Nonhuman Primate Models in the Genomic Era: A Paradigm Shift

Abstract

Because of their strong similarities to humans across physiologic, developmental, behavioral, immunologic, and genetic levels, nonhuman primates are essential models for a wide spectrum of biomedical research. But unlike other animal models, nonhuman primates possess substantial outbred genetic variation, reducing statistical power and potentially confounding interpretation of results in research studies. Although unknown genetic variation is a hindrance in studies that allocate animals randomly, taking genetic variation into account in study design affords an opportunity to transform the way that nonhuman primates are used in biomedical research. New understandings of how the function of individual genes in rhesus macaques mimics that seen in humans are greatly advancing the rhesus macaques utility as research models, but epistatic interaction, epigenetic regulatory mechanisms, and the intricacies of gene networks limit model development. We are now entering a new era of nonhuman primate research, brought on by the proliferation and rapid expansion of genomic data. Already the cost of a rhesus macaque genome is dwarfed by its purchase and husbandry costs, and complete genomic datasets will inevitably encompass each rhesus macaque used in biomedical research. Advancing this outcome is paramount. It represents an opportunity to transform the way animals are assigned and used in biomedical research and to develop new models of human disease. The genetic and genomic revolution brings with it a paradigm shift for nonhuman primates and new mandates on how nonhuman primates are used in biomedical research.

Introduction

Animal models are constantly changing; new technologies emerge, new understandings are reached, new goals arise. As a research community, we have a scientific and ethical obligation to ensure that our animal models are rigorous, appropriate, and necessary. This is particularly important in nonhuman primate research.

Model organisms are valuable because they can be controlled and manipulated in ways humans cannot; it is easier to ensure that animals share similar ages, rearing, diets, or other environmental concerns that may confound studies. In most animal models, this control extends to include genetics, through directed breeding, congenic strains, or, more recently, transgenics. Nonhuman primates have largely not seen control on this level, however. The generation times of nonhuman primates are long, and litter sizes are small, making directed breeding difficult if not impossible. This limitation extends more directly into research as well. Difficulties in animal procurement, coupled with increased ethical considerations and concomitantly higher husbandry costs, lead to necessarily smaller sample sizes.

Because of this, a greater understanding of nonhuman primate genetics is required. Small sample sizes mean confounding genetic influences impart greater variance, and tighter control over environmental variables means that genetic effect sizes are proportionally greater. For nonhuman primate studies to be most meaningful, it is necessary to take into account as much genetic information as possible, particularly functional genetic information, in the earliest stages of study design. It also means, however, that nonhuman primates offer a unique opportunity to exploit naturally occurring genetic variation and to model genotype/phenotype associations in ways that are not possible in humans or in inbred animal models. The acceleration of genomic information acquisition, particularly in rhesus macaques (Macaca mulatta), now offers an opportunity to advance and enhance the utility of the model broadly across a wide spectrum of biomedical research.

Fixed Species Differences

Species evolve through the accumulation of genetic mutations, either by chance or selection, ultimately leading to anatomic and morphologic diversity and to the more subtle physiologic and behavioral dissimilarities. When used in biomedical research, the assumption is that those species-specific differences either do not matter to our study or can be adequately controlled. Nonhuman primates are the most closely related animal models of human disease and show the fewest differences. These similarities lead to practical and ethical concerns, however. Ultimately, if it is necessary and worthwhile to understand a disease or develop a new treatment, then it should only be undertaken in a system in which those answers can be adequately resolved. If the model system selected cannot and does not answer the questions asked, particularly if they are meant to be translational, then it is not appropriate. It is in this latter case where fixed species differences are most important.

Species-Specific Adaptation

A great deal of study has gone into what sets humans apart from our relatives in the animal kingdom. This includes the very specific, such as those traits that differentiate humans and chimpanzees, as well as the more general, such as the common derived traits shared among primates. These are the factors that we must be most concerned about when we are attempting to develop translational models of human disease in animal models and where nonhuman primates are likely to be the most important and informative (Ergorul and Levin 2013; Wendler and Wehling 2010).

Broadly, species-specific adaptation can be divided into two classes, those traits that are changing across all species and those that are unique to the specific species (Vallender and Lahn 2004). The first category includes, among others, those genes and pathways modulating immune response (Hughes and Yeager 1997). By their antagonistic nature, these systems are under strong and constant selective pressures, the “Red Queen hypothesis” (Van Valen 1974). As species emerge, the selective pressures they face also change, and this rapidly results in large differences between species. Immune systems and their specific restriction factors can vary greatly between species, illustrated by the fact that pathogens often do not (easily) cross species barriers (Sawyer and Elde 2012). Studying the immune system in model organisms implicitly requires a consideration of these differences.

