Reprogramming somatic cells to pluripotency

Abstract

In 2006, Dr Shinya Yamanaka succeeded to reprogram somatic cells into pluripotent stem cells (iPSC) by delivering the genes encoding Oct4, Sox2, Klf4, and c-Myc. This achievement represents a fundamental breakthrough in stem cell biology and opens up a new era in regenerative medicine. However, the molecular processes by which somatic cells are reprogrammed into iPSC remain poorly understood. In 2009, Yamanaka proposed the elite and stochastic models for reprogramming mechanisms. To date, many investigators in the field of iPSC research support the concept of stochastic model, i.e., somatic cell reprogramming is an event of epigenetic transformation. A mathematical model, f (C_d, k), has also been proposed to predict the stochastic process. Here we wish to revisit the Yamanaka model and summarize the recent advances in this research field.

Introduction

In 2006, Dr Shinya Yamanaka demonstrated that differentiated somatic cells can be de-differentiated into induced pluripotent stem cells (iPSC) by the ectopic expression of 4 transcription factors (Oct4, Sox2, Klf4, and c-Myc).^1,2 These discoveries raise the prospect of better patient care by regenerating virtually all types of somatic cells from patient’s own tissue. However, current methods of generating iPSC are highly inefficient due to a lack of knowledge about the molecular mechanisms underlying somatic cell reprogramming. In 2009, Yamanaka proposed the elite and stochastic model³ for reprogramming mechanisms. The elite model predicts that only certain cells are possibly reprogrammed, whereas the stochastic model predicts that most or all somatic cells have the potential to become pluripotent. The stochastic model also points out the importance of epigenetic modifications and anti-senescence in iPSC generation.³ Since then, many investigators in the field of iPSC research support the concept of stochastic model, i.e., somatic cell reprogramming is an event of epigenetic transformation. In this article, we appraise recent advances and future challenges in this research field.

Stochastic Model

The stochastic model was originally developed from a cell differentiation model, as proposed by Conrad Waddington in 1957. In this model, cell differentiation is depicted as a ball rolling down on an epigenetic landscape. Briefly, the totipotent cells roll down to become pluripotent and then finally differentiate to a lineage-committed state. In normal development, pluripotent cells appear transiently, cannot stop on the slope, and are pulled by gravity to rapidly differentiate into various lineages. Based on stochastic model, the reprogramming stimuli must push the cells up into the pluripotent state by changing epigenetic status, and this is achieved by locking the iPSC in the pluripotent zone by specific epigenetic modifications. Recent evidence suggests that DNA methylation, histone acetylation, and histone methylation are involved in this process.

DNA Methylation

The promoter regions of the 4 critical transcription factors (Oct4, Sox2, Klf4, and c-Myc) that have been used to generate iPSC are heavily methylated in somatic cells, suggesting that DNA methylation prevents activation of these genes. Therefore, demethylation of the endogenous loci of the gene of these 4 factors is required in order to maintain the cells in the pluripotent state. These 4 factors themselves do not function directly in regulation of DNA demethylation. However, after several cell divisions, DNA demethylation at the promoter region of Sox2, Oct4, and klf4 may be accomplished. This may explain why iPSC generation is so slow. The finding that 5-azacytidine, a demethylation agent, promotes iPSC generation supports this model.⁴ Furthermore, it has been shown recently that DNA demethylation acquires stable pluripotency.⁵

Histone Modification

The proper histone acetylation is required for iPSC generation. For instance, histone H4 is usually hypoacetylated in somatic cells. The hyperacetylation of histone H3 and H4 in the promoter regions of pluripotency associated genes is required for iPSC generation. Only c-Myc has the capability to recruit histone acetyltransferase to target genes. A very recent finding demonstrates that c-Myc is required for early stage of iPSC generation,⁵ and valproic acid, a histone deacetylase inhibitor, increases the efficiency of iPSC generation.^6,7

In addition, histone methylation also plays an important role in iPSC generation. In iPSC, histone H3 is methylated at lysine 4 and demethylated at lysine 9 in the promoter regions of pluripotency-associated genes. In contrast, somatic cells have the opposite patterns of histone modifications. Thus, these histone methylation states have to be changed for iPSC generation. The recent discovery that histone H3 lysine 9 (H3K9) methylation is a barrier during iPSC generation favors this notion.⁸

