Author contributions: J.S.: conception and design and major part of manuscript writing; S.H.: manuscript writing; H.R.S.: conception and design, financial support, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS October 18, 2011.
Through cellular differentiation, a single cell eventually gives rise to all the various lineages of an organism. This process has traditionally been viewed as irreversible. However, nuclear transfer experiments have demonstrated that differentiated cells can be reprogrammed to form even an entire organism. Yamanaka electrified the world with the discovery that expression of only four transcription factors was sufficient to induce pluripotency in differentiated somatic cells of mammals. Expansion of this work has shown that expression of the master pluripotency gene Oct4 is sufficient to induce pluripotency in neural stem cells. In contrast to somatic cells, germline cells express Oct4 and can acquire pluripotency without the addition of exogenous transcription factors. More recently, it has been possible to also induce an alternative cell fate directly by the transdifferentiation of cells mediated by the introduction of specific transcription factors, including Oct4. Therefore, we suggest that Oct4 is the gatekeeper into a reprogramming expressway that can be directed by altering the experimental conditions. STEM CELLS2012;30:15–21
All mammalian organisms begin as a single cell—the zygote. This single cell will go on to differentiate into every cell lineage and pattern the various cells into a fully functional organism. This ability is referred to totipotency and is retained through the first few cleavage divisions. Within a few days of fertilization, the zygote has divided and differentiated to form a blastocyst, which is composed of trophectoderm and inner cell mass (ICM) cells. Even at this early stage of development, trophectoderm cells have already committed to a developmental fate and will generally not regain the potential to differentiate into other cell types. Although the ICM cells retain a broad developmental potential to form every lineage of the embryo proper, they have lost the ability to organize all the cell types independently into an organism. As such, these cells are no longer totipotent—that is, they are pluripotent. After gastrulation, all cells have committed to either a germ cell or a particular germ layer fate. Only germ cells retain the ability to form a totipotent cell through fertilization.
Although fate commitment cannot normally be reversed in vivo during development, technologies have emerged that are capable of reprogramming mammalian somatic cells to totipotency and pluripotency in vitro. Gurdon  firmly established that fate commitment is reversible by showing that nuclei from differentiating endodermal cells from different developmental stages, ranging from blastulae to swimming tadpoles, consistently gave rise to swimming tadpoles when introduced into enucleated oocytes (Table 1). In 1996, the birth of Dolly proved that reversing differentiation in mammalian species was also possible . More recently, direct induction of pluripotency in somatic cells, such as fibroblasts, has become possible [3, 19]. Finally, transdifferentiation of one cell type directly into an alternative cell lineage, such as reprogramming a fibroblast directly into a neuron, suggested that totipotent or pluripotent cells may not even be necessary intermediates. However, reprogrammed cells have been shown to retain the epigenetic memory of their tissue of origin [25, 26]. This indicates that reprogramming technologies must be improved and that careful consideration be given to the technology to be used, such as cell therapy or drug discovery, so as to obtain the appropriate result.
Table 1. Summary of reprogramming methods, species, factors used, and results
INDUCTION OF PLURIPOTENCY WITH EXOGENOUS OCT4 AND OTHER TRANSCRIPTION FACTORS
Pluripotent stem cells are also capable of reprogramming somatic cells. Embryonic stem cells (ESCs) are the best-known pluripotent stem cells and were first derived in 1981 from mouse (Fig. 1) [28, 29] and in 1998 from human blastocysts . ESCs can be grown as pluripotent cell lines without losing their differentiation potential. ESCs form teratomas when introduced into immunocompromised mice; these teratomas are composed of cells that have differentiated into derivatives of all three germ layers. In contrast to a totipotent cell, an ESC is not capable of autonomously developing into an embryo. However, ESCs readily incorporate into the ICM and form chimeras when aggregated with morula-stage embryos or injected into blastocysts. The most stringent test for pluripotency, termed tetraploid complementation, is when mice are produced entirely from cells that had been aggregated with tetraploid embryos. Whereas the tetraploid components form extraembryonic lineages, the diploid cells—if pluripotent and without major mutations—give rise to the embryo proper .
