The generation of induced pluripotent stem cells from adult somatic cells by ectopic expression of key transcription factors holds significant medical promise. However, current techniques for inducing pluripotency rely on viral infection and are therefore not, at present, viable within a clinical setting. Thus, there is now a need to better understand the molecular basis of stem cell pluripotency and lineage specification in order to investigate alternative methods to induce pluripotency for clinical application. However, the complexity of the underlying molecular circuitry makes this a conceptually difficult task. In order to address these issues, we considered a computational model of transcriptional control of cell fate specification. The model comprises two mutually interacting sub-circuits: a central pluripotency circuit consisting of interactions between stem-cell specific transcription factors OCT4, SOX2 and NANOG coupled to a differentiation circuit consisting of interactions between lineage-specifying master genes.
The molecular switches which arise from feedback loops within these circuits give rise to a well-defined sequence of successive gene restrictions corresponding to a controlled differentiation cascade in response to environmental stimuli. Furthermore, we found that this differentiation cascade is strongly unidirectional: once silenced, core transcription factors cannot easily be reactivated. In the context of induced pluripotency, this indicates that differentiated cells are robustly resistant to reprogramming to a more primitive state. However, our model suggests that under certain circumstances, amplification of low-level fluctuations in transcriptional status (transcriptional “noise”) may be sufficient to trigger reactivation of the core pluripotency switch and reprogramming to a pluripotent state. This interpretation offers an explanation of a number of experimental observations concerning the molecular mechanisms of cellular reprogramming by defined factors and suggests a role for stochasticity in reprogramming of somatic cells to pluripotency.
Citation: MacArthur BD, Please CP, Oreffo ROC (2008) Stochasticity and the Molecular Mechanisms of Induced Pluripotency. PLoS ONE 3(8): e3086. https://doi.org/10.1371/journal.pone.0003086
Editor: Nick Monk, University of Nottingham, United Kingdom
Received: May 13, 2008; Accepted: August 7, 2008; Published: August 28, 2008
Copyright: © 2008 MacArthur et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by EPSRC grant number EP/C003497/1.
Competing interests: The authors have declared that no competing interests exist.
Stem cells are present during all phases of development, from the embryo to the adult, and are characterized by their ability to self-renew indefinitely and differentiate along a variety of distinct lineages. Embryonic stem (ES) cells, which are derived from the inner cell mass of the developing mammalian blastocyst, are pluripotent: they have the ability to generate all embryonic tissues. In contrast, adult stem cells, which reside in small numbers in almost all adult tissues, are generally multipotent: their regenerative potential is tissue or germ-layer specific –. Nevertheless, since adult stem cells have the capacity to initiate specific de novo tissue regeneration subsequent to disease or trauma and may be derived by biopsy with reduced ethical controversy, they are of considerable current clinical and biological research interest.
The traditional conceptual model of cellular differentiation is discrete and hierarchical in nature . In this view, cells can be one of a number of qualitatively different types (stem, progenitor or terminal cell for instance) and differentiation proceeds through a well-defined hierarchy of increasingly committed progenitor cells which act as transit populations between the most pluripotent stem cells and terminally differentiated cells, and serve to facilitate rapid clonal expansion. This process was traditionally thought to progress through a series of irreversible gene restrictions which limited dedifferentiation to more primitive states. However, recent experimental evidence suggests that under certain circumstances “terminally” differentiated cells may retain the capacity to dedifferentiate to more primitive states and possibly trans-differentiate to alternative terminal states , –, although the molecular mechanisms by which this reprogramming occurs remain contentious .
This flexibility in the differentiation hierarchy is commonly known as lineage plasticity  and is perhaps most dramatically demonstrated by molecular reprogramming of adult somatic cells into so-called induced pluripotent stem (iPS) cells , which express the genetic and phenotypic characteristics of pluripotent ES cells. Since iPS cells potentially provide a patient-specific source of pluripotent stem cells they possess significant clinical potential . Furthermore, since they are derived from adult somatic cells which are easily harvested through biopsy, the generation and clinical use of iPS cells is not associated with the same ethical controversies as human ES cells, although they are associated with significant alternative ethical issues . However, current techniques to generate iPS cells rely on viral transfection of key transcription factors–a process which currently carries the inherent risk of insertional genetic mutations–and thus are unsuitable for use in a clinical setting. The search for alternative non-viral methods to generate iPS cells has focused research attention on the molecular basis of pluripotency and lineage plasticity.
Although pluripotency is controlled by multiple factors , in recent years the homeodomain transcription factors OCT4 and NANOG and the HMG-box transcription factor SOX2 have emerged as playing central roles in the maintenance of ES cell identity both in mice and humans –. For example, Boyer and co-workers used chromatin immunoprecipitation and genome-scale DNA microarrays to explore the core transcriptional circuitry in human ES cells , while Loh and co-workers used similar methods to identify OCT4, and NANOG targets in mouse ES cells . Taken together, three key findings emerge from these studies: (1) the architecture of the core OCT4/SOX2/NANOG pluripotency circuit is essentially conserved between mice and humans . (2) These three factors interact with each other in a coordinated manner to form a tightly regulated pluripotency circuit. In particular, the OCT4 and SOX2 proteins form a heterodimer which positively regulates the expression of all three of these core transcription factors  as well as other targets , . Similarly NANOG also positively regulates expression of all three pluripotency factors . Since the binding loci of these genes are often almost identical  and multi-protein complexes containing OCT4 and NANOG may be produced by iterative immunoprecipitation in ES cells , evidence suggests that these three factors most-likely regulate gene expression in a cooperative manner, as part of multi-protein complexes . (3) OCT4, SOX2 and NANOG co-occupy a large set of developmentally significant target genes. In particular, they repress expression of genes associated with cellular differentiation and lineage commitment while activating expression of genes associated with self-renewal and pluripotency, including key transcription factors and members of the TGF-β and Wnt signaling pathways , . As with their regulation of each other, it is likely that they also often regulate these downstream target genes in a coordinated manner . Thus, cooperative interactions between OCT4, SOX2 and NANOG are central to maintenance of the ES cell identity: appropriate expression of these 3 factors holds the cell in a pluripotent self-renewing state by activating ES cell-specific pluripotency genes and suppressing differentiation genes; while loss of expression results in up-regulation of differentiation genes, and loss of the pluripotent stem cell identity.
Commensurate with their roles in maintenance of the ES cell state, OCT4 and SOX2 also appear to play a central role in inducing pluripotency in somatic cells . In a series of break-through papers Takahashi, Yamanaka and co-workers found that retroviral infection with just four factors–OCT4, SOX2, KLF4 and c-MYC–was sufficient to transform both adult mouse and adult human fibroblastic cells to a pluripotent ES cell-like state , . In particular, they found that subsequent to transfection, drug selection for expression of the OCT4 target Fbx15  isolated a sub-population of cells which possessed many ES cell characteristics including the same gross morphology, the ability to form teratomas subsequent to subcutaneous injection in nude mice and the ability to differentiate along all 3 germ layers in vitro. However, they also found that these cells differed from ES cells in a number of crucial respects. In particular, they were unable to generate live chimeras and they differed from ES cells both in genetic and epigenetic profiles, indicating that these “first-generation” Fbx15-iPS cells are similar, but not identical to, ES cells.
Since these initial reports, the induction process has been refined by a number of groups –. In particular, iPS cells have been generated from alternative cell types including adult hepatocytes, gastric epithelial cells , and mesenchymal cells ; iPS cells have been generated without transfection with the proto-oncogene c-MYC , , ; the drug-selection process–which requires insertion of a drug-resistance gene into endogenous loci, and thus carries the risk of insertional mutations–has been replaced with selection on morphological criteria ; alternative combinations of key transcription factors which are sufficient to induce pluripotency have been identified ; and the efficiency of the reprogramming process has been improved with the use of small molecules such as DNA and histone methyltransferase inhibitors ,  and histone deacetylase inhibitors . Crucially, selection for OCT4 or NANOG expression rather than Fbx15 has been found to yield a more completely reprogrammed state , , , . The resulting iPS cells are characteristically ES-like in morphology, proliferative properties, global genetic and epigenetic status , and in their response to key factors such as retinoic acid and leukemia inhibitory factor . Additionally not only do they have the capacity to form teratomas in vivo , , they also support generation of viable live chimeras ,  and late gestation “all iPS cell” embryos through tetraploid complementation . These results currently suggest that OCT4- and NANOG-iPS cells are functionally indistinguishable from ES cells.
Taken together these reports also highlight a number of key observations which any explanation of the reprogramming process should address . (1) OCT4 and SOX2 appear to be essential to the reprogramming process, but additional factors such as c-MYC, KLF4, LIN28 appear to only to improve efficiency , , , . Furthermore, NANOG, although a core determinant of the ES cell identity, also appears to be dispensable , , , . (2) Viral gene expression is needed to induce pluripotency, however maintenance of the undifferentiated state is not reliant on continued transgene expression, but rather is maintained by endogenous gene expression , , , . (3) Epigenetic reprogramming is important in establishing and maintaining the induced pluripotent state , , .
