Given the potential limitations of in vitro experiments, it is thus sometimes most appropriate to examine cell differentiation in vivo after cell transplantation. Grafted cells experience a rich combination of signals from their corresponding local three-dimensional environment, including direct cellular interactions with adjacent host tissues, exposure to secreted morphogens, and more general homeostatic cues.
It is well-known that mammalian embryonic stem cells (ESCs) and their malignant counterparts, embryonal carcinoma (EC) stem cells, form complex teratomas when engrafted into an immune-deficient host . Studies in our laboratory and by others have used cell transplantation as an approach to test the pluri-potency of such stem cells (Fig. 2) [1, 3–6, 15]. Teratomas resulting from transplanted hESCs consist of an array of differentiated tissues representative of all three germ layers (Fig. 2). In general, these data are reasonably consistent with in vitro experiments on pluripotent stem cell differentiation as referred to earlier but also go much further. Several structures produced within teratomas derived from hESCs are highly organized and consist of ordered arrangements of different tissue types that in many ways recapitulate organogenesis within the embryo. For example, we and others have reported organized structures that have the appearance of kidney containing renal corpuscles, associated tubules, and associated vascular supply; gastrointestinal tract consisting of a simple columnar epithelium, supporting mucosa, smooth muscle layers, and neural ganglia; skin including dermal and epidermal layers, complete with stratum granulosum, keratinized cells, and hair follicles; and respiratory airway composed of pseudostratified ciliated epithelium, smooth muscle, nerves, and supporting cartilage (Fig. 2) [1, 3–6, 16]. There are also other examples of tissue types found in isolation within the body of the teratoma that may be identified using standard histological methods, including skeletal muscle, neural ganglia, pigmented cells, glands, primitive epithelium, and neuroepithelium (Fig. 2) [1, 3–6, 16]. Accordingly, the engraftment of hESCs into an appropriate host can result, in part, in the differentiation of human tissues that consist of cells in a recognized arrangement that resemble structures within the developing embryo and adult.
Figure Figure 2.. Histological analysis of differentiated tissues found in teratomas formed in the testis of severe combined immunodeficient (CB17/ICR-Prkdcscid/Crl) mice after transplantation of human embryonic stem cells. Tumors were grown for a period of 6–8 weeks. Figures show bright field micrographs of tissues prepared in Bouins fixative, embedded in paraffin wax, and sectioned (5 μm). (A): Low-power image showing tissue heterogeneity with in the tumor. (B): Wall of intestinal tract showing epithelium (ep), mucosa (m), muscle layer (mus), submucosal glands (gl), and neural ganglia (ng). (C): Higher magnification image of intestinal mucosa (m). The epithelium is typical of that found in the large intestine and consists of a single layer of columnar cells that primarily secrete mucous. (D): Smooth muscle (mus). (E): Neural ganglia (ng) linked by nerve fibers. (F): Cartilage (ctg) and bone. (G): Structures of the skin, including epidermis (ed), dermis (dm), and cornified layer (c). Note that the stratum granulosum (sg, arrow) is characterized by intracellular granules, which contribute to the process of keratinization. (H): Pseudostratified ciliated epithelium typical of a respiratory airway. (I): Kidney tissue including glomerulus (glom) with surrounding Bowman's space and adjacent tubules (tub). Note the vascular pole of the glomerulus and the presence of a blood vessel (bv). Histological staining: Weigert's (A–G) and hematoxylin and eosin (H, I). Scale bars = (A) 450 μm, (B, F) 150 μm, (C) 75 μm, (D, E, I) 50 μm, (G) 40 μm, and (H) 15 μm.
