A challenge for regenerative medicine: Proper genetic programming, not cellular mimicry


  • Angie Rizzino

    Corresponding author
    1. Eppley Institute for Research in Cancer and Allied Diseases, Department of Pathology and Microbiology, and Department of Biochemistry and Molecular Biology, University of Nebraska Medical Center, Nebraska Medical Center, Omaha, Nebraska
    • Eppley Institute for Research in Cancer and Allied Diseases, Department of Pathology and Microbiology, and Department of Biochemistry and Molecular Biology, University of Nebraska Medical Center, 986805 Nebraska Medical Center, Omaha, NE 68198-6805
    Search for more papers by this author


Recent progress in stem cell biology and the reprogramming of somatic cells to a pluripotent phenotype has generated a new wave of excitement in regenerative medicine. Nonetheless, efforts aimed at understanding transdifferentiation, dedifferentiation, and the plasticity of cells, as well as the ability of somatic cells to be reprogrammed, has raised as many questions as those that have been answered. This review proffers the argument that many reports of transdifferentiation, dedifferentiation, and unexpected stem cell plasticity may be due to aberrant processes that lead to cellular look-alikes (cellular mimicry). In most cases, cellular look-alikes can now be identified readily by monitoring gene expression profiles, as well as epigenetic modifications of DNA and histone proteins of the cells involved. This review further argues that progress in regenerative medicine will be significantly hampered by failing to address the issue of cellular look-alikes. Developmental Dynamics 236:3199–3207, 2007. © 2007 Wiley-Liss, Inc.


There are growing expectations that regenerative medicine will begin to provide new and effective cures for some of the most debilitating human diseases in the not too distant future. Ideally, regenerative medicine will evolve to the point where it will be possible to treat many age-related disorders, such as diabetes, Alzheimer's, and heart disease, by reinvigorating and even enhancing the regenerative processes involved in normal tissue repair. However, in cases where the damage is too severe, for example, after traumatic injury or advanced disease, cell and organ transplants are likely to be the best option in the foreseeable future. Although the frequency and success of cell and organ transplantation have improved dramatically over the past 50 years, each year the number of people waiting for transplants is far greater than those receiving them. This problem has led to a significant increase in efforts to develop new cell replacement therapies.

Advances in cell replacement therapy will require overcoming several major hurdles, in particular the identification of cells best suited for effective treatments. Efforts to surmount these problems have focused heavily on the isolation and characterization of stem cells from virtually every adult, fetal, and embryonic source. Although considerable progress has been made in stem cell biology over the past 20 years, the study of stem cells has generated at least as many questions as those that have been answered. Major unanswered questions include: the roles of stem cells in diseases such as cancer, the status of stem cells in the adult, the ability to reprogram cells to an earlier stem cell state and, most confusing and controversial of all, the degree of adult stem cell plasticity. This review focuses primarily on recent efforts aimed at understanding the plasticity of stem cells and the ability of somatic cells to be reprogrammed. Importantly, this review argues that many reports of transdifferentiation, dedifferentiation, and unexpected stem cell plasticity may be due to aberrant processes that lead to cellular look-alikes (cellular mimicry).


EC embryonal carcinoma ES embryonic stem HSC hematopoietic stem cell iPS induced pluripotent stem


Over the past 4 decades, stem cells have been isolated from a wide range of sources. Hematopoietic stem cells (HSC), embryonic stem (ES) cells, and their tumorigenic counterparts, embryonal carcinoma (EC) cells, were among the first stem cells to be isolated and characterized extensively. Subsequently, a wide range of stem cells, including meschenymal stem cells (Prockop,1997), neural stem cells (Reynolds and Weiss,1992), and amniotic fluid stem cells (De Coppi et al.,2007), were isolated from embryonic and adult tissues. Moreover, numerous reports have appeared within the past 10 years describing the isolation of stem cells from tumors of the breast (Al-Hajj et al.,2003), brain (Singh et al.,2003), prostate (Miki et al.,2007), and pancreas (Li C. et al.,2007), as well as several types of leukemia (Lapidot et al.,1994; Jamieson et al.,2004). Although work with acute myeloid leukemia in the 1990s (Lapidot et al.,1994) is often cited as the first definitive proof of tumor stem cells, EC cells were shown by single cell transplantation studies to be the stem cells of teratocarcinomas in the 1960s (Kleinsmith and Pierce,1964).