In this paper we shall use a computational model of stem cell differentiation to explore the molecular basis of cell fate specification and reprogramming by defined factors.
The molecular mechanisms which underpin cell fate specification are inherently complex and difficult to interpret using experiment and intuition alone. Consequently, many authors have considered computational models of various aspects of cellular differentiation (see – for instance). For example, within a theoretical context it has long been suggested that distinct cell types may correspond to attractors of (generally high dimensional) genetic regulatory networks , ; an idea which has had some recent experimental validation in mammalian cells , . Since many cell types naturally coexist in the body, the notion of a cell type as an attractor implies that the underlying dynamical regulatory system possess co-existing attractors. Consequently, differentiation has been extensively interpreted in terms of switching between co-existing attractors ,  and regulatory architectures which give rise to multistability have been discussed at length (see  and ,  and the references therein). Experimental evidence for switching between multiple discrete states during mammalian cell differentiation has also recently been provided , .
In order to better understand stem cell differentiation and reprogramming we constructed a computational model of transcriptional control of stem cell differentiation which uses these ideas and builds upon this literature. As an exemplar system we considered differentiation along the principle stromal lineages, although the regulatory architecture we shall describe occurs in a similar form in other contexts. Since mesenchymal cells have been reprogrammed to a pluripotent state  this represents a relevant example.
In addition to the core OCT4/SOX2/NANOG pluripotency circuit discussed in the introduction we also constructed an extended regulatory network for differentiation along the osteogenic (bone), chondrogenic (cartilage) and adipogenic (fat) lineages, based upon data obtained from an extensive literature search. The network we obtained consists of direct and indirect interactions between RUNX2, SOX9 and PPAR-γ, the core lineage-specifying master genes (LSMGs) for osteogenesis, chondrogenesis, and adipogenesis respectively. The architecture of this extended network is given in Fig. 1a. In order to identify dominant interactions between the core LSMGs we considered a simple coarse-graining which emerges naturally from this extended network (see Fig. 1b). In particular, each of the core LSMGs auto-activates its own production while cross-repressing expression of the other two, either directly or through molecular intermediaries.
Arrows indicate up-regulation, bars indicate down-regulation.
So, for example, BMP2 up-regulates expression of RUNX2 while simultaneously down-regulating PPAR-γ through activation of the intermediary transcription factor TAZ , . Additionally, RUNX2 expression increases sensitivity to exogenous BMP2, for example via regulation of BMP receptors or SMAD signaling . Thus RUNX2 auto-activates its own expression via a positive feedback loop with BMP2 and TAZ. Conversely, PPAR-γ while a potent activator of adipogenesis also strongly inhibits osteogenesis ,  by both direct suppression of RUNX2 expression and by altering the potential of RUNX2 to activate downstream osteogenic products , . In contrast, PPAR-γ and the CCAAT/enhancer binding protein C/EBP-α positively regulate each others expression ; thus PPAR-γ, like RUNX2, also indirectly positively regulates its own production, this time through a positive feedback loop with C/EBP-α.
Although such a coarse-graining naturally excludes effects such as time-delays produced by indirect feedback loops, it allows us to study interactions between core transcription factors in a biologically and mathematically transparent manner. Full details of the extended regulatory network for differentiation along these stromal lineages, and its coarse-graining may be found in the supplementary materials (Text S1). We note here that the mutual cross-inhibition which emerges from this coarse-graining is also seen in other examples of cell fate specification including haematopoiesis  and specification of neuronal subtypes . Additionally, auto-activation of lineage-determining genes is also common (for example, in the context of haematopoiesis it has been found that the myeloid determinant PU.1 and the erythroid determinant GATA1 are both directly auto-stimulatory ). Thus, the coarse-grained logic we consider (auto-activation and mutual cross-repression) is not specific to differentiation along the stromal lineages, but rather may represent a widespread form of transcriptional regulation of cell fate specification.
Molecular interactions between the core pluripotency factors (OCT4, SOX2 and NANOG) and the stromal lineage-specifying master genes (RUNX2, SOX9 and PPAR-γ) remain to be fully characterized. Therefore in order to connect the lineage-specifying circuit to the core pluripotency circuit, we reasoned that since RUNX2, SOX9 and PPAR-γ are master-genes associated with differentiation, they are suppressed by OCT4, SOX2 and NANOG, most likely through intermediate products ,  and similarly that RUNX2, SOX9 and PPAR-γ all suppress OCT4, SOX2 and NANOG, most likely through downstream secreted growth factors. The resulting core transcriptional circuitry for differentiation from pluripotent ES cell along the principle stromal lineages is given in Fig. 2.
Arrows indicate up-regulation, bars indicate down-regulation.
In order to study the dynamic behavior of this integrated network, we developed a computational model which accounts for the biological processes of interest. The model consists of a set of six coupled differential equations. For simplicity we present the model using an indexed notation. Hence, [Pi] gives the nuclear concentration of the product of the ith pluripotency gene (PG) where P1 = OCT4, P2 = SOX2 and P3 = NANOG. Similarly, [Li] gives the nuclear concentration of the ith lineage-specifying factor, where L1 = RUNX2, L2 = SOX9 and L3 = PPAR-γ. In order to model the effect of the extracellular environment on cell fate specification, we also accounted for the effects of various exogenous stimuli on this core circuit. Hence, si represents a combination of growth factors which stimulates differentiation along the ith lineage. So, for example, murine ES cells are stimulated to osteogenesis through up-regulation of RUNX2 by retinoic acid (RA)+BMP4; to chondrogenesis through up-regulation of SOX9 by RA+TGF-β; and to adipogenesis through up-regulation of PPAR-γ RA+Insulin . Thus, we may consider s1 = [RA+BMP4], s2 = [RA+TGF-β] and s3 = [RA+Insulin]. Note that although these three stimuli are lineage-specific, they all contain a common element (RA), thus they all partially stimulate all three lineages. Furthermore, since RA suppresses OCT4, SOX2 and NANOG expression , all three differentiation stimuli also suppress the core pluripotency circuit.
Based upon the logic above, on the structure of Fig. 2 and subsequent to biologically realistic simplifying assumptions (see supplementary Text S1) we obtained the following system of equations.(1)(2)Full details of the derivation of this model and a biological interpretation of the model parameters can be found in the supplementary materials (Text S1). A model of the core OCT4/SOX2/NANOG pluripotency circuit based upon similar assumptions has been considered elsewhere .
We may summarize the transcriptional logic given in Fig. 2 and Eqns. (1–2) as follows: the transcription of each core PG is activated by OCT4-SOX2 and OCT4-SOX2-NANOG multi-protein complexes . Conversely, the transcription of each LSMG is suppressed by the OCT4-SOX2 and OCT4-SOX2-NANOG multi-protein complexes. Additionally, each LSMG auto-activates its own production, while cross-repressing that of the other LSMGs and the core PGs. Furthermore, each LSMG is up-regulated by environmental (specific and non-specific) differentiation stimuli; while each PG is down-regulated by environmental differentiation stimuli.
Differentiation: An Irreversible Sequence of Controlled Gene Restrictions
We have conducted an extensive numerical investigation of Eqns. (1–2) and a full mathematical analysis of a simplified system (see below and supplementary materials Text S1). Here we summarize the key findings of these analyses, and discuss their biological implications. We shall focus on stable steady-state (equilibrium) solutions of this computational model. Since numerical simulations suggest that this model does not exhibit limit cycles or chaotic trajectories such equilibria represent the primary behavior of the system after any brief transients.
In accordance with the notion of a cell type as an attractor ,  we associate stable equilibria with fixed cellular phenotypes. In common with other models of cellular differentiation, the system given by Eqns. (1–2) allows multiple stable steady-states (cell types) to coexist in the same environment. This multistability derives from the positive feedback loops in the transcriptional circuitry  and results in a sequence of genetic “switches” (that is, binary responses to continuously graded stimulus), which correspond to a sequence of successive gene restrictions.
In order to consider the molecular basis of differentiation from a pluripotent state to a terminal state, we began by considering the response of a cell initially in a pluripotent state to increasing environmental differentiation stimulus. For simplicity we consider a pluripotent ES cell in culture in which osteogenesis is initiated by addition of s1 = RA+BMP4 to the culture media in increasing doses. Since the equations have the same essential form for each lineage, our conclusions also hold quantitatively for chondrogenic induction by RA+TGF-β and adipogenic induction by RA+Insulin.