Other changes are unique to individual taxa. These can occur as a result of genetic drift but are more often conceived of as being driven by selective adaptation to a novel environmental niche. Among primates generally, one usually points to the increasing importance of vision as a sensory modality and the decreasing relevance of olfaction (Heesy and Hall 2010; Williams et al. 2010); these macroevolutionary changes are reflected in widespread loss of olfactory receptors (Glusman et al. 2001; Sharon et al. 1999) and the emergence of trichromatic vision through gene duplication (Boissinot et al. 1998; Dulai et al. 1999) [but note (Matsui et al. 2010)]. In humans there are numerous examples of species-specific adaptations (Bradley 2008). Some, such as a loss of body hair, changes in sweat glands, and even skeletal changes associated with bipedalism, have drawn the interest of anthropologists but are not generally considered in the context of biomedical research. Others, most notably those associated with the brain and cognition (Sherwood et al. 2008), are intimately associated with many translational studies.

Regardless of evolutionary origin, species-specific differences must be considered when developing and interpreting animal models. If the function of a protein or expression of a gene has changed, then this, at least, must be recognized when explaining results. In the extreme case, the differences between species are so great as to invalidate the findings altogether. In these latter situations, either the model system should be changed or the question under study modified so as to be appropriate. However, even in situations less extreme, differences between species, ultimately arising at the genetic level, must be considered for informed research approaches.

Relevance to Research Models: Immunology

The most extreme example of the impact of evolutionary change on the immune system is the species specificity of numerous pathogens. The evolutionary association between host and pathogen is a close one, and the mutations associated with susceptibility or control vary greatly between even closely related taxa. For this reason, translational research on many human pathogens is only possible in closely related species. Although the examples of infectious diseases that can be studied only in primates are numerous, the most notable examples are human hepatitis C virus (HCV) and human immunodeficiency virus (HIV).

HCV has drawn attention primarily because it only affects humans and chimpanzees. It has been notoriously intractable both in cell culture and in other animal model systems, and, given the ethical and logistical difficulties of chimpanzee research, the search for alternatives to this animal model has been extensive (Bukh 2012; Sandmann and Ploss 2013). Although it is possible to investigate HCV either through comparison with genetically similar viruses or through humanized rodent systems, the translational relevance is certainly less direct and the results more potentially oblique. “Humanizing” necessarily focuses on specific aspects of the model—in this case viral entry—but it is unclear if this is wholly or solely responsible for the variability in response. In HCV research, the concerns with chimpanzee studies are significant enough that these approaches are necessary even despite their limitations. To the credit of the community, however, researchers in this field are necessarily very cognizant of the species differences with relation to HCV, and as work progresses, it will most certainly be viewed through this prism.

The importance of the rhesus macaque model in HIV research is well known. The first studies demonstrating that acquired immune deficiency syndrome (AIDS) resulted from a virus came from macaque models (Daniel et al. 1984), and today vaccine studies are conducted almost exclusively in rhesus macaques (Desrosiers 1995). Various simian immunodeficiency viruses (SIVs) are endemic in assorted primate species, but it is primarily in rhesus macaques of Indian origin that they ultimately manifest in an AIDS-like state. Other primate species, as well as more distantly related taxa with HIV-like viral relatives, control infection through host factors (Hofmann et al. 1999). Although much can be learned through the comparative studies of how other species control this virus, animal models for translational research still rely on primates. As in HCV studies, the evolution of species susceptibility has resulted in a situation in which nonhuman primates uniquely model a human disease.

Recently a high-profile article discussed species-specific differences as they related to inflammatory diseases and how this has affected translational research (Seok et al. 2013). This work showed that humans and mice responded very differently to inflammatory stress and concluded that these differences have led to the widespread failure to produce meaningful therapeutics. Although specifics and generalizability can be debated, the core concept suggests that it is a lack of a shared molecular etiology between the species that, despite similar overt responses, has been problematic. Of course, this needs not be read as a caution against animal research in general or mouse research in particular, but rather as a reminder of the need to be more precise in our animal models of disease. And it highlights a common problem. Unlike in the HCV community where the species-specific differences and their implications are so profound as to make themselves obvious to all, the effects on inflammatory response are subtle and at the molecular level and thus have largely gone unnoticed and unremarked upon.

Relevance to Research Models: Neuroscience

In much the same way, neuroscience research has felt the effects of an overreliance on rodent models (Nestler and Hyman 2010), and again the lack of a common molecular etiology for shared phenotypes may ultimately be a major confounding factor. Unlike in the immune system where differences between species can be obvious and profound, in behavioral phenotypes differences are more likely to be subtle and not immediately evident.

