongest signals of positive selection are immune related and metabolism related, that migrants can acquire selected genes from locals through admixture and that most candidate selected genes vary between populations irrespectively of the subsistence system. So there seems to be a minor overlapping on signals of adaptation across continental groups, which may reflect polygenic adaptation, calling for extended studies that provide a good population coverage. These array-based studies cannot circumvent the ascertainment bias that prioritizes common variants, so there is the urgent demand for unbiased whole-exome or whole-genome sequencing. These next generation sequencing studies are beginning to be published and providing interesting insights. Lopez et al. (34) analysed 566 high-coverage exomes and 40 low-coverage genomes from agriculturalists and hunter-gatherers and found evidence for strong signals of polygenic adaptation for height, life history (e.g. reproductive age) and pathogen-driven responses (responses of mast cells to allergens and microbes, IL-2 signalling pathway and host interactions with viruses) in hunter-gatherers. Fan et al. (4) sequenced 92 whole genomes from 44 SSA indigenous populations and found strong signals of positive selection due to local adaptation in the six meta-populations (two hunter-gatherers, one pastoralist, three agriculturalists) near genes acting on immunity, cardiovascular function and metabolism. For the immune function, the analysis highlighted antimicrobial humoral response in the central African rainforest hunter-gatherers, B cell homeostasis in the Niger-Congo agriculturalists and San hunter-gatherers, regulation of phagocytosis and chemokine signalling in the Niger-Congo agriculturalists and cytokine production in the Nilo-Saharan pastoralists. Whole genomes obtained for 25 individuals from five Khoe-San populations (35), representing the deepest lineages in the human tree, allowed the identification of past and recent selection signals overlapping immunity genes, direct evidence that immunity genes have been under selection throughout human evolutionary history. Taken together, these selection screenings are not supporting a higher pathogen burden of the agriculture transition.
Other types of omics are being conducted on SSA diverse subsistence systems. DNA methylation profiles complemented with genome-wide genotypes for 112 rainforest hunter-gatherers (w-hunter-gatherers), 94 sedentary agriculturalists occupying nearby urban deforested habitats (w-agriculturalists) and 61 agriculturalists living and regularly practicing hunting in a forested region (f-agriculturalists) of the Gabon/Cameroon area (36) revealed differential methylation variation depending on time-scale and habitat. Methylation variation associated with recent changes in habitat (w-agriculturalists × f-agriculturalists comparison) mostly concerned immune and cellular functions were due to biotic changes and not to genetic diversity (as the groups were genetically homogeneous). Whereas methylation variation associated with historical lifestyle (w-hunter-gatherers × f-agriculturalists comparison) affected developmental processes (as IGF2BP2, HOXC6 and ZNF492 related with height, age at menarche, type-2 diabetes, bone mineral density and gene–diet interactions) were due to genetic differences between the groups, and many of these variants displayed signs of positive selection among the agriculturalist populations.
The functional results from Harrison et al. (37) even oppose the agricultural hypothesis expectations. They isolated peripheral blood mononuclear cells (PBMCs) from the rainforest Batwa hunter-gatherers and their neighbour Bakiga agriculturalists, from Uganda, exposed them to gardiquimod (simulates infection with a single-stranded RNA virus) or to lipopolysaccharide (mirrors a Gram-negative bacteria) stimuli and evaluated the transcriptional response. The authors verified that positive natural selection shaped differently the early immune transcriptional response to viruses (interferon-γ and interferon-α responses) between hunter-gatherers and agriculturalists, being signals more frequent in hunter-gatherers. Responses to bacteria were not affected.