In the pluripotent state all three PGs are active (in Boolean terms, they are on) and all three LSMGs are inactive (in Boolean terms, they are off): numerical simulations suggest that this state is stable as long as the environmental stimulus (in this case the concentration of RA+BMP4) is not too high. However, as the environmental stimulus is increased the PGs are increasingly suppressed, and their expression level gradually falls. This continues until a threshold is reached, at which point the PGs can no longer maintain suppression of the LSMGs any further and one or more of the LSMGs switch on.
Numerical simulations suggest that at this point the system can display a number of different behaviors, depending upon the parameter regime. However, in all cases, we found that differentiation proceeds through a sequence of irreversible switches, corresponding to irreversible gene restrictions. Similar irreversible switches due to positive feedback loops have been observed both theoretically and experimentally in other developmental contexts (for an excellent example which combines both experiments and modeling see ). Fig. 3 shows examples of the various gene restriction sequences which this model exhibits for different parameter regimes. We have also included video animations of these bifurcation diagrams in the supplementary materials (Videos S1, S2, S3, S4), which present these switching dynamics in a more intuitive manner.
In all panels blue indicates equilibrium OCT4/SOX2 expression; red indicates equilibrium NANOG expression; green indicates equilibrium SOX9 expression; dashed purple indicates equilibrium PPAR-γ expression; and black indicates equilibrium RUNX2 expression. The vertical grey dotted lines mark the points when a differentiation event occurs. These figures should be read left to right since they illustrate stimulus increasing with time. (a) Differentiation straight from a pluripotent state to a terminal ostoblastic state (model parameter values: k0 = 10, k11 = 1, k12 = 0.5, k2 = 1, k3 = 0.9, kPL = 2.5, , , kLL = 1, m = 0.875, b = 0.1); (b) Differentiation straight from a pluripotent state to a terminal ostoblastic state via a primed state (model parameter values: k0 = 5, k11 = 1, k12 = 0.5, k2 = 5, k3 = 0.9, kPL = 0.5, , , kLL = 1, m = 0.625, b = 0.1); (c) Differentiation from a pluripotent state to a terminal osteoblastic state through a hierarchy of increasingly committed tissue-specific progenitors (model parameter values: k0 = 25, k11 = 1, k12 = 0.5, k2 = 3, k3 = 0.9, kPL = 10, , , kLL = 1, m = 0.75, b = 0.1); (d) Differentiation from a pluripotent state to a terminal osteoblastic through a hierarchy of increasingly committed tissue-specific progenitors via a primed state (model parameter values: k0 = 20, k11 = 1, k12 = 0.5, k2 = 7, k3 = 0.9, kPL = 0.1, , , kLL = 0.75, m = 0.575, b = 0.1). Video versions of these bifurcation diagrams are given in the supplementary materials (Videos S1, S2, S3, S4). Details of the biological meaning of each of the model parameters are given in the supplementary materials (Text S1).
For some parameter values we found that the cell differentiates directly from a pluripotent state to a terminal osteoblastic state in which RUNX2 alone is active (see Fig. 3a); while for other parameter values we found that differentiation proceeds through a sequence of intermediary states in which competing LSMGs are co-expressed before the terminal state is obtained (see Fig. 3c,d). Promiscuous expression of competing master genes has been discussed previously and associated with developmental multipotency prior to lineage commitment . For example, concurrent expression of RUNX2, SOX9 and PPAR-γ has been observed in the early stages of murine ES cell differentiation along the osteogenic lineage and associated with osteogenic/chondrogenic/adipogenic multipotency . Similarly, co-expression of GATA1 and PU.1 has also been observed during haematopoiesis and similarly associated with multipotency . In accordance with these observations, we associated the state in which all three LSMGs are concurrently expressed (but the PGs are off) with a tripotent state, and the state in which all two of the three LSMGs are concurrently expressed (but the PGs are off) with a bipotent state. In both the tripotent and bipotent states, pluripotency has been lost but some limited tissue-specific regenerative potential is retained.
For some parameter regimes, we found that the system can also adopt a state in which the PGs and LSMGs are co-expressed at a low level. Of all the switches we observed, only the switch from the pluripotent state to this PG/LSMG co-expressing state, was reversible: removal of stimulus from a cell in this state results in immediate transition back to the pluripotent state. Thus, we interpret the PG/LSMG co-expressing state as corresponding to a primed state in which the cell is preparing to commit to differentiation but commitment to a differentiating state has not yet been made. Such primed states, in which multiple LSMGs are co-expressed, have been suggested as a characteristic feature of tissue-specific (haematopoietic) stem cells , , ; similarly, evidence that key genes may be primed for activation in ES cells has also been presented , .
In order to investigate the nature of the switches which were observed during differentiation, we examined the stability of the various cell types–pluripotent, tripotent, bipotent and terminal–as environmental stimulus is varied. Fig. 4. shows the stability of the various cell types for a biologically representative parameter regime. Importantly, for low levels of stimulus, all cell types are concurrently stable. As environmental stimulus is increased, the cell types lose stability sequentially in order of developmental potency. Thus, for high levels of stimulus only the terminal state is stable; while for intermediate levels of stimulus the tripotent, bipotent and terminal states are concurrently stable; and for low stimulus all four cell states are concurrently stable.
The y-axis denotes total equilibrium gene expression ([P1]+[P2]+[P3]+[L1]+[L2]+[L3]). Bold lines indicate stable solutions; dotted lines indicate unstable solutions. Note that the cell types lose stability sequentially in order of developmental potency. Note also that for low-levels of stimuli all four cell types are concurrently stable: thus, the sequence of gene restrictions we observe are irreversible. Model parameter values are as in Fig. 3c.
In order to investigate the nature of the switches which occur during differentiation more rigorously we also considered behavior of a related (but highly simplified) model analytically. In this simplified model the core PGs are inactive and the interactions between the LSMGs as given in Fig. 1b in response to a generic differentiation stimulus a are considered. Thus, this model focuses on the series of secondary bifurcations which occur after the core pluripotency circuit has been switched off. In particular, we considered the following system of equations:(3)Full details of our mathematical analysis of this model are provided in the supplementary materials (Text S1). A similar model of transcriptional control of differentiation is considered in , .
In supplementary Text S1 we show analytically that this simplified model supports a monotonic increasing sequence of bifurcation points, in which cell types lose stability sequentially: For 0≤a<a1 the multipotent state in which all three LSMGs are simultaneously expressed is stable; for 0≤a<a2 (where a2>a1) the bipotent states in which 2 of the LSMGs are co-expressed are stable; and for 0≤a<a3 (where a3>a2) the terminal states in which only one LSMG is expressed are all stable. Consequently, this simplified model exhibits multistability: for 0≤a<a1 all three cell types (multipotent, bipotent and terminal) are simultaneously stable. However, as stimulus is increased above the threshold a1 the multipotent state becomes unstable, gene expression is restricted and the cell is forced to adopt a bipotent state. For 0≤a<a2 the bipotent and terminal states are simultaneously stable but the multipotent state is unstable. However as stimulus is increased above the threshold a2 the bipotent state also becomes unstable, gene expression is further restricted and the cell is forced into a terminal state. Since the terminal states are stable for a<a3, removal of environmental stimulus does not result in dedifferentiation to the bipotent or multipotent states. Similarly, since a2>a1, once the cell has adopted a bipotent state it cannot dedifferentiate to a multipotent state. We believe that a similar monotonic increasing sequence of bifurcation points underpins the irreversible switching dynamics of the full model and in particular that the full system does not exhibit hysteresis within a biologically relevant parameter regime.
The central point to note from this discussion is that the molecular switches which arise from feedback loops in the core regulatory circuitry naturally give rise to a well-defined sequence of successive and irreversible gene restrictions corresponding to a directed differentiation cascade in response to specific environmental stimuli. However, the analysis we have given so far does not account for stochastic fluctuations in transcriptional status. In fact, gene expression is an inherently stochastic process , and transcriptional “noise” can profoundly affect cell fate decisions –. In the following section we explore a role for transcriptional noise in triggering stochastic transitions between co-existing attractors. In particular, we focus on exploring noise-driven transitions between the terminal state and the pluripotent state with the aim of better understanding the molecular basis of induced pluripotency.
Cellular Reprogramming: A Role for Stochasticity
In order to consider the effect of molecular noise on system dynamics it is convenient to think of the various cell types we have identified as local minima of an energy landscape which is continually being shaped by external stimuli, and the behavior of the cell as a particle moving through this landscape to minimize its energy –. In this view, the local minima corresponding to the pluripotent, multipotent and bipotent cell types become increasingly shallow as differentiation stimulus is increased, until at the critical threshold points they disappear in consecutive order. Thus, a cell initially at the minimum corresponding to the pluripotent state is forced to transition through a hierarchy of local minima as environmental stimulus is increased, until finally it arrives at the “terminal” minimum corresponding to the fully differentiated state. If, at this point, environmental stimulus is gradually removed, then the local minima corresponding to the primitive cell types reappear sequentially in reverse order. However, since the switches we have identified are irreversible, the terminal minimum does not disappear upon removal of stimulus and the cell must therefore overcome an energy barrier if it is to escape the terminal state.