Download figure to PowerPoint
Obviously there are limitations to the xenograft approach in that transplanted hESCs form tumors that do not show the ordered and appropriate placement of tissues as would be found in normal embryogenesis. There are undoubtedly many critical factors as to why tissues within xenograft tumors are arranged in a haphazard manner. One of the primary reasons is likely to be due to the lack of information concerning the body plan and that the folding and apposition of alternative tissue types to enable appropriate cell–cell interactions is physically restricted, thus limiting the extent to which true development can occur. Nonetheless, there are certain aspects of tissue development within teratomas that suggest the in vivo environment is providing transplanted hESCs with the opportunity to display certain aspects of the developmental behavior characteristic of cells found in the embryo. Many investigators have used the teratoma approach as a means to verify the pluripotential nature of newly derived hESC lines [1, 3–6], but few have closely examined the structure of such teratomas in any detail. Accordingly, there is a distinct lack of information concerning the characterization, regulation, progression, or outcome of hESC differentiation after transplantation.
Transplantation of hESCs into Mice, a Model of Cell Differentiation In Vivo
Interspecies transplantation has permitted the study of both differentiation and therapeutic potential of stem cells. The development of animal recipients for the xenotransplantation of human cells, to investigate disease and study the mechanisms that control cell growth and differentiation, has been studied for many years . Much of this work has been pioneered by research in hematopoiesis and the discovery of immune-deficient scid (C.B-17-Prkdcscid) mice . Subsequently, many attempts have been made to develop more efficient modified severe combined immunodeficiency (SCID) mice, including the NOD/ShiJic-scid with γcnull (NOD/SCID/γcnull) mouse that has multiple immunological dysfunctions and has been shown to be an excellent recipient for the engraftment of human cells . Immune-deficient nude (C57BL/6J-Hfh11nu) and scid mice have proven to be a long-established system for tissue transplantation to examine the ability of cells to engraft and function in a mammalian host [20–22]. SCID-beige mice (C.B-Igh-1b GbmsTac-Prkdcscid-Lystbg N7), a strain of double-mutant mice with impaired lymphoid development and reduced natural killer cell activity, have commonly been used as recipients of hESC grafts and the growth of teratomas [6, 16]. Alternatively, other researchers investigating the potency of hESCs have used mice with only the scid mutation and have reported the growth of teratomas with similar features to those in SCID-beige animals [3–5]. Currently, there seems to be little evidence as to whether the strain of immune-deficient animal has an effect on the growth of hESCs after transplantation, although subtle effects between such experiments cannot yet be dismissed.
Recent work has shown that hESCs injected into the leg muscle of immune-competent (CD-1) mice failed to induce an immune response within 48 hours after injection . Given the short-term nature of these in vivo experiments, it is not possible to comment on whether mature hESC-derived teratomas provoke an immune response. Furthermore, in vitro analyses showed that undifferentiated hESCs and hESCs differentiated into embryoid bodies failed to stimulate proliferation of alloreactive primary human T cells . Collectively, these data suggest that hESCs may possess unique immune-privileged properties. Certainly immunity and tissue rejection is an important concern during allogenic transplantation. This is particularly relevant to the ability of hESCs to differentiate after transplantation and will no doubt warrant further investigation.
Characterization of Tissues Produced from Transplanted hESCs
Routine histological methods have primarily been used to examine the structure of hESC-derived xenograft tumors grown in mice [1, 3–6, 15]. Histological staining is useful for morphological visualization of tissues but is limited and is not always specific enough to be informative about the true identity of a tissue or cell type. Modern molecular approaches examining gene and protein expression are only just beginning to be used to further characterize hESC-derived xenograft tumors. Immunological techniques with antibodies specific for human-specific nuclear antigen and human E-caderin are now routinely used to distinguish between host and transplanted tissues [1, 24]. Furthermore, genetic methods such as fluorescence in situ hybridization can be used to determine the contribution of tissues derived from the implanted hESCs in relation to those originating from the murine host . This approach has been useful to show that the structure of blood vessels within hESC-derived teratomas consists of both host and donor cells, demonstrating the cooperation between these cell types to establish functional units .
