Manipulating cell fate through reprogramming: approaches and applications

ABSTRACT

Cellular plasticity progressively declines with development and differentiation, yet these processes can be experimentally reversed by reprogramming somatic cells to induced pluripotent stem cells (iPSCs) using defined transcription factors. Advances in reprogramming technology over the past 15 years have enabled researchers to study diseases with patient-specific iPSCs, gain fundamental insights into how cell identity is maintained, recapitulate early stages of embryogenesis using various embryo models, and reverse aspects of aging in cultured cells and animals. Here, we review and compare currently available reprogramming approaches, including transcription factor-based methods and small molecule-based approaches, to derive pluripotent cells characteristic of early embryos. Additionally, we discuss our current understanding of mechanisms that resist reprogramming and their role in cell identity maintenance. Finally, we review recent efforts to rejuvenate cells and tissues with reprogramming factors, as well as the application of iPSCs in deriving novel embryo models to study pre-implantation development.

Summary: This Primer explores different approaches to reprogram somatic cells to induced pluripotent stem cells (iPSCs), mechanistic insights into the reprogramming process, and applications of iPSC technology in development and rejuvenation.

Introduction

Cell fate decisions are guided by dynamic changes in epigenetic modifications, including histone and DNA methylation (Jambhekar et al., 2019; Smith et al., 2024). These modifications typically collaborate with lineage-specific transcription factors (TFs) to control gene expression programs that establish and subsequently maintain cellular identity during development and terminal differentiation (Isbel et al., 2022). As cellular plasticity gradually declines with development, it was historically thought that differentiated cells are irreversibly committed to a certain cell fate. However, the cloning of animals from differentiated cells by somatic cell nuclear transfer (SCNT) demonstrated that even terminally differentiated cells can be reprogrammed into a totipotent state, which is capable of giving rise to all embryonic and extra-embryonic tissues (Campbell et al., 1996; Gurdon et al., 1958; Hochedlinger and Jaenisch, 2002; Wakayama et al., 1998). Subsequently, Yamanaka and colleagues showed in a seminal study that TFs are sufficient to induce pluripotency in somatic cells, which provided a molecular logic for earlier cloning and cell fusion experiments, and established a tractable system with which to study the process of reprogramming (Takahashi et al., 2007; Takahashi and Yamanaka, 2006). The discovery of induced pluripotency also reinvigorated efforts to directly convert (or ‘transdifferentiate’) one differentiated cell type into another differentiated cell type using lineage-specific TFs (Wang et al., 2021b). These approaches have offered versatile experimental systems to explore the basic principles by which TFs control cell fate, and rewire epigenetic and transcriptional programs.

Motivated by basic scientific questions and potential therapeutic opportunities of induced pluripotent stem cell (iPSC) technology, the field has seen major technological advances within the past decade (Hochedlinger and Plath, 2009; Stadtfeld and Hochedlinger, 2010; Takahashi and Yamanaka, 2015; Yamanaka, 2020). For example, improved reprogramming methods allow the generation of high-quality, integration-free iPSCs using diverse sets of TFs in combination with small molecules that modulate specific epigenetic or signaling pathways (Gonzalez et al., 2011; Wang et al., 2023b). iPSCs generated with TFs or with small molecule-based approaches are thought to be molecularly and functionally highly similar to embryonic stem cells (ESCs). However, variability in the developmental potential of iPSC clones has been observed and correlated with genetic and/or epigenetic aberrations accrued during the reprogramming process (Liang and Zhang, 2013). Depending on the application of iPSCs, it is therefore crucial to use common standards for the molecular and functional assessment of pluripotency.

TF-induced reprogramming has also been leveraged as a tractable tool for studying the mechanisms that maintain differentiated cell identities. By studying the transcriptional and epigenetic changes that occur during reprogramming, and by using unbiased gain- and loss-of-function screens for enhancers of reprogramming, researchers have identified previously unrecognized mechanisms whose disruption increases cellular plasticity. Accordingly, these mechanisms play a role in other experimental and physiological cell fate transitions, and are often corrupted in diseases such as cancer (Brumbaugh et al., 2019a). These observations highlight the power of iPSC technology to gain insights into fundamental physiological and pathological processes.

Most recently, iPSC technology has been applied to aging research with the goal to rejuvenate old tissues or regenerate damaged tissues, and TF-based reprogramming has been used to generate embryo models that recapitulate early stages of human development. Together, these advances in reprogramming technology and its cellular products provide new opportunities for the study of embryonic development, aging and rejuvenation.

In this Primer, we discuss recent advances in reprogramming approaches using TFs and small molecules to consistently generate bona fide pluripotent stem cells. Additionally, we highlight progress in applying reprogramming technology for the study of cell identity maintenance, tissue regeneration and rejuvenation, as well as early embryonic development (Fig. 1).

Fig. 1.

Different strategies to produce iPSCs and applications of iPSC technology. Diverse somatic cell types can be reliably reprogrammed into iPSCs using either transcription factors, small molecules or a combination of both. Reprogramming technology has been leveraged for diverse applications, including disease modeling and studies aimed at identifying fundamental mechanisms that maintain cell identity or facilitate cellular plasticity. More recently, modified reprogramming strategies have been employed in vivo to study tissue regeneration and aging, as well as in vitro to assemble versatile models of early embryogenesis. iPSC, induced pluripotent stem cell; Oct4, octamer-binding transcription factor 4; Klf4, Krüppel-like factor 4; Sox2, sex-determining region Y-box 2; c-Myc, cellular myelocytomatosis oncogene.

