The elite and stochastic model for iPS cell generation: Multilineage-differentiating stress enduring (Muse) cells are readily reprogrammable into iPS cells


  • Shohei Wakao,

    1. Department of Stem Cell Biology and Histology, Tohoku University Graduate School of Medicine, Sendai 980-8575, Japan
    Search for more papers by this author
  • Masaaki Kitada,

    1. Department of Stem Cell Biology and Histology, Tohoku University Graduate School of Medicine, Sendai 980-8575, Japan
    Search for more papers by this author
  • Mari Dezawa

    Corresponding author
    1. Department of Stem Cell Biology and Histology, Tohoku University Graduate School of Medicine, Sendai 980-8575, Japan
    2. Department of Anatomy and Anthropology, Tohoku University Graduate School of Medicine, Sendai 980-8575, Japan
    • Department of Stem Cell Biology and Histology, Tohoku University Graduate School of Medicine, Sendai, Miyagi 980-8575, Japan
    Search for more papers by this author


Induced pluripotent stem (iPS) cells have attracted a great deal of attention, although the mechanism by which they are generated is still not fully understood. Currently, two theories, the stochastic and elite models, have been proposed. Some reports provide theoretical support for the stochastic model. Other reports, however, support the elite model. For example, some human fibroblasts, such as Multilineage-differentiating stress enduring (Muse) cells, are reported to be pluripotent and a primary source of iPS cells. Thus, the mechanism of iPS cell generation continues to be debated. In this review, we discuss the properties of the original cell source, such as the components of the original populations and the potential of each population to become iPS cells, and further discuss the implications of the two theories for iPS cell research. © 2012 International Society for Advancement of Cytometry

The generation of artificially made pluripotent stem cells, induced pluripotent stem (iPS) cells, was first reported in mouse fibroblasts by introducing exogenous Oct3/4, Sox2, Klf4, and c-Myc (the so-called four Yamanaka factors) (1). Since then, they have attracted a great deal of attention for their potential in medical research and clinical applications. Embryonic stem (ES) cells, well known pluripotent stem cells, have been studied for long time but ethical concerns relating to the use of fertilized eggs posed limit in their practical application. As iPS cells can be generated from somatic cells, they can avoid ethical concerns (2), and thus they are expected to be applied to cell therapy.

The character of iPS cells is similar to those of ES cells. Both express pluripotency markers, self-renew, differentiate into cells representative of all three germ layers, show unlimited proliferative activity and form teratomas upon transplantation (3). However, recent reports revealed differences between iPS and ES cells in terms of epigenetic modification, heterogeneity, and differentiation potential. For example, not only iPS cells exhibit distinct epigenetic differences from ES cells (4), they harbor residual DNA methylation signatures, namely “epigenetic memory,” characteristic of their somatic tissue of origin, which favors their differentiation along lineages related to the donor cell (4–6). Not only in epigenesis, iPS cells differ from ES cells in immuno-tolerance. In contrast to ES cell derivatives, iPS cells derived from autologous cells demonstrated that abnormal gene expression in some cells differentiated from iPS cells induced T-cell-dependent immune responses in syngeneic recipients (7).

These reports raise a basic questions what is the mechanism of generating iPS cells. To date, two mechanistic theories of iPS cell generation, the stochastic and elite models, have been proposed (8). The stochastic model argues that every cell type can potentially be reprogrammed to become an iPS cell by introducing factors such as Oct3/4, Sox2, Klf4, c-Myc, Nanog, and Lin28 (2, 9). In contrast, the elite model proposes that iPS cells can be generated from only specific subset of cells (8). The correct model remains an open question and the mechanism of iPS cell generation is still remains an enigma.

Two Models, The Stochastic and Elite Models In iPS Cells

iPS cells have been generated from various cell sources, such as skin fibroblasts (2); keratinocytes (10), mesenchymal cells from fat tissue (11), oral mucosa (12), dental pulp (13), cord blood cells (14), peripheral blood cells (15), including T-cells (16). From these reports, iPS cells look like capable of being generated from any cell type, supporting the stochastic model. In regard to this model, Jaenisch and coworkers argued that the number of cell divisions is a key parameter driving epigenetic reprogramming to pluripotency, and thus that theoretically almost all cells eventually give rise to iPS cells with continued growth and transcription factor expression (9). Nishino et al. reported that iPS cell generation is regulated by stochastic epigenetic events that lead to a cell condition that epigenetically more closely resembles that observed in ES cells (17). These reports theoretically and logically support the stochastic model of iPS cell generation, however, rigorous proof that all cell types are in a strict sense able to become iPS cells is still awaited.

