The common marmoset (Callithrix jacchus) is a small New World primate; it originally comes from the Atlantic coastal forests in northeastern Brazil. It has been attracting much attention in the biomedical research field because of its size, availability, and unique biological characteristics. Its endocrinological and behavioral similarity to humans, comparative ease in handling, and high reproductive efficiency are very advantageous for neuroscience research. Recently, we developed transgenic common marmosets with germline transmission, and this technological breakthrough provides a potential paradigm shift by enabling researchers to investigate complex biological phenomena using genetically-modified non-human primates. In this review, we summarize recent progress in marmoset research, and also discuss a potential application of genome editing tools that should be useful toward the generation of knock-out/knock-in marmoset models.
In the human brain, there are two major functional domains. One has been conserved in all mammals through evolution and governs fundamental functions such as reward, emotion and memory; the other is unique to primates, and is acquired through the enlargement of the cerebral cortex governing special functions such as tool use, language, and self-awareness. Thus, to properly understand these brain functions, we need appropriate animal models for studying each function. Animal models that are used to analyze brain functions are different in each case. In the former, a reductive approach is adopted based on gene manipulation using models such as genetically-modified fish and rodents, while in the latter, the approach involves complex behavior analysis using non-human primates such as macaque monkeys. Many researchers believed that the complementary nature of genetic engineering technologies in rodent and fish models and cognitive neuroscience techniques in primate research would lead to progress in this research field. However, due to the lack of appropriate animal models that can be analyzed in both aspects of the brain's functions, contact points between these two approaches have been limited.
The development of genetically engineered non-human primates has attracted attention for its potential to connect the two research fields. Recently, we succeeded in creating the world's first transgenic primate using marmosets (Sasaki et al. 2009). This technological breakthrough provides a potential paradigm shift by enabling researchers to analyze both the brain functional domains using various model marmosets.
In this review, we summarize recent updates on marmoset research and also discuss an application of an emerging technology: genome editing tools, which researchers can use to create a generation of knock-out/knock-in marmoset models.
The common marmoset (Callithrix jacchus) is a small New World primate (Fig. 1) that is native to the Atlantic coastal forests in northeastern Brazil in South America (Abbott et al. 2003; Mansfield 2003; Carrion & Patterson 2012; Okano et al. 2012; Tokuno et al. 2012). Adult marmosets have an average height of 20–30 cm, and weigh 350–400 g on average. Common marmosets have white ear tufts and relatively long banded tails, and the fur is multicolored (brown, gray, and yellow).
In the wild, marmosets primarily eat plant exudates and insects along with fruits, seeds, flowers, snails, lizards, and infant mammals (Abbott et al. 2003). Like humans, but unlike many other nonhuman primates, marmosets live in stable extended families of about 10 members (Tardif et al. 2003), and have well-developed vocal communication (Pistorio et al. 2006). A marmoset family usually consists of 1–2 breeding females, a breeding male, their offspring, and their adult relatives. Most marmosets are monogamous (Mansfield 2003). Female marmosets commonly give birth to two non-identical twins per delivery, and are ready to breed again around 10 days after giving birth. Because female marmosets need to nurse infant marmosets during the following pregnancy, the male partner and other members of a group care for infants together.