Cellular reprogramming with transcription factors

In a groundbreaking study, Takahashi and Yamanaka showed that the forced expression of the pluripotency-associated TFs Oct4, Klf4, Sox2 and c-Myc (OKSM), from individual retroviral vectors is sufficient to reprogram murine fibroblasts into iPSCs (Takahashi and Yamanaka, 2006). iPSCs generated in this way are molecularly and functionally highly similar to ESCs. For example, iPSCs self-renew indefinitely, undergo transcriptional and epigenetic rewiring towards an ESC-like state, and support the development of an entire animal, including germ cells. Remarkably, the same factors are sufficient to generate iPSCs from a host of other species, including humans (Maherali et al., 2008; Park et al., 2008; Takahashi et al., 2007; Yu et al., 2007), indicating that these factors activate conserved transcriptional circuits that are essential for the establishment and maintenance of pluripotency.

Early reprogramming studies showed that constitutive retroviral and lentiviral vectors expressing OKSM consistently produce iPSC-like colonies (Maherali et al., 2008; Takahashi and Yamanaka, 2006; Wernig et al., 2007; Yu et al., 2007). However, these vectors are often not silenced in iPSCs or are reactivated in mature progeny, which can compromise the differentiation and developmental potential of iPSCs, and sometimes lead to tumors in vivo (Okita et al., 2007). Consequently, researchers developed various virus- and integration-free approaches to deliver OKSM, including episomal- (Okita et al., 2011), adenoviral- (Stadtfeld et al., 2008), transposon- (Kaji et al., 2009; Woltjen et al., 2009) and RNA-based vectors (Warren et al., 2010). Although the efficiency of iPSC generation via vector-free methods is typically lower and more variable than that of integrating vector systems, the commercialization of these technologies has made both approaches more readily accessible and reproducible. Thus, the choice of vector system is usually driven by the specific type of application (Table 1). For therapeutic applications of iPSC technology, it is imperative to use vector-free systems to avoid rare instances of insertional mutagenesis or incomplete vector silencing (Yamanaka, 2020). By contrast, studies focusing on mechanistic aspects of reprogramming benefit from the use of integrating polycistronic (i.e. multiple coding sequences combined into a single transcript) vector systems expressing OKSM or other TF combinations, ideally under the control of a doxycycline inducible promoter (Carey et al., 2009; Sommer et al., 2009; Stadtfeld et al., 2010b). These inducible systems allow the homogeneous induction of OKSM and temporal control of factor expression, which is helpful for the identification of fully reprogrammed cells that sustain pluripotency in the absence of exogenous OKSM expression. Notably, for polycistronic vectors, the order of TFs, the way in which TFs are linked [e.g. via self-cleavable peptides or IRES (internal ribosome entry site) elements] and overall TF levels can also impact reprogramming efficiency and iPSC quality (Carey et al., 2011; Polo et al., 2010), indicating that the absolute levels and the stoichiometry of reprogramming factors matter.

Table 1.

Advantages and disadvantages of representative approaches for the generation of iPSCs

The type of somatic cell used for iPSC experiments also influences the efficiency of reprogramming. In general, highly proliferative and undifferentiated cells are more efficient donors for reprogramming than slowly dividing and differentiated cells. For example, mouse hematopoietic stem and progenitor cells, including granulocyte-monocyte progenitors and progenitor-B cells, yield iPSCs up to 300 times more efficiently than terminally differentiated B and T cells (Eminli et al., 2009; Guo et al., 2014). Consistent with this notion, the reprogramming of postnatal murine cortical neurons into iPSCs requires inactivation of the p53 (Trp53) tumor suppressor to stimulate proliferation (Kim et al., 2011). By contrast, retinal neurons give rise to iPSCs without disruption of the p53 pathway, but this protocol requires co-culture with unperturbed retinal cells to avoid apoptosis of explanted neurons (Hiler et al., 2015). Reprogramming potential is further influenced by the endogenous expression levels of the reprogramming factors in the starting cells. For example, iPSCs can be generated from mouse neural progenitors with exogenous Oct4 expression alone, as these cells already express Sox2, Klf4 and c-Myc endogenously (Kim et al., 2009).

Each of the four Yamanaka factors can be replaced by alternative TFs or small molecules during the generation of iPSCs, highlighting a remarkable degree of flexibility in the way pluripotency is induced (Table 2). For example, the estrogen-related receptor Esrrb replaces Klf4 (Feng et al., 2009), L-Myc substitutes for c-Myc (Nakagawa et al., 2008) and Nr5a2 serves as a replacement for Oct4 (Heng et al., 2010) during the induction of pluripotency. Strikingly, fibroblasts can even be reprogrammed into iPSCs without the use of Oct4 (i.e. only with SKM) (Velychko et al., 2019), which was long thought to be the only essential reprogramming factor, or with completely different sets of TFs [i.e. Sall4, Nanog, Esrrb and Lin28 (SNEL) or Sall4, Jdp2, Glis1 and Esrrb] that act downstream of OKSM (Buganim et al., 2014; Wang et al., 2023a). Together, these studies underscore the notion that different combinations of TFs can cooperatively activate the pluripotency circuitry (Buganim et al., 2012). A common characteristic of reprogramming-proficient TFs appears to be their ability to access closed chromatin (Soufi et al., 2012), which is a feature shared among so-called pioneer TFs that are also important in early development (Freund et al., 2024).