The elite model proposed that iPS cells are generated only from a particular subset of cells and thus the cells contribute to iPS cell generation is distinct from those do not. If iPS cells are generated exclusively from single sorted cells while rest of the cells do not contribute to the generation, this will suggest the elite model. There are a couple of studies showing the generation of iPS cells from single sorted cells; human and mouse Wnt1(+) melanocytic cells demonstrated higher generation efficiency than fibroblasts, and iPS cells generated from adult human CD34(+) cells showed higher differentiation ability to hematopoietic cells (15, 18, 19). However, none of these studies demonstrated that selected Wnt1(+) or CD34(+) cells are the primary source of iPS cells. In association with the elite model, some groups discuss that iPS cells are the result of the procurement of tumorigenic proliferative activity in adult stem cells (20, 21). At any rate, the elite model had been disputed, but recently, Byrne et al. reported important finding that only SSEA-3-positive cells among human dermal fibroblasts cells could generate iPS cells (22). While this article suggest the elite model, the characteristics of the original SSEA-3-positive cells were not fully evaluated, and thus the reason why only SSEA-3-positive cells could raise iPS cells was not illustrated clearly. More recently, a paper strongly supporting the elite model demonstrated that, at least in the case of human fibroblasts, iPS cells are generated only from a subset of stem cells that are already pluripotent, namely Multilineage-differentiating stress enduring (Muse) cells (23). These cells normally reside in human mesenchymal tissues (such as dermis and bone marrow) and mesenchymal culture cells [such as fibroblasts and bone marrow-stromal cells (BM-MSCs)], and exhibit the characteristic properties of pluripotent stem cells while they do not show tumorigenic properties. The relation of Muse cells and iPS cells is discussed below.

The Relation Between Mesenchymal Cells And iPS Cells

Because the properties of cells of origin greatly affect the fate of iPS cell generation, a clear understanding of the originating cells is crucial. Fibroblasts are the most popular cell source for generating iPS cells and will therefore be used as an example (1, 2). Fibroblasts are usually collected from the dermis as adherent cells. However, we must remind that the dermis comprises various cell types so that the generally called cultured fibroblasts are not consisted of a single cell type. Indeed, fibroblasts are the major cellular component, but also contain blood vessel-associated cells such as endothelial cells and pericytes as well as several types of stem or progenitor cells, such as skin-derived precursors, neural crest-derived stem cells, melanoblasts, perivascular cells, endothelial progenitors, and adipose-derived stem cells (24–31). Correspondingly, when primary cultured dermal cells are analyzed, cells positive for CD117 (a marker for melanoblasts), CD146 (perivascular cells and adipose-derived stem cells), CD271 (neural crest-derived stem cells), Snai1 (skin-derived precursors), and Slug (skin-derived precursors) are detected (23). Thus, the repertoire of the cells contained in fibroblasts is diverse and thus fibroblasts comprise heterogeneous cell populations.

The same is true for another mesenchymal lineage cells, BM-MSCs. They are usually collected as adherent cells from bone marrow aspirates and are known to be heterogeneous. BM-MSCs are positive for general mesenchymal markers such as CD29, CD44, CD71, CD90, and CD105, but the ratio of each marker differs and the cells are not always 100% positive (32, 33).

Thus, MSCs are generally a crude cell population because they are usually harvested as adherent cells from mesenchymal tissues such as the dermis, bone marrow, adipose tissue, and umbilical cord. Therefore, it must be kept in mind that each population comprising mesenchymal cells often differs with regard to its origin, phenotype, and differentiation state, and above all, its potential to become iPS cells.