The common marmoset is an ideal non-human primate model for biomedical research
The common marmoset has attracted considerable attention as a potentially useful animal model in fields such as neuroscience, stem cell research, drug toxicology, immunity and autoimmune diseases, reproductive biology and regenerative medicine, because of its size, availability, and unique biological characteristics, such as a strong relationships between parents and offspring, and a vocal form of social communication (Eliades & Wang 2008; Dell'Mour et al. 2009; Gordon & Rogers 2010). New World primates (platyrrhines) including the common marmoset diverged from Old World primates (catarrhines) occurred approximately 35 million years ago, resulting in the New World species developing adaptations to the neotropical environment along with other distinct differences in physiology and disease susceptibility (Abbott et al. 2003; Mansfield 2003). Although macaques are evolutionally closer to humans, some of the traits of the marmosets are more similar to humans than macaques due to geometrical separation (Mansfield 2003). Therefore, for researchers in various fields, the common marmoset has become attractive as an animal model for biomedical research:
In aspects of endocrinology, marmosets are more similar to humans than rodents. For pharmacokinetic and toxicological screening, new pharmaceuticals should be evaluated in a relevant animal species in which the test material is pharmacologically active, because the mechanism of action is dependent on specific receptors or epitopes. In Europe, the marmoset has been used as a non-rodent second species in drug safety assessment and pharmaceutical toxicology; the marmoset has the closer phylogenetic relationship to humans than other second species, such as the dog (Smith et al. 2001). Marmoset cells exhibit cross-reactivity with human cytokines or hormones (t'Hart et al. 2003, 2012; Kitamura et al. 2011). The common marmoset has a disease susceptibility profile that is similar to humans (Neubert et al. 2002; Carrion & Patterson 2012), thus making it a suitable model system for toxicology screening, drug development and infectious disease research. The necessity of an ideal animal model that mimics human metabolism as closely as possible became important after the thalidomide tragedy (Poswillo et al. 1972; Merker et al. 1988; Neubert et al. 1988, 1999). The teratogenic effects of experimental compounds have been shown to differ significantly between rodent and primate species. Thalidomide administration during early gestation results in specific and dramatic limb defects in primates, whereas the mouse and rat are not sensitive to the teratologic side effects (Scott et al. 1977). It is thought that the species difference is attributable to difference in materno-fetal transfer of thalidomide in the placenta between primates and rodents (Schmahl et al. 1996; Beghin et al. 2010). Indeed, the marmoset has been used to comprehensively investigate the mechanism of thalidomide teratogenesis (Poswillo et al. 1972; Merker et al. 1988; Neubert et al. 1988, 1999). Those studies suggest that metabolites of thalidomide may cause downregulation of surface adhesion receptors, thereby altering cell-to-cell and cell-to-extracellular matrix interactions in the developing limb bud (Neubert et al. 1996). Furthermore, common marmosets have been used extensively in infectious disease research due to their unique susceptibility to a number of important human infective agents. Marmosets have been used in models of hepatitis A virus infection (Pinto et al. 2002), hemorrhagic fever from Junin virus (Weissenbacher et al. 1979), malaria (Mitchell et al. 1988), measles (Kobune et al. 1996), and prion diseases (Baker et al. 1993).
Common marmosets can be handled with comparative ease. The choice of a laboratory animal for biomedical research is based on a consideration of animal welfare and practicality as well as scientific suitability. The marmoset has advantages over the macaques in terms of animal welfare and practicality. They are available for laboratory use from well-established captive colonies in national primate research centers, academic institutions, and commercial breeding facilities. Unlike macaques, marmosets do not carry herpes b virus (Macacine herpesvirus 1) (Mansfield 2003), which is beneficial for their handlers. Their small relative size can also translate into lower caging and feeding costs and reduced floor space requirement when compared to the needs of macaque species. As a result, common marmosets generally cost less to obtain and house than macaques, resulting in substantial cost saving when performing equivalent experiments (Smith et al. 2001). Because of these advantages, the common marmoset has been used as an affordable and versatile model system of spinal cord injury for a wide range of preclinical testing (Iwanami et al. 2005; Kitamura et al. 2011; Kobayashi et al. 2012).
Breeding of marmosets in laboratories is relatively easy and effective. The common marmoset ovarian cycle lasts approximately 28 days, with ovulation occurring around day 10, showing similarities to the human ovarian cycle in terms of sex hormone profiles (Summers et al. 1985). Common marmosets become sexually mature with an early onset of puberty (around 1.5 years old) (Abbott & Hearn 1978). As common marmosets are monogamous (unlike mice and macaques), male and female marmosets who get along with each other are mated in a cage. The gestation period of marmosets is relatively short, 145–148 days. A female marmoset usually gives birth to two to three offspring per delivery, and can deliver twice a year. Therefore, they have a relatively large number of deliveries (20–30 during the course of their lifespan) and offspring (40–80 during their lifespan). This reproductive advantage is obvious when comparing other non-human primates. For example, it takes about 5 years for macaques to become sexually mature, and they deliver offspring per year (Austad & Fischer 2011). Compared to a macaque, for example, from which one could obtain three pups in 3 years, one female marmoset could reasonably expect to provide 14 marmoset pups over the same period (Okano et al. 2012). Such remarkable reproductive efficiency in marmosets provides a strong advantage in terms of the development of transgenic animals.