Table 2.

Representative transcription factor and/or small-molecule combinations that are used to induce pluripotency

Remarkably, lineage-specific TFs unrelated to pluripotency can also modulate the reprogramming process or induce pluripotency. A striking case in point is the myeloid TF C/EBPα, which, when ectopically expressed in mouse pre-B cells alongside OKSM, facilitates their synchronous transition into iPSCs within a few days (Di Stefano et al., 2014). Mechanistically, C/EBPα cooperates with Tet2 to transiently induce chromatin remodeling at the regulatory regions of pluripotency genes, rendering them accessible to OKSM binding (Di Stefano et al., 2016). This observation highlights the fact that cell type-specific TFs unrelated to iPSCs/ESCs can prime somatic cells for efficient reprogramming by rewiring the epigenetic landscape. Perhaps even more surprisingly, the key reprogramming factors Oct4 and Sox2 can be replaced by the counteracting lineage specifiers Gata6 and GMMN (GATA3 and ZNF521/PAX6/OTX2 in human), which normally drive mesendodermal and ectodermal lineage specification, respectively (Montserrat et al., 2013; Shu et al., 2013). The authors of these provocative studies proposed a ‘seesaw model’ where the balance between antagonistic TFs driving early cell fate decisions in the embryo are sufficient to instate pluripotency in somatic cells (Shu et al., 2013). Although these studies provided proof-of-principle evidence for the sufficiency of lineage specifiers to induce pluripotency, it is important to note that C/EBPα only enhances reprogramming in B cells, and the efficiency of reprogramming cells with Gata6 and GMNN is orders of magnitude lower than that of OKSM. Moreover, C/EBPα and GATA factors have much more subtle effects in human than in mouse reprogramming experiments, potentially limiting their general utility in reprogramming experiments (Bueno et al., 2016; Montserrat et al., 2013). Thus, from a practical point of view, OKSM is still the most common choice for reprogramming experiments as it works across diverse species and cell types, and reliably produces iPSCs at efficiencies ranging from 0.1-3% (Stadtfeld and Hochedlinger, 2010).

Cellular reprogramming with small molecules

Soon after the discovery of iPSCs, numerous studies focused on identifying small molecules that could enhance the recovery of iPSCs and/or replace individual TFs (Table 2). For example, ascorbic acid (vitamin C) supplementation of OKSM-expressing cultures enhances the generation of iPSCs by modulating the activity of alpha-ketoglutarate-dependent dioxygenases such as the ten-eleven translocation (Tet)-dependent DNA demethylases and histone H3 lysine-36 (H3K36) demethylases, which are important in early stages of reprogramming (Chen et al., 2013a; Esteban et al., 2010; Wang et al., 2011). Likewise, 5-Aza-cytidine, a cytidine analog that inhibits genomic methylation, boosts the formation of iPSCs by facilitating the activation of pluripotency genes (Mikkelsen et al., 2008), although this treatment often kills cells due to its cytotoxic properties. Histone deacetylase (HDAC) inhibitors such as valproic acid (VPA) and trichostatin A (TSA) reportedly also boost reprogramming by promoting the activation of pluripotency-associated genes (Huangfu et al., 2008). Additionally, VPA can substitute for c-Myc (Huangfu et al., 2008). Moreover, when ascorbic acid is combined with the Gsk3β inhibitor CHIR99021 (Bar-Nur et al., 2014) and optionally with the TGFβ inhibitor RepSox (Vidal et al., 2014), fibroblasts and hematopoietic cells acquire pluripotency more rapidly and near synchronously.

Remarkably, Deng and colleagues identified a cocktail of small molecules that bypasses the requirement for exogenously expressed TFs altogether and thus is sufficient to directly reprogram mouse fibroblasts into so-called chemically induced iPSCs (CiPSCs) (Hou et al., 2013). This cocktail comprises small molecules that replace c-Myc (LSD1 inhibitor tranylcypromine), Oct4 (S-adenosylhomocysteine hydrolase inhibitor 3-deazaneplanocin A and the cAMP agonist forskolin) in addition to VPA, CHIR99021 and 616452 (a TGFβ inhibitor), abbreviated as VC6TFD (Table 2). VC6TFD also reprograms neural stem cells and intestinal epithelial cells into CiPSCs (Ye et al., 2016), although the number of independent studies that have used this cocktail remain limited (Cao et al., 2018; Chen et al., 2023; Fu et al., 2018; Long et al., 2015). Most recently, the Deng lab and others extended these observations to human reprogramming, generating the first human CiPSCs (Guan et al., 2022; Nansubuga et al., 2024 preprint; Zhang et al., 2024 preprint). In contrast to mouse cells, human reprogramming requires the application of different small molecules at different stages of pluripotency induction (Table 2; Fig. 2A). Although chemical reprogramming in both mouse and human cells is substantially less efficient and more protracted than TF-mediated reprogramming, reprogramming efficiency and kinetics can be enhanced by modulating pathways associated with cell proliferation, metabolism (Liuyang et al., 2023) or chromatin (Chen et al., 2023). Collectively, these experiments demonstrate that small molecules are sufficient to induce pluripotency in somatic cells. Although these strategies are conceptually intriguing and potentially valuable for therapeutic applications in the future, their limited use in the literature implies that chemical reprogramming is not yet as robust as TF-based methods.