Stochastic fluctuations in transcriptional status (transcriptional “noise”) may be thought of, in the first instance, by analogy to the temperature of the particle in the energy landscape. If the terminal minimum is deep and/or fluctuations in transcriptional status are small (for example, if they are suppressed by epigenetic or other means) then stochastic transitions from the terminal state to alternative (more primitive) states will be rare . However, if the terminal minimum is shallow and/or fluctuations in transcriptional status are large then stochastic transitions from the terminal state to alternative states will be more common. Thus, the frequency and type of noise-driven transitions between cell states depends both upon the structure of the attractor landscape and the form and amplitude of noise in the system. The notion of noise-driven transitions between co-existing attractor states has been well explored theoretically and experimentally in model organisms (see the recent reviews ,  and references therein); and has recently had some experimental demonstration in mammalian cells .
In the previous section we examined the behavior of a cell initially in a pluripotent state to a gradual increase in environmental stimulus and found that differentiation occurs through an irreversible sequence of gene restrictions. In this section, we examine the behavior of a cell initially in a differentiated state to removal of differentiation stimulus, and investigate a role for stochasticity in triggering transitions from a differentiated state to the pluripotent state.
Fig. 4 illustrates the stability of the various cell types for varying environmental stimulus and shows that at low levels of environmental stimulus the differentiated (osteoblastic) cell adopts a state in which RUNX2 is expressed at a low level (and all other PGs and LSMGs are OFF); and that upon complete removal of stimulus RUNX2 expression also tends to zero. Linear stability analysis shows that in the absence of environmental stimulus this state (the origin) is stable for all model parameter values as long as the protein decay-rate b>0. Furthermore, analysis of Eqns. (1–2) shows that in the absence of environmental stimulus, only the origin and the pluripotent state are stable.
In order to investigate the effect of transcriptional noise on system dynamics we included a time-dependent stochastic term in Eqns. (1–2). In vector notation we denote the state of the system by x = [P1, P2, P3, L1, L2, L3] and the right-hand-side of Eqns. (1–2) by F(x). Thus, we considered the following set of stochastic differential equations:(4)Here σ is a diagonal matrix of constants representing the amplitude of noise, where σii gives the amplitude of the noise in expression of the ith gene, and W denotes a Weiner process (Brownian motion).
Fig. 5 shows some representative simulations of Eqns. 4 illustrating the behavior of a terminally differentiated osteoblast upon complete removal of differentiation stimuli (that is, starting from the origin, in the absence of environmental stimulus). In each panel 50 separate simulations are shown–each of which may be thought of as the behavior of an individual cell within an isogenic population–in which all genes are subjected to the same level of noise (σii = σ for all i). For low-levels of transcriptional noise stochastic transitions to the pluripotent state do not occur (Fig. 5a). However, as the level of noise in the system is increased, stochastic transitions to the pluripotent state increase in frequency (Fig. 5b–d). Finally, almost all simulations result in a transition to the pluripotent state (Fig. 5d).
In each panel 50 representative simulations are shown in which the expression levels of the LSMGs are given in red; the expression of NANOG is given in blue; and the expression levels of OCT4 and SOX2 are given in black. In each panel the same amplitude of noise is applied to all 6 genes (in Eqns. 4 σii = σ for all i). (a) σ = 0.01; (b) σ = 0.025; (c) σ = 0.1; (d) σ = 0.25. Model parameter values are as in Fig. 3c.
In order to further explore the role of stochasticity in cellular reprogramming we investigated how reprogramming efficiency varies with both the form and amplitude of transcriptional noise. Fig. 6 shows the results of these investigations. Fig 6a illustrates reprogramming efficiency (the fraction of simulations which resulted in a noise-driven transition to the pluripotent state) in the presence of noise in the expression of the PGs only (blue); and in the presence of noise in the expression of OCT4 and SOX2 only (red). This figure shows that only very low-level stochastic fluctuations in PG levels are needed to efficiently reprogram terminally differentiated cells to the pluripotent state. The fact that reprogramming efficiency does not change significantly when fluctuations in NANOG are suppressed suggests that NANOG is not required to reactivate the pluripotency switch, and therefore is not required for reprogramming. Fig. 6b illustrates reprogramming efficiency in the presence of noise in all 6 genes (blue), and in the presence of noise in the LSMGs only (green), and NANOG only (red). This figure shows that noise in the expression of the LSMGs or NANOG alone is unable to trigger stochastic transitions to the pluripotent state. However, it also illustrates that amplification of noise in the expression of all 6 genes is sufficient to trigger transitions to the pluripotent state, albeit less efficiently than by targeted amplification of OCT4 and SOX2 noise (compare Figs. 6a–b). This reduction in efficiency occurs since fluctuations in the PGs and the LSMGs antagonize each other: fluctuations in OCT4/SOX2 expression serve to activate the pluripotency switch while fluctuations in the LSMGs serve to suppress the pluripotency switch.
(a) Noise on all PGs or OCT4/SOX2 alone (but not the LSMGs) results in reprogramming to the pluripotent state; (b) Noise in expression of the LSMGs or NANOG alone does not result in reprogramming; noise in expression of all 6 genes results in reprogramming albeit less efficiently than by targeted amplification of OCT4 and SOX2 noise. In each case, the fraction of cells reprogrammed after 20000 time-steps is given and the results from 1000 simulations are shown, except for the case in which there is noise on all 6 genes, where the results from 5000 simulations are shown.
Taken together Figs. 5–6 illustrate 2 phenomena. (1) in the absence of environmental stimulus to differentiate (for example in the ES cell culture conditions), targeted upregulation of OCT4 and SOX2 protein levels is necessary and sufficient to reprogram a differentiated cell to the pluripotent state. (2) In the absence of environmental stimulus to differentiate, non-specific widespread amplification of transcriptional noise–in this case, amplification of noise in the expression of all 6 genes–is also sufficient to trigger stochastic transitions to the pluripotent state. Thus, these simulations suggest that elevated levels of OCT4 and SOX2 protein levels are required for reprogramming. However, they also suggest that it is not necessary to target OCT4 and SOX2 specifically, but rather reprogramming may also be achieved by widespread non-specific amplification of transcriptional noise.
The molecular mechanisms of cell fate specification and reprogramming are clearly much more complex than accounted for in our simple mathematical model. Nevertheless, simplified models, such as the one presented, can be useful in conceptualizing the behavior of complex systems and can lead to insight not immediately available by experiment alone. With this caveat in mind, our results offer an explanation of a number of experimental observations concerning the molecular basis of differentiation and reprogramming by defined factors. In particular, the model we have presented shows that the molecular switches which arise from feedback loops within core transcriptional circuitry naturally give rise to a well-defined sequence of successive irreversible gene restrictions which correspond to a controlled differentiation cascade in response to environmental stimulus. Thus, in the absence of molecular fluctuations in transcriptional status, differentiation is unidirectional and cells are strongly resistant to reprogramming to a more primitive state. However, we found that under the right environmental circumstances, both targeted and non-specific amplification of stochastic fluctuations in transcriptional status was sufficient to trigger cellular reprogramming to a pluripotent state.
In the context of induced pluripotency, a central conclusion of this work is that in order to reprogram a terminally differentiated cell to a more primitive state it is necessary to supply it with sufficient “energy” to overcome the barrier holding it in the differentiated state. Current reprogramming techniques achieve this by inducing high levels of OCT4 and SOX2 expression from viral transgenes which in turn reactivate the endogenous pluripotency switch. However, our results suggest that since only low-level transient fluctuations in OCT4 and SOX2 protein levels are required, it is possible that this may be achieved through alternative, less invasive, methods. The fact that transient expression of the OCT4 and SOX2 proteins is sufficient to reestablish sustained OCT4, SOX2 and NANOG gene expression explains the experimental observation that transient retroviral gene expression is needed to establish to reprogrammed state, yet pluripotency is maintained by sustained endogenous gene expression , , , . Similarly, this result also explains why ectopic expression of OCT4 and SOX2 appear necessary for the reprogramming process but NANOG appears dispensable , , , .
Previous authors have observed that subsequent to early cellular differentiation core PGs such as OCT4 undergo a robust multi-step silencing procedure beginning with transcriptional repression followed by an increase in histone H3 methylation and local heterochromatinization , . This stable form of epigenetic silencing is important since it prevents harmful ectopic reactivation of PGs (ectopic activation of OCT4 results in dysplasia in gastric epithelial tissues , for example). However, it also ensures that the low-level fluctuations in OCT4 and SOX2 protein levels needed for reprogramming are not easy to induce from endogenous genes. Thus, stable epigenetic silencing of key PGs such as OCT4 and SOX2 effectively suppresses spontaneous reprogramming by heavily silencing molecular fluctuations in the nuclear levels of these proteins.
