Studying human and non-human primate evolutionary biology with powerful in vitro and in vivo functional genomics tools

Abstract

In recent years, tools for functional genomic studies have become increasingly feasible for use by evolutionary anthropologists. In this review, we provide brief overviews of several exciting in vitro techniques that can be paired with ‘-omics’ approaches (e.g., genomics, epigenomics, transcriptomics, proteomics, metabolomics) for potentially powerful evolutionary insights. These in vitro techniques include ancestral protein resurrection, cell line experiments using primary, immortalized, and induced pluripotent stem cells, and CRISPr-cas9 genetic manipulation. We also discuss how several of these methods can be used in vivo, for transgenic organism studies of human and non-human primate evolution. Throughout this review, we highlight example studies in which these approaches have already been used to inform our understanding of the evolutionary biology of modern and archaic humans and other primates while simultaneously identifying future opportunities for anthropologists to use this toolkit to help answer additional outstanding questions in evolutionary anthropology.

1. Introduction

Evolutionary anthropologists seek to understand the proximate and ultimate mechanisms of variation and adaptation within and among humans and other primate species^1–8. Anthropologists can now employ an arsenal of functional genomics techniques to help identify the causative genetic underpinnings of this variation as a means of testing hypotheses about the origins and evolutionary trajectories of the traits of interest along the primate lineage (Figure 1). We review a number of these approaches here.

Figure 1. The potential of functional genomics techniques for investigating questions in evolutionary anthropology.

Individual images: Shutterstock and Phylopic.

2. Functional Genomics Techniques

The foundations of functional genomics as we use the term are the ‘-omics’ fields, collectively: genomics, epigenomics, transcriptomics, proteomics, and metabolomics (Figure 2). Genomics refers to the underlying nucleotide sequences and organizational structure of genomes, while epigenomics is the genome-scale study of reversible processes like methylation and histone modification that can alter gene expression. The large-scale analysis of gene expression variation is called transcriptomics. Proteomics and metabolomics, in turn, refer to approaches for studying the translation and/or modification of proteins and the products of biochemical or metabolic activity, respectively.

Figure 2. The ‘-omics’.

A schematic representation of the ‘-omics’ fields of genomics, epigenomics, transcriptomics, proteomics, and metabolomics, including depictions of the origins of the biological molecules that are the focus of each field of study and some of the technologies through which these data are produced. Individual images: BioRender.

Omics research produces large-scale data through technologies including high-throughput, massively-parallel sequencing, fluorescent microarray, and mass-spectrometry. As the financial cost of producing and computational cost of analyzing -omics data has decreased, these technologies are more readily available for incorporation into research by scientists of all fields, including evolutionary anthropologists interested in tracing the histories of key human or non-human primate traits and testing hypotheses about their evolution. These -omics technologies, when combined with the functional genomics experimental approaches described below, represent synergistic opportunities to identify and characterize targets of selection and inform our understandings of human and non-human primate variation and evolution.

2.1. Cell Culture.

In vivo experiments – or experiments using living humans and other primate species – to investigate the evolutionary basis of phenotypic differences are often limited by ethical, technical, financial, and other difficulties. However, excitingly, anthropologists can still make substantial advances by using proxy in vitro studies, in which researchers conduct experiments with components of an organism isolated from the whole. One common in vitro technique is cell culture, or the growth and division of cells in an artificial environment. This artificial environment generally consists of a substrate or media with essential nutrients like vitamins, amino acids, and carbohydrates, growth factors, and gases like CO₂ or O₂. Once an organism’s cells can be maintained in this fashion, a figurative world of possible functional genomics experiments opens to scientists. Various types of human and other primate cells can be cultured for observing biological responses with the benefit of one or more -omics technologies. Once collected or generated, most of the different types of cells can be preserved indefinitely with liquid nitrogen for future work. What makes one cell type a better choice than others for a particular experiment might depend both on the specific tissues of origin and variation in the number of times the cells can divide in cell culture.

One general class of cultured cells are ‘primary’ cells. Primary cells are taken directly from the living tissue of an organism, via a blood sample, biopsy, or other means, and are otherwise not modified in any way. Because primary cells closely mimic the physiology of cells in vivo, they are ideal for exploring the phenotypic consequences of genetic variation, which depend on the tissue and/or cell type or a specific biological state⁹. Some types of primary cells, such as peripheral blood mononuclear cells (PBMCs) cannot divide in culture, whereas others like epithelial or endothelial cells can be cultured through fewer than ten cell divisions as ‘cell lines’. Still others, e.g., fibroblasts, can be cultured for even longer, up to 50 divisions. Because all primary cell samples or cell lines are established from blood samples or biopsies, in some cases they may be difficult to obtain from larger numbers of individuals in a species or for multiple different species, especially in quantities per individual sufficient for performing multiple experiments. Furthermore, after a limited number of divisions, primary cells enter a ‘replicative senescence’ stage, in which they exhibit substantial changes in morphology and gene expression patterns and no longer divide, and after which they should probably not be used for functional genomics experiments.

If long-term or extensive use of identical cells for many different or replicative experiments is needed, researchers may instead perform experiments in culture with ‘immortalized’ cell lines, which are comprised of differentiated cells that can divide indefinitely. Once established, these cell lines are often relatively easier to maintain than primary cells, and thus altogether, they are ideal for longer term experiments or for comprehensive projects that include many experiments across various conditions. Immortalized cell lines originate from either tumor tissue (benign, or cancerous like the ‘infamous’ HeLa cells obtained from patient Henrietta Lacks), or primary cells that have been ‘immortalized’ by purposeful infection with Epstein-Barr Virus (EBV)¹⁰, monkey virus SV40, or other viruses that effect continuous cell division, or by artificially integrating the DNA from these viruses into the genomes of the primary cells¹¹.