Fig. 2.

Comparison of chemical- versus transcription factor-based reprogramming approaches. (A) Small molecule-based reprogramming occurs in a stepwise manner involving distinct intermediate stages that include an epithelial and extra-embryonic endoderm (XEN)-like state. XEN cells are cultured stem cells representative of the primitive endoderm of blastocyst embryos. (B) Transcription factor (TF)-based reprogramming using fibroblasts also passes through characteristic intermediate stages, including an early mesenchymal-to-epithelial transition (MET) and the late activation of pluripotency genes. (C) During TF-induced reprogramming, the somatic program is first extinguished by OKSM-mediated sequestration of somatic TFs, leading to somatic enhancer decommissioning, as well as a gain of repressive H3K27me3 at associated promoters via PRC2 (left panel). Pluripotency networks are subsequently activated via Tet2-dependent active demethylation of pluripotency-associated regulatory elements, leading to their activation (right panel). The dotted pink lines represent RNA being transcribed. PRC2, polycomb repressive complex 2; Tet2, ten-eleven translocation methylcytosine dioxygenase 2; H3K27me3, histone H3 lysine-27-trimethylation; O, octamer-binding transcription factor 4; K, Krüppel-like factor 4; S, sex-determining region Y-box 2; M, cellular myelocytomatosis oncogene. (D) Factors that resist reprogramming are often shared with mechanisms that maintain cell identity. For example (1), nucleosome assembly via the histone chaperone CAF-1, and histone methylation at histone H3K9 and H3K36 residues, maintain somatic gene expression programs and/or repress the acquisition of pluripotency-associated programs. Moreover (2), Nudt21 controls the protein levels of fate-instructive chromatin regulators by modulating the use of alternative polyA sites of associated RNAs. (3) Lysine SUMOylation modifies the activity or stability of regulatory factors by attaching SUMO peptides via the conjugating enzyme Ubc9. (4) Chromatin remodeling via the NuRD/Mbd3 histone deacetylase complex reportedly prevents deterministic reprogramming of somatic cells towards iPSCs by keeping pluripotency genes silenced. Mbd3, methyl-CpG binding domain protein 3; NuRD, nucleosome remodeling and deacetylation; CAF-1, chromatin assembly factor 1; H3K9me, histone H3 lysine-9 methylation; H3K36me, histone H3 lysine-36 methylation; Nudt21, nudix hydrolase 21; SUMO, small ubiquitin-like modifier; Ubc9, ubiquitin-conjugating enzyme 9.

Molecular mechanisms and barriers of reprogramming

Over the past two decades, several labs have elucidated the mechanisms of reprogramming by leveraging multi-omics approaches such as RNA-sequencing, and the mapping of chromatin accessibility and chromatin marks, 3D chromatin structure and TF binding in intermediates poised to form iPSCs (Chronis et al., 2017; Mor et al., 2018; Polo et al., 2012; Rais et al., 2013; Schwarz et al., 2018; Soufi et al., 2012). On a cellular level, these studies revealed that fibroblasts progress through an early mesenchymal-to-epithelial transition (MET) (Li et al., 2010; Liu et al., 2013; Samavarchi-Tehrani et al., 2010; Unternaehrer et al., 2014), which is limiting for successful reprogramming (Fig. 2B). Consistent with this notion, mathematical modeling suggested that the reprogramming process involves both stochastic and deterministic phases that are amenable to acceleration by manipulating individual genes or pathways (Hanna et al., 2009).

On a molecular level, OKS factors bind to and redirect somatic TFs from their cognate binding sites to regions elsewhere in the genome, resulting in somatic enhancer decommissioning (Chronis et al., 2017) (Fig. 2C). This step appears to be crucial for successful reprogramming, as the overexpression of somatic TFs, such as members of the AP-1 and RUNX family, significantly impairs reprogramming by competing with OKS (Chronis et al., 2017). In parallel to somatic enhancer decommissioning, promoters of somatic genes gain H3K27me3 via the polycomb repressive complex 2 (PRC2), which ensures the stable silencing of somatic gene expression programs (Fragola et al., 2013; Onder et al., 2012) (Fig. 2C). Interestingly, de novo DNA methylation is not required to silence the somatic program, as Dnmt3a and Dnmt3b are dispensable for the generation of iPSCs (Pawlak and Jaenisch, 2011), suggesting that gene repression mediated by transcriptional and chromatin-based mechanisms is sufficient to extinguish the somatic program.

Once somatic enhancers and promoters have been silenced, OKS are redirected to pluripotency enhancers to gradually establish the pluripotency program (Chronis et al., 2017). In contrast to somatic enhancer decommissioning, the activation of pluripotency enhancers depends on DNA demethylation driven by Tet enzymes, particularly Tet2 (Doege et al., 2012; Hu et al., 2014; Sardina et al., 2018) (Fig. 2C). Tet2-mediated enhancer demethylation/activation is also required during the conversion of fibroblasts to myogenic stem-like cells (Yagi et al., 2021), suggesting a common requirement for DNA demethylation in settings that involve an experimentally induced gain in developmental potential. The fourth original reprogramming factor, c-Myc, is dispensable for reprogramming but enhances the speed and efficiency of iPSC formation in combination with OKS (Nakagawa et al., 2008; Wernig et al., 2008). The inclusion of c-Myc has been associated with an increased frequency of incompletely reprogrammed iPSC-like colonies, which may be due to the aberrant recruitment of NCoR/SMRT corepressors including HDAC3 to pluripotency loci late in reprogramming (Zhuang et al., 2018). However, c-Myc supports early stages of reprogramming by facilitating the silencing of somatic genes (Sridharan et al., 2009) and by enhancing the expression of biosynthetic pathways crucial for the acquisition of pluripotency (Zviran et al., 2019).

