Moving toward totipotency: the molecular and cellular features of totipotent and naive pluripotent stem cells

Abstract

BACKGROUND

Dissecting the key molecular mechanism of embryonic development provides novel insights into embryogenesis and potential intervention strategies for clinical practices. However, the ability to study the molecular mechanisms of early embryo development in humans, such as zygotic genome activation and lineage segregation, is meaningfully constrained by methodological limitations and ethical concerns. Totipotent stem cells have an extended developmental potential to differentiate into embryonic and extraembryonic tissues, providing a suitable model for studying early embryo development. Recently, a series of ground-breaking results on stem cells have identified totipotent-like cells or induced pluripotent stem cells into totipotent-like cells.

OBJECTIVE AND RATIONALE

This review followed the PRISMA guidelines, surveys the current works of literature on totipotent, naive, and formative pluripotent stem cells, introduces the molecular and biological characteristics of those stem cells, and gives advice for future research.

SEARCH METHODS

The search method employed the terms ‘totipotent’ OR ‘naive pluripotent stem cell’ OR ‘formative pluripotent stem cell’ for unfiltered search on PubMed, Web of Science, and Cochrane Library. Papers included were those with information on totipotent stem cells, naive pluripotent stem cells, or formative pluripotent stem cells until June 2024 and were published in the English language. Articles that have no relevance to stem cells, or totipotent, naive pluripotent, or formative pluripotent cells were excluded.

OUTCOMES

There were 152 records included in this review. These publications were divided into four groups according to the species of the cells included in the studies: 67 human stem cell studies, 70 mouse stem cell studies, 9 porcine stem cell studies, and 6 cynomolgus stem cell studies. Naive pluripotent stem cell models have been established in other species such as porcine and cynomolgus. Human and mouse totipotent stem cells, e.g. human 8-cell-like cells, human totipotent blastomere-like cells, and mouse 2-cell-like cells, have been successfully established and exhibit high developmental potency for both embryonic and extraembryonic contributions. However, the observed discrepancies between these cells and real embryos in terms of epigenetics and transcription suggest that further research is warranted. Our results systematically reviewed the established methods, molecular characteristics, and developmental potency of different naive, formative pluripotent, and totipotent stem cells. Furthermore, we provide a parallel comparison between animal and human models, and offer recommendations for future applications to advance early embryo research and assisted reproduction technologies.

WIDER IMPLICATIONS

Totipotent cell models provide a valuable resource to understand the underlying mechanisms of embryo development and forge new paths toward future treatment of infertility and regenerative medicine. However, current in vitro cell models exhibit epigenetic and transcriptional differences from in vivo embryos, and many cell models are unstable across passages, thus imperfectly recapitulating embryonic development. In this regard, standardizing and expanding current research on totipotent stem cell models are essential to enhance our capability to resemble and decipher embryogenesis.

Graphical Abstract

Pluripotent stem cells progressively obtain characteristics resembling totipotent embryos and exhibit bidirectional developmental potential. ICM, inner cell mass; TE, trophectoderm.

Open in new tabDownload slide

Introduction

Early embryonic development of mammals is accurately regulated by intricate molecular networks. The fusion of an oocyte with a sperm leads to the formation of a totipotent zygote, which is endowed with the remarkable ability to generate all embryonic and extraembryonic tissues. Understanding the mechanisms regulating embryo development has profound implications, as early developmental arrest can lead to infertility, and developmental aberrations can contribute to malformations and miscarriages. Insights into embryonic lineage segregation enhance our knowledge of organogenesis and support the development of organ models, such as humanized organs, generated by fusing high-potential stem cells with animal embryos (Wang et al., 2023a). Thus, a thorough understanding of the intricate machinery underpinning embryonic development is paramount for promoting human reproductive health.

The mammalian zygote undergoes continuous cell divisions to generate 2-cell (2C), 4-cell (4C), and 8-cell (8C) embryos (Gafni et al., 2013; Yeh et al., 2021). Subsequently, the embryo undergoes compaction to form a morula and initiates the first lineage differentiation, resulting in the formation of a blastocyst (Gamow and Prescott, 1970; White et al., 2018). This structure is composed of two distinct cell types: the inner cell mass (ICM) and the trophectoderm (TE) (Petropoulos et al., 2016; Zhang et al., 2018). Before implantation, the embryo undergoes a second lineage differentiation, giving rise to the epiblast (EPI), primitive endoderm (PrE), and TE (Fig. 1) (Petropoulos et al., 2016; Zhang et al., 2018). In mice, the zygote and 2C stage embryos both fulfill the strictest criteria of totipotency (Tarkowski, 1959; Suwinska, 2012), as a single cleavage cell has the capacity to develop into a complete organism in vivo, known as ‘in vivo canonical totipotency’. In humans, totipotency theoretically extends up to the 4C–8C stage (Braude et al., 1988; Van de Velde et al., 2008; Guo et al., 2014; Smith et al., 2014; Mazid et al., 2022). Studies have shown that blastomeres from human 4C embryos can independently develop into both ICM and TE lineages of blastocysts in vitro (Van de Velde et al., 2008). However, due to ethical constraints, it is not possible to transplant human blastomeres into the maternal uterus to confirm their in vivo canonical totipotency. Additionally, a broader definition of totipotency, termed ‘in vitro experimental totipotency’, encompasses the capacity of cells to differentiate into both embryonic and extra-embryonic lineages in vitro (Riveiro and Brickman, 2020). Current stem cells with certain totipotent characteristics are considered to exhibit in vitro experimental totipotency. However, these cells have not displayed the ability to organize into a complete organism according to a specific temporal or spatial sequence (Shen et al., 2021; Mazid et al., 2022; Li et al., 2024).

Zygotic genome activation (ZGA) marks the beginning of the zygotic transcript production for a totipotent embryo (Taubenschmid-Stowers et al., 2022; Ji et al., 2023). ZGA is initiated in a minor wave and is subsequently followed by a major wave. In mice, the minor wave of ZGA occurs from the late zygote to early 2C stage embryo, whereas the major wave of ZGA occurs during the middle-to-late 2C stage (Aoki et al., 1997; Abe et al., 2018). This process involves significant epigenetic remodeling, characterized by rapid chromatin decondensation, extensive reprogramming of post-translational modification of core histones, and global DNA demethylation (Wu et al., 2016; Zheng et al., 2016; Wang et al., 2021c). Additionally, mouse 2C stage embryos exhibit increased chromatin mobility and slow DNA replication fork speed (Boskovic et al., 2014; Nakatani et al., 2022). During the remodeling of histone modification and global DNA demethylation, transposable elements such as LINE-1 and endogenous retroviruses (ERVs) are highly expressed (Eckersley-Maslin et al., 2016; Jachowicz et al., 2017). LINE-1 elements regulate global chromatin accessibility during early mouse embryonic development, establishing a relatively accessible chromatin state (Jachowicz et al., 2017; Wu et al., 2018). Moreover, mouse ZGA is facilitated by the activities of transcription factors of the DUX family (mouse DUX), which activate downstream ERVs and ZGA-associated genes, including Zscan4, Sp110, and Cml2 (De Iaco et al., 2017; Hendrickson et al., 2017). Recent reports have highlighted the role of OBOX, a PRD-like homeobox domain transcription factor family, which can bind and activate several ZGA genes and MERVL repeats (Ji et al., 2023; Lai et al., 2023).

In humans, the minor wave of ZGA occurs at the 2C–4C stage, while the major wave takes place at the 4C–8C stage (Wu et al., 2018). Human and mouse early embryonic development both exhibit similarities and species-specific differences. For example, DUX4, the human ortholog of mouse Dux, activates the expression of numerous early human ZGA target factors during the 4C–8C stage, including ZSCAN4 and HERVL retrotransposons. Essential factors such as CTCF and KLF exhibit high conservation between humans and mice, whereas ESRRB, and NR5A2, pivotal in mouse ZGA, show comparably low expression levels in humans, underscoring species specificity (Wu et al., 2018; Chen et al., 2019b; Gassler et al., 2022; Ji et al., 2023). Similarly, several PRD-like homeobox genes such as TPRX1, TPRX2, and TPRXL play specific roles in humans (Zou et al., 2022). Although significant advances have been made in understanding the characteristics of totipotency in mice, there remain huge gaps in our knowledge of human totipotency. Further exploration is still necessary to elucidate the precise mechanisms governing the maintenance of the totipotent state.

Given the transience of totipotency during embryonic development and the ethical challenges associated with obtaining embryos for study, researchers are developing more stable in vitro models to investigate totipotency. In recent years, remarkable breakthroughs have been made in the in vitro study of totipotency stem cells in both humans and mice. Notably, various systems of 2C-like cells (2CLCs) in mice or 8C-like cells (8CLCs) in humans have been successfully established in vitro (Grow et al., 2021; Shen et al., 2021; Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Xu et al., 2022; Hu et al., 2023). Concurrently, significant progress in pluripotent stem cell research across diverse species has been supported by the derivation of stem cells from mice and humans (Chen et al., 2015, 2020; Honda et al., 2017). In this review, we focused on comparing the molecular and biological characteristics of different pluripotent stem cells, with particular emphasis on totipotent stem cells. We also introduced the applications of these stem cell models in embryo-like models and X chromosome inactivation (XCI) research. We highlight the challenges associated with stem cell models and the necessity for comprehensive functional validation of established models, and we advocate for the creation of dependable datasets for public research applications. Furthermore, we compared stem cell studies across different species for future investigation.

Methods

Search strategy

A comprehensive search of PubMed, Cochrane Library, and Web of Science was undertaken to retrieve research articles published until 1 June 2024. Search items included ‘totipotent’ OR ‘naive pluripotent stem cell’ OR ‘formative pluripotent stem cell’. The articles were manually filtered by searching the selected articles and references to identify additional studies. The flow diagram of the selection procedure, including the number of articles filtered is shown in Fig. 2.

Selection criteria

The full text of the abstract of the literature title was screened independently by the two authors (L.H. and Y.P.) using the following eligibility criteria: 1. studies investigating models of totipotent, naive pluripotent, or formative pluripotent stem cells (fPSCs); 2. research on human, mouse, porcine, and cynomolgus macaques; 3. accessible work published in English; and 4. research articles with full text. Literature that did not meet these standards was excluded. For example, we excluded articles that lacked a focus on stem cells, despite mention of the topic (Illmensee and Mintz, 1976).

Data extraction and quality of evidence assessment

The results of the literature search were extracted into Endnote 20.0 software (Clarivate, PA, USA) and duplications were manually removed. The initial search yielded 7484 results in general, which were subsequently applied to Endnote 20. Following the removal of 2624 duplicate results, the rest of the articles were included for further analysis. There were 19 articles were included in other ways. A comprehensive analysis of the remaining articles was conducted based on their titles and abstracts, leading to the exclusion of 4727 articles based on non-mammalian, non-English, non-full text, non-original articles, and content that was unrelated to the topic. Finally, 152 articles were included, as presented in Supplementary Table S1. The two authors independently extracted key information from these articles and rated the importance of the evidence as ‘high’, ‘moderate’, and ‘low’ data.

Synthesis of results

The data were narratively synthesized by categorizing the stem cell models according to the species of cells used in the studies. Two authors independently evaluated all included articles to minimize bias. In the event of disagreement, a third author conducted an assessment, and final decisions were made through unanimous agreements to each on their inclusion. Specifically, 67 studies used human stem cells, 70 studies used mouse stem cells, 9 studies used porcine stem cells, and 6 studies used cynomolgus stem cells. These categories were then further subdivided based on the developmental potency of the cells.

Additional analysis

Due to the differences in study designs and methods, it was not feasible to conduct a meta-analysis.