In situ hybridization and immunocytochemistry will be particularly useful to identify tissues that do not have an easily recognizable structure, such as in those circumstances in which tissues are in their early stages of development and do not yet possess a morphology characteristic of their differentiated phenotype (Fig. 3). For example, many primitive embryonic tissues consist of little more than layers or aggregations of cells of similar type. Identification of these structures may be achieved by molecular analysis of gene and/or protein expression such as markers for derivatives of ectoderm (brain-neurofilament 200 or β-tubulin-III, skin-keratin, adrenal-DßH), mesoderm (muscle-enolase, bone-CMP, kidney-renin or kallikrein, uro-genital tract-WT1, heart-cardiacACT, hematopoietic tissue-δglobulin or βglobulin), or endoderm (liver α1AT, pancreas-amylase [exocrine], or insulin [endocrine], visceral endoderm αFP). Moreover, many of these biomarkers have the potential to be used to more precisely determine the type of tissue that arises from within a particular germ layer, for example, heart-cardiac actin. Similarly, combinations of biomarkers, including HNF3-alpha, αFP, and Fox2a, can be used to test for the presence of extraembryonic tissues.
Figure Figure 3.. Immunocytochemical identification of ectodermal tissues formed from human embryonic stem cells (hESCs) transplanted into the testis of severe combined immunodeficient (CB17/ICR-Prkdcscid/Crl) mice (data from clone hESC-NCL1 shown). Tumors were grown for a period of 6–8 weeks. Serial sections (5 μm) of paraformaldehyde-fixed tissue were taken through a region of neural differentiation, showing the typical morphology of neuroepithelium organized as neural rosettes (nr) (A, stained with hematoxylin and eosin) and localization of the neuroprogenitor marker, (B) nestin, and (C) β-tubulin-III. Note the absence of β-tubulin-III expression in the proliferative center of the neural rosette where nestin staining is strongest. (D): More mature neural tissues show high levels of β-tubulin-III immunoreactivity. Other epithelia with a distinctly different morphology express high levels of human epidermal keratin (E–G; note that F is a phase-contrast image). Such epithelia do not stain for neural markers, whereas expression of epidermal proteins is absent in neural structures. Scale bars = (A–C) 150 μm, (D, F, G) 50 μm, and (E) 300 μm.
Download figure to PowerPoint
It is important to appreciate, however, that there are few bio-markers that show specific expression in only a single cell type and many molecules share expression patterns with alternative kinds of cell. For example, nestin, a marker commonly used to identify neuroprogenitor cells, is also expressed in pancreas and testis [25–27]. Biomarker expression is becoming a critical issue in the determination of cell identity in stem cell biology. It is now generally accepted that a combination of the presence or absence of several biomarkers is required to more accurately identify a particular tissue type. To characterize differentiating hESCs in vivo, this can be achieved either by dual-labeling or the probing of consecutive tissue sections. Such protein or gene expression should also always be linked to the morphology of the tissue concerned. Furthermore, when a particular tissue is positively identified within a teratoma, it will also be important to assess and record the nature of the surrounding tissues to determine whether there is a correlation during the formation of adjacent tissue types. Such experiments will contribute to our understanding of the development of ordered structures and how tissues form together, as in organogenesis.
In a recent report, Gertow et al.  began the process of looking more closely at the differentiation of hESC-derived teratomas by the immunocytochemical examination of a large range of general markers. Analysis and identification of these tissues showed that the transplanted hESC line, HS181, predominantly formed ectodermal derivatives, although evidence of other tissue types was recorded . This study represents the most in-depth characterization of such tumors to date and again comprehensively demonstrates that hESCs are capable of some organized development after transplantation. However, to extend these findings and improve our understanding of tissue development in humans, we need to more precisely characterize the formation of these organized tissues within such tumors and begin to examine some of the key regulatory processes that are known to mediate cell differentiation. We currently do not know, for example, whether a recognized control pathway such as bone morphogenetic protein signaling functions to regulate the formation of ectodermal tissues after the transplantation of hESCs.