Previous authors have hypothesized that additional factors such as c-MYC and KLF4 facilitate reprogramming by modulating accessibility of the OCT4 and SOX2 loci , . For example, c-MYC occupation is associated with genome-scale alteration of chromatin structure and histone accessibility , , possibly by stimulation of DNA replication  or by binding to multiple sites ; while KLF4 regulates histone acetylation , . Recent reports demonstrating the ability of small molecules such as DNA/histone methyltransferase inhibitors and histone deacetylase inhibitors to improve reprogramming efficiency – appear to support this view. Our results also support this hypothesis and suggest that by regulating OCT4 and SOX2 accessibility, additional reprogramming factors such as these increase sensitivity of the core pluripotency circuit to OCT4 and SOX2 protein levels thus increasing the probability of transitions to the pluripotent state.
In addition to confirming a central role for OCT4 and SOX2 in reprogramming by defined factors, our model also suggests that non-specific widespread amplification of transcriptional noise may aid reprogramming of somatic cells to a pluripotent state. Rather than targeted activation of the core pluripotency circuit by defined factors, this approach may be thought of as giving a non-specific “shake” to the system in order to realign it to the pluripotent ground state. Since a role for transcriptome-wide noise in defining mammalian cell fates has recently been experimentally demonstrated , we surmise that noise-processing mechanisms such as the proteasome  (which is involved in protein homeostasis and is used to target pre-initiation complexes and minimize noise in differentiation genes in ES cells ) and the Wnt signaling pathway (which has been implicated in filtering transcriptional noise ) may prove fruitful targets for future research into improving the efficiency of reprogramming protocols. Such approaches are currently under experimental investigation in our laboratory.
We conclude by noting that although iPS cells are not associated with the same ethical concerns as the derivation of human ES cells, they are nevertheless associated with their own set of significant ethical considerations . It is important that these are addressed concurrently with research in this area.
Transcriptional control of the mesenchymal lineages; Derivation of model equations; Mathematical details
(0.16 MB PDF)
A video version of Fig. 3a. Green indicates gene expression, red indicates [RA+BMP4].
(1.27 MB AVI)
A video version of Fig. 3b. Green indicates gene expression, red indicates [RA+BMP4].
(1.30 MB AVI)
A video version of Fig. 3c. Green indicates gene expression, red indicates [RA+BMP4].
(1.04 MB AVI)
Conceived and designed the experiments: BDM. Analyzed the data: BDM CPP ROO. Wrote the paper: BDM. Helped write the paper: CPP ROCO. Oversaw the project: ROCO. Conceived, designed and analyzed the mathematical model: BDM.
- 1. Gage FH (2000) Mammalian neural stem cells. Science 287: 1433–1438.FH Gage2000Mammalian neural stem cells.Science28714331438
- 2. Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, et al. (1999) Multilineage potential of adult human mesenchymal stem cells. Science 284: 143–147.MF PittengerAM MackaySC BeckRK JaiswalR. Douglas1999Multilineage potential of adult human mesenchymal stem cells.Science284143147
- 3. Wagers AJ, Weissman IL (2004) Plasticity of adult stem cells. Cell 116: 639–648.AJ WagersIL Weissman2004Plasticity of adult stem cells.Cell116639648
- 4. Hall PA, Watt FM (1989) Stem-Cells - the Generation and Maintenance of Cellular Diversity. Development 106: 619–633.PA HallFM Watt1989Stem-Cells - the Generation and Maintenance of Cellular Diversity.Development106619633
- 5. Blau HM, Brazelton TR, Weimann JM (2001) The evolving concept of a stem cell: Entity or function? Cell 105: 829–841.HM BlauTR BrazeltonJM Weimann2001The evolving concept of a stem cell: Entity or function?Cell105829841
- 6. Graf T (2002) Differentiation plasticity of hematopoietic cells. Blood 99: 3089–3101.T. Graf2002Differentiation plasticity of hematopoietic cells.Blood9930893101
- 7. Theise ND, Wilmut I (2003) Flexible arrangement. Nature 425: 21.ND TheiseI. Wilmut2003Flexible arrangement.Nature42521
- 8. Tosh D, Slack JMW (2002) How cells change their phenotype. Nature Reviews Molecular Cell Biology 3: 187–194.D. ToshJMW Slack2002How cells change their phenotype.Nature Reviews Molecular Cell Biology3187194
- 9. Phinney DG, Prockop DJ (2007) Concise review: Mesenchymal stem/multipotent stromal cells: The state of transdifferentiation and modes of tissue repair - Current views. Stem Cells 25: 2896–2902.DG PhinneyDJ Prockop2007Concise review: Mesenchymal stem/multipotent stromal cells: The state of transdifferentiation and modes of tissue repair - Current views.Stem Cells2528962902
- 10. Takahashi K, Yamanaka S (2006) Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell 126: 663–676.K. TakahashiS. Yamanaka2006Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors.Cell126663676
- 11. Hanna J, Wernig M, Markoulaki S, Sun CW, Meissner A, et al. (2007) Treatment of sickle cell anemia mouse model with iPS cells generated from autologous skin. Science 318: 1920–1923.J. HannaM. WernigS. MarkoulakiCW SunA. Meissner2007Treatment of sickle cell anemia mouse model with iPS cells generated from autologous skin.Science31819201923
- 12. Editorial (2008) New sources of sex cells. Nature 452: 913.Editorial2008New sources of sex cells.Nature452913
- 13. Kim JW, Chu JL, Shen XH, Wang JL, Orkin SH (2008) An extended transcriptional network for pluripotency of embryonic stem cells. Cell 132: 1049–1061.JW KimJL ChuXH ShenJL WangSH Orkin2008An extended transcriptional network for pluripotency of embryonic stem cells.Cell13210491061
- 14. Avilion AA, Nicolis SK, Pevny LH, Perez L, Vivian N, et al. (2003) Multipotent cell lineages in early mouse development depend on SOX2 function. Genes & Development 17: 126–140.AA AvilionSK NicolisLH PevnyL. PerezN. Vivian2003Multipotent cell lineages in early mouse development depend on SOX2 function.Genes & Development17126140
- 15. Boyer LA, Lee TI, Cole MF, Johnstone SE, Levine SS, et al. (2005) Core transcriptional regulatory circuitry in human embryonic stem cells. Cell 122: 947–956.LA BoyerTI LeeMF ColeSE JohnstoneSS Levine2005Core transcriptional regulatory circuitry in human embryonic stem cells.Cell122947956
- 16. Chambers I, Colby D, Robertson M, Nichols J, Lee S, et al. (2003) Functional expression cloning of Nanog, a pluripotency sustaining factor in embryonic stem cells. Cell 113: 643–655.I. ChambersD. ColbyM. RobertsonJ. NicholsS. Lee2003Functional expression cloning of Nanog, a pluripotency sustaining factor in embryonic stem cells.Cell113643655
- 17. Jaenisch R, Young R (2008) Stem cells, the molecular circuitry of pluripotency and nuclear reprogramming. Cell 132: 567–582.R. JaenischR. Young2008Stem cells, the molecular circuitry of pluripotency and nuclear reprogramming.Cell132567582
- 18. Loh YH, Wu Q, Chew JL, Vega VB, Zhang WW, et al. (2006) The Oct4 and Nanog transcription network regulates pluripotency in mouse embryonic stem cells. Nature Genetics 38: 431–440.YH LohQ. WuJL ChewVB VegaWW Zhang2006The Oct4 and Nanog transcription network regulates pluripotency in mouse embryonic stem cells.Nature Genetics38431440
- 19. Mitsui K, Tokuzawa Y, Itoh H, Segawa K, Murakami M, et al. (2003) The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells. Cell 113: 631–642.K. MitsuiY. TokuzawaH. ItohK. SegawaM. Murakami2003The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells.Cell113631642
- 20. Nichols J, Zevnik B, Anastassiadis K, Niwa H, Klewe-Nebenius D, et al. (1998) Formation of pluripotent stem cells in the mammalian embryo depends on the POU transcription factor Oct4. Cell 95: 379–391.J. NicholsB. ZevnikK. AnastassiadisH. NiwaD. Klewe-Nebenius1998Formation of pluripotent stem cells in the mammalian embryo depends on the POU transcription factor Oct4.Cell95379391