Results

Mouse naive pluripotent stem cells

Mouse embryonic stem cells (mESCs) were first generated in 1981 as the first mouse naive pluripotent stem cell. These cells have the potential to differentiate into all embryonic lineages and are capable of forming chimeras (Evans and Kaufman, 1981). Typically, mESCs are derived from the ICM of pre-implantation embryos at around E3.5–4.5 days and cultured on feeder cells (Fig. 3A) (Evans and Kaufman, 1981; Boroviak et al., 2015; Martinez-Val et al., 2021). mESCs are characterized by their dome-shaped colonies, active X chromosomes in female stem cells, and a transcriptomic signature that is largely similar to that of pre-implantation EPIs (Schulz et al., 2014; Sperber et al., 2015). mESCs rely on the LIF signaling, which can activate the JAK-STAT3 signaling pathway, promoting the expression of target genes Esrrb, Klf4, Nanog, C-Myc, and Tfp2l1, thereby maintaining the naive pluripotent state (Welling and Geijsen, 2013; Qiu et al., 2015). In the absence of feeders, BMP signaling can also maintain naive pluripotency, supporting the self-renewal of mESCs through activating the Smad signaling pathway (Smith et al., 1988; Nichols and Smith, 2009; Onishi et al., 2014). mESCs can also be maintained in ‘2iLIF’ medium, consisting of PD0325901 (a MEK inhibitor), CHIR99021 (a GSK3 inhibitor), and LIF (Ying et al., 2008). Cells cultured in 2iLIF medium exhibit greater stability and lower cellular heterogeneity, and are less prone to spontaneous differentiation (Kolodziejczyk et al., 2015). The naive state can also be achieved from primed pluripotent stem cells through the overexpression of naive pluripotency transcription factors such as Bmp4, Klf4, Klf2, Nanog, Esrrb, Sall1, Tfcp2l1, Nr5a1, and Nr5a2, or partial inhibition of transcription factors such as Zfp281, Zfp706, and Erk5 (Guo et al., 2009; Guo and Smith, 2010; Festuccia et al., 2012; Ye et al., 2013; Leeb et al., 2014; Stuart et al., 2014; Yeo et al., 2014; Williams et al., 2016; Mayer et al., 2020; Yu et al., 2020).

Human naive pluripotent stem cells

More than a decade after the derivation of mESCs, researchers successfully derived human embryonic stem cells (hESCs) from human embryos (Evans and Kaufman, 1981; Thomson et al., 1998). Unlike mESCs, which form dome-shaped and exhibit naive pluripotent stem cell features, hESCs are more similar to primed pluripotent stem cells, developing into flat, epithelial-like colonies (Evans and Kaufman, 1981; Thomson et al., 1998). Furthermore, the naive induction system (2iLIF) effective for mESCs fails to support the derivation of human naive stem cells (Ying et al., 2008; Hanna et al., 2010). Naive human pluripotent stem cells (hPSCs) were first reported in 2009 and described to resemble EPI cells of E6–E7 human preimplantation blastocysts (Fig. 4A) (Li et al., 2009). Subsequently, various induction methods have been developed to improve the derivation efficiency of naive hPSCs, involving the overexpression of exogenous transcription factors and modulation of key signaling pathways (Chan et al., 2013; Ware et al., 2014; Guo et al., 2016, 2017; Zimmerlin et al., 2016; Szczerbinska et al., 2019; Taei et al., 2020; Bayerl et al., 2021). By employing different combinations of transcription factors and modulating key pathways, several culture systems, including 3iL, NHSM, t2iLGö, 5iLAF, PXGL, HENSM, and ReST, have been established to induce naive hPSCs (Chan et al., 2013; Gafni et al., 2013; Theunissen et al., 2014; Takashima et al., 2015; Kilens et al., 2018; Bredenkamp et al., 2019; Bayerl et al., 2021). The induced naive hPSCs exhibit a dome-shaped morphology and a transcriptomic signature resembling that of pre-implantation EPIs. Expression levels for naive markers such as REX1, KLF2, KLF17, and DPPA3 are upregulated, while lineage-specific factors such as OTX2 and SOX11 are downregulated. Additionally, these cells display X chromosome reactivation, global DNA demethylation, reduced levels of H3K27me3, and H3K9me3 in epigenetic regulation. Furthermore, naive hPSCs exhibit higher levels of mitochondrial oxidative phosphorylation (OXPHOS) metabolism (Duggal et al., 2015; Takashima et al., 2015; Theunissen et al., 2016; Wang et al., 2018; Pontis et al., 2019; Szczerbinska et al., 2019; Bayerl et al., 2021).

Several studies have reported that the efficiency of differentiating naive hPSCs directly into the three germ layers is lower compared to that for primed hPSCs. Human naive pluripotent stem cells need to be transformed into a primed state before differentiation into the three germ layers (Guo et al., 2017; Liu et al., 2017; Sahakyan et al., 2017; Smith, 2017; Rostovskaya et al., 2019). Nonetheless, naive hPSCs have been demonstrated to show enhanced differentiation potential toward the extraembryonic lineage compared to primed hPSCs (Cinkornpumin et al., 2020; Dong et al., 2020; Liu et al., 2020; Io et al., 2021). Previous research has revealed the expression of trophoblast-related transcription factors, such as ELF3 and TFAP2C, in naive hPSCs (Theunissen et al., 2016). Recent research found that inhibition of TGFβ signaling in naive hPSCs could induce their conversion into trophoblast stem cells (TSCs) (Osnato et al., 2021).

It is worth noting that considerable differences are evident in naive hPSCs that are induced by various methods. In terms of genomic stability, naive hPSCs cultured in 5iLAF are more prone to aneuploidy, while those derived using NHSM/PXGL/t2iLGö/ReST systems exhibit karyotypic stability (Liu et al., 2017; Warrier et al., 2017). In terms of epigenetic stability, cells cultured in 5iLA and t2iLGö show significant loss of methylation at imprinting regions, which persists even after conversion to a primed state. Although NHSM and ReST culture conditions also result in the demethylation of some imprinting regions, the DNA methylation levels in major imprinting regions remain unchanged (Pastor et al., 2016; Theunissen et al., 2016; von Meyenn et al., 2016; Liu et al., 2017). Female mESCs are intriguing models for studying XCI due to their two active X chromosomes (XaXa) and the occurrence of random XCI during differentiation. In contrast, although both X chromosomes are active in female naive hPSCs, they display heterogeneous XIST states and non-random XCI upon transition to the primed state, limiting their availability as a model to study human XCI. By completely blocking autocrine FGF signaling, naive hPSC that are more uniform and experience random XCI can be induced (Vallot et al., 2015; An et al., 2020).

In summary, while naive stem cells can be induced by various methods, there is still a lack of uniform standards in culture conditions and evaluation criteria for characterizing these stem cells. Further refinement of induction methods that optimally preserve genomic stability and maintain epigenetic features of naive stem cells would ultimately be desirable to the research community.

Mouse formative pluripotent stem cells

Mouse naive pluripotent stem cells resemble the pre-implantation EPI, while mouse primed pluripotent stem cells are analogous to the E7.5 post-implantation EPI in vivo. The intermediate state between these two pluripotent stages remains poorly characterized due to the absence of suitable cellular models. In 2011, Hayashi et al. (2011) induced the first mouse fPSCs known as EPI-like cell (EpiLC) from mESCs for the first time through treatment with basic fibroblast growth factor (bFGF) and Actin. These cells are intermediate between the naive and primed states, possess formative pluripotency, and can directly respond to primordial germ cell-like cell (PGCLC) induction (Fig. 3A) (Hayashi et al., 2011). Notably, mouse primed pluripotent stem cells were found to respond to PGCLC induction, albeit in a low efficiency, whereas naive pluripotent stem cells lack the capability (Hayashi and Surani, 2009; Ohinata et al., 2009; Hayashi et al., 2011). fPSCs like EpiLCs, which can directly respond to PGCLC induction, are positioned between these two states (Hayashi et al., 2011). EpiLCs exhibit a flat epithelial structure similar to mouse EPI cells, and their transcriptomic signatures are similar to those of cells of the mouse E5.75 EPI. However, EpiLCs are transient and undergo extensive cell death after 2 days of induction (Hayashi et al., 2011). In 2020, Neagu et al. induced rosette-like stem cells (RSCs) by treating mESCs and blastocysts with a combined inhibition of Wnt and MEK signaling. RSCs exhibit a rose-like morphology, can be stably passaged for 28 generations, and resemble the mouse E5.0 EPI (Neagu et al., 2020). Subsequently, mouse FTW cells, formative stem (FS) cells, fPSCs, and EPI-like stem cells (EpiLSCs) were successively induced in vitro. These cells can be stably passaged and directly respond to PGCLC induction (Kinoshita et al., 2021; Wang et al., 2021a; Yu et al., 2021b; Luo et al., 2023).

Morphologically, mouse RSCs, FTW cells, and fPSCs have rosette-like structures, whereas FS cells form flat colonies and EpiLSCs form dome-shaped colonies (Neagu et al., 2020; Kinoshita et al., 2021; Wang et al., 2021a; Yu et al., 2021b; Luo et al., 2023). Transcriptomic analyses indicate that RSCs, FTW-ESCs, FS cells, and fPSCs resemble mouse E5.0 EPIs, E5.0-6.0 EPIs, E5.5-6.0 EPIs, and E6.0-6.5 EPIs, respectively. Moreover, mouse EpiLSCs have been found to represent an intermediate between the state of FTW and FS cells, and show similarity to cells of the mouse E5.5 EPI (Neagu et al., 2020; Kinoshita et al., 2021; Yu et al., 2021b; Luo et al., 2023). Meanwhile, mouse FS cells and fPSCs exhibit relatively low expression of mouse naive genes and are close to primed pluripotent stem cells, whereas mouse FTW cells and RSCs have high levels of naive gene expression and are proximal to naive pluripotent stem cells (Neagu et al., 2020; Kinoshita et al., 2021; Wang et al., 2021a; Yu et al., 2021b; Luo et al., 2023). In female stem cells, mouse RSCs and FTW cells display two active X chromosomes, whereas FS cells and fPSCs exhibit one active X chromosome (Neagu et al., 2020; Wang et al., 2021a; Yu et al., 2021b; Luo et al., 2023). The differences in the induced formative cell states may arise from variations in Wnt pathway stimulation. Of note, FTW cells and EpiLSCs show Wnt pathway activation, while FS cells and fPSCs feature Wnt signaling inhibition (Neagu et al., 2020; Wang et al., 2021a; Yu et al., 2021b; Luo et al., 2023). Understanding the specific properties of different models can facilitate the exploration of the continuous pluripotent state transition process from naive to primed state.

Human formative pluripotent stem cells

Formative hPSCs also capable of forming all three germ layers, as well as inducing the formation of PGCLCs. In 2021, two types of formative hPSCs, human formative pluripotent stem cells (FS cells) and FTW-hPSCs, were successfully constructed and utilized induction methods based on mouse FS cells (Kinoshita et al., 2021; Yu et al., 2021b). Kinoshita et al. (2021) derived human FS cells from naive hPSCs by inhibiting the WNT and retinoic acid (RA) signaling pathways. Human FS cells display downregulation of naive markers KLF4, KLF17, and TFCP2L1 and upregulation of the primed markers such as SOX1 and OTX2 (Kinoshita et al., 2021). Transcriptomic analysis positioned human FS cells in close proximity to the post-implantation EPI, whereas primed hPSCs were more similar to early gastrula cells (Kinoshita et al., 2021). Moreover, human FS cells exhibit the ability to directly respond to PGCLC induction (Kinoshita et al., 2021). Another formative hPSCs, referred to as FTW-hPSCs, were reported through the activation of FGF, TGF-β, and WNT/β-catenin signaling pathways in primed hPSCs. These cells can be stably passaged (∼30 generations), display multilineage differentiation potential, and retain the capacity to generate PGCLCs. The transcriptomic profile of FTW-hPSCs resembles human E8 EPI cells, characterized by the low expression of naive markers and high expression of formative specific genes such as SOX12 and FZD2 (Yu et al., 2021b). Morphologically, FTW-hPSCs show a rosette structure while human FS cells exhibit a tightly arranged polarized epithelioid structure (Kinoshita et al., 2021; Yu et al., 2021b). In short, both of them possess the capability to generate PGCs and exhibit distinct characteristics compared to naive hPSCs. Notably, primed hPSCs also exhibited the potential to respond to BMP4 signaling to form PGCLCs (Jo et al., 2022; Overeem et al., 2023). Therefore, more comprehensive research is necessary to thoroughly distinguish between human primed pluripotent stem cells and fPSCs.