The stochastic model predicts that cellular senescence is a major obstacle for the success of iPSC generation. Recent findings that disruption of the p53 (senescence regulator) network enhances the production of iPSC support this hypothesis.^9-13 Furthermore, iPSCs are separated into 2 categories based on the expression pattern of p53 and the microRNAs, and only those cells with low p53 activity are more proliferative and able to enter the pluripotent zone.¹⁴ Moreover, recent studies show that vitamin C enhances iPSC generation at least in part by facilitating the function of histone demethylases Jhdm1a/1b,¹⁵ which suppresses senescence regulator Ink4a/Arf.¹⁰ This is one of the potential mechanisms by which vitamin C enhances the generation of iPSC.¹⁶

microRNA and MET

The role of microRNA and mesenchymal-to-epithelial transition (MET) was not addressed in the stochastic model. However, recent discoveries demonstrate that microRNAs and MET are important players of the regulatory network that maintains pluripotency of iPSC. These findings further enforce the main idea of the stochastic model that somatic cell reprogramming is a process of epigenetic transformation.

microRNAs are short endogenous non-coding RNAs that bind to the 3′UTRs of target mRNAs leading to their cleavage or translational repression. Overexpression or suppression of individual microRNAs has profound effects in colony formation efficiency during iPSC generation.¹⁷ microRNAs can also regulate generation of iPSC without added transcription factors.^18,19 microRNAs enhance iPSC generation by removing epigenetic repressors. Studies show that microRNA cluster 302–367 removes epigenetic repressors in the promoter regions of Sox-2 and Oct-4 genes, rendering them accessible to transactivators.^17,20 Importantly, microRNA-induced reprogramming has unique advantages over other modulators, as the method relies entirely on the activation of endogenous pathways, thus epigenetic transitions may occur more smoothly.¹⁷ It is also known that certain microRNAs enhance the histone acetylation induced by valproic acid treatment. Therefore, microRNAs can be used as epigenetic modifiers for achieving high reprogramming efficiency.

The generation of iPSC also requires MET which is accomplished by suppressing epithelial-to-mesenchymal transition (EMT) mediators (TGFβ1, TGFβR2, and Smad) and activating MET mediator (E-cadherin) inside the cells.^21,22 Functional analysis revealed that MET plays an important role in the initiation phase and has to be completed to allow self-sustainable progression to a pluripotent state during iPSC generation.²³ Several studies indicate that MET can be achieved by epigenetic modification. ChIP-on-chip study has shown that somatic cells have opposite pattern of histone H3 methylation in the promoter of E-cadherin compared with iPSC. c-Myc, the key regulator in the early phase of reprogramming, enhances MET by shutting down TGFβ1, TGFβR2, and Smad cascade via epigenetic regulation. Furthermore, Onder et al. discovered that Dot1L, a H3K79 methyltransferase, modulated reprogramming by altering the expression of MET/EMT signaling molecules such as Snail, Slug, and TGFβ2.²⁴ These studies indicate that the completion of MET needs epigenetic transformation, which is the key issue in the stochastic model.

Mathematical Models

The critical question in iPSC generation is how to find key factors to reduce the stochastic process, or enable the stochastic events to occur earlier. This question cannot be answered by Yamanaka models. Recently, a mathematical model, f (C_d, k), has been proposed, which predicts the critical factors based on preliminary data, thus provides a clue to reduce the stochastic process.²⁵ Here, C_d is the number of cell divisions; k is the cell-intrinsic reprogramming rate per cell division. Cell division (C_d) is a key parameter driving epigenetic reprogramming to pluripotency in the mathematic model,²⁵ and the rate-limiting epigenetic event for reprogramming may be the reactivation of the key endogenous stem cell regulatory circuitry that maintains iPSC state. Accelerating reprogramming in a cell-division-rate-dependent manner requires many cell divisions (bigger C_d) but occurs earlier in time because cells divide faster, whereas in the cell-division-rate-independent mode, the cell-intrinsic rate (k) reflecting the occurrence of an unknown stochastic event(s) is enhanced, and reprogramming is achieved within a lower number of cell divisions (smaller C_d). Using mathematical model, Hanna et al. predicted that p53/p21 inhibition could reduce the stochastic process, and they further showed that p53/p21 inhibition (reducing senescence) accelerates the reprogramming process, as the cells divide more rapidly (bigger C_d), resulting in the stochastic events occurring at early time.²⁵

An advantage of applying mathematical model is that it can provide future direction for iPSC generation. One mathematical model (τ = ∫₀^t N[t′]dt) has been developed based on the theory that cell division rate can control the cell number (N). Thus, population re-scale time can affect the observed reprogramming rate (τ).²⁵ Using this model, we can predict the reprogramming time for any given numbers of cells placed in each well. This formula will help researchers to choose optimal time point and cell number required for each experiment.