Not all cell types can be immortalized in this fashion, and the infection and genome integration processes do alter cell biology and gene expression patterns of immortalized cells compared to their in vivo phenotypes^12–14. Additionally, cell lines can potentially experience problems related to over-passaging (e.g. potentially resulting in the accumulation of new mutations that might alter function) or contamination with bacteria or even other cell lines¹⁵. Although researchers need to consider this complication when planning experiments and interpreting results, the benefits of long-term division and relative ease of use have resulted in a long and productive history of the use of immortalized cell lines in functional genomic studies. Additionally, immortalized cell lines are commercially available for many primate species, including but not limited to chimpanzees, macaques, baboons, vervets, and even squirrel monkeys (e.g., from NIH Nonhuman Primate Reagent Resource, University of Massachusetts Medical School, Worcester, Massachusetts, USA or American Type Culture Collection, Manassas, VA, USA) as well as from various human populations^16,17, making for a good starting point for many evolutionary anthropology studies.

To partially obviate some of the limitations of using primary or immortalized cell lines for functional genomics studies, one may now choose to perform experiments with ‘induced pluripotent stem cells,’ or iPSCs. These cells possess the ability to differentiate into various cell types (Figure 3). As such, iPSCs have the potential to help revolutionize the study of human and non-human primate evolution, because anthropologists can use these cells to study tissue-specific biological processes otherwise difficult or impossible to interrogate from in vivo experiments with humans and other non-human primates, particularly endangered species. For example, it is theoretically possible, now or in the future, to generate orangutan neurons, Delacour’s langur chondrocytes, tarsier photoreceptor cells, or blue-eyed black lemur testicular cells. Additionally, researchers can manipulate and sample different types of cells across different cellular developmental timepoints, from embryonic to progenitor cells or even fully differentiated cell types.

Figure 3. Induced pluripotent stem cells (iPSCs).

By collecting differentiated cells such as fibroblasts or melanocytes from humans or non-human primates and exposing them to reprogramming factor proteins, scientists can now create iPSCs, which can then be subsequently differentiated into many types of cells in the body for tissue-specific experiments and functional genomic analyses. Individual images: Shutterstock and Phylopic.

iPSCs have already been generated for both humans and a few non-human primate species^18–21, and similar resources could be created for any species from which fresh blood or skin tissue samples^18,22 can be collected from at least one individual. Specifically, once a fresh sample is obtained and living, nucleated cells have been isolated, scientists can then reprogram these differentiated cells to induce pluripotency in a number of ways – with the most common and efficient method involving the activation of critical ‘pluripotency genes,’ or Yamanaka factors, typically KLF4, SOX2, MYC, and POU5F1 (also known as Oct4)²³. These genes are introduced into the cell via, for example, transfection of a viral plasmid. These genes are then expressed and translated by the machinery of the host cell, leading to cellular reprogramming and recapture of pluripotency¹⁸. Once iPSCs have been established, the stem cells can then be differentiated into various cell types via activation of different sets of genes, with the particular set depending on the cell type desired by the researcher (Figure 3). The possibilities of iPSC usage are vast, yet the generation of iPSCs and subsequent differentiated cell types can be technically challenging, requiring significant time and resources²⁴.

Although most iPSC-based research to date has focused on understanding diseases like Huntington’s, Parkinson’s, schizophrenia, and others²², the recent generation of non-human primate iPSCs from a select few species (still limited in taxonomic diversity) opens a new world of ‘cellular anthropology’ for the discovery and validation of cell type-specific changes that occurred during primate and human evolution²⁵..

2.2. Experimental Cell Culture Assays.

Because cultured cells – whether primary, immortalized, or differentiated from iPSCs – are simple systems in which all environmental conditions can be relatively controlled, researchers can use in vitro cell culture experiments to test how small components of a system function or interact under different experimental conditions. At the most basic level, cells from different individuals or species can be exposed to a molecule to observe and quantify how their responses differ. For example, cells from different human populations or primate species can be exposed to different pathogens or pathogen peptides like lipopolysaccharides (LPS)²⁶ to explore cellular immune responses. Similar experiments can be conducted with cellular exposure to endogenous hormones (e.g., testosterone, cortisol, growth hormone, oxytocin, or serotonin)²⁷ or exogenous pollutants (e.g., arsenic, lead, pesticides, or BPA)^28,29. Alternatively, environmental conditions such as temperature sensitivity or oxidative stress can be investigated. Combining in vitro experiments with transcriptomics, proteomics, and even genome editing (see below) is a powerful approach for generating and/or testing specific hypotheses related to the underlying genomic causes of phenotypic differences among human populations or non-human primate species.

2.3. Protein Reconstruction and Resurrection.

Functional genomics techniques also include investigating the functionality and evolutionary trajectory of proteins. To determine or estimate the ancestral sequence of a gene or a protein, scientists can either directly study the ancient proteins with special mass spectrometry, liquid chromatography and other methods being developed by the new field of paleoproteomics^30,31, sequence ancient DNA directly with paleogenomic approaches^32,33, or use sequence data from extant taxa to computationally infer the DNA sequence of the coding region of a gene interest that was likely present in various ancestral taxa^34–36 (Figure 4). Paleoproteomics and ancient DNA sequencing are generally limited by sample availability and preservation quality, while incomplete or uncertain phylogenetic data for extant taxa can otherwise challenge the computational inference of ancestral sequences³⁷. However, once the ancestral sequence is determined, various technological approaches can be used to study the functionality of that protein. Most simply, a researcher can synthesize the gene sequence and let the machinery of a bacterial cell produce the encoded protein. After extracting the translated protein, we can then, for example, measure receptor activation upon the binding of a ligand or quantify precursor and metabolite amounts of an enzyme to assess functionality³⁴. Alternatively, the synthesized gene sequence can be inserted into a host cell via transfection for further experimentation in that environment^34,38, or the sequence of the gene of interest can even be directly incorporated into the genome of a cell or organism via genome editing³⁴ (see below). Excitingly, anthropologists have begun to use the available genome sequences for Neandertals³² and Denisovans³³ to functionally assess distinct proteins that were present in those lineages³⁹.

Figure 4. Protein Reconstruction and Resurrection.