The low efficiency and slow kinetics of reprogramming have spurred efforts to identify additional roadblocks to reprogramming whose manipulation could facilitate the recovery of iPSCs. Importantly, this effort might also elucidate fundamental mechanisms that maintain somatic cell identity (Brumbaugh et al., 2019a). Using unbiased screens or candidate approaches, several labs, including ours, have discovered potent reprogramming barriers associated with epigenetic (e.g. NuRD complex, BAF-1 chromatin remodeler and CAF-1 dependent chromatin assembly), post-transcriptional (alternative polyadenylation, alternative splicing) and post-translational (SUMOylation, histone modifications) processes (Borkent et al., 2016; Brumbaugh et al., 2018; Buganim et al., 2013; Cheloufi et al., 2015; Papp and Plath, 2013; Rais et al., 2013; Singhal et al., 2010) (Fig. 2D). These mechanisms have recently been reviewed elsewhere (Brumbaugh et al., 2019a) and will therefore not be discussed in detail here. An emerging theme from these studies is that reprogramming barriers affect common molecular pathways whose manipulation also impacts cellular plasticity in other experimental, physiological and pathological cell fate transitions. A case in point is the histone chaperone CAF-1, which our lab identified as a reprogramming roadblock (Cheloufi et al., 2015). Notably, CAF-1 inhibition also facilitates the transdifferentiation of fibroblasts to neurons (Cheloufi et al., 2015), the reversion of ESCs to two-cell-like totipotent cells (Ishiuchi et al., 2015) and the differentiation of premalignant myeloid progenitors (Franklin et al., 2022).

Although histone modifications dramatically change during reprogramming and normal development, histone-modifying enzymes are typically under-represented in unbiased screens aimed at finding roadblocks to reprogramming. This may be because histone marks are regulated by redundant enzymes, which are expected to compensate for one another in the context of CRISPR or siRNA screens (Hyun et al., 2017). To overcome these limitations and probe the functional role of histone marks in reprogramming, our lab has recently employed histone H3 lysine-to-methionine (K-to-M) mutants (Brumbaugh et al., 2019b; Hoetker et al., 2023) that dominantly block specific histone modifications across the genome (Serdyukova et al., 2023). We found that H3K4 methylation is required for reprogramming, whereas H3K9 methylation inhibits reprogramming, consistent with previous studies that simultaneously depleted multiple redundant histone-modifying enzymes (Chen et al., 2013b; Soufi et al., 2012; Sridharan et al., 2013) (Fig. 2D). Strikingly, we uncovered a profound effect of the H3K36M mutant on reprogramming, with nearly all cells acquiring pluripotency within 7-10 days, suggesting that H3K36 methylation is a major roadblock to reprogramming (Hoetker et al., 2023) (Fig. 2D). Mechanistically, H3K36M facilitates the decommissioning of somatic enhancers acting downstream of TGFβ signals, as well as the concomitant activation of pluripotency enhancers via Tet-dependent DNA demethylation. Our work uncovers a previously unknown dual role for H3K36 methylation in the control of cell identity by integrating a crucial developmental pathway into continual expression of cell type-specific programs, and by opposing the expression of alternative lineage programs via enhancer regulation. Similar to the broad effects of CAF-1 suppression on cellular plasticity, manipulation of H3K36 methylation levels impacted other cell fate transitions such as ESC differentiation, fibroblast-to-neuron and fibroblast-to-muscle transdifferentiation, underscoring the utility of the reprogramming assay to uncover general mediators of cellular plasticity.

Criteria and approaches for assessing bona fide iPSCs

Assays to probe the developmental potential of iPSCs depend on the pluripotent state of the respective iPSCs (e.g. naïve versus primed) (Box 1), the species (e.g. mouse versus human) and the type of application (e.g. in vitro differentiation versus developmental studies) (Maherali and Hochedlinger, 2008; Stadtfeld and Hochedlinger, 2010). To assess the potential of mouse or human iPSCs to spontaneously differentiate into the three germ layers (i.e. endoderm, mesoderm and ectoderm), pluripotent cells are typically coaxed into embryoid bodies (EBs) in vitro or teratomas in vivo (Maherali and Hochedlinger, 2008). These assays are not very stringent but provide a rapid functional readout of iPSCs maintained in either a naive or primed pluripotent state. In human, more stringent assays include differentiating iPSCs into specific cell types using targeted in vitro differentiation protocols (e.g. diverse types of neurons, cardiomyocytes and pancreatic β-cells) or 3D organoid models recapitulating aspects of tissue development comprising multiple cell types [e.g. brain (Lancaster et al., 2013), intestine (Eicher et al., 2022), esophagus and stomach (McCracken et al., 2017, 2014; Trisno et al., 2018)]. In mouse, more stringent assays include the injection of iPSCs into diploid or tetraploid host blastocysts to derive chimeric mice or entirely iPSC-derived mice, respectively (Kang et al., 2009; Boland et al., 2009; Zhao et al., 2009). The latter method is considered the most stringent developmental assay and may only be useful in experiments requiring an assessment of the full developmental potential of mouse iPSCs.