Another mathematical model has also been developed. V(S) = f (gene regulatory architecture, culture conditions), which represents “energy barriers” between transitions from somatic cells to iPSC,²⁶ and future efforts should be aiming to reduce “energy barriers”. Culture condition can be quantitatively adjusted; therefore, an “ideal” reprogramming environment should allow reprogramming to proceed with no or very brief stochastic phase.

Certainly, at this time, it is difficult to use mathematical model alone to predict iPSC generation; however, mathematical model can provide future direction for iPSC generation.

Epigenetic Memory

Based on the main concept of the stochastic model, somatic cell reprogramming can be achieved by epigenetic transformation. Therefore, a more practical approach to make or maintain iPSCs is to regulate gene expression via epigenetic modifications. Examples of such modifications are histone post-translational modifications, such as lysine methylation and acetylation. It has been shown that the epigenetic signature of the somatic cell must be erased during the reprogramming process, and the concept of “epigenetic memory” has emerged in iPSC research.

“Epigenetic memory” highlights the notion that erasing epigenetic pattern of donor cells, generating and keeping epigenetic pattern of iPSC are key issues for iPSC generation. Previous studies suggest that donor cells have a unique epigenetic signature. The expression of certain microRNAs, different patterns of promoter methylation, and histone methylation of donor cell affect several critical factors in iPSC generation: (1) the overall transcriptional pattern of resultant iPSC; (2) the overall epigenetic pattern of resultant iPSC; (3) the differentiation potentials of iPSC. Early-passage iPSCs retain a transient epigenetic memory of their donor cells, and continuous passaging of iPSCs largely attenuates these epigenetic differences.^27,28

Regarding mouse embryonic fibroblasts (MEFs) or neural progenitor cells (NPCs) as donor cells, the epigenetic differences between these 2 cell types are largely unclear at this time. However, NPCs express endogenous Sox2 genes,^28,29 and Sox2 can induce histone H3K4 methylation,^{30, 31} which can enhance iPSC generation via increasing proliferation, pluripotency, and MET gene expression,³² whereas MEF needs ectopic overexpression of Sox2 in order to induces histone H3K4 methylation. Thus, the generation of iPSC from NPC compared with MEFs is relatively easy. These studies also support that epigenetic memory of original cell type is a key issue in iPSC generation.

Concluding Remarks

In summary, the original Yamanaka model predicts that reprogramming factors lift the cells up into the pluripotent state by changing epigenetic status, and retain the cells in the pluripotent zone by altering specific epigenetic status, such as DNA demethylation, histone acetylation, and histone methylation. Since then, many studies support the concept of stochastic model that somatic cell reprogramming is an event of epigenetic transformation. Several mathematical models also have been proposed to predict the stochastic process. Importantly, recent studies, suggest that iPSC formation follows an early and late deterministic phase, separated by a more stochastic phase.^5,31,33 Therefore, at this time, a fresh look at the Yamanaka model revealed that the reprogramming process can be separated into 3 stages: initiation (early deterministic phase, parameter: senescence, and MET), maturation (stochastic phase, parameter: specific microRNA expression, histone modification, and DNA methylation) and stabilization (late deterministic phase, parameter: stem cell circuitry activation) (Fig. 1). Moreover, the reactivation of the endogenous stem cell regulatory circuitry is important, and that the process is achieved by many epigenetic changes.

Figure 1. Somatic cells undergo 3 stages on their way to become iPSC. The initiation stage or early deterministic phase is controlled by senescence and MET. Cells that are able to pass the initiation state will enter the maturation or the stochastic phase which is characterized by specific microRNA expression, histone modification, and changes in DNA methylation. The final stabilization stage or late deterministic phase is characterized by stem cell circuitry activation.

10.4161/cc.26952

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.