Cell fusion experiments first demonstrated that pluripotent cells were capable of reprogramming somatic cells (Table 1). Miller and Ruddle  demonstrated that when embryonic carcinoma (EC) cells, which are related to ESCs but are derived from tumors, were fused with thymocytes, the resulting hybrid cells morphologically resembled the EC cells and had a silenced thymocyte marker Thy1. When ESCs were fused with thymocytes, the somatic nucleus adopted characteristics of the pluripotent cells, including X-chromosome reactivation (in female cells), early replication timing, unstable Xist transcription, and Oct4 promoter utilization. The ES–thymocyte hybrids formed teratomas, confirming their pluripotency .
As ESCs can reprogram somatic cells by cell fusion, Takahashi and Yamanaka  sought to identify genes expressed in ESCs that would be sufficient to induce the formation of induced pluripotent stem cells (iPSCs) (Table 1). An initial list of 24 candidate genes was compiled from existing data and these genes were cloned into retroviral expression vectors. When embryonic fibroblasts were infected with these expression vectors, iPSCs were generated. These cells expressed stage-specific embryonic antigen-1 (SSEA-1) and Nanog, formed teratomas when injected into immune compromised mice, and contributed to different tissues of developing embryos on blastocyst injection. Of significance, the retroviral transgenes of these iPSCs were methylated and their expression was silenced. However, these iPSCs also showed aberrant expression of key pluripotency genes, as well as incomplete demethylation of pluripotent gene promoters, and failed to either generate full-term chimeras or give rise to germ cells. Just a couple of months later, iPSCs selected using either the Nanog or Oct4 promoter were shown to more closely resemble ESCs than the iPSCs originally generated by Takahashi and Yamanaka both in gene expression and chimera formation [4, 5]. After 2 years, viable mice were generated entirely from iPSCs following tetraploid complementation [32–34].
In 2007, iPSCs were successfully derived from human fibroblasts through expression of the four Yamanaka factors Oct4, Sox2, Klf4, and c-Myc, as well as by the combination Oct4, Sox2, Nanog, and Lin28—the latter is the only protein of the combination that is not a transcription factor [9, 10]. These human iPSCs closely resemble human ESCs in gene expression, promoter methylation, and differentiation potential. To date, iPSCs have been derived from numerous somatic cell populations .
An important topic in reprogramming is identifying the minimum number of transgenes required for iPSC formation. Takahashi  derived iPSC-like colonies from mouse fibroblasts using Oct4, Sox2, and Klf4, (no c-Myc), or Oct4, Klf4, and c-Myc (no Sox2). Kim et al.  demonstrated that by starting with neural stem cells, which exhibited endogenous expression of two of the four Yamanaka factors at levels comparable to ESCs, the expression of only two genes, Oct4 and Klf4, was sufficient to induce iPSC formation. Through further optimization, Oct4 alone was found to be sufficient to induce iPSC formation in both mouse and human neural stem cells [7, 35].
Even different strategies used to induce reprogramming have consistently found that only a small fraction of cells will become iPSCs. Therefore, a major goal in reprogramming research is to increase the efficiency of iPSC derivation. Reduction of p53 signaling, which acts as a barrier to reprogramming by limiting cell cycling and inducing apoptosis, has been reported to significantly increase the efficiency of reprogramming [35, 36]. By screening nuclear fractions from extracts of pluripotent mouse cells, Singhal et al.  identified the ATP-dependent Brg1/Brm-associated factor (BAF) chromatin-remodeling complex as a factor that substantially increases reprogramming efficiency when used together with the four factors.