Some neuropsychiatric disorders, including schizophrenia and autism, have been suggested to be human specific (Bradshaw and Sheppard 2000; Dumontheil et al. 2008; Teffer and Semendeferi 2012). That is, changes in the brain during human evolution have uniquely created opportunities for the dysfunction that manifests as these diseases. This can be difficult to prove or disprove because human symptomology can be extremely difficult to translate into animal models, but it serves as a useful launching point for a broader conception of the role of evolution in neuropsychiatric disease and its effect on the translational relevance of our animal models. Although the causes, genetic or otherwise, of schizophrenia are far from understood, a significant handful of genes have been associated with the disease. Interestingly, these genes have also been shown to be under positive selection in humans, suggesting functional changes differentiating them from their orthologues in other species (Crespi et al. 2007).

The human brain shows increased cortical folding and an enlarged neocortex (particularly the prefrontal cortex) compared with other species, an elaboration on a trajectory of changes seemingly common in primates (Passingham 1973; Teffer and Semendeferi 2012). This change in the brain is necessarily the result of changes at the genetic level, and these genes have been under increasing scrutiny (Sherwood and Duka 2012; Somel et al. 2013; Vallender 2012). How these or other differences affect neuropsychiatric disorders remains unclear, but that they do should not be a surprise, particularly for neurodevelopmental disorders, and this has important implications for model development. Animal models of human neuropsychiatric disease are needed that take these genetic differences into account, minimizing their effects or placing them in proper context. Nonhuman primate models, by their evolutionary similarities to humans, can ameliorate these differences somewhat. Nonhuman primates show neuroanatomic and behavioral similarities to humans that other species do not, and although this does not, in and of itself, imply translational validity, it improves chances of success.

Species-Specific Functional Variation

Genetic variation provides for much of the phenotypic variation between individuals. This is strongly recognized in humans where we are perceptive to even the most subtle of differences between us. It is also true in outbred animal species, even where we cannot or do not fully appreciate the differences between individuals. In fact, owing to a relatively recent population bottleneck, human genetic variation is less than what we observe in many species, including those that dominate our nonhuman primate models (Yuan et al. 2012).

In humans, nearly all genetic variation is species specific; that is, it is found only in humans and not in even our closest extant relatives. Theoretically, cases may arise where variation is identical by state but not by descent, with two distinct evolutionary origins, but that is exceedingly rare (Leffler et al. 2013). There are very few trans-species polymorphisms in humans and chimpanzees (Hodgkinson and Eyre-Walker 2010); those identified are associated with immune response (Cagliani et al. 2008; Cagliani et al. 2010; Cagliani et al. 2012) and, most notably, the major histocompatibility complex (MHC; also known as the human leukoctye antigen). In other genera (the Macaca and Papio genera are conspicuous in this regard), lines may be blurred between genus, species, and subspecies, and greater levels of trans-species polymorphism may be found [Macaca: (Higashino et al. 2012; Li et al. 2009; Satkoski et al. 2013; Street et al. 2007), Papio: (Charpentier et al. 2012; Keller et al. 2010; Zinner et al. 2013)].

For the greater part, species have their own constellation of allelic variation. This can manifest in ways that do not matter to researchers, such as eye or hair color, or in ways that merely inconvenience, such as aggressive personalities. Of particular interest, however, is when species genetic variation impacts research, creating subpopulations that are appropriate, or inappropriate, for use as models.

Impact of Variation on Animal Models

One of the major strengths of rodent models is their inbreeding. Strains are genetically identical, allowing researchers one fewer variable in their studies. In humans and in outbred animal models, genetic variation is always present in study populations, reducing statistical power and potentially confounding interpretation of results when testing discrete hypotheses. But although unknown genetic variation is a hindrance in research studies in which animals are allocated randomly, recognized genetic variation can offer new opportunities through targeted allocation of animals to research studies.

In inbred rodent strains, genetic variation can only be studied through introduced variation, most commonly by transgenics. This, of course, is an immensely powerful tool for understanding the specific role of the locus under investigation, but it requires an a priori hypothesis or focus on a candidate gene. Recombinant inbred lines have been useful in larger genome-wide unbiased approaches to interpreting the genetic basis of phenotypes (Williams et al. 2001). Notably, the large Collaborative Cross (Churchill et al. 2004; Threadgill and Churchill 2012) and related Mouse Diversity Outbred population (Churchill et al. 2012; Svenson et al. 2012) have been developed to use admixture between eight parental strains as a tool for facilitating quantitative trait loci analyses. Necessarily, however, these populations harbor only a subset of variation derived from the parental strains, trading comprehensiveness for practicality and statistical power.

In nonhuman primates or other outbred populations, naturally occurring variation is more extensive. Where polymorphism is known to affect phenotypes, taking this into account can significantly improve study design and can give much greater statistical power to the variables that are being tested. Genetic variation in outbred populations can also serve as fodder for linkage or association studies. Although genome-wide linkage or association studies can be difficult to perform in nonhuman primate models because of the numbers of animals needed, they are not impossible, and new tools are constantly emerging (Barr 2009). These approaches aim to either focus on phenotypes for which collecting large sample sizes are increasingly possible (Gray et al. 2011; O'Sullivan et al. 2013; Szabo et al. 2013) or to improve the methodologies with which pedigrees can be analyzed (Fears et al. 2009; Jasinska et al. 2012). Control over environmental factors can significantly increase the effect size of genetic variation, and large pedigrees can support more powerful linkage studies.