We often experience that when culturing cells from other organs and tissues such as from muscle, liver, and kidney, fibroblasts are easily mixed into the primary culture. Even in immune systems such as the spleen, primary cultured cells contain fibroblasts. In other words, contamination of mesenchymal cells is unavoidable in the primary culture and collection of a single population is not guaranteed unless the cells are strictly labeled by cell surface markers and collected by cell sorting. More importantly, almost all organs contain connective tissue, and thus mesenchymal cells will easily penetrate into the primary culture from almost any organ harvested. Because mesenchymal cells are a crude cell population, primary cultures of cells harvested from organs or tissues in most cases contain a heterogeneous mesenchymal cell population, implying that those primary cultured cells become more heterogeneous. Even peripheral blood is not free from mesenchymal cells; several studies indeed demonstrated that MSCs with multilineage differentiation ability appear in the blood under disease or injury (34–38). When the original population is heterogeneous, the potential of the cells for reprogramming will also be heterogeneous and diverse. In regard to this, Hochedlinger's group demonstrated that the differentiation stage of the starting cell influences the efficiency of reprogramming into iPS cells. In mouse hematopoietic cells, cells in an undifferentiated state are better able to generate iPS cells (4).

Although iPS cells appear to be generated from terminally differentiated cells from various organs such as the liver (39), spleen (40), or peripheral blood (16), these results may not strictly rule out the possibility that iPS cells are generated from cells other than terminally differentiated cells unless those terminally differentiated cells are strictly identified and selected, e.g., using FACS, to eliminate other kinds of cells before subjecting the cells to the iPS cell-generation procedure.

Pluripotent Stem Cells are Present in Mesenchymal Cells

Mesenchymal tissues such as dermis, bone marrow, adipose tissue, and umbilical cord are known to contain tissue stem cells, namely mesenchymal stem cells (MSCs). MSCs are distinct from general adult stem cells such as hematopoietic and neural stem cells since they differentiate into broad spectrum of cell types not only into the same mesodermal-lineage, such as bone, cartilage, and adipocytes, but also into the other lineages, ectodermal and endodermal cells. For example, when MSCs are treated with a certain sets of cytokines or with transient gene introduction, they differentiate in vitro into cell types of cardiac muscle (41), skeletal muscle (42), hepatocytes (43), neuronal cells (44), endothelial cells (45), peripheral glial cells (46), insulin-producing cells (47), and epithelial cells (48). While with a very low frequency, differentiation of MSCs in vivo into cardiac muscles (49), hepatocytes (50) and keratinocytes (38) is also recognized in disease models. From these phenomena, it is speculated that MSCs contain cells resembling pluripotent stem cells.

Recently, adult human mesenchymal cells such as BM-MSCs and dermal fibroblasts were reported to contain Muse cells, a type of pluripotent stem cell (33). These cells are unique in that they have the properties of both mesenchymal stem cells and pluripotent stem cells. This property is evidenced by their marker expression profile; they express mesenchymal markers such as CD29, CD90, and CD105, while at the same time they are positive for the pluripotency marker stage-specific embryonic antigen-3 (SSEA-3), a marker for undifferentiated human ES cells (Fig. 1). It is noteworthy that Muse cells both from bone marrow or dermis do not express other stem cell markers; i.e., markers for hemaopoietic stem cells (CD34 and CD117), skin-derived precursors (Snai1 and Slug), neural crest-derived stem cells (CD271 and Sox10), perivascular cells (NG2 and CD146) or endothelial progenitors (CD31 and von Willebrand factor), suggesting that Muse cells are distinct from these known stem cells (23, 33). When Muse cells are cultured in adherent cell cultures, they appear the same as general fibroblasts, but when they are transferred to a single cell-suspension culture, the cells proliferate and form cell clusters that resemble the embryoid bodies of ES cells (Fig. 2). Single cell-derived Muse cell clusters are positive for alkaline phosphatase; express the pluripotency markers Nanog, Oct3/4, and Sox2; and differentiate into endodermal-, ectodermal-, and mesodermal-lineage cells when transferred to gelatin cultures, indicating that single Muse cells are able to generate cells representative of all three germ layers (Fig. 2). Muse cells are also able to self-renew (33). The detailed method for isolating Muse cells is described in Supporting Information.

Figure 1.

Muse cells are positive for both mesenchymal and pluripotency markers. Human dermal fibroblasts contain ∼1% of SSEA-3+ cells. When these cells were further examined for the expression of mesenchymal markers, 100% of those positive for SSEA-3(A) were positive for CD105 (B), CD90 (C), and CD29 (D). (Data from Ref. 23.) [Color figure can be viewed in the online issue, which is available at]

Figure 2.