Important basic research tools for marmosets have been developed. It is critical for an animal model system to establish various fundamental research tools. First, the genome information of the common marmoset has not been completed yet. Currently, the draft sequence assembled by the Washington University St. Louis (WUSTL) School of Medicine Genome Sequencing Center in St. Louis, Missouri, USA and the Baylor College of Medicine (BCM) Human Genome Sequencing Center in Houston, Texas, USA is available through NCBI Genbank (https://www.hgsc.bcm.edu/non-human-primates/marmoset-genome-project). The common marmoset genome was sequenced to 6X coverage using DNA from a female marmoset provided by the Southwestern National Primate Research Center in San Antonio, Texas, USA, assuming 95% coverage of the whole genome. In addition to the marmoset genome information based on the marmoset used in the USA, our group also sequenced the genome of a marmoset that came from the colony at the Central Institute for Experimental Animals (CIEA) in Kawasaki, Japan, to 30× coverage, and are now assembling the sequence data to establish the genome database for Japanese colony of marmosets (unpublished data).
Second, non-invasive imaging techniques are also important for marmoset research. Conventional histological techniques, such as sectioning of tissues, require the destruction of specimens of sacrificed animals. As a result, it is impossible to monitor anatomical changes of the same animal over long periods by conventional methods. In addition, for primate research, the number of animals available for research is another issue, because of cost and ethical issues. Magnetic resonance imaging (MRI) is a non-invasive imaging technique to visualize internal structures of the body in detail, and is used for medical diagnosis. Various MRI techniques for common marmoset have been developed, including diffusion tensor tractography (DTT) for in vivo tracing of axonal fibers of the spinal cord, visual system and fetal brain (Fujiyoshi et al. 2007; Yamada et al. 2008; Konomi et al. 2012; Hikishima et al. 2013), and voxel based morphometric (VBM) analysis (Hikishima et al. 2011). DTT is a non-invasive imaging technique that can track neuronal fiber pathways of the brain. This technique visualizes neuronal fiber tracts by detecting water diffusion in the brain (Ito et al. 2002; Masutani et al. 2003). Usually, water diffuses in all directions. When there is a barrier, such as a neuronal fiber, water will diffuse unevenly. The direction of fastest diffusion corresponds to the length of the neuronal fiber pathways. Conventional MRI techniques show the white matter of the brain as a homogeneous tissue, even though it actually contains a complex array of specifically oriented nerve fibers. In our previous study, we succeeded in visualizing both intact and surgically disrupted spinal long tracts in the adult common marmoset by DTT (Fujiyoshi et al. 2007), demonstrating the utility of this imaging tool for evaluating axonal conditions in the injured spinal cord. VBM analysis is also an advanced MRI technique applied to cognitive brain research in both human subjects and non-human primates (Ashburner & Friston 2000; Hikishima et al. 2011). VBM is a neuroimaging technique that allows investigation of focal differences in brain anatomy, using the statistical approach of statistical parametric mapping. Compared to traditional morphometry, in which volume of the brain region is measured by drawing regions of interest (ROIs) on images, VBM is more accurate, and can detect small structural differences. VBM can be used to analyze structural MR images to identify differences in brain anatomy between control and experimental groups, or to detect longitudinal changes within groups. Recently, we created a population-averaged brain template for the common marmoset for precise VBM analysis (Hikishima et al. 2011). This system will facilitate the use of marmosets in MRI studies, and potentially enable the detection of subtle anatomical differences associated with transgenic manipulation or physiological changes.