Mouse-extended pluripotent stem cells

Unlike mouse naive, formative, and primed pluripotent stem cells, which contribute exclusively to embryonic lineages, mouse-extended pluripotent stem cells (mEPSCs) can contribute to both embryonic and extraembryonic tissues. These cells can be derived from mESCs, blastocysts, and mouse 8C embryos (Table 1) (Yang et al., 2017a,b, 2019; Du et al., 2019; Dong et al., 2022). Compared with mESCs, mEPSCs exhibit better genomic stability and can be maintained for more than 20 generations (Du et al., 2019; Li et al., 2019a). The most widely used induction media for mEPSCs is EPSCM and LCDM, which enable induction of extended pluripotent stem cells from Liu’s lab (L-EPSCs) and extended pluripotent stem cells from Deng’s lab (D-EPSCs), respectively (Yang et al., 2017a,b). L-EPSCs can be induced by inhibiting key signaling pathways crucial for the first cell lineage commitment, such as MAPK, Src, and Wnt/Hippo/TNKS1/2 (Yang et al., 2017a, 2019). Similarly, derivation of D-EPSCs can be achieved through the suppression of MAPK, Wnt, and Parp signaling pathways (Yang et al., 2017b; Du et al., 2019; Li et al., 2019a; Dong et al., 2022). Specifically, studies have demonstrated that the inhibition of MAPK and its downstream Src pathways disrupts blastocyst formation, highlighting the essential role of MAPK signaling in the first cell fate decision (Qi et al., 2004; Yang et al., 2019). Moreover, the Hippo pathway is active in cells that will become the ICM but inactive in cells destined to become the TE and determining cell fate (Nishioka et al., 2009). Notably, the expanded potential in D-EPSCs relies on the inhibition of Parp signaling. Previous research indicated that mESCs acquired trophoblast differentiation potential after knockout of Parp1 (Hemberger et al., 2003; Yang et al., 2017b).

mEPSCs highly express mouse 2C genes such as Atrx, Esrp1, and Lin28 (Yang et al., 2017a,b; Dong et al., 2022). Furthermore, mEPSCs display less accessible chromatin regions compared to mESCs (Dong et al., 2022). Recent studies have identified transcription factors that influence the developmental potential of mEPSCs, such as YY1 and ETV5 (Zhu et al., 2020; Dong et al., 2022). YY1 contributes to the enrichment of H3K4me3 and H3K27ac in promoters of mEPSC-specific genes such as Gja1, so as to regulate its expression levels (Dong et al., 2022). Additionally, ETV5 activates the ERK signaling pathway by regulating Fgf2 expression, and depletion of Etv5 significantly impairs the efficiency of TSCs that are differentiated from mEPSCs (Zhu et al., 2020). mEPSCs exhibit extended developmental potency, particularly in extra-embryonic lineages. Previous research has shown that mEPSCs can differentiate into TSCs and XEN stem cells in vitro (Yang et al., 2017a; Du et al., 2019). Moreover, mEPSCs can contribute to both the ICM and TE lineages in chimeric embryos, and further contribute to post-implantation embryonic and extra-embryonic lineages, including placenta and yolk sac (Yang et al., 2017a,b; Du et al., 2019). Moreover, gene-targeted mEPSCs facilitate the rapid production of mouse models through tetraploid complementation due to their high developmental potential (Li et al., 2019a).

While these observations indicate that mEPSCs closely resemble totipotent stem cells from the embryo, significant differences are evident. For example, mEPSCs typically maintain the expression of many pluripotent-specific genes such as Oct4, Rex1, Nanog and Sox2 (Yang et al., 2017a,b; Du et al., 2019). The transcriptome of mEPSCs displays a closer resemblance to cells of the late EPI stage (E4.5/5.5) compared with mouse totipotent stem cells (Posfai et al., 2021b). Moreover, there is ongoing debate regarding the ability of EPSCs to directly generate TSCs in vitro, as well as their capacity to contribute to the TE lineage in the blastoid (Posfai et al., 2021b). Immunofluorescence staining and transcriptome analysis of D-EPSCs-blastoids revealed a small group of specific PrE-related cells predominantly constituting the ‘TE-like structures’, suggesting that such D-EPSCs-blastoids do not have a bona fide TE lineage (Liu et al., 2023a).

Human-extended pluripotent stem cells

Human-extended pluripotent stem cells (hEPSCs) have gained considerable attention for their remarkable potential to differentiate into both embryonic and extraembryonic lineages. hEPSCs can be induced from primed hPSCs or from human 4C embryos, and can maintain a normal karyotype for over 20 generations (Yang et al., 2017b; Gao et al., 2019; Tan et al., 2021; Zheng et al., 2021; Liu et al., 2021a; Chen et al., 2023). Current culture conditions for induction of hEPSCs have been primarily derived from studies on mouse EPSCs, including the use of LCDM from D-EPSCs and EPSCM from L-EPSCs (Tables 1 and 2) (Yang et al., 2017b; Gao et al., 2019; Zheng et al., 2021; Liu et al., 2021a; Chen et al., 2023). The modulation of signaling pathways involved in ICM/TE cell mass segregation may serve as a crucial factor for facilitating EPSC derivation in both human and mouse. Notably, inhibition of SRC, PARP, GSK3, and MAPK signaling pathways was found to be critical for halting blastomere differentiation, thereby facilitating the transition of PSCs into an EPSC state (Yang et al., 2017b; Gao et al., 2019; Tan et al., 2021; Zheng et al., 2021; Liu et al., 2021a; Chen et al., 2023). These methods underscore cross-species similarities among mammals and may serve as an effective approach for EPSC derivation in other species (Gao et al., 2019). Additionally, alternative methods to generate hEPSCs without feeder cells have emerged recently (Zheng et al., 2021; Liu et al., 2021a). Researchers have explored the use of catalase and vitamin C to promote hEPSCs survival and proliferation in feeder-free conditions (Liu et al., 2021a). Moreover, researchers found the use of glycolysis inhibitors could stimulate the expression of early-stage genes in these feeder-free cells (Zheng et al., 2021; Liu et al., 2021a).

HEPSCs exhibit upregulation of certain human ZGA-related genes such as ZSACN4 (Gao et al., 2019; Zheng et al., 2021; Chen et al., 2023). However, hEPSCs maintain part of transcriptome features of naive hESCs, resembling those of human E5–E7 EPI cells (Table 2) (Liu et al., 2021a; Chen et al., 2023). Most hEPSCs maintain the gene expression profile of pluripotent stem cells, characterized by the upregulation of core pluripotency genes, such as OCT4, NANOG, and SOX2, and the downregulation of lineage differentiation genes including GATA4, TEAD1, LAMA5, etc. (Yang et al., 2017b; Gao et al., 2019; Tan et al., 2021; Zheng et al., 2021; Liu et al., 2021a; Chen et al., 2023). Additionally, hEPSCs display global DNA hypermethylation (Yang et al., 2017a; Gao et al., 2019; Malik et al., 2022; Mazid et al., 2022), and show active H3K4me3 marks in the promoter regions of naive pluripotency factors (Gao et al., 2019; Chen et al., 2023).

HEPSCs share bidirectional developmental potential in blastoid assays, and the generation of interspecies chimera (Table 2) (Yang et al., 2017b; Gao et al., 2019; Sozen et al., 2021; Tan et al., 2021; Zheng et al., 2021; Liu et al., 2021a; Chen et al., 2023). Researchers have discovered the integration of hEPSC derivatives into extraembryonic tissues (e.g. placenta and yolk sac) of mouse E10.5 day conceptuses (Yang et al., 2017b; Zheng et al., 2021; Liu et al., 2021a). Notably, hEPSCs exhibit interspecies chimeric ability not only in mouse but also in pre- and post-implantation monkey embryos (Yang et al., 2017b; Tan et al., 2021; Zheng et al., 2021; Liu et al., 2021a). Compared with their contribution to mouse chimeric embryos, hEPSCs contribute less to the TE lineage of monkey chimeric embryos, exhibiting evolutionary divergence (Tan et al., 2021). Moreover, hEPSCs were found to have the potential to produce PGCLCs in vitro.

Mouse totipotent stem cells

Mouse 2CLCs

In 2012, Macfarlan et al. (2012) reported a unique cell population called 2CLCs within mESCs (a typical mouse naive pluripotent stem cell) (Fig. 3A). Subsequent studies confirmed that mouse 2CLCs share characteristics with mouse 2C embryos, including the loss of chromocenters, increased chromatin mobility, slower replication fork speed, and metabolic reprogramming (Ishiuchi et al., 2015; Hu et al., 2020; Rodriguez-Terrones et al., 2020). Moreover, mouse 2CLCs highly express the 2C-specific transcription factor DUX, which triggers the activation of the downstream retrotransposon MERVL and 2C-related genes. Additionally, mouse 2C-specific genes including Zscan4, Zfp352, Tcstv1/3, and Dub1 are upregulated in 2CLCs compare with mESCs.

Mouse 2CLCs is a metastable cell population identified in mESCs and mouse-induced pluripotent stem cells (miPSCs) (Table 3) (Ishiuchi et al., 2015; Eckersley-Maslin et al., 2016; De Iaco et al., 2017; Hendrickson et al., 2017; Hu et al., 2020; Grow et al., 2021; Iturbide et al., 2021; Nakatani et al., 2022; Zuo et al., 2022). The spontaneous emergence of mouse 2CLCs in mESCs is reported to typically be ∼0.5% but emergence rates rise to 10% through the application of alternative induction methods (Macfarlan et al., 2012; Dan et al., 2013; Eckersley-Maslin et al., 2016; De Iaco et al., 2017; Hendrickson et al., 2017; Hu et al., 2020; Sun et al., 2022; Zhao et al., 2022). Also, it has been found that the regulation of multiple epigenetic and chromatin factors is critical to facilitate the transition from pluripotency to totipotency in mESCs. Specifically, the histone deacetylase (HDAC) inhibitor trichostatin A (TSA) enhances the generation of mouse 2CLCs, emphasizing the importance of chromatin relaxation. Moreover, the inhibition of several epigenetic regulators (e.g. CAF-1, CTCF, and EP400–TIP60 complexes, etc.) can impact the chromatin state of mESCs to promote their conversion to mouse 2CLCs (Macfarlan et al., 2012; Ishiuchi et al., 2015; Choi et al., 2017; Rodriguez-Terrones et al., 2018; Yang et al., 2020; Olbrich et al., 2021; Sun et al., 2022). Additionally, functional experiments have identified other critical factors in 2CLCs, such as miR-34a (Choi et al., 2017), LIN28A (Sun et al., 2022), TRIM28 (De Iaco et al., 2017), and NELFA (Hu et al., 2020), that are important to this process. Recent studies provide evidence that regulating RA signaling and the p53 signaling pathway, as well as inhibition of glycolysis, can induce a mouse 2C-like state (Hu et al., 2020; Grow et al., 2021; Iturbide et al., 2021; Wang et al., 2021b). Among various induction methods, direct overexpression of Dux achieves a 70% success rate for 2CLC induction in mESCs, however, their 2CLC state is not sustainable upon subsequent passages (Hendrickson et al., 2017). The regulation of Dux transcription reactivates endogenous MERVL elements and promotes the expression of mouse 2C-specific genes such as Zscan4 (Choi et al., 2017; De Iaco et al., 2017; Hendrickson et al., 2017; Eckersley-Maslin et al., 2019; Hu et al., 2020; Yang et al., 2020; Grow et al., 2021; Nakatani et al., 2022; Sun et al., 2022; Zhao et al., 2022; Zuo et al., 2022). The MERVL 5′ LTR can function as an alternative promoter to drive transcription of a subset of 2C-genes (e.g. Zscan4, P4ha2, and Zfp352) and maintain a mouse 2C-like state (Macfarlan et al., 2012; Choi et al., 2017; Yang et al., 2020). Evidence also suggests that Zscan4 could be essential for genomic stability and telomere elongation in 2CLCs (Dan et al., 2013; Eckersley-Maslin et al., 2016; Rodriguez-Terrones et al., 2018; Grow et al., 2021). Notably, the expression of MERVL and Zscan4 is normally restricted to mouse 2C embryos, making the MERVL-Zscan4 dual reporter system the most commonly used tool for identifying mouse 2CLCs (Macfarlan et al., 2012; Ishiuchi et al., 2015; Eckersley-Maslin et al., 2016, 2019; Hendrickson et al., 2017; Hu et al., 2020; Yang et al., 2020; Grow et al., 2021; Iturbide et al., 2021; Nakatani et al., 2022; Sun et al., 2022; Zhao et al., 2022; Zuo et al., 2022).