Box 1. Using iPSCs as tools to capture pre-implantation stage-like cells.

2i/LIF culture (Ying et al., 2008) and serum/LIF culture (Nichols et al., 1990) capture mouse pluripotent stem cells in a naïve state that resembles the pre-implantation epiblast. Additionally, serum/LIF cultured ESCs contain extremely rare (∼1%) cells with a two-cell embryo-like identity (Macfarlan et al., 2012), which can be further expanded by manipulating specific genes, including CAF-1 (Ishiuchi et al., 2015), miR-34 (Choi et al., 2017b) and Dux (De Iaco et al., 2017; Hendrickson et al., 2017). Two-cell-like cells are thought to be totipotent, as they incorporate into both the epiblast and trophectoderm of blastocysts, although they cannot be maintained in vitro long term.

In contrast to two-cell-like cells, totipotent mouse expanded potential stem cells (EPSCs) can be maintained long term, contribute more broadly to both embryonic and extra-embryonic tissues in chimeric mice, and support the development of entirely EPSC-derived mice using tetraploid embryo complementation (Yang et al., 2017a,b). Similarly, totipotent blastomere-like stem cells (TBLCs) have been generated by treating mouse and human ESCs/iPSCs with the spliceosome inhibitor pladienolide B (PlaB) (Li et al., 2024; Shen et al., 2021). TBLCs robustly contribute to embryonic and extra-embryonic lineages, and appear to be transcriptionally even more similar to two-cell embryos than EPSCs.

Unlike mouse ESCs/iPSCs, human ESCs/iPSCs are thought to represent a more advanced (primed) pluripotent state resembling the post-implantation epiblast, which motivated efforts to capture human ESCs/iPSCs in a more primitive state (Weinberger et al., 2016). Indeed, different protocols have been developed to derive human cell lines resembling a naive pluripotent epiblast-like state (Guo et al., 2017; Takashima et al., 2014; Theunissen et al., 2014; Bayerl et al., 2021) or a totipotent eight-cell-like state (Mazid et al., 2022).

Importantly, only (mouse) iPSCs cultured in naive conditions [i.e. serum/LIF (leukemia inhibitory factor) (Nichols et al., 1990) or 2i (MEK inhibitor and Gsk3β inhibitor)/LIF (Ying et al., 2008)] will contribute to animals upon blastocyst injection, as iPSCs maintained in primed conditions (i.e. bFGF/activin) are too advanced and no longer contribute to development when injected into blastocysts. Moreover, iPSCs can only generate the three embryonic germ layers consistent with their pluripotent state. By contrast, expanded potential stem cells (EPSCs) and blastomere-like stem cells (TBLCs) generated from iPSCs using small molecules (Box 1) possess the potential to develop into extra-embryonic as well as embryonic tissues, consistent with the acquisition of a totipotent state. Thus, environmental signals can profoundly alter the developmental state and differentiation potential of iPSCs.

iPSCs generated with non-integrating vectors or inducible systems are transcriptionally, epigenetically and functionally highly similar to fertilization-derived ESCs (Choi et al., 2015), which are considered the gold standard in the field. However, iPSCs do occasionally exhibit epigenetic abnormalities that may impact their developmental potential. For example, reprogramming to pluripotency is sometimes accompanied by the aberrant gain of DNA methylation at so-called imprinting control regions (ICRs) that are important for fetal development (Bar et al., 2017; Stadtfeld et al., 2010a; Yagi et al., 2019, 2017b). Similarly, low-passage iPSCs sometimes exhibit incomplete erasure of somatic DNA methylation patterns, leading to a persistence of epigenetic marks associated with the starting somatic cell, which has been referred to as ‘epigenetic memory’ (Bar-Nur et al., 2011; Kim et al., 2010; Lister et al., 2011). Conversely, maintenance of established iPSCs (or ESCs) in certain naive culture conditions (i.e. 2i/LIF) representing the pre-implantation epiblast (Box 1) typically leads to global DNA hypomethylation and defects in the maintenance of genomic imprinting, as well as chromosomal aberrations upon prolonged culture (Choi et al., 2017a; Pastor et al., 2016; Theunissen et al., 2016; Yagi et al., 2017a). As epigenetic and genomic aberrations are inherited in mature progeny derived from iPSCs (Lister et al., 2011), such changes can potentially interfere with certain applications of iPSCs and warrant careful consideration.

Crucially, suppression of de novo DNA methyltransferases, particularly DNMT3A, or treatment with small molecules such as ascorbic acid, can prevent mouse and human iPSCs from acquiring aberrant hypermethylation at ICRs during reprogramming (Stadtfeld et al., 2012; Yagi et al., 2019). Likewise, modifications to naïve culture conditions that prevent severe hypomethylation, including the titration of MEK inhibitor or the use of alternative inhibitors (e.g. Src inhibitor) can preserve a more normal imprinting status and reduce genomic instability in established mouse and human iPSCs (Bayerl et al., 2021; Choi et al., 2017a; Di Stefano et al., 2018; Khan et al., 2021; Yagi et al., 2017a). Recent work has suggested that the transient exposure of human reprogramming cultures to naïve conditions may be an efficient way to prevent epigenetic and transcriptional abnormalities, and this approach does not seem to disrupt genomic imprinting (Buckberry et al., 2023). As most of these studies have focused on iPSCs generated by transcription factors (OKSM), it will be crucial to carefully evaluate these epigenetic processes in CiPSCs, particularly since several of the utilized small molecules target epigenetic pathways.