As our understanding of stem cells has grown, the definition of a stem cell has become increasingly complex. In general, it is accepted that stem cells have two main properties. First, and foremost, they have the ability to undergo self-renewal, that is, to generate at least one undifferentiated daughter cell when they divide. Second, stem cells have the ability to differentiate into one or more cell types, depending on their developmental potential. Totipotent stem cells have the ability to form any embryonic or extraembryonic cell type, including germ cells, whereas pluripotent stem cells, which can differentiate into cells from each of the three embryonic germ layers, and multipotent stem cells, which are able to form multiple organ specific cell types, are progressively restricted in their potential to differentiate. Others have argued that the definition of stem cells must include the ability of stem cells to reconstitute the relevant tissue functionally when transplanted into a suitable host (Verfaille,2002). This third criterion is especially relevant when discussing stem cells in the context of regenerative medicine.

Lastly, our concept of stem cells should account for how stem cells are affected by their cellular environment, i.e., the effect of the niche (anatomical site) on the ability of stem cells to self-renew and differentiate. Although the importance of the stem cell niche is now widely recognized (Scadden,2006), the extent to which the niche can affect the differentiation of stem cells has generated a major scientific controversy that is unlikely to be resolved any time soon. This debate is centered on the plasticity of stem cell differentiation, in particular on studies which have raised the possibility that multipotent stem cells, such as HSC, can differentiate into a much wider range of cells than previously thought. Although there can be little doubt that the niche of stem cells influences their differentiation, the degree to which different niches can direct differentiation is far from clear.

The problem surrounding stem cell plasticity is compounded by its definition. Plasticity is a general term that relates to the capacity of a cell from one lineage to convert to a cell type of another developmental lineage. Importantly, no specific mechanisms or cellular processes are inferred by this definition. Although it is suggested, it is not typically stated explicitly that the conversion of one cell into another should be complete. This review focuses on the issue of complete conversion in an attempt to provide a fresh perspective on the matter of stem cell plasticity. Specifically, this review argues that some, perhaps many, reports of unexpected plasticity are due to the presence of mosaic cells that form when migrating stem cells respond, albeit aberrantly and incompletely, to physiological cues found in niches other than their own (Fig. 1). More specifically, it is proposed that the propensity of stem cells to differentiate enables them to differentiate abnormally into cells that exhibit some, but not all, of the properties of the differentiated cells normally produced by another lineage. We refer to this process as “cellular mimicry” (Fig. 1). It quacks like a duck, but “it's a bird of a different feather.” Importantly, as discussed below, the possible contribution of cellular mimicry needs to be addressed in model systems being used to study regeneration, stem cell plasticity, and somatic cell reprogramming.

Figure 1.

Cellular mimicry generated due to stem cell A inappropriately localized to niche B. Stem cell A (black oval) in niche A (black arch) gives rise to cell type A (black rectangle). Stem cell B (white oval) in niche B (white arch) gives rise to cell type B (white triangle). Stem cell A (black oval) misplaced in niche B (white arch) gives rise to a cellular mosaic (black isosceles trapezoid).


Work with amphibians, in particular limb regeneration in salamanders, has captivated the interest of developmental biologists for several generations. Many have hoped that understanding regeneration in lower vertebrates will someday help transform regenerative medicine from an experimental science into a clinical branch of medicine able to offer treatments for a wide range of human disorders. Thus far, our understanding of the mechanisms responsible for regeneration is, at best, rudimentary. Even tissue regeneration, which takes place in the skin and the gut as part of the normal repair processes of the body, is poorly understood. Not surprisingly, the mechanisms responsible for regeneration of whole body parts (epimorphic regeneration), such as limb regeneration in lower vertebrates, are an even bigger mystery. Efforts to understand epimorphic regeneration have led many to conclude that the processes of dedifferentiation, transdifferentiation, and pattern formation are involved (Tsonis,2007). However, due to the complexity of the events involved, much of the published work has yet to provide a clear picture of either the cellular processes involved or the molecular mechanisms responsible for complex regeneration. Consequently, the use of terms such as transdifferentiation and dedifferentiation to describe these processes provides a seemingly simple description for a very complex set of processes.