Similar to mouse 2C embryos, mouse 2CLCs highly express totipotency-related genes and the MERVL retrotransposon, whereas the expression levels for pluripotency genes (e.g. Oct4, Sox2, and Nanog) are downregulated (Macfarlan et al., 2012; Dan et al., 2013; Ishiuchi et al., 2015; Choi et al., 2017; Hendrickson et al., 2017; Eckersley-Maslin et al., 2019; Hu et al., 2020; Yang et al., 2020; Grow et al., 2021; Iturbide et al., 2021; Nakatani et al., 2022; Zuo et al., 2022). To recapitulate the mouse 2C-like state, the chromatin of mouse pluripotent stem cells needs to be remodeled to a transcriptionally permissive state. This process involves chromatin decondensation, increased histone mobility, and enhanced chromatin accessibility for the key genes and regulatory elements (Macfarlan et al., 2012; Boskovic et al., 2014; Akiyama et al., 2015; Ishiuchi et al., 2015; Eckersley-Maslin et al., 2016; Grow et al., 2021; Nakatani et al., 2022; Zhao et al., 2022). Furthermore, mouse 2CLCs undergo global DNA demethylation and increased active histone modifications at specific genomic loci to promote totipotent-related gene transcription (Macfarlan et al., 2012; Dan et al., 2013, 2017; Eckersley-Maslin et al., 2016; Grow et al., 2021; Nakatani et al., 2022; Zhao et al., 2022). However, unlike mouse 2C, specific retrotransposons in mouse 2CLCs, such as IAP, LINEs, and SINEs, remain less active (Ishiuchi et al., 2015; Choi et al., 2017; Yang et al., 2020; Nakatani et al., 2022).

Mouse 2CLCs exhibit bidirectional developmental potential, encompassing both embryonic and extraembryonic lineages. Notably, 2CLCs significantly retain their efficiency for somatic-cell nuclear transfer (SCNT) reprogramming when compared to mESCs, suggesting their broader developmental potential (Ishiuchi et al., 2015; Nakatani et al., 2022). This bidirectional developmental capability of 2CLC-derived cells has been further substantiated through studies using teratoma and embryoid body (EB) formation assays (Choi et al., 2017; Grow et al., 2021). Additionally, fluorescence-labeled 2CLCs contribute to both the ICM and TE lineages in chimeric blastocysts (Table 3) (Macfarlan et al., 2012; Ishiuchi et al., 2015; Choi et al., 2017; Hu et al., 2020; Yang et al., 2020; Nakatani et al., 2022; Zuo et al., 2022). Following the implantation of chimeric blastocysts into the mouse uterus, 2CLCs contribute to the formation of the three germ layers and extraembryonic lineage, including the yolk sac and placenta (Macfarlan et al., 2012; Choi et al., 2017; Yang et al., 2020; Grow et al., 2021; Zuo et al., 2022).

Although mouse 2CLCs display certain totipotency characteristics similar to mouse 2C embryos, they have noteworthy differences. For example, mouse 2CLCs are induced by various methods constitute a low proportion and represent a transient state that is unstable across cell passages (Yang et al., 2020; Zhu et al., 2021). Most mouse 2CLCs retain histone modifications resembling mESCs, a pattern distinct from that observed in mouse 2C embryos (Zhang et al., 2021; Yang et al., 2022). In addition, compared to mouse 2C embryos, mouse 2CLCs lack the expression of many minor ZGA genes and maternal genes (Yang et al., 2022).

Other mouse totipotent stem cells

Recent studies have identified additional mouse totipotent stem cell models, that exhibit greater developmental potential and closer resemblance to mouse 2C embryos than 2CLCs. These include totipotent blastomere-like cells (TBLCs), totipotent-like stem cells (TLSCs), totipotent potential stem (TPS) cells, and chemically induced totipotent stem cells (ciTotiSCs) (Table 3) (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). Interestingly, mouse TPS cells and TLSCs can be derived directly from mouse embryos at the 2C, 4C, and 8C stages (Xu et al., 2022; Yang et al., 2022). The proportion of these mouse totipotent stem cells derived from their sources is generally higher than that of mouse 2CLCs. In contrast to 2CLCs, these totipotent stem cells exhibit more robust stability and can be maintained in vitro for multiple cell passages (Supplementary Fig. S1) (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023).

Mouse TBLCs induction primarily relies on the presence of spliceosome inhibitors. Spliceosome repression results in extensive inhibition of splicing for pluripotent genes with more and longer introns, while totipotent genes with fewer and shorter introns undergo effective splicing and transcriptional activation (Braunschweig et al., 2014; Shen et al., 2021). Another type of mouse experimental totipotent stem cell, known as ciTotiSCs, was induced by RA and WNT signaling activation (Iturbide et al., 2021; Hu et al., 2023). Activation of the WNT signaling pathway facilitates the proliferation and self-renewal of totipotent stem cells (Xu et al., 2022; Hu et al., 2023). Totipotency can also be achieved by inhibiting DOT1L, KDM5B, and GLP, which leads to chromocenter dissolution, the formation of broad H3K4me3 domains, and the remodeling of H3K9me2/3 domains at the loci of totipotent genes. This strategy was successfully applied for the induction of mouse TLSCs (Yang et al., 2022). In addition, mouse TPS cells were induced by the combination of chromatin state remodeling through inhibiting DOT1L and HDAC as well as RA signaling pathway activation (Fig. 3B) (Xu et al., 2022).

Mouse TBLCs, ciTotiSCs, TPS cells, and TLSCs remain stable through passages and highly express 2C-specific genes, such as Zscan4s, and Zfp352, as well as transposable elements MERVL (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). Moreover, the expression levels of pluripotency genes including Oct4, Sox2, and Nanog are downregulated (Shen et al., 2021; Xu et al., 2022; Hu et al., 2023). Furthermore, these cells display comparable epigenetic features, such as global DNA hypomethylation, open chromatin structure, and active histone profiles, resembling those of mouse 2C embryos (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). Mouse TLSCs and ciTotiSCs exhibit metabolic gene expression patterns similar to mouse 2C embryos, characterized by reduced oxygen and glucose consumption, indicative of a more reductive metabolic state (Brinster and Troike, 1979; Leese, 1995; Houghton et al., 1996; Yang et al., 2022; Hu et al., 2023). Regarding developmental potential, aside from mouse TBLCs, three other types of experimental mouse totipotent stem cells have been demonstrated to have the capacity to induce TSCs and induce the expression of specific trophoblast markers (Fig. 3B) (Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). The three totipotent stem cell types are proficient in generating blastoids and teratomas featuring both embryonic and extraembryonic lineages. Upon injection into mouse 8C embryos, all these four types of experimental totipotent stem cells actively participate in chimeric embryo development, contributing to the formation of the ICM and TE during the blastocyst stage (Fig. 3B) (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). Furthermore, they play integral roles in shaping embryonic and extraembryonic ectoderm lineages within the E7.5 mouse embryo in vivo. Remarkably, these cells also contribute significantly to placental and yolk sac development at E13.5 days of implantation (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). These compelling lines of evidence underscore the bidirectional developmental potential in each type of experimental totipotent mouse stem cell model to give rise to both embryonic and extraembryonic lineages (Fig. 3B).

Overall, these above models display partial totipotent characteristics regarding bidirectional developmental potential, transcriptional regulation, and chromatin accessibility profiles of totipotency. Nevertheless, disparities exist among various mouse stem cell models. For example, in terms of transcriptomic features, mouse ciTotiSCs, TPS cells, TLSCs, and TBLCs are reported to be similar to mouse 2C/4C blastomeres (Shen et al., 2021; Xu et al., 2022; Yang et al., 2022; Hu et al., 2023). However, among these stem cells displaying higher totipotency compared with mouse 2CLCs, ciTotiSCs were more closely related to the mouse early 2C embryo, while TLSCs and a subset of TPS cells resembled the mouse late 2C embryo state (Fig. 3C). Moreover, mouse TBLCs are more closely aligned with an intermediate state between pluripotency and totipotency compared with other mouse totipotent stem cells (Hu et al., 2023). In terms of genomic DNA methylation profiles, TPS cells, ciTotiSCs, TBLCs were found to be hypomethylated, similar to mouse 2C embryos. In addition to transcriptome and methylome features, ciTotiSCs also exhibit metabolic features similar to mouse 2C embryos, and prefer a more reductive state of one-carbon metabolism, while pluripotent mESCs and blastocysts show higher purine metabolism and TCA cycle metabolites, indicating greater oxidative activity. However, there is currently a lack of systematic comparisons of mouse totipotent stem cell models. Future studies should comprehensively compare novel and previous mouse totipotent stem cells across multiple omics dimensions to determine which derivation method yields cells most similar to natural mouse 2C embryos, with better genetic stability and developmental potential.

Human totipotent stem cells

Human 8CLCs

The initiation of ZGA in humans takes place during the 8C embryo stage, which is in contrast to ZGA initiation in mice that is observed at late 2C stage. And, while mouse totipotent stem cells were reported almost a decade ago, it is only recently that human 8CLCs have been characterized and reported (Fig. 4A) (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). These 8CLCs mirror the cellular characteristics of human 8C embryos, exhibiting high expression levels of ZGA genes and global DNA hypomethylation.

Several recent reports have described the derivation of human 8CLCs from naive hPSCs or primed hPSCs (Fig. 4B and Table 4) (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). In an observation reminiscent of the spontaneous derivation of 2CLCs from mESCs, human 8CLCs are detectable in naive hPSCs, with a low proportion capable of undergoing mutual transition between pluripotent and totipotent states (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022; Hu et al., 2023). The discovery of 8CLCs in human naive stem cells indicates that the spontaneous transition between pluripotent and totipotent-like states is not exclusive to mice but may represent a conserved feature across mammalian species. In human 8C embryos, both LEUTX and TPRX1 are highly expressed, making them suitable reporters for identifying 8CLCs (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). Based on the successful induction systems employed in mouse totipotent stem cells, various strategies have been adapted to enhance the proportion of human 8CLCs. These strategies include synergistic applications including spliceosome inhibition, genetic interference, epigenetic regulation, and modeling of signaling pathways (Fig. 4B) (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). For instance, similar to its application in mouse totipotent stem cells, the spliceosome inhibitor pladienolide B (PlaB) significantly increased ZGA-like transcription of naive hPSCs (Shen et al., 2021; Taubenschmid-Stowers et al., 2022). In another approach, genetic interference strategies such as through the overexpression of human DUX4, analogous to mouse Dux, markedly enhanced the expression of human ZGA-specific genes (LEUTX and ZSCAN4) in naive hPSCs and expanded the proportion of 8CLCs (Hendrickson et al., 2017; Yu et al., 2022). Additionally, the application of the H3K27 methyltransferase EZH2 inhibitor DZNep and the class I HDAC inhibitor TSA has been demonstrated to facilitate human 8C gene expression and promote the transition from pluripotency to totipotency, potentially through epigenetic remodeling (Mazid et al., 2022). Transcriptomic analyses between human 8C embryos and ICM highlight the crucial roles of MEK and WNT signaling pathways, and their inhibition enhances stem cell self-renewal and promotes human 8CLCs expansion in culture (Fig. 4B) (Yu et al., 2022).