- 21. Nishimoto M, Fukushima A, Okuda A, Muramatsu M (1999) The gene for the embryonic stem cell coactivator UTF1 carries a regulatory element which selectively interacts with a complex composed of Oct-3/4 and Sox-2. Molecular and Cellular Biology 19: 5453–5465.M. NishimotoA. FukushimaA. OkudaM. Muramatsu1999The gene for the embryonic stem cell coactivator UTF1 carries a regulatory element which selectively interacts with a complex composed of Oct-3/4 and Sox-2.Molecular and Cellular Biology1954535465
- 22. Yuan HB, Corbi N, Basilico C, Dailey L (1995) Developmental-Specific Activity of the Fgf-4 Enhancer Requires the Synergistic Action of Sox2 and Oct-3. Genes & Development 9: 2635–2645.HB YuanN. CorbiC. BasilicoL. Dailey1995Developmental-Specific Activity of the Fgf-4 Enhancer Requires the Synergistic Action of Sox2 and Oct-3.Genes & Development926352645
- 23. Wang JL, Rao S, Chu JL, Shen XH, Levasseur DN, et al. (2006) A protein interaction network for pluripotency of embryonic stem cells. Nature 444: 364–368.JL WangS. RaoJL ChuXH ShenDN Levasseur2006A protein interaction network for pluripotency of embryonic stem cells.Nature444364368
- 24. Yamanaka S (2007) Strategies and new developments in the generation of patient-specific pluripotent stem cells. Cell Stem Cell 1: 39–49.S. Yamanaka2007Strategies and new developments in the generation of patient-specific pluripotent stem cells.Cell Stem Cell13949
- 25. Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, et al. (2007) Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131: 861–872.K. TakahashiK. TanabeM. OhnukiM. NaritaT. Ichisaka2007Induction of pluripotent stem cells from adult human fibroblasts by defined factors.Cell131861872
- 26. Tokuzawa Y, Kaiho E, Maruyama M, Takahashi K, Mitsui K, et al. (2003) Fbx15 is a novel target of Oct3/4 but is dispensable for embryonic stem cell self-renewal and mouse development. Molecular and Cellular Biology 23: 2699–2708.Y. TokuzawaE. KaihoM. MaruyamaK. TakahashiK. Mitsui2003Fbx15 is a novel target of Oct3/4 but is dispensable for embryonic stem cell self-renewal and mouse development.Molecular and Cellular Biology2326992708
- 27. Huangfu D, Maehr R, Guo W, Eijkelenboom A, Snitow M, et al. (2008) Induction of pluripotent stem cells by defined factors is greatly improved by small-molecule compounds. Nature Biotechnology. D. HuangfuR. MaehrW. GuoA. EijkelenboomM. Snitow2008Induction of pluripotent stem cells by defined factors is greatly improved by small-molecule compounds.Nature Biotechnology
- 28. Mikkelsen TS, Hanna J, Zhang X, Ku M, Wernig M, et al. (2008) Dissecting direct reprogramming through integrative genomic analysis. Nature. TS MikkelsenJ. HannaX. ZhangM. KuM. Wernig2008Dissecting direct reprogramming through integrative genomic analysis.Nature10.1038/nature07056:. 10.1038/nature07056:.
- 29. Shi Y, Do JT, Desponts C, Hahm HS, Scholer HR, et al. (2008) A combined chemical and genetic approach for the generation of induced pluripotent stem cells. Cell Stem Cell 2: 525–528.Y. ShiJT DoC. DespontsHS HahmHR Scholer2008A combined chemical and genetic approach for the generation of induced pluripotent stem cells.Cell Stem Cell2525528
- 30. Park IH, Zhao R, West JA, Yabuuchi A, Huo HG, et al. (2008) Reprogramming of human somatic cells to pluripotency with defined factors. Nature 451: 141–146.IH ParkR. ZhaoJA WestA. YabuuchiHG Huo2008Reprogramming of human somatic cells to pluripotency with defined factors.Nature451141146
- 31. Aoi T, Yae K, Nakagawa M, Ichisaka T, Okita K, et al. (2008) Generation of pluripotent stem cells from adult mouse liver and stomach cells. Science. T. AoiK. YaeM. NakagawaT. IchisakaK. Okita2008Generation of pluripotent stem cells from adult mouse liver and stomach cells.Science10.1126/science.1154884:. 10.1126/science.1154884:.
- 32. Maherali N, Sridharan R, Xie W, Utikal J, Eminli S, et al. (2007) Directly reprogrammed fibroblasts show global epigenetic remodeling and widespread tissue contribution. Cell Stem Cell 1: 55–70.N. MaheraliR. SridharanW. XieJ. UtikalS. Eminli2007Directly reprogrammed fibroblasts show global epigenetic remodeling and widespread tissue contribution.Cell Stem Cell15570
- 33. Meissner A, Wernig M, Jaenisch R (2007) Direct reprogramming of genetically unmodified fibroblasts into pluripotent stem cells. Nature Biotechnology 25: 1177–1181.A. MeissnerM. WernigR. Jaenisch2007Direct reprogramming of genetically unmodified fibroblasts into pluripotent stem cells.Nature Biotechnology2511771181
- 34. Nakagawa M, Koyanagi M, Tanabe K, Takahashi K, Ichisaka T, et al. (2008) Generation of induced pluripotent stem cells without Myc from mouse and human fibroblasts. Nature Biotechnology 26: 101–106.M. NakagawaM. KoyanagiK. TanabeK. TakahashiT. Ichisaka2008Generation of induced pluripotent stem cells without Myc from mouse and human fibroblasts.Nature Biotechnology26101106
- 35. Okita K, Ichisaka T, Yamanaka S (2007) Generation of germline-competent induced pluripotent stem cells. Nature 448: 313–317.K. OkitaT. IchisakaS. Yamanaka2007Generation of germline-competent induced pluripotent stem cells.Nature448313317
- 36. Wernig M, Meissner A, Cassady JP, Jaenisch R (2008) c-Myc is dispensable for direct reprogramming of mouse fibroblasts. Cell Stem Cell 2: 10–12.M. WernigA. MeissnerJP CassadyR. Jaenisch2008c-Myc is dispensable for direct reprogramming of mouse fibroblasts.Cell Stem Cell21012
- 37. Wernig M, Meissner A, Foreman R, Brambrink T, Ku MC, et al. (2007) In vitro reprogramming of fibroblasts into a pluripotent ES-cell-like state. Nature 448: 318–324.M. WernigA. MeissnerR. ForemanT. BrambrinkMC Ku2007In vitro reprogramming of fibroblasts into a pluripotent ES-cell-like state.Nature448318324
- 38. Yu JY, Vodyanik MA, Smuga-Otto K, ntosiewicz-Bourget J, Frane JL, et al. (2007) Induced pluripotent stem cell lines derived from human somatic cells. Science 318: 1917–1920.JY YuMA VodyanikK. Smuga-OttoJ. ntosiewicz-BourgetJL Frane2007Induced pluripotent stem cell lines derived from human somatic cells.Science31819171920
- 39. Laslo P, Spooner CJ, Warmflash A, Lancki DW, Lee HJ, et al. (2006) Multilineage transcriptional priming and determination of alternate hematopoietic cell fates. Cell 126: 755–766.P. LasloCJ SpoonerA. WarmflashDW LanckiHJ Lee2006Multilineage transcriptional priming and determination of alternate hematopoietic cell fates.Cell126755766
- 40. Chickarmane V, Troein C, Nuber UA, Sauro HM, Peterson C (2006) Transcriptional dynamics of the embryonic stem cell switch. Plos Computational Biology 2: 1080–1092.V. ChickarmaneC. TroeinUA NuberHM SauroC. Peterson2006Transcriptional dynamics of the embryonic stem cell switch.Plos Computational Biology210801092
- 41. Cinquin O, Demongeot J (2005) High-dimensional switches and the modelling of cellular differentiation. Journal of Theoretical Biology 233: 391–411.O. CinquinJ. Demongeot2005High-dimensional switches and the modelling of cellular differentiation.Journal of Theoretical Biology233391411
- 42. Cinquin O, Page KM (2007) Generalized, switch-like competitive heterodimerization networks. Bulletin of Mathematical Biology 69: 483–494.O. CinquinKM Page2007Generalized, switch-like competitive heterodimerization networks.Bulletin of Mathematical Biology69483494
- 43. Furusawa C, Kaneko K (2001) Theory of robustness of irreversible differentiation in a stem cell system: Chaos hypothesis. Journal of Theoretical Biology 209: 395–416.C. FurusawaK. Kaneko2001Theory of robustness of irreversible differentiation in a stem cell system: Chaos hypothesis.Journal of Theoretical Biology209395416
- 44. Glauche I, Cross M, Loeffler M, Roeder I (2007) Lineage specification of hematopoietic stem cells: Mathematical modeling and biological implications. Stem Cells 25: 1791–1799.I. GlaucheM. CrossM. LoefflerI. Roeder2007Lineage specification of hematopoietic stem cells: Mathematical modeling and biological implications.Stem Cells2517911799
