Human embryonic stem cells as a model for reproductive toxicity screening
hESCs for male reproductive toxicity screening
Human embryonic stem cells, infertility, toxicity screening, embryoid bodies, in vitro model
Male reproductive toxicity examines harmful effects of various agents on all aspects and developmental stages of the reproductive system, including germ cell development and spermatogenesis. In developing a model for reproductive toxicity screening it is important to define the developmental stage that this model is going to recreate in vitro and to identify critical molecular targets of this stage. In this review we focus our discussion on potential for using embryonic stem cells (hESC) as a model for male reproductive toxicity screening. The rationale for developing novel toxicity models is that despite significant advances in our biological understanding and clinical treatments of infertility, many unresolved cases still remain. This is likely due to our lack of knowledge about environmental influences on the critical gamete developmental stages. Many practical and ethical difficulties are associated with the collection of human tissue samples to explore the unknown causes of infertility. A readily available in vitro model that mimics the human gamete development would be an extremely valuable research tool for novel toxicity assays. ESCs exhibit high degree of similarity with primordial germ cells (PGCs) at the level of gene expression and molecular signaling. In addition, recent evidence shows that ESCs in culture can be differentiated to PGCs and spermatids. Multiple lines of evidence point to the differences between mouse and human ESCs (hESCs). In light of these data, we present the case that hESCs are better suited in vitro toxicity screening models than their mouse counterparts. We then, describe some of the most promising hESC-based systems that are used today to model certain aspects of male gamete development and that have a potential to be used for toxicity screening in the near future. We conclude by discussing potential of these existing models in toxicology studies and possibilities for their improvement in the future.
Exposure to chemicals and drugs, physical or biological insults is unavoidable part of modern human's life style and can contribute to increasing rate of infertility, a medical problem of widespread proportions that affects one in seven couples. The primary causes of infertility are poor gamete quality, failure of fertilization, impaired development of the pre-implantation embryo and failed implantation. Although assisted reproductive technologies, such as in vitro fertilization and intracytplasmic sperm injection overcome some of these problems, many infertility cases remain unresolved and are classified as “infertility of unknown etiology” (MIller RK, 2007; Society, 2005). Growing evidence suggests that environmental pollutants affect the development of gametes at different stages (Clementi et al., 2008; Hauser and Sokol, 2008; Swan, 2006). Several environmental toxicants such as fungicides, may affect normal germ cell development and function leading to infertility (Anway et al., 2005). Indeed, animal studies showed that exposure to fungicides and vinclozolin in early gestation increased the incidence of male infertility in offspring (Anway et al., 2005). Furthermore, human exposure to lead and cadmium affects the spermatogenesis in terms of sperm count, motility and morphology (Telisman et al., 2000). Exposure to endocrine disrupting chemicals may interfere with development of early germ cells and increase the chances of infertility in adulthood (Skakkebaek et al., 2001). For example, phthalates increased the apoptosis in cultured human fetal testicular cells (Habert et al., 2009). Sulfasalazine, a drug commonly used for the treatment of inflammatory bowel diseases, causes male infertility with abnormal spermatogenesis (Toovey et al., 1981). In addition, there is increasing data in support of effects of aberrant primordial germ cell and gonocyte development as a cause of testicular cancers (Rajpert-De Meyts et al., 2003a; Rajpert-De Meyts et al., 2003b).
Animal experiments have provided evidence of adverse effects of environmental toxicants on the development of male gametes. In addition, mutagenicity testing and toxicity screening were conducted in multiple well-established laboratory animal-based systems (Scholz et al., 1999; Vogel, 1993). However, animal testing remains problematic due to numerous species-specific differences complexity and higher costs associated with animal experiments. In addition, some of the genetic and epigenetic changes induced during the development of gametes by environmental insults are passed through generations. Hence, it is crucial to explore the molecular mechanisms of germ cell specification and gamete development, and the factors that might disrupt these processes in vitro. To avoid the use of animals for toxicity screening, alternative in vitro cellular models using primary cultures or permanent cell lines have been developed (Scholz et al., 1999; Spielmann et al., 1998). Cell lines established from specialized somatic cells are not identical with germ stem cells and do not recapitulate their differentiation processes. Therefore, pluripotent mouse ESCs (mESC) were introduced to test in vitro mutagenic and cytotoxic effects of xenobiotics (Rohwedel et al., 2001; Scholz et al., 1999; Seiler et al., 2004). Pluripotency, unlimited proliferation, clonogenicity and ability to differentiate into germ cells are only some of the characteristics of ESC which make them serious candidates for replacement of more complex animal experiments.
