VIII. REQUIREMENTS OF STEM CELL-BASED THERAPIES
Today's most urgent problem in transplantation medicine is the lack of suitable donor organs and tissues, and treatments to replace, repair, or enhance the biological function of damaged tissue through cell transplantation/replacement therapy have until recently been limited to a few systems (41; review in Refs. 132, 384). Potential sources of cells for repair are self (autologous), same species (allogeneic), different species (xenographic), primary or immortalized cell lines, and adult stem cell-derived donor cells. The ability to cultivate, multiply, and manipulate these cell types has either limited or encouraged their use in specific treatment protocols (132). Presently, only allogeneic or matched donor-derived stem cells have been used in human cell-grafting therapies. While the differentiation potential of some adult stem cells (hematopoietic and mesenchymal) are well-characterized in vivo (HSC) or in vitro (MSC), the transdifferentiation potential of most adult stem cells remains controversial (235, 378), partly as a consequence of culture conditions (175) and contaminations or cell fusion events (3, 358). Regardless of these limitations, it is to be anticipated that human (embryonic and adult) stem cell research may help millions of people who are affected by a wide range of intractable human ailments (Parkinson's disease, spinal cord injuries, heart failure, and diabetes; see Table 5).
The in vitro developmental potential and the success of ES cells in animal models demonstrate the principle of using hES-derived cells as a regenerative source for transplantation therapies of human diseases. Before transfer of ES-derived cells to humans can proceed, a number of experimental obstacles must be overcome. These include efficient derivation of human ES cells in the absence of mouse feeder cells, and an understanding of genetic and epigenetic changes that occur with in vitro cultivation. It will be necessary to purify defined cell lineages, perhaps following genetic manipulation, that are suitable for cell-based therapies. If manipulated, then it will be important to guard against karyotypic changes during passaging and preparation of genetically modified ES-derived cells. Once introduced into the tissue, the cells must function in a normal physiological way. Finally, assurances against the formation of ES cell-derived tumors and donor/recipient immunocompatibility are additional requirements of stem cell-based therapies. As pointed out, significant progress has been made in the isolation of defined cell lineages in mouse, and important advances have already been seen with hES cells. Before therapeutically applicable, any ES-based treatment must, however, show limited potentials for toxicity, immunological rejection, or tumor formation, and at present, human ES cell research has not reached this threshold.
A. Genetic and Epigenetic Concerns
About 70 human ES cell lines (excluding those held in the private sector and established more recently) have been described that are available for research, but at present, only 
22 of them can be propagated in culture (see http://escr.nih.gov/). Although some of the hES cell lines can be cultivated indefinitely and demonstrate a normal chromosomal complement after 2 or more years of passaging, this does not necessarily mean that these cells are genetically stable during long-term culture (and correspondence by 62, see Ref. 100). In somatic cells from humans and other animals, approximately one mutation occurs every cell division. A cell that has divided 200 times in culture would therefore be expected to contain 200 mutations (195). The majority of these mutations may occur without consequence, but in those instances where protooncogenes or regulatory sequences are affected, the consequences may render the cells unsuitable for therapeutics.
Epigenetic modifications, such as DNA methylation, acetylation, histone modification, and other changes in chromatin structure that do not alter the genomic sequence, would also be expected to play an important role in the developmental potential of ES cells. We have already described how batches of serum or serum withdrawal, which causes epigenetic modifications (30), can affect the differentiation potential of mES cells and how altered functional levels of Oct-3/4 would be expected to modify development (see sect. II). In fact, epigenetic changes that decrease Oct-3/4 levels cause a decrease in cell number in mouse clone blastocysts that would be expected to adversely affect development (44). The fact that the vast majority of cloned embryos die during embryonic development, despite their normal chromosome complement, also suggests that epigenetic reprogramming in reconstructed oocytes is incomplete (297). The consequences of uncontrolled epigenetic modifications are only now being analyzed in hES cells.