Compared to naive and primed hPSCs, human 8CLCs display distinct features and share certain molecular characteristics with human 8C embryos (Table 4). For example, the expression of transcription factors and retrotransposon elements associated with ZGA are significantly upregulated in human 8CLCs, including TPRX1, DUX4, ZSCAN4, histone variant H3.Y, and MLT2A1 (Supplementary Fig. S2) (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). TPRX1 is a critical transcriptional regulator in human ZGA and is prominently expressed in human 8C embryos. By binding to the PRD-like TF-binding motif within the promoters, as well as enhancers of ZGA genes, TPRX1 activates a subset of ZGA genes including ZSCAN4 and DUX4 (Mazid et al., 2022; Zou et al., 2022). MLT2A1 and MLT2A2 were identified as potential targets for ZGA transcription factor DUX4 to increase the expression of ZGA transposable elements HERVL (Goke et al., 2015; Hendrickson et al., 2017; Liu et al., 2019). These genes hold considerable potential for investigating transcriptional events involved in ZGA in vitro and may serve as valuable markers for identifying human 8CLCs. In contrast, lineage differentiation genes such as NANOG, and SOX2, as well as markers associated with naive pluripotency including SUSD2 and ALPP, and the primed pluripotency marker CD24, are expressed at low levels in human 8CLCs (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). Interestingly, some naive pluripotency genes such as KLF17 and DPPA3 have been found to be highly expressed in human 8CLCs and the expression of DPPA3 is necessary for the conversion of naive hPSCs to 8CLCs (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). Notably, human DPPA3 exhibits only 35% identity and 53% similarity with its mouse homologous protein (Payer et al., 2003). DPPA3 plays a crucial role in inducing DNA demethylation during the conversion of hPSCs into 8CLCs. Genetic knockout of DPPA3 results in a notable increase in the methylation levels of totipotency loci, and this phenotype is accompanied by a decrease in totipotency gene expression. Interestingly, the inhibition of DPPA3 in mouse stem cells does not impair their ability to transform into mouse 2CLCs (Mazid et al., 2022). These disparities highlight a human-specific mechanism in embryogenesis and emphasize putative differences between mice and humans. In terms of epigenetic features, DNA methylation in human 8CLCs was comparable to that in human 8C embryos, showing reduced methylation levels at pluripotency loci and transposable elements (Mazid et al., 2022). Furthermore, human 8CLCs displayed increased chromatin accessibility near transposable elements and ZGA markers (Supplementary Fig. S2) (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022). In summary, the molecular and epigenetic characteristics of human 8CLCs exhibit similarities to those of human 8C embryos.

Human 8CLCs contribute to embryonic and extraembryonic lineages in vivo (by forming teratomas and interspecies chimera) and in vitro (as shown in blastoid formation assays) (Fig. 4B and Table 4) (Mazid et al., 2022; Xu et al., 2022; Yu et al., 2022). Traditionally, blastoids are generated by co-culturing naive hPSCs with TSCs (Rivron et al., 2018). However, human 8CLCs exhibit the ability to self-organize into blastoids and further contribute to TE and ICM lineages (Fig. 4B). The efficiency of sorted 8CLCs to form blastoids has been reported to be ∼58% (Mazid et al., 2022). During the self-assembly process in blastoid formation, human 8CLCs exhibit noticeable compaction and polarization (Yu et al., 2022). Intriguingly, similar to human blastocysts, blastoids composed of human 8CLCs possess the potential for differentiation into both hESCs and human TSCs (hTSCs) (Xu et al., 2022). Furthermore, hTSCs can be directly derived from 8CLCs in vitro and differentiated into syncytiotrophoblasts and extravillous trophoblasts (Mazid et al., 2022). When human 8CLCs were injected into mouse 8C embryos to evaluate interspecies chimera competency, cells originating from 8CLCs were evident in both ICM and TE at E3.5. Following transplantation of chimeric blastocysts into pseudo-pregnant mice, cells originating from human 8CLCs were identified in both embryonic and extraembryonic tissues, including the placenta and yolk sac (Fig. 4B) (Mazid et al., 2022; Yu et al., 2022). In teratoma formation assays evaluating the in vivo developmental potential of PSCs, sorted human 8CLCs exhibited contribution to the villous cytotrophoblast and placental endothelial within extraembryonic trophoblast lineage (Fig. 4B) (Mazid et al., 2022). These findings show that human 8CLCs exhibit notable totipotent characteristics and can contribute to the development of embryonic and extraembryonic lineages.

Given that the human 8CLCs re-establish a ZGA-like transcriptome and epigenetic program to achieve totipotency (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Xu et al., 2022; Yu et al., 2022), they represent an optimal model for investigating the mechanisms underlying human ZGA and lineage segregation in vitro. Nevertheless, research on human 8CLCs remains limited, and there is currently a lack of standardized protocols for reliably inducing and maintaining 8CLCs for investigation. Indeed, human 8CLCs that are derived using existing protocols cannot be stably passaged (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022; Yu et al., 2022), while chemically induced 8CLCs (ci8CLCs) can only be sustained for 2–3 weeks (Yu et al., 2022). There is only one comprehensive report evaluating the developmental potential of human 8CLCs to generate TSCs and sublineages, blastoids, teratomas, and interspecies chimeras (Mazid et al., 2022). Further evaluations are required to elucidate the molecular characteristics, developmental potential of these human 8CLCs, and will be helpful in clarifying the similarities and differences between human 8C stage embryos and human totipotent stem cells.

Human TBLCs

Recent studies have reported the successful induction of stable human TBLCs (hTBLCs) in vitro. By applying a similar approach to induce spliceosome repression as employed to derive mouse TBLCs, human PSCs can be converted into a unique state distinct from human 8CLCs, known as ZGA-like cells (ZLCs) (Shen et al., 2021; Li et al., 2024). Through continuous treatment with a specific medium containing spliceosome inhibitor PlaB, ZLCs further transition into hTBLCs, representing a pre-ZGA state and exhibit feature of human zygotes/2C–4C blastomeres (Fig. 4A and Table 4) (Li et al., 2024).

Interestingly, although ZLCs exhibit similar molecular characteristics to human 8C embryos, they possess distinct molecular features compared to human 8CLCs. Both ZLCs and 8CLCs co-express 253 ZGA genes, such as HIST1H2BG/BK, GADD45A/B, and GPATCH3. However, ZLCs do not express 596 classic ZGA genes, including DUXA/B, ZSCAN4/5B, LEUTX, and TPRX1, which are highly expressed in 8CLCs. In contrast, ZLCs specifically express 280 ZGA genes, such as ZNF23/34, FAM32A, HBEGF, and MED26. Moreover, unlike 8CLCs that retain some pluripotency gene expression, ZLCs exhibit a more thorough eradication of pluripotency genes such as DNMT3L, TBX3, and GPX2 (Li et al., 2024).

Despite these differences in gene expression traits, ZLCs induced by high concentrations of spliceosome repression were found to be in a transient cell state that was ultimately unstable over time in culture. By incorporating a series of chemical compounds that promote stem cell self-renewal, including minocycline, Y27632, and CHIR99021, and combining these with a low concentration of PlaB treatment, an MYCP medium was developed (Fig. 4B and Table 4). MYCP medium was found to be capable of inducing hPSCs and human-induced pluripotent stem cells (hiPSCs) to form hTBLCs that can be maintained stably over 20 passages. hTBLCs lack expression of ZGA stage genes; instead, they highly express specific genes and transposons from human zygote and 2C–4C embryos, such as ZBTB16, ZNF337, L1P5, LTR18A, and HERVK14/L40-int (Supplementary Fig. S2). Compared to primed hPSCs, hTBLCs exhibit a global decrease in H3K27me3 modifications, resembling the epigenetic erasure state around ZGA stage (Li et al., 2024).

hTBLCs possess bidirectional developmental potential both in vitro and in vivo (Fig. 4B). During spontaneous differentiation without spliceosome repression, hTBLCs undergo a process that mirrors pre-implantation embryo development, initially transitioning from a pre-ZGA state to a ZGA-like state, and subsequently differentiating into embryonic and extra-embryonic cell lineages resembling EPI, PrE, and TE cells in vitro (Fig. 4B). Additionally, hTBLCs were reported to have the capability for self-assembly into blastoids in common human embryo culture medium, without external factors (Fig. 4B). The cellular compositions of EPI, PrE, and TE in these blastoids were found to more closely approximate human blastocysts than to 8CLCs and naive stem cells. hTBLCs contribute to the three germ layers, as shown in EB differentiation and teratoma assays. Furthermore, hTBLCs exhibit a comparable ability to contribute to both embryonic and extra-embryonic tissues in chimeric mice (Fig. 4B) (Li et al., 2024).

Naive pluripotent stem cells in other species

Porcine naive pluripotent stem cells

Pigs are ideal experimental animals for human medical research owing to their genetic and physiological similarities to humans compared to mice (Zettler et al., 2020). The establishment of naive porcine PSCs has been valuable for evaluating stem cell-based therapies, production of transgenic pigs, as well as for xenotransplantation studies. Research on naive porcine PSCs enhances our understanding of pluripotency and cellular reprogramming.

Current induction methods for human naive PSCs using NHSM (Ying et al., 2008; Gafni et al., 2013) and for mouse ESCs using 2i/LIF (Ying et al., 2008; Yang et al., 2010) have also been found to apply effectively to porcine naive pluripotent stem cells, which suggests mammalian conservation in the maintenance of naive pluripotency across these species (Rodriguez et al., 2012; Choi et al., 2016; Yuan et al., 2019). Zhang et al. (2015) utilized human and mouse naive pluripotent stem cell induction methods to induce porcine naive pluripotent stem cells, termed hpiPSCs and mpiPSCs, respectively. From the perspective of signaling pathway induction, inhibiting the WNT, MEK, GSK3β, MAPK, BMP, and TGF-β/Activin/NODAL pathways facilitates porcine naive pluripotent stem cell induction (Rodriguez et al., 2012; Choi et al., 2016; Yuan et al., 2019; Chen et al., 2020). Regarding transcription factors, similar to the induction of human and mouse naive pluripotent stem cells, the Yamanaka factors OCT4, SOX2, KLF4, and c-MYC also induced pluripotency in the porcine naive ESCs derivation (Liu et al., 2012; Park et al., 2018; Zhang et al., 2019). However, porcine naive pluripotent stem cell induction may additionally require supplementation with bFGF (Yuan et al., 2019), a growth factor additive that is not essential for mouse and human naive pluripotent stem cells. In fact, in naive human ESCs, autocrine FGF2 signaling induces significant heterogeneity in both X chromosome status and pluripotency. By employing specific inhibitors to block FGF signaling, a more homogeneous population of naive hESCs can be achieved (An et al., 2020). Moreover, IRF-1 is expressed in the ICM of early porcine blastocysts and its overexpression in piPSCs enhances naive pluripotency via promoting the JAK-STAT signaling pathway (Shi et al., 2020). However, certain transcription factors negatively regulate porcine naive pluripotency, as overexpression of OTX2 reduces the self-renewal capacity of piPSCs and the expression of pluripotency genes NANOG, OCT4, and ESRRB (Wang et al., 2016). OTX2 is highly expressed in human primed pluripotent stem cells, whereas it is downregulated in human naive pluripotent stem cells (Bi et al., 2022). Overexpression of three transcription factors, including OTX2, can effectively induce the differentiation of hiPSCs into retinal pigment epithelium (Dewell et al., 2021). Besides regulating signaling pathways and transcription factors, HDAC inhibitors, which alter chromatin structure by removing acetyl groups from histones, also promote the derivation and long-term feeder-free culture of porcine iPSCs. This is consistent with the use of HDAC inhibitors in human and mouse pluripotent stem cell culture to enhance mouse somatic cell reprogramming efficiency (Huangfu et al., 2008) and promote the transition of hPSCs to a more naive state (Guo et al., 2017).