Historically, most large-scale studies of variation in nonhuman primates, and indeed in most outbred populations, have focused on neutral variation as a means for assessing geographic origins, relationship status, and inbreeding or for modeling historic demographics (Kimura 1968). Studies of functional genetic variation have been fewer and farther between and generally have emerged on an ad hoc basis in response to specific needs of the animal model. In rhesus macaques, as in most nonhuman primates, the primary driving factor for many years was understanding variability in immune response.

Relevance to Research Models: MHC

As mentioned, one of the most important uses of rhesus macaques is as an animal model of HIV/AIDS. Over decades of using this model, researchers have come to understand the importance that rhesus macaque genetic variation plays in their studies. Early on, species differences in response to the virus were recognized, leading to the development of the macaque model (Chakrabarti et al. 1987), but it was not long before scientists also noticed a differential response in rhesus macaques of different geographical origin; Indian-origin rhesus macaques respond more similarly to humans than their Chinese-origin brethren (Ling et al. 2002; Trichel et al. 2002). This followed the recognition that Indian-origin rhesus macaques with certain alleles at the MHC locus, notably A*01 (Miller et al. 1991), B*08 (Loffredo et al. 2007), and B*17 (Mothe et al. 2002), were more resistant to SIV infection.

Understanding this genetic variation certainly helps scientists to understand the mechanism of infection of HIV and the body's natural response, but it also has another important role. In the quest to eradicate AIDS, many new therapeutics are being developed to prevent HIV infection or to contain and control its spread. Whether these therapeutics take the form of vaccines, antivirals, or other prophylactics, the initial studies of efficacy often are performed in rhesus macaque models. It is essential that in both control groups and experimental groups natural genetic variation, which affects the same dependent variables, is taken into account a priori. Moreover, because of the costs associated with the research, it is often beneficial to exclude “naturally resistant” animals from these studies. Indeed, this is why Chinese-origin rhesus are not widely used and why increasingly investigators on SIV projects request animals without specific MHC alleles or host restriction factor genotypes. More recently, it has been shown that polymorphism at the rhesus macaque TRIM5 locus can differentially affect SIV replication (Newman et al. 2006), and it is increasingly more and more common that TRIM5 genotyping is performed before assignment of animals to studies to either balance representation or eliminate particular genotypes from research studies a priori. Going forward, the importance of host restriction factor polymorphism ensures that the SIV research community will continue to be early adopters.

In many ways, therefore, SIV investigators were precocious in arriving at the importance of genetic variation in their animal models for interpreting their work. What was nonexistent is now commonplace. Candidate genes such as the MHC were known, and the large number of studies and large number of animals facilitated the identification of phenotypic differences that could then be studied more in depth (Bontrop and Watkins 2005; Sauermann 2001). This reverse genetics approach has yielded important information and insight into the significance and strength of understanding functional genetic variation in the context of phenotype and illustrates just how essential a priori genetic characterization is to the interpretation and validity of the research model. It illustrates the need to prioritize systematic phenotyping along with advancing genomics as a new modus operandi.

Relevance to Research Models: Cytochrome P450 Proteins

Cytochrome P450 proteins (CYPs) are xenobiotic metabolism enzymes that catalyze more than 75% of drug transformation events (Guengerich and Rendic 2010). In humans, genetic variation at the P450s, in particular CYP2C9, CYP2C19 and CYP2D6, is associated with differential drug metabolism, efficacy, and toxicity (Johansson and Ingelman-Sundberg 2011). To highlight the breadth of diseases and treatments this polymorphism effects, it is useful to consider associations between CYP2C19 and efficacy of antimalarials (Kaneko et al. 1997), CYP2D6 and opiate (Kirchheiner and Brockmoller 2005) and ecstasy (Carmo et al. 2006) toxicity, and CYP2B6 with tolerance of highly active antiretroviral therapy treatment in HIV (Haas et al. 2005; Rotger et al. 2005). Overall, the genetic effects of the cytochrome P450 genes and other xenobiotic metabolism enzymes are so pronounced as to drive drug development and represent some of the strongest and most far-reaching pharmacogenetic effects yet observed (Sim et al. 2013).

Macaques, particularly the cynomolgus macaque (Macaca fascicularis), are commonly used in drug metabolism studies to understand dosing and toxicity effects before clinical trials (Boelsterli 2003). Indeed, the recent publication of the cynomolgus macaque genome was completed by private industry to facilitate its role in this capacity (Ebeling et al. 2011). There have been significant efforts to identify the full complement of CYPs in nonhuman primates as well as to determine their organization and expression (Ebeling et al. 2011; Uno, Iwasaki et al. 2011). Pharmaceutical development is benefited by identifying what differences exist in drug metabolism between primate species; indeed, it is these very differences that have driven the use of nonhuman primates in coordination with other animal models, and there is a great potential to advance their utility through resolving and modeling genetic influence on drug metabolism (Emoto et al. 2013; Nishimuta et al. 2011; Yoda et al. 2012).