- 45. Huang S, Guo YP, May G, Enver T (2007) Bifurcation dynamics in lineage-commitment in bipotent progenitor cells. Developmental Biology 305: 695–713.S. HuangYP GuoG. MayT. Enver2007Bifurcation dynamics in lineage-commitment in bipotent progenitor cells.Developmental Biology305695713
- 46. Kauffman SA (1993) The origins of order: self-organization and selection in evolution. Oxford: OUP. SA Kauffman1993The origins of order: self-organization and selection in evolutionOxfordOUP
- 47. Keller AD (1995) Model Genetic Circuits Encoding Autoregulatory Transcription Factors. Journal of Theoretical Biology 172: 169–185.AD Keller1995Model Genetic Circuits Encoding Autoregulatory Transcription Factors.Journal of Theoretical Biology172169185
- 48. Kirkland MA (2004) A phase space model of hemopoiesis and the concept of stem cell renewal. Experimental Hematology 32: 511–519.MA Kirkland2004A phase space model of hemopoiesis and the concept of stem cell renewal.Experimental Hematology32511519
- 49. Preisler HD, Kauffman S (1999) A proposal regarding the mechanism which underlies lineage choice during hematopoietic differentiation. Leukemia Research 23: 685–694.HD PreislerS. Kauffman1999A proposal regarding the mechanism which underlies lineage choice during hematopoietic differentiation.Leukemia Research23685694
- 50. Roeder I, Glauche I (2006) Towards an understanding of lineage specification in hematopoietic stem cells: A mathematical model for the interaction of transcription factors GATA-1 and PU.1. Journal of Theoretical Biology 241: 852–865.I. RoederI. Glauche2006Towards an understanding of lineage specification in hematopoietic stem cells: A mathematical model for the interaction of transcription factors GATA-1 and PU.1.Journal of Theoretical Biology241852865
- 51. Roeder I, Loeffler M (2002) A novel dynamic model of hematopoietic stem cell organization based on the concept of within-tissue plasticity. Experimental Hematology 30: 853–861.I. RoederM. Loeffler2002A novel dynamic model of hematopoietic stem cell organization based on the concept of within-tissue plasticity.Experimental Hematology30853861
- 52. Kauffman S (1969) Homeostasis and differentiation in random genetic control networks. Nature 224: 177–178.S. Kauffman1969Homeostasis and differentiation in random genetic control networks.Nature224177178
- 53. Huang S, Eichler G, Bar-Yam Y, Ingber DE (2005) Cell fates as high-dimensional attractor states of a complex gene regulatory network. Physical Review Letters 94: 128701-1–128701-4.S. HuangG. EichlerY. Bar-YamDE Ingber2005Cell fates as high-dimensional attractor states of a complex gene regulatory network.Physical Review Letters94128701-1128701-4
- 54. Chang HH, Hemberg M, Barahona M, Ingber DE, Huang S (2008) Transcriptome-wide noise controls lineage choice in mammalian progenitor cells. Nature 453: 544–547.HH ChangM. HembergM. BarahonaDE IngberS. Huang2008Transcriptome-wide noise controls lineage choice in mammalian progenitor cells.Nature453544547
- 55. Thomas R (1978) Logical Analysis of Systems Comprising Feedback Loops. Journal of Theoretical Biology 73: 631–656.R. Thomas1978Logical Analysis of Systems Comprising Feedback Loops.Journal of Theoretical Biology73631656
- 56. Monod J, Jacob F (1961) Teleonomic mechanisms in cellular metabolism, growth and differentiation. Cold Spring Harb Symp Quant Biol 26: 389–401.J. MonodF. Jacob1961Teleonomic mechanisms in cellular metabolism, growth and differentiation.Cold Spring Harb Symp Quant Biol26389401
- 57. Angeli D, Ferrell JE, Sontag ED (2004) Detection of multistability, bifurcations, and hysteresis in a large class of biological positive-feed back systems. Proceedings of the National Academy of Sciences of the United States of America 101: 1822–1827.D. AngeliJE FerrellED Sontag2004Detection of multistability, bifurcations, and hysteresis in a large class of biological positive-feed back systems.Proceedings of the National Academy of Sciences of the United States of America10118221827
- 58. Soule C (2006) Mathematical approaches to differentiation and gene regulation. Comptes Rendus Biologies 329: 13–20.C. Soule2006Mathematical approaches to differentiation and gene regulation.Comptes Rendus Biologies3291320
- 59. Chang HH, Oh PY, Ingber DE, Huang S (2006) Multistable and multistep dynamics in neutrophil differentiation. Bmc Cell Biology 7: HH ChangPY OhDE IngberS. Huang2006Multistable and multistep dynamics in neutrophil differentiation.Bmc Cell Biology7
- 60. Becskei A, Seraphin B, Serrano L (2001) Positive feedback in eukaryotic gene networks: cell differentiation by graded to binary response conversion. Embo Journal 20: 2528–2535.A. BecskeiB. SeraphinL. Serrano2001Positive feedback in eukaryotic gene networks: cell differentiation by graded to binary response conversion.Embo Journal2025282535
- 61. Hong JH, Hwang ES, McManus MT, Amsterdam A, Tian Y, et al. (2005) TAZ, a transcriptional modulator of mesenchymal stem cell differentiation. Science 309: 1074–1078.JH HongES HwangMT McManusA. AmsterdamY. Tian2005TAZ, a transcriptional modulator of mesenchymal stem cell differentiation.Science30910741078
- 62. Hong JH, Yaffe MB (2006) TAZ - A beta-catenin-like molecule that regulates mesenchymal stem cell differentiation. Cell Cycle 5: 176–179.JH HongMB Yaffe2006TAZ - A beta-catenin-like molecule that regulates mesenchymal stem cell differentiation.Cell Cycle5176179
- 63. Yang SY, Wei DY, Wang D, Phimphilai M, Krebsbach PH, et al. (2003) In vitro and in vivo synergistic interactions between the Runx2/Cbfa1 transcription factor and bone morphogenetic protein-2 in stimulating osteoblast differentiation. Journal of Bone and Mineral Research 18: 705–715.SY YangDY WeiD. WangM. PhimphilaiPH Krebsbach2003In vitro and in vivo synergistic interactions between the Runx2/Cbfa1 transcription factor and bone morphogenetic protein-2 in stimulating osteoblast differentiation.Journal of Bone and Mineral Research18705715
- 64. Lecka-Czernik B, Moerman EJ, Grant DF, Lehmann JM, Manolagas SC, et al. (2002) Divergent effects of selective peroxisome proliferator-activated receptor-gamma 2 ligands on adipocyte versus osteoblast differentiation. Endocrinology 143: 2376–2384.B. Lecka-CzernikEJ MoermanDF GrantJM LehmannSC Manolagas2002Divergent effects of selective peroxisome proliferator-activated receptor-gamma 2 ligands on adipocyte versus osteoblast differentiation.Endocrinology14323762384
- 65. Moerman EJ, Teng K, Lipschitz DA, Lecka-Czernik B (2004) Aging activates adipogenic and suppresses osteogenic programs in mesenchymal marrow stroma/stem cells: the role of PPAR-gamma 2 transcription factor and TGF-beta/BMP signaling pathways. Aging Cell 3: 379–389.EJ MoermanK. TengDA LipschitzB. Lecka-Czernik2004Aging activates adipogenic and suppresses osteogenic programs in mesenchymal marrow stroma/stem cells: the role of PPAR-gamma 2 transcription factor and TGF-beta/BMP signaling pathways.Aging Cell3379389
- 66. Jeon MJ, Kim JA, Kwon SH, Kim SW, Park KS, et al. (2003) Activation of peroxisome proliferator-activated receptor-gamma inhibits the Runx2-mediated transcription of osteocalcin in osteoblasts. Journal of Biological Chemistry 278: 23270–23277.MJ JeonJA KimSH KwonSW KimKS Park2003Activation of peroxisome proliferator-activated receptor-gamma inhibits the Runx2-mediated transcription of osteocalcin in osteoblasts.Journal of Biological Chemistry2782327023277
- 67. Khan E, bu-Amer Y (2003) Activation of peroxisome proliferator-activated receptor-gamma inhibits differentiation of preosteoblasts. Journal of Laboratory and Clinical Medicine 142: 29–34.E. KhanY. bu-Amer2003Activation of peroxisome proliferator-activated receptor-gamma inhibits differentiation of preosteoblasts.Journal of Laboratory and Clinical Medicine1422934
- 68. Rosen ED, Hsu CH, Wang XZ, Sakai S, Freeman MW, et al. (2002) C/EBP alpha induces adipogenesis through PPAR gamma: a unified pathway. Genes & Development 16: 22–26.ED RosenCH HsuXZ WangS. SakaiMW Freeman2002C/EBP alpha induces adipogenesis through PPAR gamma: a unified pathway.Genes & Development162226