Porcine naive pluripotent stem cells display a compact, dome-like morphology reminiscent of human naive PSCs and mESCs, and were found to maintain their undifferentiated state over extended passages. In particular, porcine naive-like ESCs (nESCs) can be obtained through reprogramming factor-assisted strategies to maintain naive state for at least 130 passages in vitro (Zhang et al., 2019). In addition, porcine naive pluripotent stem cells express high levels of naive gene DPPA3, REX1, ESRRB, OCT4, and SSEA-1 (Rodriguez et al., 2012; Fujishiro et al., 2013; Hall and Hyttel, 2014; Hou et al., 2016; Zhang et al., 2019). Porcine naive pluripotent stem cells form embryoid bodies and teratomas and differentiate into the three germ layers both in vitro and in vivo (Rodriguez et al., 2012; Fujishiro et al., 2013; Park et al., 2018; Yuan et al., 2019; Zhang et al., 2019). Under defined conditions, they can also be induced to undergo directional differentiation into neural and kidney precursors (Zhang et al., 2019). Recently, scientists generated a humanized mesonephros in nephric-defective pig embryos by overexpressing MYCN/BCL2 in human iPSCs with 4CL medium (Wang et al., 2023a), advancing the development of human regenerative medicine. Establishing and studying porcine naive pluripotent stem cells advance regenerative medicine, offer a valuable model for human disease and development, and enable potential xenotransplantation applications.

Cynomolgus naive pluripotent stem cells

As a non-human primate, the cynomolgus monkey offers advantages over rodent models due to its evolutionarily relatedness to humans, in both genetic and physiological traits (Chan, 2013; Chen et al., 2015; Honda et al., 2017; Kang et al., 2018). Although stable mESCs have been established in mouse, it remains unknown whether naive pluripotent stem cells can be achieved in non-human primates. Initially, when primed-state monkey iPSCs were transferred to the traditional induction medium used for mouse naive pluripotent stem cells, they failed to transition to a naive state. Instead, these monkey iPSCs rapidly differentiated and lost their pluripotency (Fang et al., 2014), and these results suggested that protocols for rodent naive stem cell induction were insufficient to induce naive states in monkey iPSCs.

Subsequent research found that monkey PSCs can be induced into naive-like PSCs using the same induction conditions as the NHSM medium for human naive PSCs. This induction process involves the concerted inhibition of MEK/GSK3/P38/JNK/ROCK/PKC pathways (Fang et al., 2014; Chen et al., 2015; Fu et al., 2020; Cao et al., 2023). Subsequently, a series of human naive induction medias, including RSeT, 5iLAF, PXGL, and 4CL, were systematically evaluated for their derivation efficiency on monkey naive PSCs (Cao et al., 2023). The findings indicated that 4CL and RSeT could stably convert monkey-primed PSCs to a naive state and maintain them for over 20 passages. In contrast, stem cell culture with PXGL and 5iLAF media gradually induced loss of pluripotency upon passaging. Comprehensive molecular analyses revealed that monkey naive PSCs cultured in 4CL medium exhibited more balanced genome-wide DNA demethylation, maintained genomic stability, and exhibited higher expression levels of naive pluripotency genes (Cao et al., 2023).

Cynomolgus naive PSCs exhibit dome-shaped morphology, are capable of forming teratomas, possess the ability to differentiate into three germ layers, and exhibit limited chimeric contribution (Chen et al., 2015; Honda et al., 2017; Kang et al., 2018). Naive factors such as KLF4 and ZNF534 are significantly upregulated in naive monkey PSCs compared to the primed cells, while lineage-determining factors such as OTX2, FOXA2, DUSP6, CER1, and GATA6 are significantly downregulated (Chen et al., 2015; Honda et al., 2017; Kang et al., 2018). Mouse naive genes Tfcp21l1, Essrb, and Tbx3 remain at low levels of expression in both naive and primed pluripotent stem cells in monkeys, suggesting that these genes have different functions in monkey compared to mouse (Chen et al., 2015).

In terms of developmental potential, the majority of cynomolgus naive PSCs induced by these culture media exhibit limited chimeric contribution. For example, the chimeric ratio of cynomolgus naive PSC was found to be extremely low after being injected into monkey morula embryos (Chen et al., 2015; Honda et al., 2017). When cynomolgus naive PSCs were injected into mouse 8C embryos, almost all chimeric embryos degenerated or underwent developmental arrest at the implantation stage (Honda et al., 2017). Further investigation revealed that the limited chimeric potential observed in monkey naive pluripotent stem cells was attributable to apoptosis, G1 phase arrest, and premature differentiation (Aksoy et al., 2021). Recently, Li et al. (2023) induced naive cynomolgus monkey ESC (cyESC) based on the human-naive induction system and found these could differentiate into extraembryonic lineages in vitro and establish cynomolgus monkey blastoids. Notably, the blastoids established by naive cyESC could simulate gastrula formation in vitro and, when transplanted into a surrogate cynomolgus monkey recipient, it could form a gestational sac and induce an early pregnancy response (Li et al., 2023). Moreover, recent research has indicated that culture with 4CL medium can induce high-potential naive cyESCs, which could contribute significantly to cells within chimeric blastocysts and to tissues of live-born monkey chimeras (Cao et al., 2023). The derivation of high-potential monkey stem cells holds significant promise as an experimental platform through which to investigate the mechanisms of non-human primate development, and as an experimental model to better understand the pathogenesis of human disease.

Application of the stem cell models

Currently, we have a deeper understanding of the molecular mechanisms and cellular behaviors of pluripotent stem cells with diverse developmental potentials across species. High-potency stem cells possess the remarkable ability to self-organize into blastoids and post-implantation embryo models, advancing our collective understanding of the mechanisms for pluripotency and lineage restriction. Indeed, high-potency stem cell models are recognized to be valuable for investigating XCI and hold significant promise for various clinical applications. In the following sections, we will discuss the latest advancements and clinical implications of the models derived from pluripotent stem cells.

Blastoids

Developing a robust in vitro model can greatly facilitate research in embryonic development. Blastoids, in particular, have gained significant attention as a powerful tool for investigating early human embryonic development, such as the first lineage differentiation and embryo implantation. Utilizing blastoids or gastruloids to screen for factors that influence reproductive toxicity represents an attractive, scalable platform for discovery that is a viable alternative to more traditional approaches, such as animal experiments (Mantziou et al., 2021). Additionally, blastoids are pivotal in uncovering the root causes of clinical challenges such as recurrent implantation failure, congenital anomalies, and embryonic demise (Niethammer et al., 2022; Terhune et al., 2022).

There are mainly naive hPSC-blastoid, hEPSC-blastoid, 8CLC-blastoid, and hTBLC-blastoid models in humans (Fan et al., 2021; Sozen et al., 2021; Yanagida et al., 2021; Yu et al., 2021a, 2022; Liu et al., 2021b; Kagawa et al., 2022; Mazid et al., 2022; Li et al., 2024). Upon inhibition of the Hippo, TGF-β, and ERK signaling pathways, human naive pluripotent stem cells acquire the potential to self-organize into blastoids, contributing to both TE and ICM lineages (Yanagida et al., 2021; Kagawa et al., 2022). hEPSCs can be used in combination with hEPSC-derived TE-like cells to construct blastoids (Fan et al., 2021). Furthermore, human 8CLCs and hTBLCs, characterized by their experimental totipotency, exhibit efficient self-organization into blastoids (Mazid et al., 2022; Yu et al., 2022; Li et al., 2024). Human 8CLCs demonstrated a higher efficiency in blastoid induction when treated with the same inhibitors targeting ERK and TGF-β signaling pathways as those used for naive hPSCs (Yanagida et al., 2021; Yu et al., 2021a; Mazid et al., 2022). The lineage distribution within blastoids derived from human 8CLCs was found to more closely mirror that of human blastocysts compared with naived hPSCs (Mazid et al., 2022; Li et al., 2024). In contrast, naive hPSC-blastoids exhibit a higher proportion of cells of the EPI lineage and a lower proportion of cells of the TE lineage compared to blastocysts (Li et al., 2024). Conversely, blastoids derived from human 8CLCs presented a lower proportion of PrE-lineage cells (Li et al., 2024). Notably, hTBLCs can self-assemble into blastoids in embryo culture medium (G2-PLUS medium) without the need for pathway inhibition (Li et al., 2024). These blastoids exhibit a cell composition highly similar to human blastocysts, indicating that stem cells closer to a totipotent state in vivo are more likely to generate structures closely resembling natural embryos.

The early models primarily rely on aggrewell for 3D in vitro self-assembly to generate blastoids, with relatively low reported efficiencies. The use of inverted microwell in 3D culture can facilitate cell-to-cell communication and promote self-assembly (Fan et al., 2021; Sozen et al., 2021; Yanagida et al., 2021; Yu et al., 2021a; Liu et al., 2021b; Mazid et al., 2022; Guo et al., 2024). Notably, Kagawa et al. (2022) reported that the induction efficiency of blastoids within hydrogel microwells could reach 70%. Recently, Yu et al. (2023) reported blastoid efficiencies in culture that exceeded 80%, achieved through optimized eHDM and eTDM culture conditions. Transcriptomic analyses reveal blastoid clustering patterns were comparable to those of natural blastocysts (Fan et al., 2021; Sozen et al., 2021; Yanagida et al., 2021; Yu et al., 2021a, 2022; Zhang et al., 2021; Liu et al., 2021b; Kagawa et al., 2022; Mazid et al., 2022). Blastoids can induce other stem cells such as hTSCs, naive hPSCs, and primed hPSCs (Fan et al., 2021; Yu et al., 2021a, 2022; Liu et al., 2021b; Kagawa et al., 2022). Interestingly, when cultured in vitro, blastoids are capable of recapitulating post-implantation development, such as the process of blastocyst attachment, growth, and hCG secretion (Fan et al., 2021; Sozen et al., 2021; Yu et al., 2021a; Liu et al., 2021b). Kagawa et al. (2022) also established an endometrial-like organ system and co-cultured it with blastoid to simulate in vivo development following attachment to the endometrium.

Blastoids are capable of producing pre-implantation lineages that resemble ICM, TE, and peri-implantation EPI, TE, and PrE-like cells (Rivron et al., 2018; Kime et al., 2019; Sozen et al., 2019; Li et al., 2019b; Xu et al., 2022; Yang et al., 2022; Zhang et al., 2023). Moreover, in vitro culture of mouse blastoids developed beating heart-like structures and brain-like tissues (Amadei et al., 2022). After implantation into a pseudopregnant mouse uterus, mouse blastoids were found to successfully undergo decidualization (Rivron et al., 2018; Kime et al., 2019; Sozen et al., 2019; Li et al., 2019b; Xu et al., 2022; Yang et al., 2022; Zhang et al., 2023). However, the resulting decidua were smaller and more fragile than those formed by blastocysts (Yang et al., 2022; Zhang et al., 2023).