Allelic variation of specific cytochrome P450 genes in cynomolgus monkeys macaques (Macaca fascicularis) has been identified that affects drug metabolism, including CYP1A2 (Uno and Osada 2011), CYP1B1 (Uno et al. 2011), CYP1D1 (Uno et al. 2011), CYP2C7 (Uehara et al. 2012), CYP3A4 (Uno et al. 2010), and CYP3A5 (Uno et al. 2010) variations, has been identified. By expanding our understanding of polymorphism in non-human primates at the CYPs, we both develop better models of human pharmacogenetics associated with drug metabolism and ensure that ongoing studies of drug safety and risk are fully informed and not influenced by false findings resulting from cryptic genetic variability. A cynomolgus macaque model, for instance, that metabolizes a drug in ways or at rates that a human would not is not helpful for the drug dosing and toxicity studies for which it is meant. This cryptic functional variation, therefore, could be responsible for inappropriately removing efficacious and safe drugs from the development pipeline or, worse, inappropriately moving potentially harmful drugs forward.

Parallel Functional Variation across Species

Although genetic variation may be distinct, not identical by descent or state, it can still result in similar functional effects. In humans, this can manifest as the genetic heterogeneity underlying diseases. Cystic fibrosis, for instance, is caused by dysfunction in a single ion channel, but numerous different causative mutations have been identified in the gene (Kerem et al. 1989), and similar shared effects of allelic heterogeneity have been identified in the serotonin transporter (Sutcliffe et al. 2005). This understanding can be broadened and expanded across species as well. Just as two variants in humans can ultimately lead to similar effects, distinct variation in different species has the potential to show similar results.

As previously mentioned, naturally occurring genetic variation among rhesus macaques bred in captivity is extensive compared with either human or rodent models. High levels of genetic diversity are driven in part by demographic factors, including large effective population sizes, but, as generalists, the range and breadth of ecologic niches in which macaques find themselves may also contribute to a broad functional genetic profile (Richard et al. 1989; Suomi 2006). This allows for unique opportunities for exploitation of naturally occurring variation. Unlike transgenics, where variation is artificially introduced, existing nonhuman primate genetic variation can be used to model human variation and take reverse genetic approaches.

The functional commonality between variation in rhesus macaques and in their human orthologues can extend from the level of the gene or protein to observable phenotypes, physiologic responsivity, and pharmacogenetic response. By comparing the functionality and phenotypic association of particular alleles in rhesus macaques to humans, comparable yet distinct similarities are identifiable, enabling the selection of cohorts of rhesus macaques specially and naturally suited for modeling the genetic contribution of candidate genes to phenotypic or disease variance, increasing translational relevance of the animal model and deepening our understanding of disease.

Functional Similarity from Genotype to Phenotype

When we speak of functional similarity, there are several different moieties to consider. The most basic level, the nucleotide or amino acid, is at the same time the simplest and most difficult to define. As mentioned previously, true trans-specific polymorphisms are rare. We don't expect, nor do we observe, exactly the same mutations in humans and nonhuman primates. Rather we look to variation that is in the same functional domain for similarity. The difficulties arise, of course, because defining functional domains is not necessarily a simple task and defining specific functional positions is more complicated still. Not all variation within a given domain is meaningful, but it can be functionally similar. To understand functional similarity, it is necessary to look to molecular studies: Do polymorphisms change ligand binding or macromolecular interactions? Protein folding or localization? Temporal, spatial, or intensity of expression patterns? All of these factors can be accomplished through multiple means, allowing distinct variation to share a functional effect.

These molecular effects can then be carried further, to the organismal level, where parallel associations can be observed. Ultimately mutations of similar functional effect should have the same phenotypic results. Although imprecise, one manifestation of this is to see similar genetic association or linkage. This understanding has long, if implicitly, underlain naturally occurring animal models of disease. When prostate cancer is studied in dogs or retinal degeneration is studied in cats, we do so with the belief that the underlying genetic causes are similar, but this is a forward genetics approach. By first identifying shared functional variation between humans and nonhuman primates, we have the ability to take a reverse genetics approach, identifying animals for study a priori, minimizing our animal usage and maximally ensuring our power to detect effects.