Potential harmful effects of compounds on replication and differentiation of mESCs into germ cells can be screened using the established mESC test (EST). Initially, this test was based on 3 end points: 1) effects on ESC growth, 2) effects on ESC differentiation into three germ layers and 3) the ability of mESC to differentiate specifically into cardiomyocytes (Genschow et al., 2000; Scholz et al., 1999). Introduction of additional end points such as gene expression by real time PCR, assessment of mESC differentiation into multiple cell lineages including germ cells and application of more sophisticated statistical models to classify the toxic potential of chemicals has much improved the sensitivity of this test (zur Nieden et al., 2004). As a result, EST was expected to be a potential method for large scale throughput screening of toxicity based on multiple end points that will mirror growth and differentiation processes essential for evaluation of toxic potential of an agent.
After the successful derivation of hESC, the question arose whether hESC are better suited than mESC in evaluating toxicity for humans. The adoption of hESC-based systems has been slow due to two main reasons: a) requirement of highly skilled staff and complex culture systems for their successful continuous cultivation and b) ethical and legal issues due to the use of human embryos for derivation. Recent development of defined culturing systems and novel method of hESC derivation from biopsied blastomeres while preserving the embryo diminish these concerns (Chung et al., 2008; Ilic et al., 2009).
Human genome sequencing data, development of gene and protein arrays, new generation RNA sequencing, miRNA RT-PCR arrays, and proteomic approaches have opened a new era in toxicity screening by creating a molecular framework for assessment of the well-being of human. However, the selection of the most appropriate model for reproductive toxicity screening remains to be established.
To establish hESC as a better suited model than mESC for reproductive toxicity screening in humans, it is important to address the following questions: 1) are mESC substantially different from hESC, and 2) can hESC and their derivatives model specific stages of gamete development.
mESC vs. hESC
Transcriptomes of human and mouse ESCs contain both evolutionary conserved and divergent functional modules (McCarroll et al., 2004; Stuart et al., 2003). The expression pattern of core biological network that is fundamental to ESC phenotype is a conserved module and shows positive cross-species correlation. In contrast, the expression pattern which is indicative of species-specific differences in regulation of ESCs is a divergent module and shows negative or no correlation (Sun et al., 2007). Expression of several genes and signaling pathways that regulate self-renewal and survival of ESC are distinct or exclusively important only in one of the two species (Rao, 2004). For example, leukemia inhibiting factor signaling is critical for ESC self-renewal in mice but it does not seem to be critical or even required for hESC (Ginis et al., 2004; Rao, 2004; Sun et al., 2007). Similarly, the FGF pathway, which functions differently in the renewal of ESCs in human and mouse, is significantly up-regulated in mESCs relative to hESCs (Ginis et al., 2004; Rao, 2004; Sun et al., 2007). In contrast, cytokine secretion and hormone biosynthesis are up-regulated in hESCs compared to mESCs (Ginis et al., 2004; Rao, 2004; Sun et al., 2007).
Moreover, hESC and mESC mRNA profiles indicated low overall concordance rate (e.g. in one comparison was around 40%) compared to that seen in human-to-human cell comparisons (90% between hESC samples) suggesting a high possibility of additional differences (Rao, 2004). Approximately 25% of all genes considered as expressed genes in human cells were with unknown function (Brandenberger et al., 2004) indicating that there are additional, still uncharacterized, genes that are differentially expressed in human ESC whose function may turn out to be important but divergent.
Micro RNAs (miRNAs) are non-translated short RNAs that bind to complementary mRNAs and regulate the expression of genes. A total of 579 and 721 miRNA have been identified in mouse and human genome, respectively (http://microrna.sanger.ac.uk, ; Wang et al., 2009). Although extensive comparative data is still lacking, miRNA profiling of mESC and hESC showed both common and distinct features. Genome-wide expression profile revealed that although 13 mESC lines, 21 mouse EB, and 6 mouse somatic tissues expressed 425 miRNA with 14 specific to mESC, most of human miRNA are not expressed at all (Chen et al., 2007). Distinct miRNA profiles were also observed in different ESC lines (Chen et al., 2007; Houbaviy et al., 2003; Landgraf et al., 2007; Laurent et al., 2008; Mineno et al., 2006; Morin et al., 2008; Suh et al., 2004; Wang et al., 2009). hESC miRNA expression signature of 31 miRNAs has been identified so far (Landgraf et al., 2007; Morin et al., 2008; Suh et al., 2004; Wang et al., 2009).
