Hallmarks of Totipotent and Pluripotent Stem Cell States

Abstract

Though totipotency and pluripotency are transient during early embryogenesis, they establish the foundation for the development of all mammals. Studying these in vivo has been challenging due to limited access and ethical constraints, particularly in humans. Recent progress has led to diverse culture adaptations of epiblast cells in vitro in the form of totipotent and pluripotent stem cells, which not only deepen our understanding of embryonic development but also serve as invaluable resources for animal reproduction and regenerative medicine. This review delves into the hallmarks of totipotent and pluripotent stem cells, shedding light on their key molecular and functional features.

Short summary for the e-table of contents:

In this review, Peng Du and Jun Wu present a detailed overview of the growth conditions, molecular characteristics, and functional attributes of totipotent and pluripotent stem cells generated in vitro. Additionally, they discuss the limitations, challenges, emerging concepts, and future prospects associated with stem cells generated from early embryos.

1. Introduction

Mammalian early embryonic development is characterized by a high degree of self-organization and autonomy, crucial for the formation of a complete organism. This process begins with fertilization and progresses to gastrulation, involving a sequence of cell-fate decisions and symmetry breaking events. Post-fertilization, the embryo undergoes multiple cleavage divisions, evolving into a pre-implantation blastocyst. The blastocyst comprises the epiblast, primitive endoderm, and trophectoderm, three foundational tissues of a developing conceptus¹. Following blastocyst implantation, substantial tissue reorganization and lineage development occur. Notably, epiblast cells undergo epithelialization and influenced by extraembryonic tissues, form the anterior-posterior axis. This leads to gastrulation, a process marked by complex cellular activities and the formation of the three primary germ layers: ectoderm, mesoderm, and endoderm², which are instrumental in the development of various tissues and organs.

Pre-gastrulation development encompasses the stages of embryonic totipotency and pluripotency. In a natural embryo, totipotency is a trait of the blastomeres, whereas pluripotency is characteristic of the epiblast cells. A single blastomere, present from the two-cell stage up to the four-cell stage (though this can vary among species), possesses the remarkable ability to generate all embryonic and extraembryonic tissues and develop into a complete organism if isolated. On the other hand, epiblast cells, emerging in later stages of pre-implantation development, are pluripotent. They can differentiate into almost all cell types within the organism, except for extraembryonic tissues such as the placenta. Totipotency and pluripotency can be recapitulated in vitro in dynamic states of totipotent³ and pluripotent stem cells⁴ under specific conditions. The ability of these stem cells to maintain aspects of totipotency and pluripotency outside the natural embryonic context underscores their significance as powerful tools in basic and translational research.

Despite the potential, however, notable differences exist between in vivo and in vitro conditions. In vivo, cells are tightly regulated by the embryonic microenvironment, existing in a state of continuous development and actively transitioning through various stages. In contrast, cultured stem cells are typically maintained in nutrient-rich environments designed to promote proliferation and prevent differentiation, settings that are quite different from the natural embryonic environment. Despite ongoing efforts to develop conditions that more closely mimic the in vivo embryonic niche and support more authentic totipotent and pluripotent cell states in vitro, iťs important to recognize that significant differences in transcriptional, epigenetic, and metabolic characteristics exist between embryonic cells in their natural state and those cultured in laboratory settings.

Embryonic cells are characterized by their remarkable plasticity, allowing them to transdifferentiate or dedifferentiate under certain in vitro conditions into cell types not typically observed in vivo. This plasticity, which should be distinguished from intrinsic cell potency (a range of cell types a progenitor or stem cell can develop into), reflects how a cell reacts to environmental stimuli or experimental conditions, revealing adaptability that goes beyond its inherent potency. For instance, epiblast cells in embryos and their in vitro counterparts, pluripotent stem cells (PSCs), can generate extraembryonic trophoblast cells under certain in vitro conditions. This capability, usually absent in developing embryos, indicates that external factors can markedly influence the behavior and and developmental path of a pluripotent cell. The plasticity of embryonic cells, together with their spatiotemporal characteristics, contributes to the array of totipotent and pluripotent stem cell states that are maintained in vitro under various culture conditions, which should be carefully evaluated and differentiated from stem cell potency.