A) On the left, if DNA can be extracted and sequenced from ancient hominin or extinct non-human primate samples, then the amino acid sequences of ancient proteins can be inferred relatively directly based on gene coding region DNA sequences. On the right, if ancient genomes are unavailable, ancestral protein sequences can be reconstructed through computational analysis of sequences from extant species. B) The ancestral DNA sequence can then be synthesized and either inserted into an organism’s genome directly by genome editing to observe phenotypic effects or inserted into bacterial or mammalian cells in culture for experimental assays and characterization, either after extraction and purification of the protein or in the in vitro cell culture environment. Individual images: Shutterstock and Phylopic.

2.4. Genome Editing with CRISPR-Cas9.

Ideally, scientists would perform functional experiments after precisely modifying the endogenous genome of a cell or organism at only the segment of the genome or the specific nucleotide position being tested. After doing so, the direct link between genotype and cellular or organismal phenotype could be confirmed, without the confounding factor of other genomic differences between the cells of two individuals or species.

A very exciting, recently emergent functional genomics technology is CRISPR-Cas9 genome editing technology⁴⁰. CRISPR, or Clustered Regularly Interspaced Short Palindromic Repeats, when coupled with the protein Cas9 to create the CRISPR-Cas9 complex, is a highly programmable system whereby researchers can identify or design a small piece of sequence to serve as a ‘guide’ for editing a specific section of the genome. This ‘guide,’ comprised of synthesized RNA that is complementary to the gene or sequence of interest, is incorporated into the Cas9 protein that then, as a complex, binds to that researcher-chosen segment of DNA in the genome of the cell(s) being manipulated. The Cas9 enzyme cuts the DNA at that location to create a double-stranded break in the DNA, after which the cell’s own repair machinery is used to add or delete selected genetic material or to replace an existing DNA segment with a different one (Figure 5; for more details on the discovery and methodology, see refs.^41,42). CRISPR-Cas9 can even be used to edit multiple genes or genetic loci simultaneously^43,44. Directed genome editing with CRISPR-Cas9 is a relatively new technology that still suffers from challenges like the potential for off-target mutations and difficulty with delivery of the CRISPR-Cas9 system into specific desired cells or tissues⁴⁵. Thus far, CRISPR-Cas9 has been used primarily for disease modeling in human cell lines^46,47 and to create transgenic organisms in non-primate models (e.g., Drosophila⁴⁸ and mice^49,50) and species of agricultural importance (e.g., cattle⁵¹).

Figure 5. How CRISPR-Cas9 can be used to edit genomes.

Schematic representation of how CRISPR gene-editing technology operates. (A) The Cas9 enzyme is guided by a ‘guide RNA (gRNA)’ in pink, which is specific to the target DNA sequence of interest. The Cas9 enzyme then creates a double-stranded break (DSB) in the unedited DNA (in green) at the site specific to the gRNA, which is then repaired in one of two natural, DNA-repair mechanisms: non-homologous end joining (NHEJ) or homology-directed repair (HDR). In NHEJ, the ends of the DSB are joined back together without the need for a template, often resulting in the insertion or deletion of various numbers of nucleotides (in black and gray). Thus, NHEJ typically causes a frameshift mutation, knocking out the function of the gene. In contrast, HDR uses a template, or donor DNA provided by the scientist (in purple), as a guide to repairing the DSB. If the ends of the donor DNA match the sequence of DNA around the DSB, then HDR will use that donor DNA as a template, resulting in the desired insertion or edit (in purple). (B) CRISPR-Cas9 genome editing techniques can be used in cell culture or embryonic genomes. Individual images: Shutterstock.

2.5. Transgenic organisms.

Although in vitro techniques are invaluable for dissecting mechanisms of genetic action, they, of course, cannot capture the full biological complexity of an organism and how that complexity may affect the ultimate phenotype. If it is altogether technically, ethically, and financially possible, researchers can observe the phenotypic consequences of a genetic variant (or variants) of interest with an initial in vivo transgenic organism study, or they might confirm major conclusions from their in vitro experiments with such a study. In this technique, DNA sequence variants, whole genes, or multiple variants from across the genome from one species replace the orthologous (corresponding) sequences of the transgenic organism. In order to produce an adult transgenic organism, scientists must make these genomic edits to the genome of a single-celled embryo prior to insertion into the mother to develop normally.

Since CRISPR-Cas9 techniques vastly improve the feasibility of in vivo work relative to previously-available genome-editing methods (e.g., TALENS)⁴⁰, experiments with transgenic organisms, especially non-primate models like mice, are becoming less likely to be out of technical or financial reach for evolutionary anthropologists. Moreover, while one might theoretically prefer to observe in vivo phenotypic changes from genome editing in the species of interest (e.g., in the case of anthropologists, humans or non-human primates), many biological systems are highly conserved across mammalian taxa, meaning that the insertion of primate DNA into a mouse or zebrafish model may still be useful to help assess or confirm the phenotypic outcomes of human or non-human primate genomic variation. For example, genome editing of mice has already been used to screen the functional effects of sets of fixed human-chimpanzee genetic differences hypothesized to partially underlie human-specific morphological changes^52–54. ‘Humanized’ mice have also been used to investigate the molecular biology of human skin color variation⁵⁵, loss of body hair⁵⁶, susceptibility to pathogens⁵⁷, and even brain metabolism⁵² and size⁵⁸. While we expect the use of transgenic primates for in vivo evolutionary anthropology studies to remain rare due to the combination of technical difficulties and ethical concerns⁵⁹, other scientists have already started generating transgenic primates using CRISPR-Cas9^60,61, for example, to study incomplete sexual maturation in rhesus macaques⁶².