Muse cell collection and characterization. Muse cells can be collected from cultured mesenchymal cells (fibroblasts or BM-MSCs) and mesenchymal tissues (fat tissue, dermis, and bone marrow aspirates) as cells double-positive for SSEA-3 and CD105. Collected Muse cells express SSEA-3 on their surface, similar to human ES (hES) cells. After isolating Muse cells by FACS, single Muse cells cultured in suspension (single cell-suspension culture) generate characteristic clusters that are very similar to the embryoid bodies formed by human ES cells. When the cell clusters are transferred onto gelatin culture and spontaneous differentiation is induced, cells with endodermal- (positive for α-fetoprotein [αFP] and cytokeratin7 [Ck7]), ectodermal- (positive for neurofilament), and mesodermal- (positive for desmin and smooth muscle actin [SMA]) lineage cells are observed. Bars = 10 μm. (Pictures reproduced from Refs. 23 and33.)

The existence of pluripotent cells in MSCs has long been suggested, but to date there have been no reports clearly demonstrating self-renewal and tri-lineage differentiation at a single cell level, so the pluripotency in MSCs has remained controversial (51, 52). Most importantly, single Muse cells are able to generate mesodermal-lineage (osteocytes, adipocytes, chondrocytes, skeletal muscle cells, smooth muscle cells), ectodermal-lineage (neuronal cells, glial cells, epidermal cells), and endodermal-lineage cells (hepatocytes, biliary system cells) in vitro, and keep self-renew, so that they are considered pluripotent stem cells (33).

ES cells and iPS cells, artificially established pluripotent stem cell, are known to form teratomas upon transplantation. In contrast to these pluripotent stem cells, however, Muse cells do not undergo tumorigenic proliferation, and do not develop into teratomas when transplanted into immunodeficient mouse testes (33). Consistently, ES cells and iPS cells have high levels of telomerase activity as well as high expression levels of genes related to cell-cycle progression, whereas Muse cells have low levels of both of these activities, the same level as in naive fibroblasts (23). The nontumorigenicity of Muse cells is consistent with the fact that they reside in normal adult mesenchymal tissue.

A “pluripotent” cell is defined as that having the ability to give rise to cell types of all three embryonic germ layers, namely endodermal, mesodermal, and ectodermal cells (53). When we call “pluripotent stem cells,” the concept “stem cell” applies not only to tri-lineage differentiation, but also the ability to self-renew. In many cases, pluripotent stem cells show generation of germ line transmitting chimeras and/or teratoma formation in addition to the above two requirements to mimic normal development (53, 54). However, epiblast stem cells, a type of pluripotent stem cell, do not form teratomas under certain circumstances (55). Therefore, pluripotent stem cells do not always show generation of germ line transmitting chimeras or teratoma formation.

As mentioned above, MSCs differentiate into a broad spectrum of cells that crosses the oligolineage boundaries between mesodermal and ectodermal or endodermal lineages, and such differentiation ability is often called “multipotency” (28, 29). However, the term multipotency is not adequate to describe the high differentiation ability of these cells, which are able to generate cells representative of all three germ layers. In fact, such cells are often called ‘pluripotent’ to describe their high differentiation ability (56–59). Furthermore, generation of germ line transmitting chimeras are not possible in the case of human cells. Overall, “self-renewal” and “tri-lineage differentiation” are essential and common requirements for pluripotent stem cells, and these two properties are sufficiently comprehensive to represent the high differentiation ability of MSCs rather than setting limits by including generation of germ line transmitting chimeras and/or teratoma formation abilities. Therefore, in this review, we define “pluripotent stem cells” as cells having the ability to self-renew and to differentiate into cells representative of all three germ layers.

The behavior of Muse cells in vivo is interesting. Without induction, naive Muse cells act as “repair cells” when infused into the peripheral bloodstream of acute injury models. This was verified by the infusion of green fluorescent protein (GFP)-labeled naive human Muse cells into an immunodeficient mouse model with spinal cord injury, skeletal muscle degeneration, or fulminant hepatitis. Infused Muse cells differentiated into neuronal cells (neurofilament-positive cells), skeletal muscle cells (human dystrophin-positive), or hepatocytes (human albumin- and human anti-trypsin-positive cells) after integration into these damaged tissues (23, 33). The results demonstrated that Muse cells can integrate as functional cells into damaged tissue and differentiate into ectodermal- (neuronal cell), endodermal- (hepatocytes), and mesodermal-lineage cells (skeletal muscle cells) according to the site of integration, and contribute to tissue reconstruction.