Problems with epigenetic memory appear to be a general feature of reprogramming. Nuclear transfer of B6C3F1 female mice resulted in abnormal obesity not found in the donor mice . After subsequent mating, the obesity phenotype was not transmitted to the progeny, which suggests that it was an epigenetic error that occurred during reprogramming. A hypomorphic DNA methyltransferase 1 (DNMT1) allele, which reduced epigenetic memory by decreasing global DNA methylation, significantly improved the efficiency of blastocyst formation after nuclear transfer . Similarly, residual DNA methylation signatures have been found in iPSCs, which lead to restricted differentiation into cells with a different fate from the tissue of origin . Interestingly, these problems are most prominent in iPSCs of an early passage and are largely attenuated upon further passaging .
Recently, Bock et al.  systematically compared the genome-wide gene expression and DNA methylation of 20 human ESC and 12 iPSC lines. They found that the vast majority of genes exhibiting significant variability between iPSC lines were similarly variable between ESC lines. No specific locus that discriminated ESCs and iPSC could be detected. Using a statistical model, those authors concluded that somatic memory does not contribute to more than 0.01%–0.001% of the variation seen in human iPSC lines . Therefore, many of the findings regarding epigenetic memory in iPSCs appear to result from the epigenetic diversity inherent to pluripotent stem cells.
INDUCTION OF PLURIPOTENCY WITHOUT EXOGENOUS TRANSCRIPTION FACTORS IN CELLS WITH ENDOGENOUS Oct4
In contrast to somatic cells, germ cells retain the ability to form pluripotent cells through embryogenesis. Transcription factors required for pluripotency, such as Oct4, are already expressed within cells of the germ lineage and do not need to be added exogenously to induce pluripotency. Therefore, germline cells are potentially a rich source of patient-specific pluripotent stem cells that, by their very nature, retain no epigenetic memory of the somatic cells, are likely to have fewer mutations than somatic cells, and their derivation requires no genetic manipulation. The relative ease of inducing pluripotency in germline cells has enabled the derivation of such cells almost 15 years before pluripotency could also be induced in somatic cells (Fig. 1). Because of the amazing reprogramming capacity and capability of germline cells in vivo (establishment of totipotency after fertilization and induction of pluripotency in the preimplantation embryo) and in vitro in unipotent germ cells (see below), we consider the germline to represent a reprogramming expressway. The reprogramming power of germline cells is also highlighted by the transfer of somatic cell nuclei into oocytes and by the dominant nature of pluripotent cells in fusion experiments as described above.
In 1992, two groups reported that pluripotent stem cells could be generated from primordial germ cells (PGCs) derived from 8.5-day-old mouse embryos [13, 14]. PGCs are unipotent cells in vivo, as they only differentiate to form germ cells. However, in contrast to embryonic fibroblasts, which require exogenous transcription factors to induce pluripotency, embryonic PGCs can be converted into pluripotent stem cells in culture through the addition of specific growth factors, such as Fgf2, leukemia inhibitory factor (LIF), and Steel, with an efficiency of about 5%. The resulting cells, termed embryonic germ (EG) cells, are morphologically indistinguishable from ESCs. Moreover, both EG cells and ESCs express markers, such as SSEA-1 and alkaline phosphatase, and both form teratomas composed of cells from all three germ layers after injection into immunocompromised mice. When introduced into blastocysts, EG cells readily form chimeras comparable to ESCs. In 1998, Shamblott et al.  demonstrated that EG cells could be derived from human PGCs using conditions similar to those for EG derivation in the mouse.
In 2004, Kanatsu-Shinohara et al.  generated pluripotent stem cells from neonatal mouse testis. Although pluripotent EG cells can be derived from PGCs, these cells are only available from embryos. Mouse spermatogonial stem cells can be derived from mouse testis and directed to self-renew in vitro as germline stem cells (GSCs). Under these conditions, GSCs are unipotent and are only able to differentiate into sperm. On transplantation into the seminiferous tubules of infertile mice, GSCs are capable of engrafting, reconstituting the testicular tissue with new gonocytes, and forming fully functional germ cells that are in turn capable of fertilizing oocytes. Teratomas are not observed, which demonstrates that GSCs are not pluripotent. However, when testis cells were cultured under ESC conditions, pluripotent stem cells were obtained in 4 of 21 experiments. The overall frequency of formation ES-like cells was rare, at 1 in 1.5 × 107, which is the equivalent of about 35 newborn testes. Removal of the gene p53 increased the efficiency of derivation of ES-like cells from neonatal testis and enabled the derivation of ES-like cells from adult testis. These cells expressed all of the markers of pluripotency comparable to ESCs and formed teratomas after transplantation, instead of sperm. Like ESCs, germline-derived pluripotent stem (gPSCs) were capable of forming chimeras. In contrast to ESCs, tetraploid complementation was not successful. This could have been due to either a male imprinting pattern or an aberrant DNA methylation at some imprinted loci such as Peg10 in the gPSCs.