Effect of Site-Specific Transplantation on the Formation of hESC-Derived Teratomas
Evidence indicates that local environmental cues from the tissue into which cells are transplanted influence the ability of the grafted cells to differentiate. Wakitani et al.  reported that murine ESCs produced a greater ratio of cartilaginous tissues in teratomas grown in the knee joint compared with cells grafted into the subcutaneous space. These data suggest that implanted cells are more likely to differentiate into tissues resembling their surroundings and such effects are probably mediated by local environmental signals.
Most experimental evidence for this notion has been generated by studies investigating the integration of immortalized neural cell lines, neural progenitor cells, or murine ESCs into the brain [29–37]. For example, immortalized primary neural cells transplanted into adult or neonatal rat hippocampus and cortex exhibited morphologies similar to cells of the transplant site [35, 36]. However, this is not always consistent, as Lundberg et al.  reported that the plasticity of similar immortalized neural cells was more restricted when transplanted into the adult environment. Determination of cell identity in response to local cues has also been demonstrated with nonmodified neural precursors from the striatum, where it has been shown that positional-specific cues play an instructive role in determining the regional phenotype of cells in the forebrain . Similarly, the generation of neural precursor cells from murine ESCs indicates that such neurons have the ability to participate in neural development and functionally integrate with the host brain [29, 33, 34].
The proliferation, differentiation, and death of human EC cells also seem to be regulated by the anatomical site into which they were implanted [38, 39]. NTERA2.cl.D1 human EC cells transplanted into subarchnoid space and superficial neocortex of adult immune-deficient mice proliferated and formed bulky tumors that were lethal within 70 days after implantation. In contrast, NTERA2.cl.D1 cells grafted into the caudoputamen ceased proliferating and formed neural-like tissues that seemed to integrate with the host nervous system. The authors suggest that tissues of the caudoputamen have the ability to effect the differentiation of implanted cells, possibly by a secreted soluble factor, but this was not identified [38, 39].
Deacon et al.  observed that nondifferentiated murine ESCs were able to form neural derivatives after transplantation, but, according to their approach, transplanted ESCs did not always differentiate into phenotypes corresponding to those of the implantation site. Given their more undifferentiated, primitive state, it is possible that ESCs may not be as responsive to as wide of a range of signals as are slightly more differentiated cells. Alternatively, ESCs retain the ability to form the full spectrum of somatic cell types, whereas committed precursors are more limited in their scope for differentiation.
Comparatively little has been done to determine whether transplantation of hESCs into different anatomical locations effects their ability to differentiate into tissues corresponding to their surroundings. Most investigators have reported the growth of hESC-derived tumors within the subcapsular compartment of either the testis or the kidney [1, 3–5, 16]. This is primarily to ensure that it is straightforward to accurately implant cells into several individual animals at the same location and that the implanted cells remain and are contained at that particular implantation site. Most solid teratomas grown in these sites and other locations display evidence of containing tissues representative of all three germ layers (Fig. 3) [1, 3–6, 16], but it is really not known how these different tumors compare because there has been no thorough characterization of their structure. On occasion, solid tumors fail to develop at all and fluid-filled cysts form that often push aside and destroy much of the surrounding host tissues (Fig. 4) . The reason for this is unknown but may involve differences in the exact location cells were released into the host or the exact status of the hESCs at the time of transplantation. A detailed investigation is required to determine whether site-specific cues influence the differentiation of a particular hESC line after transplantation. Site-dependent differences in the ability of transplanted hESCs to show their developmental potential may be subtle but nonetheless important. It has been shown that hESC xenografts grown in the testis predominantly form derivatives of the ectoderm and mesoderm ; however, would similar grafts into a different site predominantly form tissues of other germ layers? More detailed analysis of tumor formation in alternative sites will provide useful information about the role of the host tissues in the regulation of tissue differentiation by hESCs.