The ratio of Muse cells is less than 1% in cultured BM-MSCs and a small percentage in commercially obtained fibroblasts, but it is very low (1 out of 3,000 mononucleated cells) in the fresh human bone marrow mononucleated cell fraction (33). Histologically, Muse cells locate sparsely in the connective tissues of organs (Fig. 3) (23).

Figure 3.

Localization of Muse cells in adult human skin. Muse cells labeled by SSEA-3 are sparsely observed in the connective tissue of the dermis, fat tissue, sweat glands, and hypodermis. Bars =50 μm. (Pictures reproduced from Ref. 23.)

Muse Cells Support the Elite Model of iPS Cell Generation

Apart from the stochastic model, the possibility of the elite model has also been argued. Recently, the result supporting the elite model was reported in Muse cells. At least in the case of human fibroblasts, iPS cells are generated only from Muse cells, but when Muse cells were removed from human fibroblasts, the remaining cell population was unresponsive to the four Yamanaka factors and failed to generate iPS cells (Fig. 4) (23).

Figure 4.

Muse and non-Muse cells in iPS cell generation. When human fibroblasts were separated into Muse and non-Muse cells and each population were subjected to iPS cell generation, iPS cells are generated only from Muse cells and never from non-Muse cells. The properties of each cell population are shown in the boxes. [Color figure can be viewed in the online issue, which is available at]

When human fibroblasts were separated into Muse cells and non-Muse cells and the four Yamanaka factors were added, both populations generated colonies but far fewer were generated by non-Muse cells. Importantly, iPS colonies showing human ES cell-like morphology and TRA-1–81 positivity [a marker for promising iPS colonies (60)] were only generated from Muse cells but not from non-Muse cells (Fig. 4). Consistently, endogenous Sox2 and Nanog, the master genes of pluripotent stem cells, were detected in cells and colonies derived from Muse cells, but never in those derived from non-Muse cells (Fig. 4) (23).

Colonies generated from Muse and non-Muse cells were further picked up and passaged to establish iPS cell lines. Only colonies picked from Muse cells established iPS cells (Muse-iPS cells), and colonies originating from non-Muse cells (non-Muse colonies) were unlike human ES in their morphology. Muse-iPS cells differentiated into endodermal-, ectodermal-, and mesodermal-lineage cells in vitro, and formed teratomas after injection into immunodeficient mouse testes (23).

Chan et al. reported that incompletely reprogrammed colonies can be divided into Types I and II colonies; Type I colonies, which do not express Rex1, Abcg2, Dnmt3b, and Cdx2, remain in the incompletely reprogrammed state and do not develop into iPS cells, whereas Type II colonies, which do not express Rex1, Abcg2, or Dnmt3b, but do express Cdx2, occasionally spontaneously transit to iPS cells (61). When investigating non-Muse colonies, they were all negative for Rex1, Abcg2, Dnmt3b, and Cdx2 correspond to Type I colonies that stay arrested at an early stage of iPS cell generation, remain in the incompletely reprogrammed state, and thus do not develop into iPS cells (Fig. 4) (23).

The inability of non-Muse cells to respond to the four Yamanaka factors could also be seen in the methylation state of the promoter regions of Nanog and Oct3/4. In the naive state, the Nanog and Oct3/4 promoter regions are more methylated in non-Muse cells than in Muse cells. In Muse cells, those partly methylated promoter regions become completely demethylated when they develop into Muse-iPS cells, while such demethylation never observed in the Nanog and Oct3/4 promoter regions of non-Muse cell-derived colonies but stayed methylated (23).

Gene expression profiles demonstrated that the expression “level” of genes related to pluripotency is lower in naive Muse cells but is substantially elevated when they become to Muse-iPS cells (Fig. 5). Most importantly, the expression “pattern” of genes related to pluripotency, namely the repertoire of genes expressed, is nearly the same between naive Muse cells and Muse-iPS cells. In contrast, naive non-Muse cells do not express genes related to pluripotency, and both the expression level and pattern show no substantial changes even after receiving Yamanaka's four factors, namely in non-Muse colonies (Fig. 5). In the case of genes related to cell-cycle progression, they were mostly upregulated in Muse-iPS cells as compared with naive Muse cells (Fig. 5). This is consistent with the fact that naive Muse cells have lower telomerase activity and do not form teratomas after transplantation into immunodeficient mouse testes, while Muse-iPS cells formed teratomas. In non-Muse cell-derived colonies, some of the genes related to cell-cycle progression were upregulated compared with those in naive non-Muse cells, but the upregulation was marginal and not as extensive as in Muse-iPS cells (Fig. 5) (23).