3. Functional genomics and evolutionary anthropology

Scientists have begun using the functional genomics techniques described above as parts of their explorations of fundamental anthropological questions. Yet we note that even when these tools are not applied directly, prior functional genomics studies have still provided the groundwork for major anthropological insights^63–69 via the construction and the public availability of general gene functional information databases, e.g. Gene Ontology (GO)^70,71. We first review how functional genomics has been used to study modern human adaptation to the local environment, followed by the contributions of functional genomics to our understanding of archaic human biology and the potential contribution of introgressed archaic hominin genetic material to modern human phenotypes and disease risk. Our survey also includes functional genomics studies of phenotypic differences between hominins and non-human primates, and of dietary specialization and host-pathogen co-evolution in non-human primates.

3.1. Using functional genomics to study local adaptation in modern humans

Modern humans originated in Africa more than 200,000 years ago and spread across much of the globe over the last ~60,000 years⁷². Subsequently, many populations have ‘locally adapted’ in various ways to their own particular habitats, diets, and associated immune system pressures^73,74. Furthermore, and repeatedly, it seems that either the genetic changes underlying locally adaptive phenotypes or the phenotypes themselves (e.g., light skin⁷⁵, short stature⁷⁶, high fat diet⁷⁷, or malaria resistance^78,79) can have pleiotropic or byproduct effects that increase disease risk in the same organ system or in other areas of the body (e.g., skin cancer⁸⁰, heart disease^81,82, type II diabetes^77,83, and sickle cell anemia^84–86). Functional genomics offers enormous opportunities to test hypotheses related to and explore the mechanics of these ‘evolutionary medicine’ processes.

3.1.1. Diet and diet-related human diseases.

Functional genomics can be used to explore the evolutionary basis of diseases with both diet-associated and genetic risk components, like obesity and type II diabetes. Such diseases affect millions of people worldwide today and, in many populations, have been increasing in prevalence⁸⁷.

The high rates of some of these diseases could reflect, at least in part, past histories of selection pressures on genetic variants associated with a specific diet⁷⁷ or an energetically ‘thrifty’ metabolism⁸³. For example, the ‘thrifty genotype’ hypothesis⁸⁸ is generally based on the premise that our ancestors may have faced selective pressure to store energy as fat as efficiently as possible during times of food abundance in order to survive times of food scarcity. Now, with abundances of food available to most individuals living in many present-day societies, the same genetic variants associated with ‘thrifty’ metabolism that may have previously increased in frequency due to positive selection may now contribute to increased risks for metabolic disorders like obesity and type II diabetes among the carriers of those variants⁶.

Functional genomic approaches can be used to characterize underlying molecular biological processes of candidate “thrifty” genetic variants to help test and/or refine this hypothesis. For example, Minster and colleagues⁸⁹ recently identified via a genome-wide association study (GWAS) a single nucleotide polymorphism (SNP) in the gene CREBRF that appears to impart a significant and relatively large effect on body mass index (BMI) variation in Samoans, who have one of the highest rates of obesity in the world (nearly 80% and 91% of adult men and women, respectively⁸⁹). The derived allele associated with higher BMI is surrounded by a strong signature of recent positive selection, and while this allele is present at fairly high frequency (22–28%) in Samoans, it is extremely rare in European populations (< 0.1%)⁸³.

Using genome editing and mouse adipocyte cell lines to study how this allele specifically affects cellular energy storage and use, researchers observed that adipocytes with the derived Samoan variant exhibited greater accumulation of lipids, especially triglycerides, but (perhaps counter-intuitively) lower rates of glycolysis, or energy production⁸³. In other words, these functional genomics experiments revealed that, at the cellular in vitro level, the derived Samoan CREBRF variant led to increased fat storage but less energy usage overall. Importantly, this result is consistent with the in vivo observation that this derived variant is simultaneously associated with a 1.3-fold increase in obesity risk per allele in Samoans yet also a 1.6-fold decreased risk for type 2 diabetes^83,90. This unexpected pattern was recently replicated in Māori and Pacific (Polynesian) people living in Aotearoa/New Zealand⁹¹. Given that obesity is a major risk factor for type 2 diabetes, this counter-intuitive relationship – the mechanism of which has now been partially clarified by functional genomics – is inconsistent with the second premise of the thrifty genotype hypothesis and demonstrates the potential complexity of human biology that should be considered when investigating this and other evolutionary ‘mismatch’ concepts.

3.1.2. Human health & immune function.

Pathogens have long been and still continue to be strong drivers of local adaptation for human populations worldwide⁹². In fact, in one example comparing individuals with ancestries from two different continents, 10% of genes expressed in macrophages showed population-specific gene regulation in response to infection, and such genes were significantly enriched for signals of recent positive selection⁹³. As genetic variants underlying differential disease susceptibility and resistance patterns in humans are identified, the combination of functional and evolutionary genomics can help elucidate pathogen-human evolutionary relationships and investigate specific hypotheses of anthropological interest related to local adaptation, transitions to agriculture and urbanization, and more^94–97.

A diversity of human erythrocyte variants – including those resulting in the Duffy-null blood group, sickle cell disease, the thalassemias, and glucose-6-phosphatase deficiency – each provide at least some protection against malaria^85,98. Genetic mutations underlying these phenotypes vary widely in frequencies across the globe, but collectively they are most prevalent in areas where malaria was historically endemic⁸⁵. Functional genomics approaches have helped scientists begin to understand the mechanisms by which these mutations provide protection from malaria.

For example, common erythrocyte variants like HbS (sickle cell)⁹⁹, HbC¹⁰⁰, and those that cause the thalassemias¹⁰¹ protect against the complications of severe Plasmodium falciparum malaria via abnormal presentation of parasite’s erythrocyte membrane protein-1 (PfEMP-1), the major cytoadherence ligand and virulence factor. In vitro, these parasitized blood cells exhibit reduced adherence to microvacuscular endothelial cells^99–101 and reduced rosetting, or clumping, with non-parasitized cells¹⁰⁰, both well-known complications of severe malaria. Reduced rosetting is also caused by decreased expression of cell-surface complement-receptor 1 (CR1) which interacts with PfEMP1^102,103; CR1 expression is reduced in red blood cells with either α-thalassemia or polymorphisms in CR1 in vitro¹⁰³.