This review seeks to provide a comprehensive summary and detailed comparison of the dynamic states of totipotent and pluripotent stem cells identified thus far. We will focus on outlining the molecular and functional hallmarks used to define these states (Figure 1) and discuss how they correspond to their in vivo counterparts.

Figure 1. Hallmarks of totipotent and pluripotent stem cells.

A list of key molecular and functional hallmarks used to define and/or distinguish different totipotent and pluripotent stem cell states.

2. Dynamic Pluripotent Stem Cell States

In vitro cultured PSCs exhibit various pluripotency states, each with different levels of similarity in global gene expression, chromatin states, and developmental potential compared to in vivo epiblasts. Numerous culture methods have been developed to stabilize naïve, intermediate/formative, and primed PSCs in both mice and humans (Figure 2). It should be noted that PSC states are not as clearly defined in humans as in mice, with considerable variations observed across protocols used by different laboratories. Given the limited variety of PSCs reported from species beyond rodents and primates, we will only briefly touch on these.

Figure 2. Growth conditions for mouse and human pluripotent stem cells.

Culture components, such as growth factors, cytokines, and/or chemicals, included in each growth condition for mouse and human PSCs are indicated using blue color boxes. The related signaling pathways are noted on the left.

2.1. Naïve Pluripotent Stem Cell States

Naïve pluripotency is defined by an unfettered developmental capacity to differentiate into all somatic and germline lineages. Mouse embryonic stem cells (mESCs) are considered the benchmark for naïve pluripotency because they are functionally equivalent to the naïve epiblast cells in the inner cell mass (ICM) of an embryonic day 4.5 (E4.5) blastocyst⁵. mESCs express core pluripotency transcription factors (TFs) and can generate all adult lineages in a chimera⁶ and a whole mouse through tetraploid complementation⁷. mESCs were first derived from the inbred 129 strain and cultured on mouse embryonic fibroblasts (MEFs) with media containing serum to maintain pluripotency^8,9. Advances have led to feeder-free and/or serum-free cultures that require leukemia inhibitory factor (LIF)^10,11, bone morphogenetic proteins (BMPs)¹², or the use of inhibitors that block differentiation and promote self-renewal^13,14. Two most widely used growth conditions for mESCs include serum with LIF (designated as S/L herein) and a “2i/L” culture using a combination of MEK inhibition (PD0325901, or PD03) and GSK3 inhibition (CHIR99021, or CHIR) with LIF¹³. S/L-cultured mESCs exhibit variable expression of pluripotency-related TFs^15–17, suggesting they are in a metastable state¹⁸. S/L culture cannot support derivation of mESCs from non-permissive strains such as CBA, MF1 and nonobese diabetic (NOD) mice. 2i/L-cultured mESCs, in contrast, show reduced heterogeneity in pluripotency TF expression¹⁷ and can be derived from a variety of mouse strains^19–22. Interestingly, several substates, reflecting a developmental continuum of naive pluripotency, emerge in response to different combinations of extrinsic 2i/L and S/L signaling cues²³.

Despite the success, 2i/L culture is not without its limitations. Extended cultivation of mESCs in 2i/L leads to permanent changes in their genetic and epigenetic makeup, which detrimentally impacts their developmental capabilities^24,25. These deficiencies are in part due to prolonged MEK inhibition that induces alterations in DNA methyltransferase expression²⁴. In supporting this, mESCs deficient in Erk1 and Erk2 couldn’t be maintained beyond four passages and exhibited slowed growth, G1 arrest, increased apoptosis, and genomic instability²⁶. To address these limitations, several alternative conditions have been developed: 1) Tuning down the concentration of PD03²⁵; 2) Replacing the MEK inhibitor with a Src inhibitor^24,25,27,28. 3) Supplementation of vitamin C, which acts as a cofactor for Fe(ii) 2-oxoglutarate dioxygenase enzymes like the Tet family, has been shown to stimulate Tet-dependent DNA demethylation, fostering a DNA methylation landscape in mESCs that more closely resembles that of the ICM²⁹. 4) Blocking certain protein kinase C (PKC) isoforms, especially PKCζ, preserves the pluripotency of mESCs without needing to activate STAT3 or block MEK/GSK3 signaling³⁰. 5) Inhibiting mediator kinase CDK8/19 (CDK8/19i) could recapitulate naïve pluripotency in mESCs without affecting FGF/MEK signaling or global DNA methylation³¹. 6) Development of conditions for PSCs with extended or expanded potential^32,33, which initially were thought to be totipotent-like but later studies revealed they are more similar to naïve mESCs^34,35.