Taken together, these findings propose the existence of important differences between mouse and human ESC. Due to the differences in expression of miRNAs and their direct effect on expression of genes regulating key signaling pathways in mouse and human ESC, we expect that these differences may affect end points used for toxicity assessment. The more sophisticated the end points incorporated into toxicity assays, the more apparent differences between mouse and human ESCs may arise.
hESC and their derivatives as in vitro models for reproductive toxicity screening
Knowledge of the molecular mechanisms that govern the development of germ cells and gametes is crucial for the successful toxicity screening and identification of the factors that may alter these processes. With the use of current assisted reproductive technologies which bypass many natural barriers to prevent the fertilization of abnormal sperm, some of the genetic and epigenetic alterations may be transferred through male germ line to subsequent generations. Hence, it is critical to understand the regulation of germ cell development.
ESCs vs. PGCs
ESCs and PGCs share a number of common features. ESCs are defined as pluripotent stem cell lines derived from early embryos before the formation of the tissue germ layers (Smith, 2006). They are derived usually from the inner cell mass cells of the pre-implantation blastocyst (Thomson et al., 1998) or blastomeres of the 8-cell stage embryos (Ilic et al., 2009). ESCs have indefinite proliferative capacity and differentiate into all three germ layers under the appropriate conditions in vitro (Thomson et al., 1998). Primordial germ cells are small population of founder pluripotent cells segregated in the developing yolk sac of the embryo as early as 7.25 days post coitum (dpc) in mouse. They migrate to genital ridge of the developing embryo and further differentiate to produce gametes (McLaren, 2003). PGCs share with ESC indefinite proliferative capacity and expression of a number of stemness genes such as OCT4 and NANOG. In addition, several PGC.-specific genes such as DAZL, PUM2, STELLAR, PIWIL2 and TEX-14 are also expressed in ESCs (Mikkola et al., 2006). Tissue non-specific alkaline phosphatase which is normally used as a marker to monitor PGC migration is also highly expressed in ESCs (Aflatoonian and Moore, 2006). These findings indicate that PGCs and ESCs may use common signaling pathways in regulating their self-renewal. The pluripotency of ESCs is often demonstrated by chimera formation following the injection into blastocysts. Interestingly, unipotent embryonic germ cells obtained from PGCs similarly contribute to tissue formation in chimeras (Matsui et al., 1992). However, comparison between PGCs and ESCs using microarrays revealed that, in reference to universal mouse reference RNA ( Stratagene, La Jolla, CA), 382 transcripts were significantly up-regulated in ESCs and only 188 were elevated in PGCs. This indicates that the transcriptome of the PGC is distinct from that of ESC (Mise et al., 2008). At the same time, Payer et al (2006) showed that the ESC colonies displayed a well-defined subpopulation of cells expressing Stella (Dppa3) and Oct4 to indicate that some of the cells in the ESC colonies might be pre-destined to form PGCs (Payer et al., 2006). It has been recently shown that human ESCs can be differentiated into PGCs and post meiotic spermatids (Aflatoonian et al., 2009). Taken together, these results indicate that ESCs replicate some of the characteristic of PGCs and can be differentiated to germ cells and gametes to model the development of male gametes.
Derivation of male germ cells from ES cells
An interesting aspect of the in vitro generation of germ cells from ESCs is the ability to narrow down much longer timeline of the natural development from early fetal stage to puberty. This is accomplished by culturing ESC in the appropriate growth factor and hormonal microenvironment that activates the specific molecular signaling pathways supporting the differentiation towards germ cells. In the natural development, the cross talk between proximal epiblast and the extra-embryonic ectoderm is directly responsible for PGC formation. Signaling activity of bone morphogeic proteins (BMPs), the members of the transforming growth factor beta superfamily, have been known to play a key role in this differentiation process (Ying et al., 2001). BMP4 and BMP8b secreted by extra-embryonic ectoderm have been shown to induce PGC formation (Ying et al., 2000; Ying et al., 2001). BMP4 is also involved in the induction of PGC formation from ESCs (Toyooka et al., 2003). The signaling of BMPs is mediated through SMAD1 and SMAD5 in a subpopulation of proximal epiblast cells (Massague and Wotton, 2000).