Meanwhile, P. vivax malaria uses the Duffy antigen encoded by the DARC gene as a receptor for erythrocyte invasion; individuals homozygous for the Duffy-null allele do not express these antigens on the surface of erythrocytes and are generally immune to P. vivax infection¹⁰⁴. The causative mutation is in the promoter region of the DARC gene; it disrupts the binding of transcription factor GATA1 resulting in a 20-fold decrease in expression of reporter genes in vitro¹⁰⁴.

Another example is the co-evolutionary arms race between the Trypanosoma parasites that cause African sleeping sickness and the apolipoproteinL1 (APOL1) gene that helps to destroy trypanosome pathogens^57,105. Two derived APOL1 alleles appear to offer protection against the Trypanosoma subspecies T. b. rhodesiense in Africans, yet also increase the risk of experiencing kidney disease later in life, at least in African-American populations¹⁰⁶. These derived alleles occur at relatively high frequencies (mean = 10.3%, range = 0–49%) in areas of eastern and southern Africa where T. b. rhodesiense is endemic^57,105. Functional genomic experiments have revealed that mice that express the derived versions of APOL1 proteins via injection of recombinant plasmids survive over 65% longer when infected with T. brucei than mice that express the ancestral human variants⁵⁷. Further functional genomics work may help researchers understand the mechanisms leading to the concurrent increased risk for kidney disease, potentially leading to new treatments while also helping us understand better the dynamics of this evolutionary trade-off.

2.1.3. Variation in human skin color & height.

Some of the most obviously variable human traits are skin color and height, which can vary markedly among human populations from different regions of the world and affect one’s disease risk across multiple axes (e.g., skin cancer⁸⁰ and heart disease^81,82, respectively). Latitude plays a major role in skin color variation; compared to more equatorial regions (and with other factors such as altitude held constant), more northern and southern latitudes experience lower levels of solar radiation, and lighter skin allows greater absorption of otherwise limited UVB radiation than does darker skin. Although UVB radiation does cause sunburn and skin cancer, these conditions are thought to have relatively minor average effects on early-life reproductive success⁷⁵. In contrast, UVB radiation is a critical component in our skin’s ability to synthesize Vitamin D, which is a key nutrient in the development of bone, the maintenance of the innate immune system, and the normal functioning of the pancreas, brain, and heart⁷⁵ – thus, lighter skin has likely conveyed a selective advantage for individuals living in northern latitudes despite the concomitant risk of skin cancer.

Functional genomic methods have been used to confirm that depigmented skin evolved convergently, at least in part, in each of multiple different world regions rather than having evolved completely in one location and spreading elsewhere^107–109. In Southeast Asians, a nonsynonymous derived mutation in the OCA2 gene is associated with lighter skin color; this allele is present in greater than 50% of Han Chinese yet extremely rare in western Europeans and practically absent in Africans^55,110. In vitro, melanocyte cells lines homozygous for the derived OCA2 allele produce significantly less (~60%) melanin than do cells either heterozygous or homozygous for the ancestral allele. Furthermore, transgenic mice generated using CRISPR-Cas9 to edit OCA2 mimic the phenotypic variation observed in humans⁵⁵.

Elsewhere, a derived mutation in the SLC24A5 gene that contributes significantly to depigmented skin in Europeans and South Asians (but not East Asians) was initially identified via the discovery of the causal mutation for the zebrafish ‘golden’ pigmentation phenotype in the orthologous gene (slc24a5) for that species¹¹¹. In zebrafish with wildtype (non-mutant) slc24a5, these scientists were able to significantly lighten embryos coloration via knockdown of slc24a5 mRNA expression using anti-sense oligos¹¹¹. Conversely, injection of wildtype slc24a5 mRNA into zebrafish embryos with the ‘golden’ slc24a5 mutation partially restored the darker, wildtype coloration phenotype¹¹¹. In humans, a nonsynonymous SLC24A5 mutation with high levels of population differentiation (i.e., the derived allele is nearly fixed in Europeans and nearly absent in some sub-Saharan African, East Asian, and Native American populations) significantly explains skin pigmentation variation in admixed individuals after accounting for genome-wide ancestry variation¹¹¹.

Human skin color variation is also linked to genetic variation in the gene MFSD12, with the discovery of derived alleles associated with darker skin in a GWAS conducted with populations across Africa¹⁰⁹. In primary human melanocytes, these derived alleles are associated with significantly lower levels of MFSD12 expression¹⁰⁹, and in a functional genomics experiment, knocking down ~80% of the expression of the ortholog Mfsd12 in mouse melanocytes resulted in a 30–50% increase in melanin production compared to control cells¹⁰⁹. Finally, Mfsd12 null mice created using CRISPR-Cas9 had grey rather than agouti colored coats¹⁰⁹.

Variation in human height is an anthropologically interesting and medically relevant trait with extensive intra- and inter-population human variation¹¹². For example, SNPs in regulatory regions of the GDF5 gene, which encodes the growth differentiation factor 5 protein, have been linked both with variation in stature and variation in risk for osteoarthritis, such that homozygous possession of a particular derived allele at this locus correlates with ~1 cm decrease in height and a ~1.2–1.8-fold increased risk of osteoarthritis⁷⁶. Surprisingly, the alleles associated with shorter stature and a higher risk of osteoarthritis exhibit evidence of strong past positive selection in Eurasians⁷⁶. In human femoral-growth-plate chondrocytes in vitro, one of these alleles, the derived variant of a GDF5 regulatory region SNP (rs4911178), effects reduced expression of a reporter gene relative to the ancestral variant. Moreover, this allele is also associated with relatively lower expression of the GDF5 enhancer in the long bones of transgenic mouse embryos⁷⁶.

These results provide an initial functional biological basis for exploring the selective advantages that increased the frequencies of the short stature-associated GDF5 alleles in northern regions despite the commensurate increase in osteoarthritis susceptibility in later life. For example, are there additional pleiotropic phenotypes also associated with these alleles, that may have been the targets of selection rather than short stature In terms of the evolutionary tradeoff consideration, whereas osteoarthritis represents a significant health burden for billions of people across the world¹¹³, it is generally a disease of older age and thus may have been relatively easily offset in terms of purifying natural selection by any phenotype associated with earlier fitness benefits¹¹⁴, which may help explain the evolutionary history of these GDF5 alleles.