mESCs have set the standard for identifying and characterizing naïve pluripotency states in other species. 2i/L culture also enabled the derivation of germline-competent rat ESCs (rESCs)^36,37, which facilitated the generation of genetically modified rats by homologous recombination^38–40. 2i/L causes rESCs to express differentiation-related genes, such as Cdx2 and T, hinting at an overly strong inhibition of GSK3^41,42. Indeed, lowering GSK3 inhibition (titrated 2i/L, t2iL) in rESCs suppresses differentiation and promotes self-renew^41,42. These observations underscore the need for carefully balanced GSK3 inhibition to maintain stable naïve rESCs. Nevertheless, even in a t2iL condition rESCs can generate fertile offspring via tetraploid complementation only at very early passages⁴³. Interestingly, inhibition of PKC signaling (PKCi) also promotes self-renewal of rESCs without compromising their developmental potency⁴⁴. Similar to mice, PKCi culture appears to be more effective than 2i/L to repress differentiation in rESCs⁴⁴. However, it is still unclear if PKCi-cultured rESCs are competent in tetraploid complementation assays.

The generation of naïve-like human PSCs (hPSCs) was achieved by overexpression of OCT4, SOX2, NANOG, and LIN28 in fibroblasts, and culturing in a medium containing human LIF and inhibitors specifically targeting MEK (PD03), GSK3 (CHIR), and ALK5 (A83–01)⁴⁵. Other pioneering studies in generating mESC-like hPSCs also relied on the ectopic expression of known pluripotency-related TFs in fibroblasts or in conventional hPSCs^46–52, which paved the way for developing transgene-independent conditions to stabilize human naïve pluripotency in vitro. The initial cultures supporting transgene-independent naïve-like hPSCs were established by testing combined actions of different factors known to regulate naïve pluripotency or employing unbiased screens, which include NHSM⁵³, t2iLGö⁵⁰, 5i/L/A⁴⁹, 2iF⁵⁴, 3iL⁵⁵, 2i/LIF/FGF2/Ascorbic Acid/Forskolin⁵⁶, LCDM³², and hEPSCM⁵⁷. Subsequent refinements were made, leading to the evolution of t2iLGö to PXGL⁵⁸; NHSM to HENSM⁵⁹; 5i/L/A to m5i/L/A/F⁶⁰ (reduced MEK inhibition) or AXGY⁶¹. More recent progress has introduced alternative conditions, including 2iLI⁶², CDK8/19i³¹, 4CL⁶³, AIC-N⁶⁴ and prEpiSC⁶⁵ cultures. The unexpected variety of growth conditions for naïve hPSCs could be partially attributed to the greater developmental plasticity of the human naïve epiblast that retains the ability to feed into extraembryonic trophoblast lineages well past the early pre-implantation stages⁶⁶. Several naïve-like hPSC cultures, some modified, have been used on non-human primates (NHPs). These include STAT3 overexpression and culture systems such as t2iLGö, PXGL, NHSMV, E-NHSM, 5i/L/A, 4CL, and 4i/L/b^67–70. When applied to NHP PSCs they result in varied colony morphology, variable expression levels of naïve TFs, and assorted X-chromosome inactivation (XCI) states in female cells.

2.2. Primed Pluripotent Stem Cell States

During the post-implantation phase, the epiblast is subjected to inductive cues from extraembryonic tissues, effectively ‘priming’ them for differentiation. Rodent epiblast stem cells (EpiSCs) are derived from post-implantation epiblast and represent primed pluripotency. EpiSCs are typically maintained in media containing FGF2 and Activin-A (F/A) that supports their developmentally more advanced epiblast state^71,72, which bears transcriptional and functional similarity to anterior late-gastrula primitive streak cells^73,74.