Another important consideration for comparative studies between different iPSC lines and between iPSC versus ESC lines is genetic background. Variations in genetic background between pluripotent stem cells have been recognized as the main driver behind transcriptional and epigenetic differences associated with ESCs and iPSCs (Choi et al., 2015; Rouhani et al., 2014). Thus, unperturbed isogenic human iPSCs or iPSCs with a similar genetic background as the experimental human iPSCs are the ideal control for comparative studies, which is particularly relevant for patient-specific cell lines harboring potentially subtle transcriptional, epigenetic and functional differences.

In vivo reprogramming for tissue regeneration and rejuvenation

Transient OKSM expression in vitro generates some degree of plasticity at intermediate stages (Schiebinger et al., 2019). This discovery prompted researchers to test whether transient OKSM expression in vivo could be used in the context of tissue regeneration and rejuvenation, and subsequent studies indeed demonstrated that a brief (2 day) but repeated induction of OKSM in vivo bypasses teratoma formation and premature death, and ameliorates several hallmarks of aging in mice (Ocampo et al., 2016) (Fig. 3A). Importantly, transient OKSM expression appears to improve age-associated regenerative defects across diverse tissues, including the skin (Doeser et al., 2018), muscle (Wang et al., 2021a), liver (Hishida et al., 2022), heart (Chen et al., 2021) and nervous system (Xu et al., 2024). Moreover OKS(M) expression was reported to restore vision in animals that experienced optical nerve crush injuries (Lu et al., 2020) and to improve cognitive function in a progeroid mouse model (Rodriguez-Matellan et al., 2020). Although these studies are very encouraging, it is important to recognize that the mechanisms underlying these purported rejuvenation phenotypes remain, with rare exceptions (Lu et al., 2020), poorly understood and warrant further examination. Moreover, there is a remaining risk of teratoma formation, even in cases where OKSM is expressed cyclically. For example, when OKSM is induced briefly in vivo but abrogated before pluripotency is acquired, kidney tumors develop (Ohnishi et al., 2014). Thus, it will be imperative to carefully control the duration and levels of OKSM expression before applying in vivo reprogramming in a clinical setting. It will also be important to understand whether other reprogramming TFs or small molecules have similar effects on tissue rejuvenation and regeneration to OKSM, but perhaps without the associated risk of tumor formation. Indeed, two recent studies suggested that small molecules previously used to enhance iPSC reprogramming improve age-associated cellular and molecular changes, and reduce transcriptional and epigenetic aging clocks (Mitchell et al., 2024; Yang et al., 2023).

Fig. 3.

Applications of reprogramming technology in rejuvenation research and embryo modeling. (A) Partial reprogramming involves short pulses of OKSM expression in cultured cells or mice (left panel). This approach is reportedly insufficient to generate iPSCs in vitro or teratomas in vivo but sufficient to reverse some hallmarks of aging, consistent with a rejuvenation-like phenotype. Rejuvenation-like phenotypes after OKSM pulses have been observed at the molecular, tissue and organism levels (right panel), including, for example, the restoration of H3K9 and H4K20 methylation towards a youthful state, improved tissue architecture and, in some models, an extension of lifespan. H4K20me, histone H4 lysine-20 methylation; H3K9me, histone H3 lysine-9 methylation. (B) OKSM expression in fibroblast cultures reportedly gives rise to both extra-embryonic-like (blue) and embryonic-like (red) cell types that can self-assemble into 3D blastocyst-like models. Trophoblast stem cells (TSCs) are cultured stem cells representative of trophectoderm precursors of the blastocyst embryo, which give rise to differentiated cells of the placenta. TSCs can either be derived directly from fibroblast cultures undergoing reprogramming (dotted line) or indirectly from naive iPSCs using cytokines (solid line). iPSC, induced pluripotent stem cell; O, octamer-binding transcription factor 4; K, Krüppel-like factor 4; S, sex-determining region Y-box 2; M, cellular myelocytomatosis oncogene.

In vitro models of embryogenesis using reprogramming tools

Human embryonic development has been difficult to study due to the inaccessibility of early embryogenesis, limited numbers of donated embryos and derivative human ESCs, as well as ethical considerations and legal restrictions (Lovell-Badge et al., 2021; Rivron et al., 2023). However, recent advances in capturing different states of pluripotency, combined with reprogramming technology, have enabled the development of embryo models that recapitulate key aspects of pre-implantation development in mouse and human (Handford et al., 2024) (Fig. 3B).