3.2. Archaic hominin biology and genetic introgression to modern humans

The availability of Neandertal³² and Denisovan³³ genome sequences provide an enormous opportunity to develop novel insights into the evolutionary biology and behavioral ecology of these hominins¹¹⁵, complementing and extending what is possible via studies of their archeological and fossil records alone. Moreover, these paleogenomic data enable us to identify introgressed genome segments in modern humans, originally acquired through interbreeding with these archaic hominins and subsequently retained¹¹⁶. For example, Neandertal ancestry comprises 0.5–2% of modern, non-African human genomes due to admixture between modern non-Africans and Neandertals, whereas up to 5% of the genome of modern-day South Asians is introgressed via admixture from Denisovans^{32,33,117,118}. The legacy of this past admixture leads to yet another, indirect opportunity to reconstruct archaic hominin biology: by considering the functional effects of these alleles in modern humans. For example, greater numbers of Neandertal ancestry alleles at various loci across the genome have been linked to increased risks for various diseases including hypercoagulation, actinic keratosis, and even depression¹¹⁹. Experiments that functionally characterize the effects of these variants, perhaps with the direct goal of improving human health, can thus simultaneously provide indirect insights into archaic hominin biology^119,120.

3.2.1. Archaic hominin biology.

Paleogenomic data combined with functional genomics can help us to reconstruct archaic hominin traits not preserved in the fossil record, including those that affect soft tissues. For example, via ancestral protein reconstruction and in vitro cell culture methods, researchers have demonstrated that at least some Neandertals likely had red hair and pale skin. Specifically, Neandertal-specific nonsynonymous mutations in exon 8 of the MC1R gene, which encodes one of the key proteins helping to regulate pigmentation in vertebrates, caused reduced cell-surface expression of MC1R and reduced G-protein coupling efficiency when transfected into immortalized fibroblast-like monkey kidney cells in vitro¹²¹. Because MC1R is a transmembrane protein that, when activated, causes melanocytes to produce black or brown eumelanin rather than red or yellow phaeomelanin, decreased MC1R function results in pale skin and red hair in modern humans¹²¹. Interestingly, the nucleotide mutations in Neandertals and modern humans are different, suggesting that MC1R-associated red hair phenotypes evolved convergently in both lineages.

Functional genomics has also been central to initial efforts aiming to infer whether archaic hominins likely had speech production capacity. For example, among a broader set of genes associated with speech development (e.g., CMIP, KIAA0319, and ROBO1¹²²) is the highly conserved transcription factor FOXP2 or forkhead box P2¹²³. A decade ago, Enard and colleagues⁵⁴ introduced human nucleotide substitutions into mouse FOXP2 in a transgenic study — the humanized mice produced qualitatively different ultrasonic vocalizations compared to wild-type mice (although see Ref.¹²⁴), and also exhibited significant differences in dopamine levels, neuronal morphology, and even exploratory and social behavior.

Since then, in vitro assays¹²⁵, cell culture experiments¹²⁶, and other transgenic work¹²⁷ have been used to further explore how interspecific FOXP2 variation may influence brain and vocal development differences between species. For example, upstream of the FOXP2 gene, there is a derived, modern human-specific (i.e., absent from the Neandertal genome) nucleotide substitution resulting in the alteration of a transcription factor binding site upstream of FOXP2. Transfection of both the modern human and Neandertal versions of this binding site and a downstream reporter gene into immortalized human HeLa cells in vitro resulted in the Neandertal version driving relatively greater transcription of the reporter gene¹²⁵. The language ability consequences of the inferred human-Neandertal difference in FOXP2 expression are yet unclear, and other derived, hominin-specific FOXP2 nonsynonymous substitutions arose prior to the archaic hominin-modern human split (further discussed below)¹²⁸. Still, this system provides an initial step forward and an opportunity for continuing functional exploration. Furthermore, as we refine our ability to differentiate iPSCs into neural cell types and even into cerebral organoids (i.e. ‘mini brains’)^25,129, evolutionary anthropologists will be able to integrate gene editing technology with these systems to identify and/or functionally confirm the phenotypic effects of genetic changes on neural development and changes in brain size, metabolism, and encephalization.

With recent advances in technology, researchers can also obtain epigenomic data from ancient samples¹³⁰. One comparison of genome-wide methylation patterns among modern humans, Neandertals, Denisovans, and chimpanzees revealed significant methylation differences between modern humans and archaic hominins. These methylation differences were enriched within a network of genes (e.g., NFIX and SOX9) involved in the spatiotemporal patterning of skeletal development and growth, particularly in genes affecting the protrusion of the lower midface and size of the nose¹³¹. In modern humans, the NFIX gene locus is hypermethylated compared to that for archaic hominins¹³¹. Across modern human and mouse tissues, increased NFIX methylation is correlated with decreased NFIX expression, and NFIX knock-out mice have vocal tract anatomy alterations that mirror the general configuration of modern humans, with a more caudal positioning of the tongue and descent of the hyoid-larynx lower into the throat – positioning that for modern humans is critical to speech production¹³¹.

3.2.2. Phenotypic effects of adaptive introgression.

Interestingly, the pattern of introgressed archaic hominin DNA segments among the genomes of modern humans is not randomly distributed. For example, frequencies of archaic-introgressed alleles are considerably higher at some loci in the genome in some modern human populations, implying a past history of positive natural selection and adaptation¹¹⁶.