Another important source of information to this issue are gut microbiome studies. Two studies (38,39) performed 16S rRNA sequencing on traditional SSA communities from Central African Republic, Tanzania and Botswana, finding that the gut microbiomes of hunter-gatherers are phylogenetically distinct from agriculturalists, pastoralists and agropastoralists. Hunter-gatherers have increased abundance of Prevotellaceae, Treponema, and Clostridiaceae, while agriculturalists gut microbiome is dominated by Firmicutes rendering it more similar to a westernized gut microbiome (USA cohorts were used for comparison purposes). The first work (38) included metabolomics data, allowing the conclusion that agriculturalists had an increased abundance of predictive carbohydrate and xenobiotic metabolic pathways, while hunter-gatherers displayed increased abundance of predicted virulence, amino acid and vitamin metabolism functions, as well as dominance of lipid and amino acid-derived metabolites. These microbiomes seem, thus, to be adapted to the diverse diets between agriculturalists and hunter-gatherers. Rampeli et al. (40) performed additionally a ‘resistome’ analysis in Hadza hunter-gatherers gut microbiome to inform on the antibiotic resistance potential of the microbes. They noticed the presence of antibiotic resistance genes in the little antibiotic exposed hunter-gatherers microbiome, suggesting the global presence of environmentally derived resistances even in unexpected places as the SSA rain forest.
By clearing forests and constructing irrigation systems, agriculturalists may also have propitiated the increase in vectors transmitting pathogens. Plasmodium falciparum malaria is the best case study in the African context for this implication, and supportive evidence came from various works (41,42), namely Otto et al. (43) analysis upon good-quality genomes from the closest ancestor of human P. falciparum in gorillas. This analysis indicated that divergence between human and gorilla Plasmodium species occurred around 40–60 kya, followed by a period when the low human and gorilla population densities did not drive selection of host-specific vectors and parasites, a situation that changed with the expansion of the human population at the advent of farming. Farming could have driven the selection of Anopheles gambiae mosquitoes to feed primarily on humans, and also the selection of the fittest infective P. falciparum genotype. However, a recent careful (accounting for balancing selection) analysis (44) of the sickle-cell mutation that protects against malaria revealed that this mutation emerged around 22 kya in the ancestors of current SSA agriculturalists, showing a more ancient exposure to malaria than previously found (under neutrality, the estimation was around 7.3 kya; (45)). By including hunter-gatherers in their study, the authors (44) showed that they acquired the advantageous mutation through adaptive gene flow from the agriculturalists ancestors during the last 6 kya.
Overall, these results are not supporting the hypothesis that selection on the immune system driven by higher pathogen pressure was a main change in the transition to agriculturalists subsistence. In fact, some evidence even seems to contradict it and may be reflecting pre-agriculture responses to ecological differences since hunter-gatherers–agriculturalists separation 60 kya (46), including for malaria. Definitely, further investigation is needed, but the high plasticity of the immune system will be a big challenge to overcome when trying to disentangle selection events.
The hygiene or pathogenic sterilization hypothesis
Epidemiological observations in the 1980s indicated that having many siblings correlated with a decreased risk of hay fever, suggesting a protective influence of postnatal infection that might be lost in the presence of modern hygiene (47). The ‘hygiene hypothesis’ was formulated, explaining the increase of allergies, chronic inflammatory and autoimmune disorders in developed economies as resulting from the decreased exposure to pathogens, leading to an imbalanced immune response that eventually promoted chronic inflammation (24) (Fig. 2).
Population genetics findings are supporting this hypothesis. The candidate genes for susceptibility to inflammatory bowel disease, celiac disease, type-I diabetes, multiple sclerosis or psoriasis, found through genome-wide studies (GWAS), are also the ones being discovered in positive selection screenings, many showing strong (Fst values > 0.4, between SSA and European) population differentiation (48). It has been hard to identify the pathogen driving each of these selection events, but an indirect indication comes from the higher frequency of these risk alleles in populations exposed to high pathogen loads. In the pre-industrialized environments, these variants seem to have played an otherwise beneficial protective role in host defense (49).
Nedelec et al. (50) followed up on this differential population susceptibility to immune-related diseases by testing the transcriptional response of African versus European primary macrophages to live bacterial pathogens (Gram-positive Listeria monocytogenes and Gram-negative Salmonella typhimurium). The ancestry-associated differences in the gene regulatory response to infection affected 9.3% of macrophage-expressed genes, leading African ancestry to display a stronger inflammatory response and reduced intracellular bacterial growth. These differences are largely under genetic control by cis- or trans-acting expression quantitative trait loci (eQTL), which also match recent, population-specific signatures of adaptation. Interpreting these results in the light of expectations from the ‘hygiene hypothesis’: SSA populations continue to be mostly under pathogen pressure, so a higher inflammatory response was favoured allowing survival to infectious diseases; out-of-African migrants, as Europeans, adapted to a low-pathogen burden, reducing their inflammatory response, but a fraction of these individuals will still have an excessive immune response that may act against the self.