Third, marmoset embryonic stem (ES) cells and induced pluripotent stem (iPS) cells are available for both in vitro and in vivo research (Sasaki et al. 2005; Tomioka et al. 2010). Although marmosets are good model animals for cognitive science, they are not necessarily advantageous for investigation of molecular and cellular mechanisms due to the limited number of marmosets available for research. Thus, use of ES cells and iPS cells is a complementary approach for marmoset research, because these cells can differentiate into all cell types of the body. There are several important applications of these cells, including: (i) studies of early embryonic development and specification of the three germ layers of marmosets; (ii) preclinical studies of regenerative medicine; and (iii) potential tools for transgenic techniques. Our collaborative group has generated both marmoset ES and iPS cells for these purposes (Sasaki et al. 2005; Tomioka et al. 2010). However, unlike mouse ES cells, those pluripotent stem cells cannot be used for generation of knock-out animals. The growth factor requirements, gene expression profiles and morphology of marmoset ES and iPS colonies are more like human ES cells (which are Leukemia Inhibitory Factor [LIF] independent) than mouse ES cells (which are LIF dependent) (Hanna et al. 2010). Mouse ES cells retain the capacity to contribute to somatic cell lineages upon injection into blastocysts (termed as “naïve state”), whereas mouse epiblast stem cells derived from the post-implantation epiblast of the murine embryo, do not (De Los Angeles et al. 2012). Because marmoset ES and iPS cells lack the naïve state, these cells cannot be directly used for the development of knock-in/knock-out marmosets. However, recent studies have dramatically revealed the molecular mechanisms underlying the naïve state of pluripotent cells (Wray et al. 2010). If the current primate pluripotent stem cells can be converted into naïve ES/iPS cells in future, it will be possible to generate knock-out/knock-in marmosets using the same strategy as mice.
Transgenic techniques and neurological disease models
The creation of genetically-modified mice, such as transgenic and knock-out mice, was one of the major contributions to the field of life science research, because those model mice are very useful to clarify the genetic functions and molecular mechanisms underlying various pathological conditions. However, in some cases, results obtained in model mice cannot be directly applied to humans because of the many physiological, anatomical and histological differences between mice and humans (Smith et al. 2001; Mansfield 2003; Eslamboli 2005; Abbott & Bird 2009). So, there is a demand for more research in primates, which more closely resemble humans in terms of physiological function and anatomy. Until recently, efforts to generate transgenic primate animals were unsuccessful, and scientific evidence of the expression of introduced genes in the somatic tissues of any derived transgenic primate was lacking. In 2008, Yang et al. reported progress in developing a transgenic model of Huntington's disease in a rhesus macaque that expresses polyglutamine-expanded the human huntingtin (HTT) gene (Yang et al. 2008). They injected mature rhesus oocytes with high titre lentiviruses expressing exon 1 of the human HTT gene with 84 CAG repeats and lentiviruses expressing the green fluorescent protein (GFP) gene, into the perivitelline space. After fertilization by intracytoplasmic sperm injection, the injected eggs were developed to the 4- to 8-cell stage and transferred to surrogate mothers. As a result, five newborns carrying the transgenic mutant HTT and GFP gene were delivered. These monkeys display difficulty in movement coordination and involuntary movement, such as chorea and dystonia, with HTT aggregates in the brain. Although the transgene was inserted into the genome of those founder monkeys, the germline transmission of the transgene has not been confirmed yet. Without germline transmission of a transgene, it is difficult to establish and maintain a stable transgenic primate line. Actually, of the five founders, one transgenic monkey did not show any clinical symptoms, and the other four founders also showed different degrees of movement dysfunctions, showing inconsistent phenotypes among the founder transgenic primates.