In its simplest form, transdifferentiation is the “transition from one differentiated state to another” (Selman and Kafatos,1974). More specifically, transdifferentiation is the transformation of a cell type from one lineage to a differentiated cell of another lineage without reverting to a developmentally more primitive cell type (Fig. 2). In contrast to transdifferentiation, dedifferentiation is the conversion of a differentiated cell to a more primitive cell, which in theory should have the potential to differentiate into other cell types. As such, the dedifferentiated cell is transformed from a more differentiated state to an embryonic or progenitor state (Fig. 2). Of interest, many have argued for decades that tumors arise by dedifferentiation. However, there is little conclusive data to support this contention. With the reemergence of the stem cell theory of cancer, there is growing recognition that stem cells involved in normal tissue repair are likely to be the primary origin of cancer stem cells (Rizzino,1993; Song and Miele,2007).

Figure 2.

Schematic representation of stem cell self-renewal, differentiation, dedifferentiation, and transdifferentiation. Stem cells (black oval) can self-renew or differentiate into progenitors (black or white rectangles), which then differentiate into committed cells (black triangle or white octagon). Transdifferentiation between committed cells is considered to be very unlikely; whereas, transdifferentiation between progenitors may be possible. Finally, dedifferentiation of a committed cell to a much earlier progenitor is questionable.

The debate over the extent to which cells can transdifferentiate and dedifferentiate heavily influences how we approach regenerative medicine. Clearly, it could have enormous impact in a clinical setting if one could harness the purported transdifferentiation and/or dedifferentiation of adult cells. There is a large body of research, especially from work with lower vertebrates and Drosophila, that argues strongly that transdifferentiation and dedifferentiation are not rare events in nature. Many excellent reviews cover different aspects of this fascinating field (Slack and Tosh,2001; Brockes and Kumar,2005; Alvarado and Tsonis,2006; Tsonis,2007). Thus, this topic will not be discussed in depth here. However, many of these studies have left important questions unanswered. One common concern about the interpretation of the events surrounding regeneration in many of the model systems is the possible, if not likely, contribution of multipotent or even pluripotent stem cells. More specifically, are some of the critical changes being observed due to the presence of multipotent stem cells rather than transdifferentiation and/or dedifferentiation? Similarly, it is unclear whether the cellular transitions reported to be due to transdifferentiation are in fact due to dedifferentiation. Attempts to use in vitro model systems that are more amenable to experimental investigation have raised an equally vexing question concerning the degree to which the events modeled in culture faithfully reflect those that occur in animal models, where function is more readily assessed. In many of the simpler models, especially those where transdifferentiation is being studied, relatively few markers have been used to monitor the cellular events taking place (see below). Consequently, the use of only a small number of markers leaves open the obvious question of whether the transitions observed represent transdifferentiation or whether they reflect mosaic cells that possess critical properties of both the original cells and the “transdifferentiated” cells (cellular mimicry). Thus, in cases where only a few markers are examined, the findings reported should only be considered preliminary. As discussed in the next section, the same concern applies to much of the research focused on the plasticity of mammalian stem cells after transplantation into animal models.


During the past 10 years, a great deal of excitement in the field of regenerative medicine has been generated by high profile reports claiming that stem cells isolated from several different sources could each form a range of differentiated cells not previously thought to be generated by the stem cell population being transplanted. For example, there were reports in the late 1990s that genetically marked skeletal muscle and neural stem cells each possess the ability to form cells of the hematopoietic lineage (Jackson et al.,1999; Bjornson et al.,1999). Other studies described the formation of cells that exhibited liver, gastrointestinal (GI) tract, and lung phenotypes from a single HSC (Krause et al.,2001). Collectively, these and other studies raised the expectation that stem cells, in particular the HSC of the bone marrow, could differentiate into cells normally derived from each of the three embryonic germ layers when exposed to the right conditions. Moreover, many began to argue (and many still do) that lineage-restricted stem cells could be induced to transdifferentiate when exposed to different physiological signals. Curiously, dedifferentiation to a more primitive stem cell is usually not cited as a possible mechanism responsible for these unexpected findings.