Based on these data, it is likely that the current supply of human ES cell lines may be insufficient to adequately test their potential for cellular therapeutics. Additional or freshly isolated ES cell lines may be a constant requirement, but with the current legal constraints, this may not be possible in all countries. The generation of ES cell-derived germ cells (136; see Refs. 164, 365) may represent one possible alternative source for these cells, but before this can occur, it will be necessary to determine whether gametes can be obtained from hES cells that are capable of forming blastocyst-like structures. This of course brings up one additional concern: gametes generated from ES cells will have undergone prolonged cultivation times with accumulating genetic and epigenetic defects, which may render these cells of limited value, except in the context of nuclear transfer (see sect. IXB).
B. Tumorigenesis
It is well established that undifferentiated, early embryonic cells commonly generate teratomas or teratocarcinomas when transplanted to extrauterine sites (346; see sect. I). This is not surprising, because ES cells display many features characteristic of cancer cells (57) including unlimited proliferative capacity (351), clonal propagation, and a lack of both contact inhibition and anchorage dependence. Tumor growth in immunodeficient animals appears to depend primarily on the presence of an undifferentiated stem cell population. Benign teratoma formation would therefore be expected at the site of injection and potentially at other locations whenever undifferentiated ES cells are present. Short-term, tumor formation does not appear to be a significant problem; however, few long-term animal experiments have been performed to demonstrate that transplantation of ES cell-derived donor cells do not give rise to tumors. Importantly, it is not simply the transplantation of mouse (396) and human (362) ES cells that results in the growth of teratomas, but also the transplantation of ES-derived differentiated cell populations (38, 191). The passive elimination of undifferentiated cells via lineage selection protocols as described below may therefore prove insufficient to eliminate the cancer risk. It may be necessary to develop additional strategies for the active elimination of tumorigenic cells by directing the expression of suicide or apoptosis-controlling genes in graft tissue.
C. Purification and Lineage Selection
Because of the potential tumorigenicity of human ES cells (362), protocols have been established to purify committed cells of the desired phenotype and exclude nondifferentiated cells from cell grafts. In this context, early tissue-restricted stem and progenitor cells, characterized by a limited potential for self-renewal (i.e., cells may not be tumorigenic), a high proliferative capacity, and the ability to generate a number of differentiated cell progeny, are of special interest. Two major experimental schemes have been devised to isolate such progenitor or tissue-specific stem cells from differentiating ES cells: 1) selection of specified progeny through the use of cell surface markers coupled with flow cytometric fluorescence-activated (FACS) or magnetic-activated cell sorting (MACS) selection and 2) genetic manipulation to introduce selectable markers and/or therapeutic genes.
As examples, Li et al. (210) employed a drug-resistance gene under the control of a lineage-specific promoter. In this "lineage selection" experiment using mES cells, a neomycin cassette was targeted to the neuron-specific SOX2 gene. After selection with neomycin (G418), only those cells expressing the neomycin gene under the control of the SOX2 promoter survived, resulting in the development of an apparently pure population of neuroepithelial cells, which subsequently differentiated into neuronal networks. Similar strategies have been employed for the isolation of skeletal muscle cells using MyoD as a target gene (95). For the selection of cardiac cells (from a low yield of 3–5% cardiomyocytes in ES-derived populations), targeting of the cardiac 
-MHC gene promoter has yielded populations consisting of 99.6% cardiac myocytes (192). Recently, a lineage selection strategy combined with specific culture conditions was successfully employed to generate a neural progenitor population of high purity (15).
FACS sorting of cells expressing enhanced green fluorescent protein (EGFP) offers an alternative (and substitute) to drug selection and has been used to isolate cardiac myocytes from D3 ES cells expressing the EGFP under the control of the cardiac -actin promoter (193). A similar strategy has proven successful for the isolation of ventricular cells following targeting of the MLC-2v promoter by ECFP (enhanced cyan fluorescent protein) and EGFP (229, 237). The direct sorting of differentiated cells using fluorescent antibodies and magnetic microbead-tagged antibodies by MACS is especially feasible for cell types, which express defined surface antigens, as is the case for cells of the hematopoietic lineages (159).
Because no single drug-resistance or fluorescence-based enrichment procedure generates a 100% pure population of cells, it may prove useful to combine the two using antibiotic resistance and EGFP expression (by FACS or MACS) together with cultivation in specific growth factors as done by Marchetti et al. (220). Attempts are underway to test similar selection systems with hES cells.