In 2018, Rivron et al. established the first self-organizing 3D model of pre-implantation blastocysts by aggregating mouse ESCs and trophoblast stem cells (TSCs) in vitro (Rivron et al., 2018). These embryo models are termed ‘blastoids’ and have a similar morphology and transcriptional profile to murine E3.5 blastocysts. Subsequently, Li et al. demonstrated that blastoids could also be generated from EPSCs without the use of TSCs, including EPSCs derived from fibroblasts via TF-induced reprogramming (Li et al., 2019). In 2021, the first human blastoid models were developed (Liu et al., 2021; Yu et al., 2021). Specifically, Liu et al. used TF-induced reprogramming of human fibroblasts towards iPSCs and observed that cell types with molecular characteristics of epiblast, trophectoderm and primitive endoderm simultaneously emerged. When the authors allowed these reprogramming intermediates to aggregate, they self-organized into structures reminiscent in morphology and cellular composition of E5-E7 human blastocysts, including epiblast, primitive endoderm and polar and mural trophectoderm-like cells. In a parallel approach, Rivron and colleagues took advantage of the intrinsic potential of human naïve ESCs and iPSCs to give rise to trophectoderm (Kagawa et al., 2022). When naïve cells were aggregated in the presence of Hippo, TGFβ and ERK inhibitors, they gave rise to blastoids at high efficiency (∼70%) containing epiblast, trophectoderm and primitive endoderm-like cells. These in vitro-derived models have not yet been shown to develop to term in mouse, and human embryo models cannot be tested for development in vitro beyond 14 days post fertilization or the formation of the primitive streak, whichever occurs first (Rivron et al., 2023). Nevertheless, these models are extremely valuable for studying pre-implantation development as they recapitulate key aspects of fertilization-derived blastocysts, such as self-organization, cell lineage specification, embryo polarity, morphogenesis and competency for implantation (Fan et al., 2021; Sozen et al., 2019, 2021; Yanagida et al., 2021). Looking ahead, iPSC-based embryo models could enable not only the mechanistic and spatiotemporal study of early development, but also advance our understanding of infertility and diseases affecting early fetal development, which have previously been difficult to study.

Conclusions and perspectives

TF- and small molecule-based reprogramming approaches to generate iPSCs have become powerful new paradigms for basic research as well as translational applications. We have discussed in this Primer how the selection of reprogramming methods depends in large part on the specific application, with polycistronic transgenic, lentiviral or transposon-based approaches being the methods of choice for most basic research application, owing to the homogeneous expression of OKSM and a reasonable reprogramming efficiency. By contrast, virus-free integration-free methods, including RNA-, Sendai viral- or episomal-based approaches are the methods of choice for disease modeling and therapeutic applications, as they prevent permanent alterations to the genome. Although non-integrating strategies may lead to more variable reprogramming efficiencies than integrating strategies, recent technological improvements and the access to commercial kits have made these approaches more readily available and reproducible (Fusaki et al., 2009; Schlaeger, 2018; Schlaeger et al., 2015; Warren et al., 2010). An intriguing alternative approach that does not require overexpression of any gene products is chemical reprogramming. However, before using this approach in a clinical setting, it is imperative to further dissect the underlying mechanisms, improve reprogramming efficiency and speed and perform careful side-by-side comparisons with TF-based approaches in terms of ability to reprogram different cell types and to maintain epigenetic and genomic integrity, as well as the differentiation potential of resultant iPSCs.

Studies dissecting the mechanisms of reprogramming have uncovered barriers to iPSC formation, including targetable epigenetic and signaling pathways, the manipulation of which allows dramatically improved reprogramming efficiency. These insights can be leveraged for mechanistic studies to produce large amounts of intermediate cells, as well as for therapeutic applications to boost the production of patient iPSCs from rare starting materials. Remarkably, studies focused on barriers to reprogramming have elucidated general mechanisms controlling cellular plasticity in the context of other experimental and physiological cell fate transitions, as well as cancer. We believe the reprogramming assay continues to be a valuable discovery platform for regulators and mechanisms controlling cell identity and cellular plasticity.

The transient expression of OKSM has become an attractive model for promoting some aspects of tissue regeneration and alleviating certain hallmarks of aging, underscoring the utility of reprogramming factors to dissect mechanisms that modulate healthspan and lifespan. Given the oncogenic potential of the reprogramming factors, it will be important to further optimize these protocols before considering therapeutic applications using reprogramming strategies. Similarly, it will be key to define the underlying mechanisms by which the reprogramming factors reverse some of these hallmarks of aging without inducing pluripotency or changing cell identity. Such mechanistic insights have the potential to reveal targetable pathways that act downstream of OKSM but without the associated concern of promoting oncogenesis.

Finally, the use of reprogramming TFs and small molecules has enabled the reprogramming of mouse and human pluripotent stem cells to resemble the cell states present in the early embryo. In addition to providing models for studying early pluripotent cell states, these stem cell systems have enabled the assembly of embryo-like structures that recapitulate key aspects of pre-implantation development. As such, they provide a unique opportunity for studying the earliest stages of human development, as well as for understanding embryo loss associated with infertility. As these embryo models are relatively new, it will be important to further compare and contrast available protocols before concluding which strategy is best suited for which scientific question or application. Additionally, it will be important to carefully consider ethical concerns associated with the use of human models that can, in principle, implant in the uterus.

In summary, advances in iPSC technology over the past two decades have revolutionized the field of stem cell biology, provided fundamental insights into mammalian development and TF/chromatin interactions, and offered a tractable tool with which to study diseases and modulate aspects of aging. We expect that the continued mechanistic dissection of these processes has the potential to uncover novel pathways and targets for the treatment of degenerative conditions, as well as aging.