During the past 5 years, many studies have challenged the main conclusions of these studies and provided experimental findings questioning the belief that stem cells possess far more plasticity than previously thought. The experimental studies challenging the extensive plasticity of stem cells will not be described here in detail, because they have been the subject of many excellent reviews (Verfaille,2002; Bonnet,2003; Vassilopoulos et al.,2003; Moore and Quesenberry,2003; Pauwelyn and Verfaillie,2006; Stocum,2006). In general, these studies and their accompanying review articles have provided several alternative explanations for the apparent plasticity of transplanted stem cells (Fig. 3). One explanation that is not widely accepted, but nearly impossible to dismiss, is that the bone marrow and other sources of stem cells contain pluripotent stem cells that have the capacity to form nearly any cell type of the body under the right set of conditions. If the cells of this pluripotent population were free to migrate to areas of tissue damage, they could be induced by physiological cues appropriate to that niche to differentiate into the very cells needed for repair. Alternatively, bone marrow and other sources of stem cells may contain several different multipotent stem cells, which when transplanted together could give rise to cells from more than one developmental lineage. For example, one study, from the laboratory that originally reported the hematopoietic potential of skeletal muscle (Jackson et al.,1999), subsequently concluded that primitive HSC are present in muscle tissue (McKinney-Freeman et al.,2002). Attempts to directly address the possible presence of more than one stem cell in the population as the root cause for the extensive plasticity observed have involved transplanting single cells. However, these studies have not fully resolved the issue. As mentioned above, one study reported the formation of epithelial cells from the liver, GI tract, and skin after transplantation of a single HSC (Krause et al.,2001). To the contrary, others reported that HSC lack robust capacity to form nonhematopoietic cells types in vivo (Wagers et al.,2002; Lechner et al.,2004; Murry et al.,2004). In one study, where a single HSC was found to reconstitute peripheral blood leukocytes in lethally irradiated mice, only a minuscule number of genetically marked Purkinje-like and hepatocyte-like cells were found (Wagers et al.,2002). However, due to differences in the experimental design of these studies, and the difficulties of ensuring that the correct cell has been isolated and not damaged by manipulation before transplantation, some have argued that this issue remains an open question (Moore and Quesenberry,2003).

Figure 3.

Explanations for higher than expected stem cell plasticity.

Other studies have provided compelling evidence that apparent stem cell plasticity may be due, at least in part, to cell fusion (Vassilopoulos et al.,2003; Wang et al.,2003). These studies argued that the apparent formation of liver cells from transplanted bone marrow was due to formation of polyploid cells generated from the fusion of the transplanted cells with host hepatocytes. Although cell fusion is unlikely to occur at high frequency after cell transplantation, the formation of nonhematopoietic tissues from bone marrow is a very rare event, if it occurs at all (Lechner et al.,2004).

An equally likely explanation for the apparent extensive plasticity of stem cells is the formation of cellular mosaics (cellular mimicry) due to aberrant responses of lineage-restricted stem cells to inappropriate physiological signals (Fig. 1). Although a cellular mosaic can be formed by cell fusion, this is by no means the only way a cellular mosaic can be generated (see below). In theory, there are several ways to determine whether cellular mimicry is a common occurrence. Fuller characterization of the cells in question by extensive gene profiling of expressed RNA and/or extensive epigenetic analysis could help resolve this issue. More importantly, the demonstration of function, especially in response to a range of appropriate physiological signals, would be necessary to argue that the cells formed are not cellular look-alikes. Thus far, the rarity of “transdifferentiated” events has precluded extensive characterization of the unexpected differentiated cells that form during cell transplantation studies. To make matters worse, this problem could be compounded by heterogeneity in the differentiated population of cells. If a range of different markers are differentially expressed throughout the population, it could make it appear as if the “transdifferentiated cells” express the full range of expected markers.