- 69. Zhang P, Behre G, Pan J, Iwama A, Wara-Aswapati N, et al. (1999) Negative cross-talk between hematopoietic regulators: GATA proteins repress PU.1. Proceedings of the National Academy of Sciences of the United States of America 96: 8705–8710.P. ZhangG. BehreJ. PanA. IwamaN. Wara-Aswapati1999Negative cross-talk between hematopoietic regulators: GATA proteins repress PU.1.Proceedings of the National Academy of Sciences of the United States of America9687058710
- 70. Gowan K, Helms AW, Hunsaker TL, Collisson T, Ebert PJ, et al. (2001) Crossinhibitory activities of Ngn1 and Math1 allow specification of distinct dorsal interneurons. Neuron 31: 219–232.K. GowanAW HelmsTL HunsakerT. CollissonPJ Ebert2001Crossinhibitory activities of Ngn1 and Math1 allow specification of distinct dorsal interneurons.Neuron31219232
- 71. Rothenberg EV, Telfer JC, Anderson MK (1999) Transcriptional regulation of lymphocyte lineage commitment. Bioessays 21: 726–742.EV RothenbergJC TelferMK Anderson1999Transcriptional regulation of lymphocyte lineage commitment.Bioessays21726742
- 72. Kawaguchi J, Mee PJ, Smith AG (2005) Osteogenic and chondrogenic differentiation of embryonic stem cells in response to specific growth factors. Bone 36: 758–769.J. KawaguchiPJ MeeAG Smith2005Osteogenic and chondrogenic differentiation of embryonic stem cells in response to specific growth factors.Bone36758769
- 73. Ivanova N, Dobrin R, Lu R, Kotenko I, Levorse J, et al. (2006) Dissecting self-renewal in stem cells with RNA interference. Nature 442: 533–538.N. IvanovaR. DobrinR. LuI. KotenkoJ. Levorse2006Dissecting self-renewal in stem cells with RNA interference.Nature442533538
- 74. Xiong W, Ferrell JE (2003) A positive-feedback-based bistable ‘memory module’ that governs a cell fate decision. Nature 426: 460–465.W. XiongJE Ferrell2003A positive-feedback-based bistable ‘memory module’ that governs a cell fate decision.Nature426460465
- 75. Hu M, Krause D, Greaves M, Sharkis S, Dexter M, et al. (1997) Multilineage gene expression precedes commitment in the hemopoietic system. Genes & Development 11: 774–785.M. HuD. KrauseM. GreavesS. SharkisM. Dexter1997Multilineage gene expression precedes commitment in the hemopoietic system.Genes & Development11774785
- 76. Tai GP, Polak JM, Bishop AE, Christodoulou I, Buttery LDK (2004) Differentiation of osteoblasts from murine embryonic stem cells by overexpression of the transcriptional factor osterix. Tissue Engineering 10: 1456–1466.GP TaiJM PolakAE BishopI. ChristodoulouLDK Buttery2004Differentiation of osteoblasts from murine embryonic stem cells by overexpression of the transcriptional factor osterix.Tissue Engineering1014561466
- 77. Enver T, Heyworth CM, Dexter TM (1998) Do stem cells play dice? Blood 92: 348–351.T. EnverCM HeyworthTM Dexter1998Do stem cells play dice?Blood92348351
- 78. Orkin SH (2000) Diversification of haematopoietic stem cells to specific lineages. Nature Reviews Genetics 1: 57–64.SH Orkin2000Diversification of haematopoietic stem cells to specific lineages.Nature Reviews Genetics15764
- 79. Szutorisz H, Georgiou A, Tora L, Dillon N (2006) The proteasome restricts permissive transcription at tissue-specific gene loci in embryonic stem cells. Cell 127: 1375–1388.H. SzutoriszA. GeorgiouL. ToraN. Dillon2006The proteasome restricts permissive transcription at tissue-specific gene loci in embryonic stem cells.Cell12713751388
- 80. McAdams HH, Arkin A (1997) Stochastic mechanisms in gene expression. Proceedings of the National Academy of Sciences of the United States of America 94: 814–819.HH McAdamsA. Arkin1997Stochastic mechanisms in gene expression.Proceedings of the National Academy of Sciences of the United States of America94814819
- 81. Acar M, Becskei A, van Oudenaarden A (2005) Enhancement of cellular memory by reducing stochastic transitions. Nature 435: 228–232.M. AcarA. BecskeiA. van Oudenaarden2005Enhancement of cellular memory by reducing stochastic transitions.Nature435228232
- 82. Arias AM, Hayward P (2006) Filtering transcriptional noise during development: concepts and mechanisms. Nature Reviews Genetics 7: 34–44.AM AriasP. Hayward2006Filtering transcriptional noise during development: concepts and mechanisms.Nature Reviews Genetics73444
- 83. Blake WJ, Kaern M, Cantor CR, Collins JJ (2003) Noise in eukaryotic gene expression. Nature 422: 633–637.WJ BlakeM. KaernCR CantorJJ Collins2003Noise in eukaryotic gene expression.Nature422633637
- 84. Kaern M, Elston TC, Blake WJ, Collins JJ (2005) Stochasticity in gene expression: From theories to phenotypes. Nature Reviews Genetics 6: 451–464.M. KaernTC ElstonWJ BlakeJJ Collins2005Stochasticity in gene expression: From theories to phenotypes.Nature Reviews Genetics6451464
- 85. Losick R, Desplan C (2008) Stochasticity and cell fate. Science 320: 65–68.R. LosickC. Desplan2008Stochasticity and cell fate.Science3206568
- 86. Maamar H, Raj A, Dubnau D (2007) Noise in gene expression determines cell fate in Bacillus subtilis. Science 317: 526–529.H. MaamarA. RajD. Dubnau2007Noise in gene expression determines cell fate in Bacillus subtilis.Science317526529
- 87. Huang S, Ingber DE (2007) A non-genetic basis for cancer progression and metastasis: self-organizing attractors in cell regulatory networks. Breast Disease 26: 27–54.S. HuangDE Ingber2007A non-genetic basis for cancer progression and metastasis: self-organizing attractors in cell regulatory networks.Breast Disease262754
- 88. Waddington CH (1957) The Strategy of the Genes. London: Allen & Unwin. CH Waddington1957The Strategy of the GenesLondonAllen & Unwin
- 89. Goldberg AD, Allis CD, Bernstein E (2007) Epigenetics: A landscape takes shape. Cell 128: 635–638.AD GoldbergCD AllisE. Bernstein2007Epigenetics: A landscape takes shape.Cell128635638
- 90. Kramers HA (1940) Brownian motion in a field of force and the diffusion model of chemical reactions. Physica A 7: 284–304.HA Kramers1940Brownian motion in a field of force and the diffusion model of chemical reactions.Physica A7284304
- 91. Feldman N, Gerson A, Fang J, Li E, Zhang Y, et al. (2006) G9a-mediated irreversible epigenetic inactivation of Oct-3/4 during early embryogenesis. Nature Cell Biology 8: 188–194.N. FeldmanA. GersonJ. FangE. LiY. Zhang2006G9a-mediated irreversible epigenetic inactivation of Oct-3/4 during early embryogenesis.Nature Cell Biology8188194
- 92. Hochedlinger K, Yamada Y, Beard C, Jaenisch R (2005) Ectopic expression of Oct-4 blocks progenitor-cell differentiation and causes dysplasia in epithelial tissues. Cell 121: 465–477.K. HochedlingerY. YamadaC. BeardR. Jaenisch2005Ectopic expression of Oct-4 blocks progenitor-cell differentiation and causes dysplasia in epithelial tissues.Cell121465477
- 93. Knoepfler PS, Zhang XY, Cheng PF, Gafken PR, McMahon SB, et al. (2006) Myc influences global chromatin structure. Embo Journal 25: 2723–2734.PS KnoepflerXY ZhangPF ChengPR GafkenSB McMahon2006Myc influences global chromatin structure.Embo Journal2527232734
- 94. Niwa H (2007) How is pluripotency determined and maintained? Development 134: 635–646.H. Niwa2007How is pluripotency determined and maintained?Development134635646
- 95. Evans PM, Zhang W, Chen X, Yang J, Bhakat KK, et al. (2007) Krurppel-like factor 4 is acetylated by p300 and regulates gene transcription via modulation of histone acetylation. Journal of Biological Chemistry 282: 33994–34002.PM EvansW. ZhangX. ChenJ. YangKK Bhakat2007Krurppel-like factor 4 is acetylated by p300 and regulates gene transcription via modulation of histone acetylation.Journal of Biological Chemistry2823399434002
- 96. Collins GA, Tansey WP (2006) The proteasome: a utility tool for transcription? Current Opinion in Genetics & Development 16: 197–202.GA CollinsWP Tansey2006The proteasome: a utility tool for transcription?Current Opinion in Genetics & Development16197202