Under F/A conditions, EpiSCs self-renew into substates reflecting different post-implantation epiblast stages^75,76. This heterogeneity likely originates from dynamic epiblast cell populations that exist during the peri-gastrulation stage and are influenced by WNT/β-catenin signaling ⁷⁷. Inhibiting WNT signaling, either genetically or chemically (with XAV939 and IWR-1), alongside F/A, stabilizes a more uniform primed pluripotent state and enhances EpiSCs’ clonal expansion and derivation efficiency⁷⁸. Notably, EpiSCs cultured with FGF2 and IWR-1, known as region-selective EpiSCs (rsEpiSCs), show transcriptional similarity and selectively engraft to the posterior epiblast⁷⁹. This characteristic indicates that different primed conditions impart specific regional traits to EpiSCs⁴.

Conventional hPSCs, for example cultured in mTeSR1 medium⁸⁰ and categorized as 'primed' PSCs, share many characteristics with mouse EpiSCs (mEpiSCs) but have distinct molecular and functional traits. For instance, PRDM14 is essential for maintaining hPSCs, but not mEpiSCs⁸¹. Different from mEpiSCs, conventional hPSCs don’t exhibit increased FGF5 or N-CADHERIN expression but maintain the expression of E-CADHERIN and REX1^53,81. Furthermore, they can efficiently differentiate into both primordial germ cell-like cells (PGC-LCs) and amniotic epithelium, unlike mEpiSCs^82–84. Similar to mEpiSCs, inhibiting WNT signaling in hPSCs reduces heterogeneity and spontaneous differentiation, enhancing quality and clonal efficiency^79,85.

Primed PSC cultures show more consistency across species compared to naïve conditions. F/A conditions have successfully derived primed PSCs from a wide range of species, including endangered species like drills and northern white rhinoceros⁸⁶, as well as bats⁸⁷. WNT inhibition conditions are also broadly effective in various species, including mice, humans, rats^88,89, cattle^90–92, sheep^92,93, pigs⁹², rabbits⁹⁴ and NHPs^79,95. WNT-inhibited PSCs exhibit species-specific differences in colony morphology⁹³, clonogenicity⁷⁹, and responsiveness to PGC-LCs induction^89,94,95. These differences may stem from inherent species-specific traits, variations in FGF and Nodal/Activin signaling levels, or the WNT inhibitors used.

2.3. Intermediate and Formative Pluripotent Stem Cell States

Early post-implantation epiblast cells undergo substantial morphological, molecular, and metabolic changes, leading to gastrulation. This period involves shifting growth factor dependencies for epiblast self-renewal. Studies have identified various intermediate mouse pluripotency states between the naïve and primed epiblast cells^{35,75,85,96–107}. A notable intermediate state is “formative pluripotency”, found in E5-E6 epiblasts, characterized by unique molecular traits and the ability to contribute to germline chimeras and respond to PGC induction by BMP signaling¹⁰⁸.

The interplay of FGF, TGF-β/Nodal/Activin, and WNT signaling is key in defining the temporal identity of epiblast cells cultured in vitro. For maintaining mESCs in their ground state, active WNT signaling is essential, whereas FGF/ERK signaling is inhibited to block differentiation, and TGF signaling is dispensable. During early post-implantation development, epiblast cells transition from the apolar ICM to the organized rosette structures of the E5 epiblast¹⁰⁹. The combination of LIF with WNT and FGF/ERK inhibitors has enabled the derivation of rosette-like stem cells (RSCs) from both blastocysts and mESCs¹⁰¹, mimicking the E5 epiblast’s cellular architecture. RSCs express markers of both naïve and formative pluripotency (e.g. OTX2) and their chromatin landscape resembles the primed state, positioning them between naïve and primed pluripotency¹⁰¹.

TGF and FGF signaling activation is crucial for advancing pluripotency beyond the naïve state^110,111. The combination of Activin-A with BMP4, CHIR, and LIF (in ABCL medium) stabilizes advanced pluripotent stem cells (ASCs), which are developmentally more advanced than the ground state and characterized by a stable hypermethylated epigenome and preserved genomic imprints⁹⁸. Treating mESCs with FGF2 and Activin-A for 2 days produces transient formative epiblast-like cells (EpiLCs), responsive to PGC-LC induction and chimera competence, similar to E5.75 epiblasts¹⁰⁴. Another study discovered a transient intermediate cell type during EpiLC differentiation, representing a poised state of pluripotency between the naïve and formative states⁹⁹. These cells, losing the expression of naïve genes and not yet expressing formative markers, are marked by the activation of various miRNAs mediated by ISY1, resembling peri-implantation epiblasts around E4.5-E5.0. However, these cells require further characterization and have not been stably cultured yet.