Gene profiling studies have shown that, at each step along a given differentiation pathway, there are significant changes in gene expression, both qualitative and quantitative. Thus, as development proceeds through numerous developmental transitions in the course of progressing from totipotent, to pluripotent, to lineage-specific multipotent cells, there must be radical reprogramming of cells and their gene expression profiles. As a result, the batteries of genes expressed by different lineage-specific multipotent stem cells are expected to be substantially different from one another. In fact, comparative gene profiling between mouse ES cells, HSC, and neural stem cells demonstrated that their gene expression profiles are radically different. Fewer than 300 genes (approximately 5% of the genes examined) were enriched in all three populations and less than 15% of the genes were enriched in both the neural and hematopoietic cell populations (Ramalho-Santos,2002; Ivanova et al.,2002). Importantly, their gene expression profiles are not only radically different, but the genes themselves are differentially modified by highly complex combinations of epigenetic modifications (DNA methylation and histone modification), which lead to major changes in chromatin structure (Li B. et al.,2007; Reik,2007). As a result, the expression of genes in a given lineage is subject to multiple layers of regulation. While the majority of genes of a given lineage are strongly silenced, there is another group of inactive genes that is poised for expression when exposed to the signals that direct the proper function of the cells of that lineage (Lee et al.,2006). Importantly, one can expect that the genes in this inactive group differ between lineages. When viewed in this context, it would be a major challenge for lineage-specific stem cells to transdifferentiate fully and properly reprogram critical gene regulatory networks that were modified extensively over multiple developmental transitions. Hence, the reprogrammed cells would not be expected to regain all the cell surface receptors and signaling pathways required to respond properly to each of the niche signals that dictate the fate of organ-specific stem cells. This leads to the inevitable conclusion that transdifferentiation into cells of distant developmental lineages is unlikely to be complete or to occur at high frequency. Nonetheless, it would not be surprising to find that migrating stem cells that become lodged in a niche other than their own could differentiate into cells that express some, but not all, of the markers characteristic of that tissue (cellular mimicry). Importantly, in the vast majority of cases, the differentiated cells that form may exhibit a subset of the markers normally expressed by cells produced by a given niche, but they would not be able to perform the full spectrum of activities required of cells normally derived from that niche.

The molecular argument above does not mean to suggest that transdifferentiation or dedifferentiation (see below) cannot occur in nature. Transdifferentiation and dedifferentiation between closely related developmental lineages may require altering the expression of only one or two critical genes. For example, this may be true between progenitors of the liver and pancreas, which arise from adjacent regions of endoderm during development. Given their common developmental origin, it is not surprising that treatment of a pancreatic tumor cell line with high doses of dexamethasone in conjunction with oncostatin M promotes the loss of pancreatic markers (amylase) and the expression of liver-specific markers (albumin; Shen et al.,2000). Although it is unclear whether the effects of dexamethasone and oncostatin M in this cell culture model are reversible, essentially the same result was observed when the cells were transfected with an expression vector for the transcription factor C/EBPβ, which is believed to be a master switch controlling pancreatic and hepatic cell fates during development. The converse is also true. Forced expression of a chimeric form of the transcription factor Pdx1, which is required for pancreatic development, has been reported to convert liver cells, both in vitro and in vivo, into cells that exhibit pancreatic markers (Horb et al.,2003; Li et al.,2005). Although these studies make a compelling argument for transdifferentiation, insufficient markers were examined to rule out the possibility that cellular mosaics had formed, rather than fully transdifferentiated cells.