At the same period, our group independently attempted to generate transgenic primates using common marmosets. As marmosets have high reproductive efficiency and these characteristics become very advantageous for creation of transgenic primates, we decided to use marmosets for transgenic modification. We introduced the enhanced GFP (EGFP) gene into marmoset embryos, using a self-inactivating lentiviral vector (Sasaki et al. 2009). A high titre of lentiviral vectors carrying a foreign gene encoding EGFP (5.6 × 109 to 5.6 × 1011 transducing units per mL) were injected at the earliest embryonic stage possible, into either in natural or in vitro fertilization (IVF) embryos. Because the perivitelline space of the marmoset early embryo is relatively small, the embryos were first placed in 0.25 mol/L sucrose medium, which made the perivitelline space expand 1.2–7.5 fold (Fig. 2). The injected embryos were cultured in vitro for a few days, and only those fertilized embryos that expressed EGFP were selected for implantation into surrogate mothers. This selection process by EGFP was very reliable and accurate, and of five embryos implanted, exogenous EGFP expression was confirmed in various infant tissues and placenta in all the five founders. In addition, upon reaching maturity, sperm and oocytes from two of the five transgenic animals were examined and the foreign gene expression in the germ cells was confirmed. Furthermore, in vitro fertilization of wild-type marmoset oocytes with transgenic sperm resulted in healthy offspring that expressed EGFP, establishing stable transgenic lines of marmosets expressing EGFP. This is the first successful generation of transgenic non-human primates with germline transmission.
The successful generation of transgenic marmosets enables researchers to use a better model system to study life science using genetic models of primates, which are more similar to humans than rodents. For example, Alzheimer's disease is the most common neurodegenerative disease and is characterized by progressive loss of memory and other cognitive functions (Chin 2011). Neuropathological hallmarks of the disease are amyloid plaques and neurofibrillary tangles, which are accompanied by neuronal and synaptic loss (Walsh & Selkoe 2004; Tanzi & Bertram 2005). Over the last two decades, a number of mouse models have been created for understanding and developing therapies for Alzheimer's disease (Chin 2011). A number of transgenic mice have been created that overexpress amyloid precursor protein (APP) and/or presenilin containing one or more mutations linked to familial Alzheimer's disease (Chin 2011). Although these mice recapitulate many of the key features of Alzheimer's disease, neurofibrillary tangles are notably absent (Games et al. 1995; Hsiao et al. 1996; Sturchler-Pierrat et al. 1997). One of the possible explanations why those mice do not display the neurofibrillary tangles is that the lifespan of mice is too short to detect the phenotype. The greatest known risk factor for Alzheimer's disease is increasing age, and the majority of people with Alzheimer's disease are 65 and older (Tanzi & Bertram 2005). Even though genetic risks are present in patients, it takes several decades for the onset of the disease. As a result, the mouse models overexpressing mutant APP can form amyloid plaques, but would die before forming neurofibrillary tangles. If transgenic marmosets overexpressing mutant APP are generated, the model primates may display the similar long-term clinical course to human Alzheimer's disease, due to its long lifespan (approximately 15 years) (Austad & Fischer 2011).
Strategy for generation of knock-out marmosets with genome editing technology
The success of generation of transgenic marmosets enables a new era of primate research. However, the shortcomings of current transgenic technologies limit our ability to genetically modify primates. The current transgenic technique can only insert exogenous genes into the genome in a random manner (Sasaki et al. 2009). In addition, due to the usage of lentiviruses as vectors for gene delivery, the size of an inserted gene is limited up to approximately 8 kb. Thus, the current transgenic technique can produce only marmoset models of diseases that are caused by overexpression of a relatively small mutant gene, such as Parkinson's disease. However, because most human genetic diseases are attributable to either point mutations or deletions of endogenous genes, new gene modification technology against endogenous genes is necessary for broad applications of genetic marmoset models to disease research.
Gene targeting is a genetic technique that uses homologous recombination to replace an endogenous gene (Mansour et al. 1988). To target genes in mice, a targeting construct that contains a mutation in a gene of interest is introduced into mouse ES cells. After a selection process, mutated ES cells are used to contribute to part of a mouse tissue by injecting them into the blastocyst. After chimeric mice, from which reproductive cells are derived from the mutated ES cells, are selected, knock-out/knock-in mice are finally obtained via breeding. While this replacement strategy based on homologous recombination can be applied to marmosets ES cells (Shiozawa et al. 2011), these recombinant marmoset ES cells cannot contribute to development of whole organisms at this moment. As mentioned above, the existing lines of marmoset ES cells are not in the naïve state and cannot contribute to germline cells of chimeric animals (Tachibana et al. 2012).