Relevance to Research Models: 5HTTLPR, MAOA, CRH, and NPY

In both humans and rhesus macaques, the gene that encodes the serotonin transporter, SLC6A4, harbors a degenerate repeat polymorphic sequence in a promoter region containing positive regulatory elements and forming complex secondary structures (Lesch et al. 1997). Two alleles of a tandemly repeated sequence, referred to as “short” and “long,” were first identified in humans, and reporter assays in cell culture showed that the long variant had greater basal transcriptio-nal activity and greater cyclic AMP- and protein kinase C-mediated induction of gene expression (Heils et al. 1996). The long allele also showed a greater ability to suppress SV40 enhancer/promoter-driven luciferase expression, with serial deletions of repetitive elements resulting in a progressively greater impairment of promoter activity. Three additional alleles (18-, 19-, and 20-repeat) were subsequently identified (Kunugi et al. 1997; Michaelovsky et al. 1999), and short and long alleles were further differentiated by more thorough sequence analysis, revealing at least 14 different isoforms occurring at low frequencies and with ethnic stratification (Nakamura et al. 2000).

Polymorphism in this promoter region was identified in Old World and New World monkeys but was absent in prosimians and other mammals (Lesch et al. 1997). Paralleling human observations, a similar repeat polymorphism was identified in rhesus macaques, as well as in baboons, orangutans, gorillas, and chimpanzees (Lesch et al. 1997). Further analyses identified polymorphic alleles in Barbary and Tibetan macaques (Wendland, Lesch et al. 2006). Old World monkeys, macaques, and baboons harbor two allelic variants with 23 or 24 repeat elements, differing in evolutionary origin from the variation seen in humans and other apes. Functionally, however, promoter activity of the macaque long variant was greater than that of the short form and showed a similar magnitude of difference in transcriptional activity in vitro compared with that in humans (Bennett et al. 2002).

A similar situation was found in the promoter region of monoamine oxidase A (MAOALPR). A functional repeat polymorphism consists of 3, 3.5, 4, or 5 copies of a 30 bp motif in humans, and luciferase reporter assays performed in multiple cell lines demonstrated that constructs containing 3.5 or 4 repeats were expressed at higher levels than constructs containing three or five repeats (Sabol et al. 1998). The alleles containing 3.5 or 4 repeats appear to act as upstream activators of transcription, promoting transcription 2 to 10 times more efficiently than alleles with three or five copies of the repeat. A repeat polymorphism was also found in rhesus macaques, gorillas, orangutans, and Gelada baboons. However, unlike in humans, these repeat motifs were18 bp in length (Wendland et al. 2006). Rhesus macaques in particular are highly polymorphic, with alleles containing five, six, and seven repeats. The activity of both the five- and six-repeat variants were approximately 26% higher than that of the seven-repeat allele in SH-SY5Y cells, which constitutively express MAOA in vitro (Newman et al. 2005).

In humans, the promoter region of the gene encoding corticotropin-releasing hormone (CRH) is complex but with three major haplotypes predominating (Baerwald et al. 1996; Baerwald et al. 1999). In luciferase reporter assays, these promoter variants modulate CRH gene expression in a cell line–dependent manner (Wagner et al. 2006). The three haplotypes also showed varying expression levels after stimulation by cyclic AMP and differential responses to dexamethasone challenge. Five common CRH haplotypes have been identified and studied in rhesus macaques, one of which harbors a single nucleotide polymorphism (SNP) in a consensus glucocorticoid response element abolishing its function (Barr et al. 2008). Similar to the functional effect observed in humans, in rhesus macaques this haplotype does not show a decrease in reporter activity after treatment by cortisol, whereas the other four haplotypes do. In both species, haplotypes that result in attenuated sensitivity to corticosteroids in reduction of transcription of CRH appear to exist.

The neuropeptide Y gene (NPY), which encodes a neuromodulator implicated in numerous physiologic processes (Benarroch 2009), also harbors parallel functional variation in humans and rhesus macaques. A human SNP upstream of the transcribed region has been reported to influence NPY gene expression in vitro (Zhou et al. 2008) and has been associated with peptide levels in anterior cingulate cortex in human ex vivo (Sommer et al. 2010). Copy number variation (CNV) in NPY that associates with functional magnetic resonance imaging measuring response to reward and emotion processing has also been reported (Lesch et al. 2011). In rhesus macaques, an SNP was identified in the region orthologous to the human variant, a conserved portion of a NPY repressor, that disrupts a Sox5 transcription factor binding site and a preferred glucocorticoid response element (Lindell et al. 2010). Ex vivo studies in the macaque found associations between the SNP and NPY expression levels in the amygdala as well as cerebrospinal fluid levels of NPY after stress.