Within Africa, the effects of distancing from a traditional way of life are being evaluated in terms of immune-related adaptations. Probably the first impact will be an increase of exposure to new pathogens as less isolated groups (but keeping living in the same location and conditions) will have contact with more people. Evidence supporting this higher exposure was found by Owers et al. (51) when conducting a genome-wide selection screening in San groups: eight immune genes were targeted for strong selection in the ‡Khomani (abundant contact with migrants into the region), and none in the Juj’hoansi (isolated). At a second level, the transposition of people to an urban westernized way of life will eventually increase hygienic conditions and lower overall pathogen pressure. The already mentioned epigenomic survey (36) found that urban agriculturalists were enriched in differentially methylated genes associated with autoimmune disorders when compared with forest-based agriculturalists, following the same tendency of populations from western economies.
The fast pace of technological advance is allowing remarkable direct insights into the past, namely the pre-colonial SSA period through the ‘lens’ of the human gut microbiome from a mid-15th century Bantu-speaker palaeofaecal specimen (Bushman Rock Shelter in Limpopo Province, South Africa (52)). This gut microbiome matched a mixed forager-agro-pastoralist diet, testifying the dependence on a hybrid food system. But more important is that now we know the constitution of an SSA gut microbiome (to add up to one from the 5000-old Tyrolean Iceman) preceding recent adaptation to westernized diets, including the consumption of coffee, tea, chocolate, citrus and soy, and the use of antibiotics, analgesics and toxic environmental pollutants. In adaptation to these new ingredients and drugs/pollutants, extant gut microbiomes acquired new bacteria able to process/eliminate them, and even surviving hunter-gatherers did not escape these acquisitions, as already mentioned. It remains to be proved if these microbiome changes rendered us more prone to allergic reactions and autoimmune and inflammatory diseases.
Conclusions
We hope to have provided convincing evidence of the gains from pursuing varied high-throughput studies in SSA diverse groups. The theoretical impact of these discoveries will be unique in terms of knowledge on human adaptation driven by pathogens, some of which already acquired a ‘residential permit’ in human bodies, after thousands of years of coevolution. In practical terms, results from coevolution-based analyses can be used to considerably improve prognosis of infectious diseases, by performing the simultaneous genetic screening of host and pathogens (22). A first level of this prognosis could be to ascertain only ancestry backgrounds and check for matched or mismatched ancestries; a second more precise level would be to screen coevolved variants that are beginning to be discovered (23). Faecal microbiota transplantation is being used as an efficient treatment for recurrent Clostridium difficile infection, although it is still a challenge in terms of regularization and quality control (53). For sure, more information on microbiomes from traditional ways of life and from ancient faeces will provide insights into useful microbes that could help us in re-acquiring microbiome diversity and immune homeostasis. Our group (54) is exploring an adaptive strategy on SSA lipid profile regulation as a mechanism of resistance from viral diseases, in order to repurpose lipid-acting drugs as prophylactics/treatment in dengue fever. Adaptive immune responses selected the best biological mechanism that makes us resistant to diseases; once we have characterized these mechanisms, we can mimic them in terms of pharmaceutical interventions.
Acknowledgements
FCT and Portuguese Foundation for Science and Technology (PhD grants SFRH/BD/136299/2018 and SFRH/BD/145217/2019, respectively) to N.P. and R.J.P. i3S is financed by FEDER—Fundo Europeu de Desenvolvimento Regional funds through the COMPETE 2020—Competitiveness and Internationalization Operational Programme (POCI), Portugal 2020, and by Portuguese funds through FCT/Ministério da Ciência, Tecnologia e Inovação in the framework of the project ‘Institute for Research and Innovation in Health Sciences’ (POCI-01-0145-FEDER-007274).
Conflict of Interest statement. No conflict of interests.