Despite their attractiveness as a model for embryo research, current blastoids still have some limitations. Recently, analyses using an integrated human single-cell RNA sequencing dataset have revealed unidentified cell signatures for blastoids (Kagawa et al., 2022; Zhao et al., 2025). TE-like cells in preimplantation blastoids share more similarities to post-implantation amniotic ectoderm cells, indicating the risk of misannotation when reliable reference is lacking (Zhao et al., 2025). Moreover, several TE markers such as GATA2 and BIN2 showed low expression levels however markers of amnion cells including ISL1 and IGFBP7 were ectopically expressed in TE-like cells, altogether highlighting significant functional differences between current blastoid models and human blastocysts. Therefore, stringent benchmarking of embryo models is required, including comparisons of structure morphology and correct cell numbers and types between models and genuine embryos, alongside functional evaluations of cell types within embryo models (Posfai et al., 2021a).

In summary, both human and mouse blastoids hold promise as in vitro models for simulating early pregnancy-related diseases and for use in pharmacological and toxicological testing. However, developing more realistic models that accurately replicate the in vivo blastocysts remains an important area of ongoing research. The next frontier in embryo model research will be placing these models in environments more closely resembling the uterus and studying how they interact with the endometrium. While no regulations currently prohibit the transfer of non-human embryo models into live animals, except in humans, such procedures may raise ethical concerns, particularly if they result in live births (Mallapaty, 2024). Therefore, future studies involving embryo models must proceed with great caution.

Post-implantation embryo-like structures

Due to the technical challenges and ethical restrictions of obtaining post-implantation samples, human post-implantation development remains largely elusive. The recent development of stem cell research has facilitated the generation of numerous stem cell-based embryonic models (Hyun et al., 2020). Various post-implantation human embryo models that are achieved through culturing stem cells with 2D micropatterns or 3D environments (Warmflash et al., 2014; Shao et al., 2017a,b; Xue et al., 2018; Manfrin et al., 2019; Simunovic et al., 2019; Zheng et al., 2019a,b; Moris et al., 2020; Chen et al., 2021; Minn et al., 2021). Notably, hESCs and hPSCs cultured in traditional 2D conditions lack defined tissue structures. However, by constraining these cells within small-diameter 2D micropatterns to regulate their spatial distribution and activating BMP4-SMAD signaling, hESCs can be induced to form post-implantation tissue-like structures (Warmflash et al., 2014; Xue et al., 2018; Manfrin et al., 2019; Minn et al., 2021). Furthermore, stem cells cultured in 2D micropattern can differentiate into ectoderm, mesoderm, endoderm, and TE in a central-to-peripheral pattern, recapitulating the spatial organizational of cell types observed in human gastrula-stage embryos (Minn et al., 2021). This cell fate determination is influenced by the cell position and the differential localization of BMP4 receptors (Etoc et al., 2016; Moris et al., 2020; Minn et al., 2021), as well as BMP4-SMAD or BMP4-WNT/NODAL cascades (Warmflash et al., 2014; Xue et al., 2018; Manfrin et al., 2019). Through the application of a 3D culture system, hESCs or hiPSCs can self-assemble into an asymmetric structure resembling amniotic sacs, referred to as a ‘post-implantation amniotic sac embryoid’ (Shao et al., 2017a,b; Simunovic et al., 2019; Zheng et al., 2019a,b; Moris et al., 2020; Chen et al., 2021). Amniotic-like cells are located on one side of the structure, while pluripotent trophoblast-like cells surround the central cavity on the other side. Further research has established a 3D suspension culture system based on hESCs. By controlling the size of cell clusters using cones to form hESC spheres before transferring them to Geltrex for induction, researchers can achieve pre-gastrulation embryonic-like tissues that mimic post-implantation and pre-gastrulation amniotic sacs with ∼50% efficiency (Chen et al., 2021). These pre-gastrulation embryonic-like tissues can subsequently differentiate into asymmetric embryonic-like, amniotic-like, and mesodermal-like tissues (Chen et al., 2021). Weatherbee et al. (2023) generated in vitro human post-implantation embryo structures by employing models derived from pluripotent stem cells representing multiple lineages. Furthermore, several research teams reported on the use of intermediate (formative-to-primed) pluripotency states stem cells, EPSC cells, and naive pluripotent stem cells to self-assemble as models that closely mimic gastrula-like embryos (Oldak et al., 2023; Pedroza et al., 2023; Liu et al., 2023b). These models effectively recapitulate post-implantation development and cellular interactions. The development of human post-implantation models has facilitated the study of key events in embryonic development, such as axis formation and gastrulation following early implantation. Yet, current post-implantation embryo models face several limitations. The model developed by Liu et al. (2023b) does not fully replicate human embryo complexity due to the absence of trophoblast cells and a chorionic cavity. Additionally, there is variability in the efficiency of model formation among different hEPSC lines, and the reliance of hEPSCs on mouse embryonic fibroblasts (MEFs) may complicate the peri-gastruloid formation process. Additionally, the models for peri-gastruloid development are limited to 11–13 days, beyond which growth stalls. Nevertheless, despite these challenges, these models have been informative for understanding post-implantation development, particularly where post-implantation human embryos are inaccessible from ethical perspectives.

X chromosome inactivation modeling

In contrast to the well-understood mechanisms of XCI in mouse embryos, the phenomenon of XCI in human embryos remains poorly characterized. The human pre-implantation embryo lacks paternal-specific XCI and undergoes random XCI following implantation. In the pre-implantation ICM of human embryos, all X chromosomes remain active and display biallelic expression of XIST, as well as reduced expression of X-linked genes in female embryos (Okamoto et al., 2011; Petropoulos et al., 2016). Research using hPSCs have provided valuable insight into the mechanisms for X chromosome regulation during early human embryo development. Traditional human primed pluripotency stem cells exhibit substantial heterogeneity in their X chromosome status, encompassing the XaXa phenomenon characterized by the absence of XIST expression, XaXi where one X chromosome is subject to inactivation, and XaXe (XCI erosion) (Shen et al., 2008; Mekhoubad et al., 2012; Vallot et al., 2015). Naive hPSCs exhibit the activation of both X chromosomes similar to the ICM cells of embryo (Theunissen et al., 2016; Sahakyan et al., 2017; Vallot et al., 2017). Nevertheless, they show non-random XCI during differentiation (Theunissen et al., 2016; Sahakyan et al., 2017). Subsequent research has revealed that blocking autocrine FGF signaling in naive human PSCs reduces cellular heterogeneity. This approach enables the modeling of random XCI by isolating TFCP2L1 expression (HT) cells, which exhibit bi-allelic XIST expression and undergo random XCI upon differentiation (An et al., 2020). The X chromosome state in human EPSCs is significantly influenced by the primed hPSCs from which they originated, resulting in a heterogeneous X chromosome status (Wang et al., 2023b). However, the X chromosome status in hTBLCs and human 8CLCs remains undefined in current research. Given that hTBLCs and human 8CLCs closely resemble the early embryo cells, they may represent an ideal model for investigating XCI, thereby advancing our understanding of critical processes in early human embryo development.

Potential clinical applications

hPSCs have been applied in clinical research. For instance, patient-specific naive hiPSCs derived from β-Thalassemia fibroblasts can efficiently correct the gene mutation by CRISPR/Cas9, offering an improved strategy for personalized treatment of inherited genetic diseases (Yang et al., 2016). Moreover, naive hPSCs have the capability to generate a humanized mesonephros in pigs by embryo complementation, offering a potential solution to the immune tolerance problem in organ transplantation therapy (Wang et al., 2023a). Formative hPSCs can be directly induced to PGCLCs, which are of significant clinical and research value for deciphering the molecular mechanisms underlying the development of germline cells and their defects in the field of fertility research (Chen et al., 2019a; Kinoshita et al., 2021; Yu et al., 2021b). Notably, viable pups have been successfully generated through IVF using the oocytes derived from mPGCLCs (Hikabe et al., 2016; Bhartiya et al., 2017), while hPGCLC has been reported to be successfully used for the in vitro production of human primordial germ cells and spermatogonia (Murase et al., 2024), thereby showcasing its potential for therapeutic applications for treating infertility. Furthermore, high-potential stem cells possess the capability to differentiate into both extraembryonic and embryonic lineages. For instance, hTBLCs can spontaneously differentiate into both extraembryonic and embryonic cell types including the TE, PrE, and EPI lineages without any additional treatment (Li et al., 2024). Embryo models derived from naive pluripotent stem cells have demonstrated the capacity to develop structures mimicking the heart and nervous system (Sozen et al., 2019; Li et al., 2019b; Amadei et al., 2022; Zhang et al., 2023), highlighting their potential to progressively form human organ primordia. Despite significant challenges, these high-potential stem cells may exhibit the potential to generate functional organs via interspecies chimeras or in vitro embryo models, recapitulating the natural processes of in vivo organ development. Notably, while there is potential for such applications, significant technical and ethical challenges must be addressed before they can be realized. Technically, current embryo models face several challenges, including low construction efficiency, lengthy derivation processes, and discrepancies in certain cellular states compared to human real embryos (Yu et al., 2021a; Liu et al., 2021b). In addition, hPSCs may contribute to the animal germline and nervous system during the formation of interspecies chimeras, raising serious ethical concerns (Hyun et al., 2021).

Discussion

From naive pluripotent stem cells to totipotent stem cells, the cell potency displays the transition of moving toward totipotency; understanding the underlying mechanisms facilitates the investigation of key molecular events in early embryonic development, such as ZGA and cell fate decisions. However, although certain totipotency features have been captured by human 8CLCs or mouse 2CLCs, there remain unresolved matters that necessitate exploration to attain authentic totipotency in vitro.

Owing to ethical and technical constraints, research on human embryonic development and stem cells lags behind that of mouse and is largely dependent on mouse-based culture systems. Despite some conservation of ZGA-related gene activation and epigenetic reprogramming between human and mouse totipotent stem cells, significant species differences remain, for example, the expression of genes such as TPRXs, DPRX, and ARGFX of the PRD-like gene family, which are key for human ZGA and are primate-specific (Tohonen et al., 2015; Jouhilahti et al., 2016; Zou et al., 2022). Similarly, the mouse totipotency pioneer factor Nr5a2 is weakly expressed in human early embryos (Gassler et al., 2022; Zou et al., 2022). Furthermore, the functions of signaling pathways such as WNT and FGF during the induction of totipotency stem cells differ between these species (Sato et al., 2004; Linneberg-Agerholm et al., 2019; Huyghe et al., 2020; Yu et al., 2022; Hu et al., 2023). Inhibition of the WNT pathway is conducive for the induction of human 8CLCs from naive hPSCs, while activation of the WNT pathway contributes to the induction of mouse totipotent stem cells (Yu et al., 2022; Hu et al., 2023). Research on mice also contributes to the comprehension of human totipotency. For example, recent findings in mice have identified KLF17 as a messenger that recruits RNA Pol II to regulate ZGA (Hu et al., 2024). Notably, KLF17 exhibits a sequence conservation rate of 81.5% in their zinc fingers between humans and mice (van Vliet et al., 2006) and is prominently expressed in human 8CLCs (Mazid et al., 2022; Taubenschmid-Stowers et al., 2022), suggesting its potential involvement in human totipotency. Consequently, continuing research on totipotent human embryo cells to identify the key factors and pathways of human totipotency would enhance the derivation of human totipotent stem cells.