It is given that a very limited amount of source material will usually be available for the generation of patient-specific pluripotent stem cells in cell culture. Ko et al.  provided proof of principle for the conversion of adult GSCs into pluripotent stem cells. In a subsequent study, Ko et al.  demonstrated that self-renewing GSCs could be obtained even from small biopsies, at least from the mouse. These GSCs could then be reprogrammed into pluripotent stem cells under specific culture conditions, including a microenvironment dependent on the number of plated GSCs and the length of culture. The pluripotency of these gPS cells was confirmed by chimera formation and in vitro differentiation into functional neurons and cardiomyocytes. Using such an approach, pluripotent stem cells could be clonally derived from very limited source material. Therefore, in principle, such an approach could be applied to human biopsied material for the generation of patient-specific pluripotent stem cells. Although several reports have described the derivation of pluripotent cells from human testis, the results of these studies are controversial [41–45]. For example, Conrad et al.  claimed to have derived pluripotent cells from human testis, but further examination demonstrated that the cells in question were more likely to be fibroblasts or fibroblast-like cells [45–47].
It is also possible to induce cells directly into an alternative fate through transdifferentiation by introducing specific transcription factors. This was first demonstrated in 1987 when David et al. converted fibroblasts into myoblasts by expressing the transcription factor MyoD . This suggests that other lineages could also be formed through a similar approach (Table 1). Thomas Vierbuchen et al.  used exactly the same approach as Yamanaka and demonstrated that mouse fibroblasts could be directly transdifferentiated into neurons. After testing an initial collection of 19 neuronal lineage-specific transcription factors, a combination of only three factors, Ascl1, Brn2, and Myt1l, was found to be sufficient in inducing neuronal differentiation in fibroblasts, with more than 19% efficiency in about 12 days. Normal electrophysiological function was observed in the generated neurons. Interestingly, fibroblast transdifferentiation into neurons could be induced with only Ascl1, but the efficiency was significantly lower and the neurons failed to electrically mature.
Similarly, Masaki Ieda et al.  demonstrated that a combination of three cardiac-specific transcription factors, Gata4, Mef2c, and Tbx5, could directly induce mouse fibroblast transdifferentiation into cardiomyocytes. As with previous approaches, a cocktail of 14 transcription factors was initially tested for the ability to induce transdifferentiation. Gata4, Mef2c, and Tbx5 induced αmajor histocompatibility complex (MHC) - green fluorescence protein (GFP) reporter expression in more than 20% of mouse fibroblasts. A significant fraction of these αMHC-GFP-positive cells formed functional cardiomyocytes and exhibited spontaneous contraction.
Efe et al.  used an alternative strategy to directly convert mouse embryonic fibroblasts into cardiomyocytes, by overexpressing Oct4, Sox2, Klf4, and c-Myc. However, instead of the normal pluripotency cell culture conditions, the authors cultured the cells under alternative conditions that favored cardiomyocyte formation. By day 18, spontaneously beating colonies were observed. Between 15% and 20% of the cells expressed cardiac markers, such as Flk1, Nkx2.5, and cTnT. Subsequent analysis revealed that the cells had directly transdifferentiated into the cardiac lineage instead of forming a transient pluripotent intermediate.