Similar arguments apply to dedifferentiation. Full dedifferentiation of a lineage-restricted stem cell into a developmentally distant stem cell that is several steps removed would require extensive reprogramming of many gene regulatory networks. In contrast, dedifferentiation to a stem cell state only one step removed from the original cell could occur by the gain or loss of one or two critical transcription factors. Such a conclusion is supported by studies involving the conversion between B- and T-cell progenitors by the transcription factor Pax5. Like C/EBPβ and Pdx1, Pax5 functions as a master regulator and directs the commitment of lymphoid progenitors to the B lymphocytic lineage. In this case, transplantation of RAG2-deficient mice with Pax5-null pro-B cells leads to long-term reconstitution of the thymus and the formation of mature T cells that express α/β-T-cell receptors (Rolink et al.,1999). Taken together, these studies lead one to conclude that, as the developmental distance between lineages increases, the probability of full interconversion/reprogramming (transdifferentiation or dedifferentiation) between these lineages declines precipitously (Fig. 4). Given this argument, and the points raised in the previous section about cellular mimicry and plasticity, further study will be needed to determine the validity of reports claiming that bone marrow cells in experimental settings are able to form gastric tumors (Houghton et al.,2004; Pellicano et al.,2006) or regenerate oocytes (Johnson et al.,2005; Tilly and Johnson,2007).

Figure 4.

Schematic description of the likelihood of transdifferentiation and dedifferentiation occurring between cells at different stages of development. TD, transdifferentiation.


After the cloning of Dolly, several new strategies for reprogramming of cells have been proposed or tested (Fig. 5). Readers interested in this topic are referred to a recent review (Hochedlinger and Jaenisch,2006). The most successful in vitro reprogramming method involves the generation of ES cells from early embryos that had been produced by somatic cell nuclear transfer into enucleated oocytes (Trounson,2005). However, this process, referred to as therapeutic cloning, has been challenged on ethical grounds. A variation on this theme involving the transfer of a somatic nucleus into an enucleated ES cell has been proposed but not tested (Rizzino,2002). Although testing this approach would have several technical difficulties due to the large nuclear to cytoplasmic ratio of ES cells, cell fusion studies argue strongly that human ES cells do, in fact, have significant reprogramming machinery (Cowan et al.,2005). Another reprogramming strategy being tested involves incubating enzymatically permeabilized cells with nuclear extracts from oocytes or pluripotent cells (Taranger et al.,2005).

Figure 5.

Experimental approaches for reprogramming of somatic cells.

Recently, a new and exciting approach to nuclear reprogramming has been used (Takahashi and Yamanaka,2006; Okita et al.,2007; Wernig et al.,2007; Maherali et al.,2007). In these studies, pluripotent cells were produced after as few as four genes (Oct-3/4, Sox2, c-Myc, and Klf4) were introduced retrovirally into either embryonic or adult mouse fibroblasts. The induced pluripotent cells (iPS), like ES cells, contributed to each of the three embryonic germ layers after injection into mouse blastocysts and subsequent development in vivo. In the initial study, reactivation of the endogenous Fbx15 gene was used to identify and select iPS cells (Takahashi and Yamanaka,2006). The Fbx15 gene is a nonessential gene, which is expressed by ES cells, but repressed after ES cells undergo differentiation (Tokuzawa et al.,2003). Despite the similarities between ES cells and Fbx15 reactivated iPS cells (Fbx15-iPS), they were far from identical. Fbx15-iPS cells were shown to be pluripotent, but they did not produce germline chimeric mice (Takahashi and Yamanaka,2006). Furthermore, the endogenous Oct-3/4 gene, which is essential for early development and for the pluripotency of ES cells, was not properly reprogrammed in Fbx15-iPS cells (Takahashi and Yamanaka,2006). More specifically, the Oct-3/4 promoter in these cells remained methylated, unlike the Oct-3/4 promoter in ES cells, which is methylated only after ES cells differentiate and the gene is silenced (Gu et al.,2006). Other important developmentally regulated genes, such as Nanog, were expressed at lower levels in Fbx15-iPS cells than in ES cells, at least in the majority of the four-gene iPS cell clones tested. Moreover, other genes, such as Ecat1 and Sox2, were not expressed or were expressed at much lower levels in Fbx15-iPS cells than in ES cells. Thus, Fbx15-iPS cells share many properties in common with mouse ES cells, and yet they are distinctly different from ES cells (cellular look-alikes).