Figure 5.

Gene expression pattern in Muse, Muse-iPS (M-iPS), non-Muse, and non-Muse colonies (non-Muse col). The expression “pattern” of pluripotency markers in Muse cells and Muse-iPS was basically same, but expression “level” was higher in Muse-iPS cells than in naive Muse cells (1). Neither Muse nor non-Muse cells showed a change in the expression “pattern” of pluripotency markers even after receiving Yamanaka's four factors (2). While naïve fibroblasts are known to contain Muse cells, the expression pattern and level of pluripotency markers in the fibroblasts as a whole takes on the pattern of non-Muse cells because Muse cells are only up to several percentages (3). Genes related to cell cycle progression did not largely differ between Muse and non-Muse cells (4), but they were upregulated when Muse cells became Muse-iPS cells (5). (Modified version of Table 1 in Ref. 23.)

Ips Cells are The Result of Resetting Differentiated Cells Into Embryonic State—is This True?

The most impressive observation of these gene expression patterns is that, regardless of whether the cells are Muse or non-Muse cells, the expression “pattern” of genes related to pluripotency is not altered by introduction of Yamanaka's four factors. If the gene expression pattern of naive non-Muse cells show a drastic change after receiving the four Yamanaka factors and eventually exhibit the same pattern as seen in Muse-iPS cells, this would strongly support the notion that the clock of terminally differentiated cells is wired back to the embryonic state by introducing the four Yamanaka factors. If so, every cell, even terminally differentiated cells, would have the potential to become an iPS cell and this would strongly support the stochastic model.

The gene expression profile in whole human fibroblasts will be as same as that in non-Muse cells because the ratio of Muse cells contained in fibroblasts is only several percent (23). Therefore, in fibroblasts as a whole, the signal from Muse cells is drowned out by the vast majority of non-Muse cells. Therefore, when iPS cell generation procedure in fibroblasts were analyzed, it looks as if Yamanaka's four factors stochastically brought terminally differentiated fibroblasts back to a state resembling the cells of the inner cell mass. However, the differences in the results of Muse and non-Muse cell experiments clearly indicate that this did not happen in human fibroblasts.

In the framework of Muse and non-Muse cells, human fibroblasts can be divided into two populations: cells that primarily contribute to iPS cell generation and those that do not. These results demonstrate that the human fibroblast system fits into the elite model of iPS cell generation. Further studies are needed to evaluate the potential of this system in other tissues and cell types.

iPS Cells and Tumorigenicity

The original source of iPS cells is nontumorigenic somatic cells, but once they become iPS cells, they show newly acquired unlimited proliferative activity. Intracellular signaling in iPS cell generation is reported to be similar to that in tumorigenesis. For example, Lin et al. demonstrated that iPS cell generation can be divided into two phases: the initial stable intermediate phase and the final phase. The final phase is promoted by the overexpression of Sox2 to induce downregulation of transforming growth factor beta (62), a tumorigenesis suppressor (63, 64). Generation of iPS cells is enhanced when using p53-knockdown cells, and p53 acts as a critical regulator that suppresses tumorigenesis (65, 66). These studies clearly demonstrate the similarity of intracellular signaling in tumorigenesis and iPS cell generation. In terms of cancer-related epigenetic changes and regulation of genes related to cell-cycle progression, iPS cells seem to be prone to the abnormal epigenetic changes that are characteristic of neoplasia. A recent study suggested that differential changes between cancer and normal cells, and between iPS and normal ES cells, are marginal or confined to regions upstream from promoter CpG islands (67). On the other hand, another group reported that cancer-specific promoter CpG island hypermethylation recognized in the cancer cells is also observed in iPS cells, and that one essential reprogramming factor, OCT4, produces cancer-specific epigenetic changes. Cancer-related epigenetic abnormalities that arise during early reprogramming, including hundreds of abnormal gene-silencing events, patterns of aberrant responses to epigenetic-modifying drugs resembling those of cancer cells, and the presence of cancer-specific gene promoter DNA methylation alterations, are inherited to iPS cells (68).