To overcome this bottleneck for generation of knock-out/knock-in marmosets, researchers are examining the potential of genome editing tools, namely engineered nucleases (Carroll 2011; Gaj et al. 2013; Joung & Sander 2013). These artificially engineered nucleases are used for induction of specific double-stranded breaks at desired locations in the genome. Double-stranded breaks initiate the cell's endogenous mechanisms to repair the breaks by either homology-directed repair (HDR) or non-homologous end-joining (NHEJ). Through these repair processes, mutagenesis against endogenous genes can be inserted. Importantly, as HDR and NHEJ are essentially functional in all organisms, genome editing with engineered nucleases can apply to all model animals. Currently, zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and clustered regularly interspaced short palindromic repeat (CRISPR)/Cas system are mainly used as engineered nucleases. ZFNs are artificial restriction enzymes that consist of a specific DNA-binding domain, which is made of tandem zinc finger-binding motifs that are fused to a non-specific cleavage domain of the restriction endonuclease FokI (Carroll 2011). TALENs are chimeric protein generated by the fusion of a transcription activator-like effector (TALE) DNA binding domain to an endonuclease FokI (Cermak et al. 2011). Very recently, the CRISPR/Cas system was reported as a new genome editing tool (Cong et al. 2013; Mali et al. 2013). This system makes use of the RNA-based adaptive immune system in bacteria and archaea, and the simple base-pair complementarities between the engineered single-guide RNA and a target genomic DNA sequence guides the Cas9 nuclease. Using those engineered nucleases, a number of genetically-modified animals have been generated successfully (Geurts et al. 2009; Mashimo et al. 2010; Ochiai et al. 2010; Hauschild et al. 2011; Bedell et al. 2012; Sung et al. 2013; Suzuki et al. 2013; Wang et al. 2013; Yang et al. 2013).
Using the advantage of engineered nuclease technology, we are now proceeding to generate marmoset models of neurodevelopmental disorders, including Rett syndrome (Kishi & Macklis 2005; Chahrour & Zoghbi 2007) and tuberous sclerosis complex (Ess 2010). Although mice models are available for these diseases, they do not necessarily mimic the critical symptoms. For example, although male Mecp2 −/y mice are primarily used as “Rett syndrome model” due to the robust phenotypes (Chen et al. 2001; Guy et al. 2001), most patients with Rett syndrome are females who are heterozygous for the MECP2 mutation (Chahrour & Zoghbi 2007). Additionally, Mecp2 −/y mice can survive until adulthood, whereas MECP2 mutations in human males are thought to lead to embryonic lethal, or severe neonatal encephalopathy and death (Villard 2007). This substantial difference between the mouse models and the human patients suggests that a new model animal that follows as close to the clinical course of the diseases as possible is necessary for better understanding of the pathogenesis and future treatments for neurodevelopmental disorders. To this end, we have generated engineered nucleases against the causative genes, and are now injecting these nucleases into fertilized marmoset eggs (Fig. 3). By transplanting the injected embryos into surrogate mothers, genetically-modified marmoset models of neurodevelopmental disorders are supposed to be born in the near future.
The successful creation of transgenic marmosets offered a new animal model for the study of human diseases, and more and more researchers have been attracted to this new research field. However, compared to mouse genetic models, the current gene manipulation techniques of marmosets are very limited, and new manipulation methods against endogenous genes need to be developed. Emergence of genome editing tools will facilitate and accelerate the development of new gene manipulation technology of marmoset models. We believe that genetically-modified primates will offer new insights into higher cognition research, and have the potential to contribute significantly to new therapeutic strategies for addressing currently intractable neurodegenerative diseases such as Parkinson's disease and amyotrophic lateral sclerosis, as well as mental disorders such as schizophrenia and autism spectrum disorders.
This work was supported by grants from the Strategic Research Program for Brain Sciences by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan, and from the Funding Program for World-leading Innovative R&D on Science and Technology (FIRST) by the Cabinet Office of Japan, to H.O. and E.S., and the MEXT Grant-in-Aid for Young Scientists to N.K. This study was also partly supported by PRESTO of the Japan Science and Technology Agency to E.S. We thank Ms. Reona Kobayashi for her illustration of a common marmoset.