Modern humans have independently colonized multiple high-altitude habitats: the Tibetan Plateau, the Andean Altiplano, and the Simien Plateau of Ethiopia, resulting in overlapping but distinct avenues of genetic change and adaptation to hypoxic high-altitude conditions in each population¹³². In modern Tibetan populations, genomic scans for signatures of positive selection identified the EPAS1 gene locus as a top candidate¹³³. EPAS1 encodes the transcription factor HIF2α, a gene in the hypoxia response pathway that stimulates the production of red blood cells and subsequently leads to increases in blood hemoglobin concentration¹³³. A GWAS confirmed that EPAS1 genetic variation is associated with hemoglobin concentration phenotypic variation, with the derived Tibetan allele associated with relatively lower hemoglobin levels¹³³. Scientists subsequently subjected endothelial cells, which naturally produce EPAS1, from Tibetans with and without the adaptive variants to prolonged hypoxic conditions in vitro, causally linking the derived allele to down-regulation of the expression of EPAS1 under hypoxic conditions¹³⁴. Furthermore, heterozygous EPAS1-knockdown mice (i.e., their expression levels of EPAS1 are 50% lower than mice homozygous for wild-type EPAS1) exhibit a blunted response to hypoxia relative to wildtype mice, mirroring the Tibetan adaptive phenotype¹³⁴. The high frequencies of these EPAS1 alleles likely represent an adaptation to high altitude, as high hemoglobin levels are otherwise associated with increased risk of cardiac events^133–135.

Amazingly, when compared to other modern and archaic hominin genomic data, the derived EPAS1 allele and surrounding haplotype is found not only in Tibetans but also in the Denisovan genome, in a pattern strongly suggesting that the adaptive Tibetan haplotype was originally introgressed from Denisovans¹³⁶. This finding is especially striking considering the recent report of evidence for a Denisovan presence on the Tibetan plateau at least 120,000 years ago¹³⁷.

In non-African modern human populations, the set of genome-wide loci with unusually high levels of archaic hominin ancestry is significantly enriched for genes related to immune function¹³⁸. In several cases, the functional consequences of these adaptively introgressed variants have been investigated. For example, there are high levels of Neandertal ancestry in modern European populations (27.8% to 43.0%) at the oligoadenylate synthetase (OAS) locus, which consists of three genes (OAS1, OAS2, and OAS3) that play critical roles in the innate immune response to pathogens^39,139. In macrophages from European individuals cultured in vitro, the percentage of Neandertal-like sites at all three OAS loci correlates with the expression of each gene in response to exposure to various viral pathogens, including the flu and Herpes Simplex 1 and 2³⁹.

3.3. Hominin evolution using functional genomics

Classic derived hominin phenotypes include bipedality, encephalization, morphology that facilitates fine motor control in the hand and wrist, and more. Yet researchers have been challenged to identify the genetic underpinnings of these derived changes, given the ~35 million single nucleotide substitutions and ~5 million insertions or deletions that are fixed between humans and our closest evolutionary relatives, chimpanzees and bonobos¹⁴⁰. In this section, we will discuss how functional genomics techniques are being used to help make advances in this area of research.

3.3.1. Hominin morphological evolution.

Using comparative genomics, scientists have identified regions of the genome that are highly conserved (implying important functions) among all studied primates excepting humans, yet have changed substantially in the hominin lineage either via an unusually high rate of nucleotide substitutions (potential positive selection) or deletion^3,140. In one such region, HACNS1, the rate of nucleotide substitution observed along the human lineage is more than a magnitude greater than that typically observed across the genome (i.e., there are 16 human-specific nucleotide substitutions in a 546-bp element that is otherwise highly conserved among all studied terrestrial vertebrates) suggesting at least some of these SNPs may have been adaptive and were fixed by positive selection¹⁴¹. As a good candidate for the genetic basis of (unknown) hominin-chimpanzee phenotypic differences, researchers separately coupled the human, chimpanzee, and macaque versions of HACNS1 to a reporter gene (in this case lacZ), which produces a visually obvious protein when transcribed, and inserted it into the genome of a mouse embryo. In contrast to the non-human primate HACNS1 versions, human HACNS1 drove substantially higher levels of gene expression in mouse limb buds across multiple in utero developmental ages¹⁴¹. The authors hypothesized that the HACNS1 substitutions could be partly responsible for the derived morphology and unique dexterity of the human wrist.

Similarly, scientists used functional genomics to characterize the phenotypic consequences of a human-specific ~60 kilobase deletion near the androgen receptor gene¹⁴². In a reporter gene experiment, the (intact) chimpanzee and mouse versions of this genomic region drove reporter gene expression in the facial vibrissae and genital tubericles of the mouse embryos and, in 60 day old mice, also drove expression in the dermis of the penile spines⁵³. Thus, this human lineage-specific deletion appears to be at least partly responsible for the lack of sensory vibrissae and penile spines in our species.

3.3.2. Pathogen resistance differences between humans and other primates.

Functional genomics is a critical tool for understanding the mechanisms behind variation in pathogen resistance among primates, such as the cellular invasion mechanism for species-specific malaria. Here, parasite invasion studies using iPSC-derived erythrocytes or hepatocytes¹⁴³ coupled with comparative genomics and transcriptomics have yielded highly relevant insights into the co-evolutionary relationships between malaria species and their human or non-human primate hosts. P. knowlesi, a zoonotic malarial species, is partially constrained from invading human cells due to differences in cell surface receptor composition caused by the gene CMAH, which is functional in chimpanzees and macaques but not in humans¹⁴⁴. Expression of the functional chimpanzee CMAH enzyme in cultured human cells resulted in comparable P. knowlesi invasion rates between human, chimpanzee, and macaque cells¹⁴⁴. Similar experiments targeting other pathogen-interacting proteins that differ substantially among humans and non-human primates could be used to explore why other diseases that are deadly to humans but not non-human primates or vice versa (e.g., P. falciparum¹⁴⁵ or rhinovirus¹⁴⁶, respectively) do not progress similarly across host species.

3.4. Functional genomics and non-human primate evolution

To date, most non-human primate functional genomics research has focused on rhesus macaques^{58,60,62,147–149} due to their status as biomedical models for human disease and biology. However, improvements in the cost and technical efficiency of functional genomics techniques plus the availability of genomic information and cell lines from a growing number of primates are making it increasingly possible for us to apply at least the in vitro tools virtually any primate species for studies of their evolutionary biology.