More recently, three other studies described the production of iPS cells that appear to be remarkably similar to mouse ES cells. Instead of using reactivation of the endogenous Fbx15 gene to identify and select iPS cells, reactivation of the endogenous Nanog or Oct-3/4 gene was used. Although there appear to be some subtle differences between Nanog-iPS cells and Oct-3/4-iPS cells, both Nanog-iPS cells and Oct-3/4-iPS cells were capable of contributing to the germline (Okita et al.,2007; Wernig et al.,2007; Maherali et al.,2007). Arguably, contribution to the germline is the most rigorous test for reprogrammed somatic cells. Thus, it was not surprising that Nanog-iPS cells were shown to have undergone extensive epigenetic reprogramming to an ES-like stem cell state. Not only were the endogenous Oct-3/4, Sox2, and Nanog genes in these iPS cells expressed at levels similar to their endogenous counterparts in mouse ES cells, the methylation status of the Nanog and Oct-3/4 genes in the Nanog-iPS cells was virtually identical to the methylation status of these genes in mouse ES cells (Okita et al.,2007; Maherali et al.,2007). Even more impressive, ChIP-on-chip analysis (chromatin immunoprecipitation followed by DNA hybridization to a mouse promoter microarray), which was used to compare histone modifications at specific gene loci in iPS cells and ES cells, indicated that Nanog-iPS cells are nearly identical to mouse ES cells (Maherali et al.,2007). Specifically, the pattern of histone 3 K4 and K27 trimethylation at a set of signature genes (957 genes whose methylation patterns were significantly different between ES cells and MEFs) were virtually identical at nearly all of the genes (94.4%) when Nanog-iPS cells and ES cells were compared; whereas, less than 1% of the signature genes exhibited a MEF-like methylation pattern in Nanog-iPS cells. Together, these results argue strongly that judicious selection of a set of reprogramming genes, specifically Oct-3/4, Sox2, c-Myc, and Klf4 genes in the case of mouse cells, can convert somatic cells into a pluripotent stem cell state nearly equivalent to ES cells. However, the method for producing iPS cells has room for improvement. One of the studies reported that approximately 20% of the adult chimeric mice derived using donor Nanog-iPS cells developed tumors, and this appears to be due, at least in part, to the reactivation of the c-Myc transgene (Okita et al.,2007). Despite this last caveat, the development of iPS cells establishes a very promising strategy for producing reprogrammed, pluripotent stem cells. As discussed in the Perspectives section, this strategy may eventually prove very useful for future cell replacement therapies and may help mitigate the controversy surrounding therapeutic cloning. However, for this to become a reality, a different approach will be needed for inducing the expression of “reprogramming genes.”


The major premise put forward in this review argues that studies involving regeneration, cell plasticity, and reprogramming, if not properly analyzed, may lead one to conclude that cellular transitions between distant developmental lineages have occurred, when in fact only cellular look-alikes have formed. For the field of regenerative medicine to move forward, it will be important that conclusions regarding transdifferentiation, dedifferentiation, and reprogramming be supported by extensive characterization of both the starting and ending populations. At each step, conclusions drawn need to be carefully vetted. Technologies such as gene profiling, epigenetic analysis, and proteomics should be used wherever possible. However, as progress is made and we begin to consider moving cellular therapies into the clinic, even more rigorous criteria will be needed. If this is not done, clinical disasters could result from the use of cells that have not been thoroughly characterized.

Important differences noted above between ES cells and Fbx15-iPS cells make two relevant points. First, gene profiling and epigenetic analysis are needed to determine the degree to which cells have been reprogrammed. Detailed investigation demonstrated that Fbx15-iPS cells were not fully reprogrammed to an ES-stem cell state and that a refinement in the experimental design was needed. Second, there is growing recognition that the expression of master regulators, such as Oct-3/4 and Sox2 in ES cells, must be expressed at the correct levels to properly control the fate of cells. For example, less than a twofold increase in levels of Oct-3/4 radically alters the cell fate of ES cells by inducing their differentiation (Niwa et al.,2000). Similar increases in Sox2 in ES cells are enough to disrupt a gene regulatory network that heavily influences the self-renewal and pluripotency of ES cells (Boer et al.,2007). Thus, efforts to produce cells for cell replacement therapy by genetically reprogramming somatic cells will not only need to express the correct master regulators, but also ensure that they are expressed at the correct levels. This will help ensure that the reprogrammed cells can respond appropriately to physiological signals. Clearly, careful examination of cell reprogramming at the cellular and molecular levels will enhance the prospects of success when functional studies are undertaken in vivo.