D. Tissue-Specific Integration and Function
One of the critical questions concerning the potential therapeutic use of ES-derived cells is whether cells produced by a particular in vitro differentiation protocol can integrate into the recipient tissue and fulfil the specific functions of lost or injured cells. This seems to be possible for at least some mES-derived progeny, since a degree of specific function has been reported following transplantation (Table 6, see sect. IX). In pilot experiments designed to analyze the potential of human ES-derived neural progenitor cells to integrate into the developing brain, the transplanted cells integrated into the developing nervous system of mice (292, 415). Similarly, colonies of hES cells have been grafted directly adjacent to the host neural tube of chick embryos. These cells subsequently differentiate into primary structures with morphologies and molecular characteristics typical of neural rosettes and differentiated neurons (139). Although it is too early to conclude normal, full, or protracted functioning of transplants derived from hES cells, these earliest findings are clearly encouraging, but extensive experimentation in large-animal models will be required before application in humans.
E. Immunogenicity and Graft Rejection
One major problem potentially associated with the use of hES-derived cells for tissue regeneration is the immunological (in)compatibility between donors and recipients. Clearly, uncontrolled immune reactions would lead to rejection of mismatched grafts. Although the levels of MHC-I expression on hES cells are low, they increase moderately after differentiation either in vitro or in vivo, and markedly following interferon treatment (101). The absence of MHC molecules may also lead to natural killer cell rejection of the transplanted cells. Several approaches to reduce or eliminate ES-derived graft rejection have therefore been proposed.
1) One could reduce the host reactivity to allogeneic ES-derived transplants by classic immunosuppression, as is routinely employed for organ transplantation (132). Unfortunately, most of the immunosuppressive drugs currently used are associated with complications, including opportunistic infections, drug-related toxicities, skin malignancies, and posttransplantation lymphoproliferative disorders. A more specific suppression of immune rejection may be achieved by the cotransplantation of both therapeutic tissue and hematopoietic stem cells generated from the same parental ES cell line (see Ref. 252) or by preimmunization of recipients with preimplantation-stage stem cells, as has been recently reported to induce long-term allogeneic graft acceptance (110).
2) A tempting alternative to suppressing the immune rejection would be to avoid it completely by eliminating the genes responsible. The first report of successful homologous recombination in hES cells is an important step towards the generation of genetically modified ES cells for transplantations (419). One possibility is that the elimination of major histocompatibility complex (MHC) class I expression in hES cells may generate a "universal cell" that would be suitable for all patients (41, 101). Homologous recombination has been used to "knock out" MHC class I and class II molecules in mES cells; however, the consequences of such extensive gene targeting are difficult to assess (144). Additionally, loss of the MHC class I and class II molecules do not necessarily protect against rejection, because of indirect allo-recognition-mediated rejection and/or natural killer cell-mediated cell destruction.
3) Another option relies on the generation and storage of HLA-isotyped and/or genetically manipulated hES cell lines in a cell bank. Only humans with similar HLA molecules could be donors for other hosts. Practically this would require determination of allogeneic compatibility. For ES cells derived from one human individual, all HLA molecules would be clonal. As such, banks of ES cells with known HLA backgrounds could be established. According to some calculations based on organ transplantation data, a minimum of 200 or more ES cell lines generated from independent HLA subtypes would be required. The requirement of isolating multiple pure populations of ES cells with defined HLA molecules represents an enormous amount of work, may be unattainable, and under current law, i.e., in the United States, could only be performed in the private sector.
4) The fourth principal possibility relies on the generation of autologous donor cells through a process known as "therapeutic cloning" (201; see sect. IXB), which, in principle, follows the strategy used to create the sheep Dolly (392). In the therapeutic cloning approach, somatic cell nuclei of the patient would be fused to enucleated human eggs, which in vitro would be cultivated into blastocysts. From these cells, hES cell lines would be established and differentiated into the desired cell types for transplantation (201). Recently, two South Korean groups demonstrated the proof of principle for this strategy (165) (see sect. IXB). Such cells should be immunologically compatible, because they contain (except in the mitochondrial genome) the same genetic information as the patient. However, it is evident that the unlimited use of human oocytes for the generation of autologous donor cells would generate numerous ethical and legal problems (252; see also sects. IX and XI).
Answers to all your Biology Questions