Figure Figure 4.. The structure of xenograft tumors can be variable when grown in alternative anatomical locations. It is currently not certain whether the variability of tumor growth is really a result of site-specific transplantation or whether it simply shows variability in tumor type. Further investigation is required (see text for discussion). These images show examples of the gross anatomy of xenograft tumors grown as a result from the transplantation of human pluripotent stem cells (A) into the testis, (B) within the intraperitoneal cavity, (C) beneath the capsule of the kidney, or (D) subcutaneously in the flank of immune deficient mice. In all the cases above, differentiated human tissues were identified, for example, cartilage (ctg), skin (sk), and smooth muscle (sm). In some instances, tumors can be highly cystic and consist primarily of fluid-filled cavities (fc), pushing aside existing host tissues, as seen in the testis above (A, seminiferous tubules [st]). Alternatively, tumors may be solid with delicate connective tissue septa (ts). Scale bars = (A) 3.0 mm, (B) 800 μm, and (C, D) 1.2 mm.
Download figure to PowerPoint
Comparison of the Developmental Potential of Different hESC Lines In Vivo
There is little information as to whether different hESC lines possess greater or less ability for differentiation after transplantation into immune-deficient mice. More is currently known about the differences in developmental potential between human pluripotent embryonal carcinoma stem cells (which are considered to be the malignant counterparts of hESCs ) and hESCs. Analysis of teratomas derived from hESCs and human EC cells most obviously shows that all the hESC lines tested to date have the ability to form tissues representative of all three germ layers in vivo [1, 3–6, 14] whereas certain EC lineages, such as those derived from the TERA2 parent line, appear more restricted in their capacity for differentiation after transplantation [14, 15].
A recent study by Heins et al.  showed that two out of the seven hESC lines they derived formed fluid-filled cysts rather than solid tumors; however, it is not clear how many times this experiment was repeated. In our experience, the formation of such cysts is not unusual, and it is possible that a single hESC line has the capacity to produce either a solid tumor or a fluid-filled cyst in separate experiments (Fig. 4). There is a distinct lack of information concerning the developmental potential of different hESC lines and, in particular, a comparison of the teratomas produced by such hESC lineages. A thorough evaluation of the potency of independently derived hESC lines is required to fully address whether differences in developmental capacity exist between alternate hESC lines. Provided transplantation of hESCs is carried out in a controlled fashion, a difference in the ability of independent hESC lines to differentiate may be due to the original derivation of the particular cell line or how that hESC lineage is routinely grown and maintained.
Effect of Cell Development on the Formation of hESC-Derived Teratomas
The developmental status of the host may also play an important role in determining hESC differentiation after transplantation. Most hESC-derived teratomas have been grown in adult murine tissues [1, 3–6, 16]; however, it may be argued that transplantation of hESCs into an adult host is a less appropriate environment compared with grafting hESCs into embryonic tissues. Many molecular pathways associated with cellular development will be active during embryogenesis and during the formation of certain tissues in the neonate. It is thus reasonable to hypothesize that transplantation of hESCs into developing host tissues will result in alternative types of hESC differentiation compared with those reported in adult tissues. To address this issue, Goldstein et al.  investigated the integration and differentiation of hESCs transplanted into the chick embryo. The chick model is a well-characterized and accessible experimental system to study the inductive interactions between cells during differentiation. Several investigators have shown that mammalian cells and tissues transplanted into avian embryos can respond to local environmental cues and develop into tissues appropriate to their location in the host [40–42]. Initial experiments suggest that hESCs transplanted in ovo survive, proliferate, differentiate, and integrate with the host embryonic environment . However, further detailed analysis of the hESC-derived tissues is required in this avian system to determine whether differences exist in the developmental potential of hESCs engrafted into embryonic or adult chick tissues. Although technically more challenging than experiments in the chick, methods have been developed and used to study the fate of transplanted cells in rodent embryos [30, 43]. Similar approaches may be adapted to enable the study of hESC differentiation after their transplantation into rodent embryonic and postnatal tissues.