Activating WNT signaling by CHIR balances the differentiation effects induced by Activin-A and FGF2, maintaining transient EpiLCs as metastable epiblast-like stem cells (EpiLSCs) in culture. EpiLSCs have bivalent energy metabolism, distinctive transcriptomic and chromatin accessibility profiles, and the competence for PGC-LC induction¹⁰⁵. Stable and uniform populations of formative-like FTW-ESCs³⁵ or Intermediate Pluripotent Stem Cells (INTPSCs)¹⁰⁰ can be directly derived from blastocysts, converted from mESCs, or reprogrammed from MEFs when cultured on feeders with FGF2, Activin-A, and CHIR. FTW-ESCs show lower levels of naïve markers than mESCs and moderate expression of formative markers. Their gene expression profile closely resembles the E5.0-E5.5 epiblast, and they are responsive to PGC-LC induction and competent for germline chimera formation following blastocyst injection³⁵.

Subpopulations of F/A-EpiSCs and EpiSCs cultured under different conditions also reveal intermediate states between naïve and primed pluripotency. F/A-EpiSCs carrying the Oct4-GFP transgene contain both GFP-positive and GFP-negative populations. GFP-positive cells exist transiently in early passages and contribute to blastocyst chimeras, suggesting a pre-gastrulation epiblast state⁷⁵. Germline transmission, however, is not observed with these cells. A modified EpiSC medium with FGF4 stabilizes the Oct4-GFP-positive population, maintaining their chimeric potential¹⁰⁶. IWPs, a porcupine inhibitor, stabilizes EpiSCs in a pre-gastrulation state (IWP2-EpiSCs) when combined with F/A, leading to increased efficiency in reverting to naïve mESCs and chimera formation⁸⁵. The effects differ from EpiSCs cultured with XAV or IWR-1, possibly due to varying endogenous WNT signaling levels. Interestingly, combining a WNT activator (CHIR) and a WNT inhibitor (XAV939 or IWR1, but not IWP2) stabilizes EpiSCs at a stage closer to mESCs (CX/CR1-EpiSCs)¹⁰⁷. This combination’s mechanism involves containing β-CATENIN in the cytoplasm, separate from its role in regulating transcription.

Two studies have generated formative PSCs resembling E6 epiblast, using modified EpiSC culture conditions with XAV, Activin-A and FGF2. The first study, using low Activin-A stimulation, XAV, and a retinoic acid receptor inverse agonist (BMS 493) - the A_loXR condition – derived so-called formative stem (FS) cells. These cells show low naïve marker expression, high formative marker expression (Fgf5, Otx2, Oct6), XCI, direct PGC-LC responsiveness, and chimera formation¹⁰². The second study focused on post-implantation epiblast progression, culturing mESCs and E5.5–E6.5 epiblasts in 3D Matrigel with Activin-A, FGF2, and XAV. This culture condition led to epiblast-like embryoids (epiblastoids) forming rosettes and lumens, and stable formative cell lines (fPSCs)¹⁰³. These findings, along with FTW-ESCs³⁵ and EpiLSCs¹⁰⁵, suggest that the formative state spans a continuum between naïve and primed states. Future research is needed to determine whether other intermediate mouse PSCs also fit within this formative pluripotency spectrum.

Studies on cultured hPSCs also propose intermediate states of pluripotent epiblast cells^{35,102,112–115}. Reassessing hPSCs under conventional conditions revealed cells at stages earlier than primed pluripotency. One study identified a subset of NCAD+ cells at the periphery of hPSC colonies with high self-renewal capacity and a tendency towards the extraembryonic endoderm lineage¹¹². This finding suggests these NCAD+ cells represent a stage before primed pluripotency, as naïve hPSCs can differentiate into extraembryonic endoderm cells¹¹⁶, but primed hPSCs usually don’t. Another study found that lipid deprivation induces a stable intermediate pluripotency state in conventional hPSCs, characterized by high self-renewal, expression of naïve markers, and a bivalent metabolic profile¹¹³. Additionally, a distinct subpopulation at the periphery of hPSC colonies, similar to NCAD+ cells, showed potential for germline and somatic differentiation, a shorter G1 cell cycle, and a unique open chromatin profile, without lineage priming¹¹⁴. More recently, hPSCs cultured on laminin-111 in a defined primed condition displayed features similar to the early post-implantation epiblast¹¹⁵.