3.4.1. Primate dietary adaptations.

Primates exhibit an incredible range of dietary variation¹⁵⁰ and associated adaptations^151–153. For example, plant cellulose is broken down by symbiotic bacteria in the extended digestive tract of folivorous colobine monkeys through fermentation, producing large quantities of fatty acids that in turn lower the colobine small intestine pH to 6–7, compared to more typical pH ranges of 6.5–7.5¹⁵³. Fermentation simultaneously produces large quantities of bacterial RNA which must be metabolized for energy by the host’s RNASE enzyme. In most mammals, the RNASE enzyme, encoded by the gene RNASE1 encodes, operates optimally at pH 7.4–7.8¹⁵¹. Using experimental protein assays, researchers found that amino acid changes in a colobine-specific duplication of the RNASE1 gene, RNASE1B, lowered the optimal pH for enzymatic activity to 6.4–6.6, matching the colobine small intestine environment¹⁵³. Thus, in colobine monkeys, RNASE1 likely degrades double-stranded RNA throughout most of the body whereas RNASE1B degrades single-stranded RNA produced by foregut bacteria¹⁵¹.

Similar functional genomics work, specifically protein reconstruction and in vitro experimental assays, has been used to explore opsin gene spectral sensitivity variation¹⁵⁴ and the evolution of variation in color vision across the primate lineage¹⁵⁵. After identifying nonsynonymous (amino acid changing) nucleotide substitutions among the opsin genes of primates with different visual capacities, researchers have reconstructed these opsin proteins in vitro and measured their spectral tuning to determine the light wavelengths different primate species can see¹⁵⁵. By combining functional genomics with genomics information and data on activity patterns and dietary specializations, anthropologists can explore the selective pressures such as diet^156,157 or sexual selection^158–161 that led to trichromatic vision and the visual abilities of primate ancestors as well as the retention of dichromacy in nocturnal primates like tarsiers¹⁶² or aye-ayes¹⁶³.

3.4.2. Primate health & immune function.

As discussed above, understanding the evolution of primate immune systems and the co-evolutionary histories between primates and their pathogens is critical for combating zoonoses and improving our ability to prevent and treat human disease, as well as for comparative biology research and conservation purposes. In particular, some infectious agents cause disease progression in humans, yet their counterpart in non-human primates rarely produce the same disease progression — including two of the most widespread and deadly diseases in humans, HIV and malaria¹⁶⁴.

Researchers have used functional genomics to begin exploring mechanisms behind the differential HIV and SIV disease progression among primates. Infection by HIV and SIV in ‘non-natural hosts’ like humans and macaques respectively inevitably progresses to AIDS without treatment¹⁶⁵. Conversely, ‘natural hosts’ like wild vervets (genus: Chlorocebus) and sooty mangabeys (Cercocebus atys) often have high SIV viraemia, yet infected individuals apparently never progress to AIDS^165–167.

Scientists explored vervet resistance to AIDS progression by intersecting vervet genome-wide selection scan data with gene expression data from in vitro experiments of vervet and macaque immortalized cells infected with SIV¹⁶⁸. Among the set of loci identified in the selection scan, genes involved in responding to viral pathogens – and especially those with a vervet-specific SIV infection response based on the in vitro experimental data – were significantly overrepresented¹⁶⁸. This signature of potential adaptation against SIV infection suggests a longer co-evolutionary relationship between vervets and SIV compared to that between humans and HIV¹⁶⁸.

Similarly, a comparative analysis between the sooty mangabey and macaque genomes identified 34 immune system genes – many with known roles in HIV infection – with highly divergent sequences to explore for their potential roles in the differential SIV outcome between species¹⁶⁹. One strong candidate is the ICAM-2 (intercellular adhesion molecule 2) gene, which encodes a transmembrane glycoprotein involved in immune response and lymphocyte recirculation. In sooty mangabeys, this gene has a derived 499-bp deletion effecting a truncated protein; surface expression of ICAM-2 was detected in human and macaque but not in sooty mangabey immune cells¹⁶⁹. Meanwhile, TLR-4 (toll-like receptor 4), which encodes a receptor that binds bacterial proteins and activates the immune response, has a sooty mangabey-specific mutation resulting in an additional 17 amino acids at the end of the resulting protein¹⁶⁹. The authors created chimeric constructs of sooty mangabey TLR-4 without the additional 17-amino acids and wild-type macaque TLR-4 plus the additional 17-amino acids and used them in an LPS stimulation in vitro cell culture experiment to demonstrate that the protein elongation is responsible for a muted immune response to infection¹⁶⁹. As TLR-4 signaling is one of the primary mechanisms of HIV-induced chronic immune activation, this blunted infection response may, in part, underlie the non-pathogenic nature of SIV infection in sooty mangabeys.

4. Conclusion

Even with only an initial set of applied studies published to date, to us it is clear that functional genomics offers an exciting and valuable toolkit for evolutionary anthropology. Specifically, these techniques can be used to advance our understandings of the underlying genetic bases of variation in structural, functional, and certain behavioral traits among humans, hominins, and living and fossil primates. From this information we can then perhaps better estimate the timing of when these traits emerged and sometimes powerfully test adaptive evolutionary hypotheses concerning their origin and maintenance, thereby complementing and extending – for example to soft tissue and genetically-mediated behavioral traits – fossil record-based interpretations.

Biographies

Dr. Kathleen Grogan is an NIH-funded NRSA postdoctoral fellow in the Departments of Anthropology and Biology at Pennsylvania State University. Her research interests focus on the intersection of genomic variation and inter-individual differences in phenotype and fitness within the context of environmental variation.

Dr. George Perry is an Associate Professor of Anthropology and Biology and Chair of the Bioinformatics and Genomics Interdisciplinary Graduate Program at Penn State University. His anthropological genomics research group studies human evolution, evolutionary ecology, and evolutionary medicine.