Relevance to Research Models: Stress

Collectively, 5HTTLPR, MAOALPR, CRH, and NPY are key mediators of the stress response. The functionality of genetic variation identified in these genes in rhesus macaques parallels the functionality of homologous polymorphisms identified in humans that, in turn, associate with phenotypic aspects of the stress response. Stress has negative consequences on human health and is a risk factor for common somatic diseases, including cardiovascular disease, metabolic syndromes, obesity, gastrointestinal symptoms, atopic diseases, and pain (Bjorntorp and Rosmond 2000; Chrousos 2009; de Kloet et al. 2005; Holsboer 2001; Parker et al. 2001), as well as psychiatric diseases, including depression, anxiety, sleep disorders, and cognitive dysfunction (Chrousos and Kino 2005; Pariante and Lightman 2008). The stress response is a complicated system, mediated by the hypothalamic-pituitary-adrenal (HPA) axis and showing a marked interindividual variability that is shaped by both genes and environment (Kudielka et al. 2009; Zimmermann and Stansbury 2004). In humans, polymorphisms in genes that are involved in HPA axis regulation have been associated with a spectrum of stress-related disorders. These parallel functional polymorphisms found in rhesus macaques offer opportunities to develop better models of stress-associated diseases and to better understand their underlying causes and treatments.

Generally, an increase in the number of associated polymorphisms considered in an analysis renders a greater proportion of explained variance, but epistatic interactions, epigenetic regulatory mechanisms that enhance or silence allelic functionality, and the intricacies of gene–gene interaction networks that control the stress response complicate and confound candidate gene analyses. Notably, there are candidate genes implicated in the stress response that have been studied in both humans and rhesus macaques, such as tryptophan hydroxylase 2 (TPH2), where understanding similarities in variation between species is more elusive (Chen et al. 2006). On a gene network level, however, functional parallelism extends beyond individual gene function. For example, TPH2 encodes the rate-limiting synthesis enzyme for serotonin production, whereas SLC6A4 encodes the serotonin transporter, which regulates extracellular serotonin. It may be the case that underproduction of serotonin and high activity of the serotonin transporter each could result in a similar outcome of low extracellular serotonin. In this regard, genomic information on rhesus macaques will translate into a far greater understanding of functional parallelism, improving the utility, validity, and translational efficacy of the rhesus model for neuropsychiatric research. More generally, a comprehensive systems biology approach offers a significant advance in our understanding and use of nonhuman primate animal models.

Stress exposure throughout development and particularly during formative stages of development shapes individual differences in stress reactivity later in life. Physiologically, this is manifested as an altered regulation of the HPA axis. Early life stress, for example, can result in hyperactivity of central CRH neurons later in life and a sensitization of the HPA axis as well as autonomic responses to stressors (Heim et al. 2008). But, just as with genetic factors, the correlation of early life stress to later life HPA axis function is elusive; not all individuals who experienced early life stress show aberrant HPA axis function as adults. However, the interaction between environment and genetics does explain a large proportion of interindividual variability in the stress response, and it is here that nonhuman primate models can make perhaps their greatest contribution. Rhesus macaque models of stress-related disorders that incorporate genetic variants that functionally mimic those implicated in the stress response and associated diseases not only are more translational than those that do not, but they have also been useful for investigating how environmental histories affect genetic predispositions and therapeutic interventions. The ability to control the environmental and developmental histories of rhesus macaques affords opportunities for translational models of gene–environment interactions relevant to the stress response that are far less reliable or unachievable in humans.

Functional Genetics of Nonhuman Primates Moving Forward

Research into nonhuman primate genetics has existed nearly as long as human genetics. Indeed, our inferential understanding of the importance of genetic similarities and differences between the species predated even our formalized understandings of genetics or evolution. And yet, for as far as we have come, our progress will be dwarfed by the advances of the next decade. Not only is information being generated at a staggering pace, but new doors are opening that have been hitherto unconceived. This, coupled by logistic realities, is leading to a paradigm shift in the ways in which nonhuman primates are being used in biomedical research, moving from the genetics studies described above to an expansive genomics context.

New Areas of Focus for Variation

Epigenetic influences, as in the case of stress above, can also be interrogated in much the same way as genetic effects. These changes in methylation, acetylation, and the like are more easily observed in nonhuman primates than humans and a greater control over environmental influences than is possible with humans means that their proximate causes are more easily identifiable as well. Candidate gene studies examining the interplay between gene and environment, especially in early childhood, are becoming increasingly sophisticated (Kinnally et al. 2010; Lindell et al. 2012), and molecular work in this vein has begun already and is poised to expand (Provencal et al. 2012). It is likely still too soon to say with any certainty how interspecific differences affect epigenetics, although tantalizing clues are being uncovered (Farcas et al. 2009). Although the mechanisms are likely to be highly similar throughout mammals at least, it will be important to understand how similarly, or differently, epigenetic changes express themselves between species.