As for genes related to cell-cycle progression, Muse cell studies have provided some information. These genes are mostly upregulated in Muse cell-derived iPS cells compared with naive Muse cells, consistent with the fact that naive Muse cells are nontumorigenic while Muse-iPS cells are tumorigenic. Notably, Nanog and Oct-4 accelerate cell-cycle progression in pluripotent stem cells such as ES cells (69, 70). Thus, it is possible that the generation of iPS cells from Muse cells requires a much higher expression of critical transcription factors, including pluripotency markers, that leads to the activation of genes related to cell-cycle progression, and is followed by further elevation of pluripotency marker expression. Such synergistic effects may result in higher expression levels of genes related to pluripotency as well as to cell-cycle progression in Muse cell-derived iPS.

Similarities may exist between cancer cells and iPS cells in terms of epigenesis, gene expression, and cancer-related abnormalities. The mechanism of acquiring tumorigenic properties during the process of iPS cell generation, however, may differ from that of pathologic carcinogenesis and further studies are needed to clarify this issue.


iPS cells were first reported to be generated from mice and human fibroblasts with very low efficiency, nearly 0.001%, (1, 2), and following researches have made efforts to improve the generation efficiency by using valproic acid, inhibitors for transforming growth factor −beta, MAPK/ERK, or suppression of p53 (71–73). As for the use of valproic acid, the efficiency in mouse embryonic fibroblasts was increased up to ∼2% to 3%, based on Oct4-GFP quantification (74). A replication-defective and persistent Sendai virus vector containing Oct4/Sox2/Klf4/c-Myc was reported to induce iPS cell from mouse primary fibroblasts with an efficiency of ∼1%, as estimated by GFP expression driven by the Nanog promoter (75). Similarly, replacing c-Myc with Glis1 increased iPS cell generation from human fibroblasts with an efficiency of ∼0.16%, also based on Nanog promoter activity (76).

Despite these efforts, however, the generation efficiency is still far from being high enough. Even in the case of Muse cells, generation efficiency is only 0.03%, albeit counted strictly based on the expression all of Nanog, endogenous Oct3/4, Sox2, Rex1, Abcg2, Dnmt3b, and Cdx2. This efficiency corresponds to 30 times greater efficiency than naive fibroblasts (23).

As evidenced by these reports, the serious problem in iPS cell research is that the criteria for generation efficiency differ among reports. Some reports calculate generation efficiency based only on ALP staining, whereas others base generation efficiency on the expression of a single pluripotency marker such as Nanog or Oct3/4. Because of the current lack of unified criteria to identify the generation of iPS cells, the reported generation efficiencies cannot be compared with each other. It must be bear in mind that not all colonies positive for ALP staining are iPS cells, and likewise, not all colonies that are positive for only a single pluripotency marker such as for Nanog or Oct3/4 meet the proper criteria for iPS cells (60, 77). Previously, gene expression analyses in live images and quantitative polymerase chain reaction were performed both in colonies resembling and colonies not resembling ES cells. The results revealed that the expression of Nanog or Oct3/4, or positive reaction to ALP occur in various kinds of colonies other than iPS cells and thus demonstrating that both factors are unreliable markers for the identification of iPS cells (61, 78). Not confined to iPS cells, tissue stem cells are occasionally positive for Oct3/4- or Nanog, suggesting that cells other than pluripotent stem cells express these markers (79, 80). These findings indicate that the calculation of iPS cell generation based on the single expression of Oct3/4 or Nanog will lead to overestimation of the number of iPS cells, and thus unified and reliable criteria to identify iPS cells must be firmly established.

Understanding the properties of the original starting cell population for generating iPS cells is important for understanding their generation mechanism. Indeed, when the emergence of iPS cells is unforeseeable, it seems that all cells have the potential to become iPS cells and that a subset of cells look like stochastically develop into iPS cells by coincidence combined with an exquisite balance of intrinsic factors. On the other hand, cells that are already pluripotent from the beginning such as Muse cells are recognized among mesenchymal cells and/or MSCs, and iPS cells are exclusively generated from these cells; thus we now recognize that the stochastic model is not the only viable theory of iPS cell generation. Therefore, we must turn our attention to the heterogeneity and diversity of the original cell population in their property, differentiation state, and potential to become iPS cells.

Together, there is the strong need for a basic understanding of the iPS cell-generation mechanism. At any rate, the questions of how are they generated remain crucial issues to be resolved, and understanding the basic characteristics of iPS cells will advance the studies of these cells and their application to regenerative medicine.