Several markers are used for proper identification of germ cells and expressed at different stages of development. Increased expression of Fragilis (Lfitm3), a member of an interferon-inducible gene family, is observed in the founder population of PGCs at 7.25 dpc. Increased expression of another gene Stella (Dppa3) is also observed around the same time in a subpopulation of cells within the center of the Fragilis-expressing cluster and its expression is maintained during the migration of PGCs (Saitou et al., 2002). Expression of tissue non-specific alkaline phosphatase is also first detected in PGCs at 7.25 dpc (McLaren, 2003). In addition, cell surface carbohydrate SSEA1 presence is observed in PGCs migrating from hindgut endoderm to the genital ridges (Fox et al., 1981). Migratory PGCs also up regulate c-kit (kit), a receptor tyrosine kinase, which is activated by binding of stem cell factor and controls PGC growth and survival (Sandlow et al., 1999). Expression of postmigratory marker mouse vasa homolog (Mvh) is observed at the end of the PGC migration (Toyooka et al., 2000). DAZL is a specific postmigratory marker which is necessary for the development and is also highly expressed in PGCs (Ruggiu et al., 1997). In addition, the pluripotency marker Oct4 is continuously expressed and is necessary for the survival of PGCs (Kehler et al., 2004; Pesce et al., 1998). Moreover, Blimp1 (Prdm1) has also been shown to have a critical role in the induction and normal development of PGCs and is expressed in the proximal epiblast cells (Ohinata et al., 2005). Stage-specific expression of these markers renders them as useful tools for correct identification of different developmental stages of germ cells from formation of PGCs to their differentiation into immature and mature gametes.
Derivation of PGCs
In vitro differentiation of PGCs from ESCs seems to be a spontaneous process with simple differentiating culture conditions. Several studies showed that mouse (Geijsen et al., 2004; Nayernia et al., 2006; Toyooka et al., 2003) and human (Aflatoonian et al., 2009; Kee et al., 2006) ESCs can be differentiated into PGCs. The spontaneous differentiation of mouse ESCs into PGCs can be achieved by culturing cells as a monolayer (Hubner et al., 2003) or embryoid bodies (Geijsen et al., 2004) in a differentiation medium containing DMEM and 10% fetal bovine serum. Cells resembling PGCs were identified with germ cell specific markers c-kit and Mvh, and pluripotency marker Oct4 in 3-7 day monolayer differentiation culture. When ESCs were differentiated via generation of embryoid bodies, putative PGCs expressing Stella, Fragilis, and Mvh were identified after 3 days of differentiation (Clark et al., 2004; Geijsen et al., 2004; Toyooka et al., 2003). Although, ESCs and PGCs co-express genes which are used as markers for germ cells, subsequent selection confirmed that these cells are PGCs and not undifferentiated ESCs (Geijsen et al., 2004). For any practical application, it is crucial to promote differentiation of ESCs into PGCs during in vitro culture to increase the efficiency of the process since spontaneous differentiation is inefficient and unpredictable. Increased efficiency can be achieved by providing appropriate microenvironment or niches during ESC differentiation. It has been shown that culture of EBs in a medium containing retinoic acid facilitates the differentiation and proliferation of PGCs in vitro (Koshimizu et al., 1995). Since BMP signaling plays a key role in the differentiation of PGCs, co-aggregating EBs with BMP4-producing cells also facilitates the formation of PGCs (Toyooka et al., 2003). Silva et al (2009) used retinoic acid and testosterone in the ESC differentiation culture to improve the formation of PGCs (Silva et al., 2009). A recent study compared the use of BMP4 and retinoic acid during human ESC differentiation to facilitate putative PGC formation and showed that retinoic acid substantially increased the VASA expression in day 7 cultures (Aflatoonian et al., 2009).