Current approaches to derive human 8CLCs still exhibit significant differences when compared with real human 8C embryos (Mazid et al., 2022). For example, several pluripotency genes, including SALL4 and NODAL, are highly expressed in 8CLCs and not in human 8C embryos (Mazid et al., 2022; Yu et al., 2022). Additionally, methyltransferases such as DNMT3B were found to be highly expressed in human 8CLCs, which may be responsible for the different DNA methylation status in certain regions compared to human 8C embryos (Yu et al., 2022). Apart from these mentioned distinctions, potential differences in genomic histone modifications and X chromosome status between human 8CLCs and 8C embryos remain to be comprehensively evaluated. The activity of both X chromosomes in females is a key feature of human totipotent and naive stem cells, underscoring the need for a more comprehensive and detailed analysis of the transcriptional and epigenetic profile of human 8CLCs. Moreover, the epigenetic stability of imprinted genes in 8CLCs has not been characterized. Although 8CLCs partially mimic the molecular characteristics of human 8C embryos and possess both in vivo and in vitro developmental potential, their developmental ability may be impaired due to undiscovered epigenetic or regulatory abnormalities.

By the strictest definitions of in vivo canonical totipotency, a single cell is genuinely totipotent for its ability to give rise to a viable organism. In vitro experimental totipotency refers to the ability to contribute both embryonic and extraembryonic lineages. Distinguishing between in vivo canonical totipotency and in vitro experimental totipotency is crucial when describing stem cell models. Although mouse 2CLCs and human 8CLCs exhibit potential for differentiation into embryonic and extra-embryonic tissues in the embryo, they display a predilection toward different lineages, yet their capacity to give rise to complete organisms is currently limited. For example, teratomas derived from sorted human 8CLCs exhibit a greater contribution to the villous cytotrophoblasts sublineage and placental endothelial cells compared to the cytotrophoblast sublineage within the extraembryonic trophoblast lineage (Mazid et al., 2022). Moreover, the contribution of the totipotent stem cells to both embryonic and extraembryonic lineages is confined to a group of cells rather than a single cell, diverging from the canonical definition of totipotency of ‘a single cell’. Establishing a unified standard is imperative for conducting developmental potency assessment. Current totipotent stem cells lack the ability to generate an entire organism, nor do these models develop along a specific, defined temporal sequence and spatial arrangement akin to natural embryos. Furthermore, single-cell transcriptomics analysis of mouse TPS cells and human 8CLCs reveals inherent heterogeneity, with only a limited number of clusters closely resembling mouse 2C embryos or human 8C embryos, respectively (Mazid et al., 2022; Xu et al., 2022). Given the potential confounding effects of cell heterogeneity on experimental outcomes, it is critical to define the biological characteristics of the cellular populations in stem cell models. Additionally, the zygote inherits numerous maternal transcripts from the oocyte, which actively contribute to driving the process of ZGA in the embryo. Certain maternal factors, such as DPPA3, NELFA, and OBOX, are highly expressed during totipotent stem cell induction and have demonstrated significant importance in experimental studies (Hu et al., 2020; Mazid et al., 2022; Ji et al., 2023; Lai et al., 2023; Sakamoto et al., 2024). However, the expression and function of other maternal factors in pluripotent stem cells remain unexplored, and the absence of certain maternal factors may account for the limitation in the potential of in vitro totipotent stem cells.

Totipotent stem cells are an important source for the construction of blastoids and post-implantation models to investigate cell lineage segregation. The 8CLCs can self-assemble into blastoids, contributing to the ICM and TE lineages (Mazid et al., 2022), allowing the exploration of lineage segregation mechanisms without the disturbance of introducing different stem cell lines. Although the original cells (naive and primed PSCs) have undergone ZGA, the totipotent mouse 2CLCs, and human 8CLCs reactivate mouse 2C or human 8C-like transcriptional and epigenetic programs after induction, making them suitable for studying ZGA programs in vitro. In addition, totipotent stem cells hold great potential for clinical applications, including cell therapy, regenerative medicine, modeling reproductive diseases, and elucidating the underlying mechanisms of infertility. However, it is important to note that despite these promising applications, there remain significant differences between these cells and natural embryo cells upon transcriptome and epigenome. Additionally, variations in induction methods and efficiencies may lead to discrepancies in results across different models, thus limiting reproducibility. Therefore, while stem cell models offer certain advantages, they cannot yet replace natural embryos but are useful tools for reproductive research.

However, the development of totipotent stem cells is also associated with major ethical concerns. A future envisioning the emergence of fully defined totipotent stem cells that can generate an entire organism from a single cell is within reach. Therefore, ethical research needs to be cautious about whether totipotent stem cells should be considered strictly in accordance with the requirements for embryos. The International Society for Stem Cell Research (ISSCR) imposed a 14-day rule in 2006, allowing human embryos to be cultured in vitro only up to the 14th day or until the appearance of the primitive streak, whichever occurs first (Hyun et al., 2016). However, due to the distinct developmental timelines of embryos and in vitro embryo models, which are influenced by specific cell-inducing culture conditions, implementation of the 14-day rule is challenging (Hyun et al., 2020; Lovell-Badge et al., 2021). For instance, some culture systems can lead to the formation of human gastruloids within 72–96 h (Hamazaki et al., 2024), and only by terminating and examining these models can researchers determine the presence of a primitive streak structure. Moreover, researchers have proposed extending the 14-day limit of in vitro embryo culture period to 28 days, as research during these additional 2 weeks could provide valuable insights into organ development, developmental disorders, congenital abnormalities, and fertility-related issues. Simultaneously, it remains hard to preserve fetal tissue beyond the 2-week period. The 28-day limit is an artificial threshold and does not imply that embryos beyond this timeframe possess absolute moral status. In fact, it represents a balance between ethical considerations and scientific benefits (Writing Group of the ESHRE Ethics Committee, 2024). In 2021, the ISSCR relaxed the 14-day rule for embryo culture (Lovell-Badge et al., 2021). Nevertheless, specific models still necessitate individualized oversight and approval on a case-by-case basis to determine the appropriate termination point for experiments. Moreover, in vitro embryo models involving both embryonic and extraembryonic lineages, which aim to simulate complete individual development, require rigorous evaluation. Suppose an in vitro embryo model is demonstrated to undergo the same developmental stages as a natural embryo and can result in live birth in other mammalian species. In that case, it should be subject to the same ethical constraints and limitations as natural embryos (Writing Group of the ESHRE Ethics Committee, 2024). Conversely, for models that lack the capacity to form a complete embryo, relatively loose criteria are recommended. In conclusion, an ethics committee should evaluate the necessity of creating embryo models, ensuring the use of the least controversial model and the shortest developmental period required to achieve the scientific goals, in line with the subsidiarity principle (Writing Group of the ESHRE Ethics Committee, 2024).

The main challenge for researchers developing guidelines of human embryo models is determining when an embryo model should be regarded as equal to a natural embryo. This determination governs whether human embryo models are subject to the legal and ethical regulations that apply to research on natural human embryos. Current legal definitions of a human embryo vary significantly across countries, leading to international variability in research standards. For instance, in Spain, an embryo is defined based on fertilization, meaning that embryo models which are not fertilized are not classified as embryos (De Miguel Beriain et al., 2024). In contrast, in Australia, human embryo can be defined as ‘any other process that initiates organized development of a biological entity with a human nuclear genome or altered human nuclear genome that has the potential to develop up to, or beyond, the stage at which the primitive streak appears’ (Findlay et al., 2007). At present, embryo model studies are striving to closely mimic the state of natural embryos for research, yet they also need to ensure a clear distinction from natural embryos to avoid significant ethical and legal concerns. The rapid advancements in embryo model research underscore the need for new guidelines and restrictions. To address this, the ISSCR has established the Embryo Models Working Group in June 2024, tasked with evaluating the current state of the science based on models published since 2021 and revising earlier guidelines (International Society for Stem Cell Research, 2024). One strategy to alleviate ethical concerns could involve developing models that focus on the genesis of specific organs or creating embryo models that are unable to form a complete organism, such as those lacking essential genes for heart development. However, gene editing in embryo models introduces risks, such as potential off-target effects or the generation of organisms with impaired organogenesis. ISSCR guidelines prohibit the transfer of any human embryo model into a human or animal uterus (International Society for Stem Cell Research, 2023). Ethical consideration on further embryonic stem cell research should take into account both the potential biomedical benefits and the scientific quality of the research.

Conclusion

Totipotency is a complex and highly regulated process involving multiple regulatory mechanisms at both molecular and cellular levels. The development of human totipotent stem cell models holds significant potential for advancing our understanding of key events and regulatory patterns during human early embryo development. Researchers in this field must clearly distinguish between in vivo canonical totipotency and in vitro experimental totipotency, as these models exhibit distinct developmental potentials and spatiotemporal assembly capabilities.

Moreover, different human experimental totipotent stem cells display unique molecular characteristics. Human ZLCs broadly silence pluripotency genes while activating 8C stage-specific genes, which differ from those in human 8CLCs. This suggests the existence of additional regulatory networks in embryos that may govern human ZGA beyond the classical DUX/ZSCAN4/TPRX1-mediated network. While various methods can induce totipotent stem cells, they remain distinct from natural embryos, as they only replicate certain transcriptional characteristics of natural human 8C embryos. Meanwhile, 8CLCs and ZLCs lack a more comprehensive assessment, such as the description of epigenetic characteristics in ZLCs. Notably, the induction of these cells requires a prolonged induction period, yet none are capable of maintaining stable existence.

A long-term goal for totipotent stem cell research is to develop highly reproducible culture protocols that enable these cells to closely mimic natural embryos. Among these efforts, artificial intelligence-assisted high-throughput screening holds promise for the rapid prediction and identification of potential drug combinations and induction pathways. Several approaches could enhance the similarity of totipotent stem cells to natural embryo cells. Metabolically, adjusting the metabolic state of totipotent stem cells helps to simulate natural embryos and provides valuable insights into further understanding the metabolic reprogramming mechanisms of natural embryos. Additionally, the role of maternal factors in current totipotent stem cell models has not been thoroughly investigated and may contribute to their impaired developmental capacity. Physically, examining the response of totipotent stem cells to mechanical forces could elucidate the role of mechano-regulation in determining embryo cell fate. From the perspective of the microenvironment, co-culturing totipotent stem cells with other cell types or organoids may enhance their developmental potential and reveal the cellular interactions driving embryo development. Current strategies for inducing totipotent stem cells in humans and mice may also provide a foundation for converting pluripotent stem cells to totipotent stem cells in other species, potentially through mechanisms such as spliceosome inhibition. In conclusion, the major bottleneck in totipotent stem cell research is the insufficient understanding of the mechanisms underlying embryonic development, particularly regarding the mechanisms of human ZGA and totipotent regulation. Future research should integrate studies on both embryo development and totipotent stem cells, as these areas can complement and inform each other.

Although significant challenges remain, the totipotent stem cell model holds immense scientific and clinical promise for investigating developmental diseases and infertility, warranting sustained attention and dedicated research efforts. Furthermore, totipotent stem cells could potentially serve as reserve cells in the future, providing a clinical alternative for personalized medicine. It is crucial that any medical or commercial applications of totipotent stem cells are strictly regulated.

Supplementary data

Supplementary data are available at Human Reproduction Update online.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Acknowledgements

The authors would like to thank Y.L. and Y.X. for their support and advice in the preparation of this manuscript. They would also like to thank BioRender.com for helping to draw the figures.

Authors’ roles

L.H. and Y.P. contributed to the study design, the literature acquisition, and the writing of the manuscript. J.Q., L.Y., and P.Y. contributed to the design and revision of the manuscript.

Funding

National Key Research and Development Program of China (2022YFC2702200, 2023YFA1800301); National Natural Science Foundation of China (82125013, 82201838); Peking University Third Hospital Fund for Interdisciplinary Research (BYSYJC2023001); Clinical Medicine Plus X—Young Scholars Project, Peking University, Fundamental Research Funds for the Central Universities (PKU2024LCXQ005); Young Elite Scientists Sponsorship Program by China Association for Science and Technology (YESS20200398).

Conflict of interest

The authors declare that they have no conflicts of interest.

References

Author notes