In humans, personalized medicine is growing in importance, and yet there are few, if any, useful animal models, providing another area where better understandings of nonhuman primate genetic variation may be useful. One reason why pharmacogenetics can be such a perplexing field is because humans are so different from one another not only (or necessarily) genetically but also in terms of historic environmental variability. Genetic effect sizes are small relative to environmental variability associated with taking medications, much less the histories of the individuals themselves. Further, pharmacogenetics research will forever be complicated by the simple fact that it is being studied in the context of trying to solve health issues, hardly the ideal milieu for a scientific experiment. Yet in nonhuman primates, these confounding factors can be minimized and the relative size of genetic effects significantly expanded. Perhaps more important, they can be used to identify pharmacogenetic effects before drugs go to market rather than solely retrospectively.

Although much of the focus to date has been on SNPs or fairly simple short repeats, other kinds of variation are also amenable to this type of analysis. Meaningful parallel variation may be most simple and direct to observe in proteins, but variation in regulatory regions is likely to be more plentiful and perhaps more important for understanding complex disease. This extends to variation in microRNAs, the role of which, so far, has only been hinted at (Hu et al. 2011; Somel et al. 2011). Perhaps the most promising and interesting future avenue along these lines, however, is in CNVs. CNVs are regions of the genome that can vary in presence, absence, or count. Although the most well-known are associated with Mendelian disease (Vissers and Stankiewicz 2012), many appear to be neutral or to have subtle effects that are not fully understood (Freeman et al. 2006; Girirajan et al. 2012). Because of the unique mechanisms by which CNVs are created, it may be that identical, by state if not by descent, variation exists. There is already some evidence that many human CNVs can be observed in nonhuman primates (Gazave et al. 2011; Gokcumen et al. 2011; Lee et al. 2008), although it will still take considerable effort to separate the wheat from the chaff.

Advances in Practice

Of course, the major shift in the field is the transition from genetics to genomics. Genomic technologies already have made new approaches viable; the publication of non-human primate genomes and the relative inexpensiveness of sequencing are facilitating candidate gene and other genetic analyses. But soon, extremely soon, we will be discussing not simply the genotypes of an animal at a handful of candidate genes but rather the entirety of the animals genome. The number of nonhuman primates in research is small, and their importance is great. With genome prices decreasing dramatically (already the cost of a rhesus macaque genome is dwarfed by its purchase and husbandry costs), more and more genetic information is an inevitability. As with all fields, handling this deluge will be a source of both difficulty and opportunity and must be anticipated and planned for now.

Along the same lines, this genomic revolution has reverberations beyond the macaque. As more and more species are better genetically characterized, their suitability for specific kinds of biomedical research can be better appreciated. This is perhaps most notable for those species already in use but for which genetics is still nascent. Both the baboon and vervet monkey are already well along this path, but it will be of particular interest to observe the effect the genomic revolution has on New World monkeys in research. Many of the logistical difficulties in macaques and other Old World monkeys are lessened in the New World primates, and a better understanding of their functional genetics may guide scientists toward areas in which these species are more appropriate. Indeed, simply understanding interspecific differences will help us better understand which species are more likely to give translational results and where nonhuman primate studies are necessary and where they are not.

The most immediate and important change will occur in the existing use of rhesus macaques. Just as understanding of the MHC has led to directed selection of animals for SIV studies, this greater understanding of functional genetic variation will help dictate how animals can best be used across all studies. By understanding an animal's unique genetic compliment, it can be better designated for targeted assignment; genetically unique animals can be garnered, and an informed process can ensure that the animals are assigned to appropriate research trajectories. An increased awareness of genetics can also support individual studies. In the broadest sense, this can mean ensuring a balance of genotypes or relations between arms of an experimental trial, but it can also mean designing studies with an eye toward genotype that maximizes the ability to detect genetic associations. This ensures that the ethical mandate of nonhuman primate research is best achieved. Although it is unlikely that selected breeding on a large scale will be practical or even desirable, a priori genome sequencing will maximize study power, reduce the overall number of animals necessary for research, and ensure that unnecessary or uninformative use is minimized.

Conclusions

The nonhuman primates that are used in biomedical research are unique, but their genetic variation, rather than making them less useful as animal models, affords an opportunity to make them more useful. As we understand the functional genetic variation in nonhuman primates, we improve their utility as models for biomedical research. We refine them. We are able to more efficiently and appropriate use animals in studies. We reduce the total numbers that we use. Most important, the science that results is more meaningful. It is more likely to share a common molecular etiology with human disease, and it is more likely to be translational. We use animal models because they tell us information that human studies and in vitro studies cannot. Their use is warranted only if this is the case, and we have an ethical obligation to ensure that their use is held to the highest standards, practices, and principles of our current knowledge.

No longer can rhesus macaques be designated into research projects based solely on availability. Although many of our current examples derive from candidate gene studies or otherwise more targeted genetic work, the proliferation and rapid expansion of genomic data availability will inevitably encompass each animal used in academic and industry research. This represents a broad opportunity to transform the way animals are assigned and used in biomedical research. The genetic and genomic revolution brings with it a paradigm shift for nonhuman primates and, intrinsically, new mandates on how nonhuman primate use in biomedical science progresses into the future.