It is evident that the results generated thus far with iPS cells are truly impressive and offer the hope of eventually being able to produce genetically matched pluripotent cells by reprogramming somatic cells isolated from patients in need of cell replacement therapy. Aside from the cost involved, we will need to overcome many difficult hurdles before this becomes a reality. First and foremost, it remains to be determined whether human iPS cells can be produced. Assuming human somatic cells can be reprogrammed like their mouse counterparts, it remains to be determined whether the same four genes (Oct-3/4, Sox2, c-Myc, and Klf4) will properly reprogram human cells. Moreover, reprogramming of mouse somatic cells to iPS cells is very inefficient (<0.1%; Takahashi and Yamanaka,2006; Okita et al.,2007; Wernig et al.,2007; Maherali et al.,2007). Perhaps other cells, such as HSC, may prove to be a better starting population. Alternatively, there may be a subset of cells within the population of cells used thus far (e.g., a minor stem cell population) that is far more amenable to reprogramming. However, given the molecular arguments discussed earlier regarding the unlikely prospect of efficient transdifferentiation and dedifferentiation, increasing the efficiency of somatic cell reprogramming is unlikely to be achieved easily. Another major obstacle that will need to be addressed will be to determine whether any human iPS cells produced have been reprogrammed properly. Studies used to test the properties of mouse iPS cells (participation in normal development and contribution to the germline) cannot be used to test human iPS cells. Consequently, different tests will be needed for adequate characterization of human iPS cells.

Even if proper reprogramming proves to be highly efficient with human cells, one should not expect miracle cures from cell replacement therapies in the immediate future. Learning how to reprogram somatic cells or possibly identifying stem cells that do not need to be reprogrammed will only be the first step. Aside from ensuring that all pluripotent cells have been eliminated from the transplanted population (Carson et al.,2006), we will also need to determine how to instruct the reprogrammed cells to form the cell types needed for transplantation. Although significant progress has been made in learning how to direct the differentiation of mouse and human ES cells, few of the differentiated cell types produced in vitro have achieved a high degree of function after transplantation. Given the premise of this review, one should consider whether many, if not most, of the differentiated cells generated in vitro are, in fact, cellular look-alikes that will be unable to respond to the full range of physiological cues needed for proper function in vivo. As this challenge is met, we will also need to learn how to properly transfer donor cells to the host organ (the correct niche) where they can work their magic.

Finally, if human cells can be adequately reprogrammed, or if we can identify a ready source of stem cells in the adult, this should also make it possible to produce genetically matched cells that, in principle, should not be rejected after transplantation. Importantly, this would help negate some of the current ethical issues surrounding the use of human ES cells. Nonetheless, success in cellular reprogramming is unlikely to defuse this ethical debate. To determine whether reprogramming to a pluripotent stem cell state has occurred, one needs a proper point of reference. At the current time, human ES cells are considered the gold standard. Some would argue that human amniotic fluid stem cells (De Coppi et al.,2007) could serve as a suitable reference point. However, such a conclusion would warrant careful comparisons between human ES cells and human amniotic fluid stem cells. Ironically, concerns over the use of human ES cells are not only slowing progress in regenerative medicine, but they also appear to be hampering progress on the very technologies that could eliminate their use in cell replacement therapy.


Timothy McKeithan is thanked for reading this manuscript and for numerous discussions concerning its content and arguments put forward in this review. Janel Kopp is also thanked for reading this manuscript. Michelle Desler is thanked for computer assisted construction of all figures. Work in the Rizzino laboratory is supported by a grant from the National Institutes of General Medical Sciences.