It is equally important to consider the developmental status of the hESCs implanted into host tissues. For example, it is not known whether predifferentiation of hESCs toward the ectodermal lineage before transplantation will result only in the growth of differentiated human ectodermal structures in the host. Interestingly, some pluripotent human EC lines appear to be restricted in their capacity to differentiate in that so far they have only been able to produce ectodermal derivatives in vitro [15, 44, 45]. Teratomas produced from such EC cells also consist only of ectodermal tissues . Predifferentiation of cultured hESCs before transplantation into SCID mice may be used to determine whether hESC differentiation can be restricted as a result of prior manipulation in vitro. Such experiments may be based on earlier reports that certain growth factors induce the differentiation of hESCs toward particular lineages [7, 8]. Transplantation may therefore provide a powerful approach to confirming earlier in vitro work and will be useful to test whether the status of grafted hESCs influences their ability to differentiate in vivo.
Assessing the Progression of hESC-Derived Tumor Growth
hESCs grafted into immune-deficient hosts have been maintained over several weeks until, more often than not, the palpable growth of a teratoma has been detected [1, 3–6, 16]. There has been no attempt, however, to examine the progression of tumor growth during this period. Can it be assumed that transplanted hESCs move through the various stages of cell development to produce differentiated tissues? For example, do events similar to gastrulation occur within the hESC xenograft? After all, tissues representative of all three germs layers are present in teratomas that originate from an apparently homogeneous population of hESCs. Alternatively, do grafted hESCs differentiate directly into specific tissue types without passing through some of the earlier stages of cell development normally experienced by cells in the embryo? The assessment of the early stages of xenograft growth may provide the opportunity to study and model the early stages of human embryogenesis.
The maturation of tissues within the xenograft should also be considered. There are obviously issues concerning the welfare of the host, but do hESC-derived teratomas continue to grow? Will other tissues that have previously not been observed in hESC-derived teratomas (for example, cardiac muscle) appear in time? Gertow et al.  commented that the maturation level within a teratoma correlated with the time spent in vivo. Is the capacity of xenograft growth limited to the numbers of hESCs engrafted, or is there a resident population of proliferating cells that continues to contribute to tumor growth? Regions of cell proliferation have been identified within hESC-derived teratomas after 6–7 weeks of growth. However, experimental evidence shows that these cells are not remaining hESCs but seem to be precursors of immature and developing human tissues that subsequently differentiate along predictable developmental pathways . In addition, there is no evidence of any malignant tissues such as embryonal carcinoma in any of the tumor samples tested [3, 16]. However, there are some data that suggest that hESCs may undergo karyotypic changes in vitro .
There are a variety of instances in biomedical research in which terms like chimera or hybrid are used, including chimeric DNA containing sequences derived from more than one source, nuclear-cytoplasmic hybrids, chimeric early embryos consisting of cells from more than one individual, and xenografts of cells transplanted from one species into a host of another species. The transplantation of human stem cells into nonhuman hosts enables scientists to study cell development in humans without the need to use human embryos. Such experiments, however, raise ethical concerns that may create controversy over this area of research. Karpowicz et al.  have recently reviewed some of the ethical issues regarding the transplantation of human stem cells into nonhuman embryos. They argue that because the potential scientific and therapeutic benefits are strong and that human dignity to undertake such experiments would not be violated, it is ethically acceptable to conduct such experiments. This is despite those who oppose such experimentation on moral grounds. Interestingly, the study of transplanted human tissues into postnatal animal recipients has been practiced for many decades and currently seems to attract much less attention. Such experiments are subject to vigorous regulation, and animal usage is closely controlled. It is considered unsafe and therefore unethical to test human stem cells in human recipients until their function and tumorigenic capacity had been thoroughly evaluated by alternative means, including in vivo studies of human stem cells in living animals. In general, there are many instances when biomedical research has been brought into contention, raising both ethical and moral issues. This can only be healthy for the progression of research, and these debates should be considered an essential element for the advancement of science in a modern society.