Using a similar approach, Szabo et al. directly converted human dermal fibroblasts into multilineage blood progenitors . Overexpression of Oct4 resulted in a population of round hematopoietic-resembling cells expressing the hematopoietic marker CD45 but not pluripotent markers. After changing the culture conditions to those supporting early hematopoiesis, hematopoietic precursors were isolated that were capable of forming granulocytic, monocytic, megakaryocytic, and erythroid lineages, as well as supporting in vivo engraftment. A pluripotent cellular intermediate appeared not to be required to generate these hematopoietic cells (Fig. 2). It would be an amazing scientific accomplishment and potentially of enormous practical medical relevance if such intermediate cells could be not only defined but also stabilized in culture. This cell in principle could represent an artificial state not found in vivo. This is certainly true also for other cells kept in culture, the most famous example being ESCs. Indeed, it is an amazingly flexible feature of ESCs that they can be taken out and be brought back to the germline.
The power of defined culture conditions in specifying cell fate has also been demonstrated with respect to the induction of pluripotency. Depending on the culture conditions, fibroblasts can be reprogrammed by the Yamanaka cocktail to either iPSCs or induced epiblast stem cells (Fig. 1) .
Both nuclear transfer and induction of pluripotency has resulted in cells that retain the epigenetic memory of the donor cell origin. Although it is not known whether this also holds true for induced transdifferentiation, we extrapolate from these results and therefore argue that this is likely to be the case, if not even more so than with iPSCs.
OCT4 AND THE REPROGRAMING EXPRESSWAY
Because of these new reprogramming experiments, we propose that Oct4 is more than a master regulator of pluripotency—it is the master regulator all along and into the reprogramming expressway. It is well known that Oct4 is specifically expressed in pluripotent cells, and expression of Oct4 is sufficient to induce pluripotency in somatic cells . However, Oct4 is also expressed in cells committed to each of the three germ layers of gastrulation-stage embryos . This suggests that Oct4 plays an important role in the commitment of pluripotent cells to somatic lineages. Indeed, ESCs overexpressing Oct4 undergo rapid differentiation and lose pluripotency . Recently, Thomson et al.  have shown that Oct4 and Sox2 are critical for germ layer fate choice. This appears to be accomplished by differentiation signals that continuously and asymmetrically modulate Oct4 and Sox2 protein levels, thus altering their binding pattern to the genome. Therefore, Oct4 expression in somatic cells may lead to the induction of progenitor cells that are committed to a particular germ layer, as well as give rise to iPSCs when cultured under specific conditions. As such, Oct4 would not simply be a “reprogramming factor,” but rather the gatekeeper into and out of the reprogramming expressway that can be directed by altering the experimental conditions.
The results of recent transdifferentiation experiments suggest that simply modifying the experimental conditions can influence the trajectory of reprogramming. For example, when using epiblast culture conditions, which require Fgf and Activin signals, epiblast stem cells are directly formed . When Fgf is used in combination with Egf instead of Activin, neural progenitors, which are the ectodermal lineage, are readily induced in mouse fibroblasts . Efe et al.  demonstrated that the use of serum in the absence of LIF results in cardiomyocytes commitment, which are the mesodermal lineage. Therefore, by modifying the culture conditions, either iPSCs, induced epiblast stem cells, induced neural cells, or induced cardiac cells are formed using the same factor combination, which includes Oct4. We suggest that these results are likely to be extended to other lineages in the future, and that Oct4 is likely to be the key factor in inducing transdifferentiation, as it was for inducing iPSC formation.
Through reprogramming technologies—nuclear transfer, cell fusion, induced pluripotency, and transdifferentiation—various cell types can be created from donor tissues, including those from patients with known pathologies. Oct4 is capable of inducing transdifferentiation of somatic cells into multiple cell types, including committed progenitors, when used under different experimental conditions. Therefore, we argue that Oct4 is the gatekeeper into a reprogramming expressway.
We would like to give special thanks to Jeanine Müller-Keuker for helping with the figures and Areti Malapetsas for editing the manuscript.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.