Derivation of spermatogonia and spermatids
Since PGCs are the precursors of gametes, PGC differentiation is a critical first step in the derivation of spermatogonia and spermatids from ESCs. Once PGCs are derived, cells should be cultured in optimal microenvironment for male gamete development. Putative male meiotic germ cells were developed after 20-day culture of EBs in differentiation medium determined by DNA content analysis and immunostaining to exhibit characteristics similar to those of round spermatids (Geijsen et al., 2004). Nayernia et al (2006) reported that culturing ESC-derived PGCs in a growth factor cocktail containing retinoic acid resulted in formation of fertile haploid male gametes which produced viable offspring after intracytoplasmic sperm injection into oocytes (Nayernia et al., 2006). Recently, cells similar to round spermatids expressing high levels of PRM1 were produced after 14 days of hESC differentiation using EB intermediary stage in the medium containing retinoic acid (Aflatoonian et al., 2009). Although successful development of male gametes is possible using mESCs, protocols for the development of human male gametes from human ESC while showing promise still require further refinement. Characterization of the niches specific for the final developmental stages of male gametes is necessary to increase the efficiency of the development of male gametes in vitro.
hESC derived EBs as in vitro model for reproductive toxicity screening
EBs produced under strictly controlled experimental conditions from a single-cell suspension (Ungrin et al., 2008) can be used to model the development of germ cells as EBs support the formation of PGCs. The EBs generated using this technique consist of central epiblast cells (primitive ectoderm) and the surface hypoblast (primitive endoderm). The advantages of this model are that it allows generation of ultra high-throughput EBs of the same size and structure, and development of PGCs in a microenvironment similar to that of developing embryo. The novel and important characteristics of this system are: reproducibility – EBs are consistent in shape and size; order – they reflect tissue-level organization; ease of use – hundreds to thousands of EBs can be produced without unreasonable amount of labor and scalability – the production of EBs can be easily increased for several orders of magnitude. It seems that this, or similar systems, have a potential for being developed not only into a model of gamete differentiation, but also into high-throughput model for the differentiation of other tissues.
The search for a model that closely mimics normal development of human male gametes is in its early stages. Three major events mark differentiation of the human ESCs into gametes: 1) differentiation into PGCs, 2) generation of spermatogonial stem cells and 3) formation of sperm cells. These events are critical for the development of functional male gametes. We anticipate that human ESCs-based systems are the best potential candidates to model specific aspects of these processes. Validity of these models awaits further evaluation since their equivalency with reproductive tissues in situ remains to be assessed.
In this review we have described some of the most promising hESC-based systems currently used to model certain aspects of male gamete development and that we believe hold promise for development of toxicity screening assays. Recent introduction of robust, straight-forward hESC culturing techniques under feeder-free conditions and derivation of hESC lines that preserve the embryo and limit ethical concerns, promise to promote faster adoption of these systems in reproductive toxicity screening.
In summary, we predict that reproducible models for screening of potential environmental toxicants can be developed by manipulation of hESC. Establishment of such models will not only improve our understanding of molecular composition of optimal niche/microenvironment needed for germ cell generation and signaling pathways involved in this process., but also help us to better understand certain aspects of infertility and, more generally, human germ cell development.
We anticipate that the creation of more relevant in vitro models of reproductive toxicity screening will be achieved in the near future. The application of ESCs may modernize our understanding of genetic and epigenetic changes of the germ cell development at the molecular level and may lead to the identification of specific markers that can be used for screening of potentially toxic compounds. The validity of these models should be tested by using agents that are known to impair germ cell development in vitro and, if possible, in vivo. Development of new techniques such as transcriptome sequencing, miRNA RT-PCR arrays and proteomics that are now available to detect and identify markers of germ cell development, might serve to standardize and validate these ESC models. The application of such hESC-based systems to model the key differentiation events of the male gametes could significantly improve our understanding of toxic insults leading to male infertility.
In addition to the discussed advantages of human-based systems, the costs of these hESC high throughput tests will be relatively low in comparison to animal studies. These or similar in vitro assays can be validated using known toxic compounds which cause male infertility. It is likely that once these models are well established, the screening for reproductive toxicity will become more widespread and will be used to test a variety of xenobiotics present in everything from food additives to clothing and plastics used in everyday objects. We hope that the accumulation of relevant data acquired using these and similar model systems will help to predict potential toxic compounds that contribute to the increased rate of male infertility in human population in industrial countries.