Comparative transcriptomic studies show that hPSCs cultured under various naïve conditions align more with early post-implantation (E8–E10) than pre-implantation (E6) human epiblasts, indicating they are in intermediate pluripotency states³⁵. Additional culture conditions supporting hPSCs between naïve and primed states have been reported. Two mouse formative PSC conditions, FTW³⁵ and AloXR¹⁰², have been adapted for human cells. AloXR-hPSCs, not expressing naïve markers, resemble early post-implantation monkey epiblasts and share some traits with primed hPSCs¹⁰². These cells rapidly differentiate into neural, endodermal, and mesodermal lineages, but their PGC-LC differentiation competence is untested¹⁰². FTW-hPSCs exhibit gene expression profiles between naïve and primed states, resembling E8 human epiblasts³⁵, and show potential for PGC specification and interspecies chimera formation^35,117. Given the transient nature of the naïve state in human and NHP blastocysts^118,119, compared to the stable gene expression of early post-implantation primate epiblasts¹²⁰, capturing the intermediate and primed states in primates may be more feasible than capturing the naïve state.

3. Hallmarks of Pluripotent Stem Cell States

The characteristics of naïve and primed PSCs were initially identified by contrasting these two states against each other¹²¹. Yet, the uniqueness of these defining features of naïve and primed pluripotency has been challenged by the identification and characterization of PSCs in intermediate stages. Additionally, variations between species call for different benchmarks to define pluripotency across organisms. Moreover, the ethical restrictions on using hPSCs in germline chimera and tetraploid complementation assays necessitate alternative methods to functionally define pluripotency in human research. This review provides a summary of the distinguishing features, or hallmarks, of various PSC states (Figure 3). Readers seeking a more in-depth understanding are encouraged to consult comprehensive recent reviews^{4,96,122–129}. As the variety of PSCs from different species continues to grow, the characteristics used to differentiate them will need frequent updates. In this review, we include only those hallmarks that are currently applicable to the majority of mouse and human PSCs identified to date. The criteria for defining PSC states are generally classified into two main groups: molecular and functional hallmarks (Figure 3).

Figure 3. Molecular and functional hallmarks of dynamic pluripotent stem cells.

Left, A summary that highlights the selected molecular and functional characteristics useful for distinguishing mouse and human PSCs in various states of pluripotency. Right, the hallmarks each type of PSC meets are highlighted with color-coded boxes.

3.1. Molecular Hallmarks

Signaling dependency.

Most naïve PSC cultures use the 2i/L framework or its variants, highlighting the importance of LIF/STAT3 and WNT signaling, and inhibiting FGF/ERK signaling, to maintain naïve pluripotency in vitro. Recent methods have employed tankyrase inhibitors to support human naïve pluripotency^58,59,61. FGF2 is also included in various naïve hPSC cultures^{47,49,53–56}, suggesting FGF signaling might positively influence naïve hPSCs through non-ERK pathways, though the exact mechanisms are unclear. In several species, including mice, rats, bovines, marmosets, and humans, strong NANOG expression in preimplantation blastocysts occurs without FGF/ERK signaling, suggesting ERK independence as a universal feature of naïve pluripotency^130–133. Nonetheless, 2i/L lone is not enough to maintain naïve pluripotency in species beyond mice and rats, necessitating modulation of other signaling pathways. Moreover, the reliance on specific signaling pathways for naïve pluripotency varies across species; for instance, TGF signaling is dispensible for mESCs but is needed for naïve human pluripotency^134,135.

Stabilizing primed pluripotency in vitro hinges on activating FGF and TGF pathways, a common requirement across species. Adding WNT inhibitors can result in a 'ground state' of primed pluripotency with uniform pluripotent gene expression. However, intermediate/formative pluripotency, which includes several temporal states, has different signaling dependencies. Various methods are used to maintain formative PSCs, with some activating WNT signaling (as in FTW-PSCs and EpiLSCs) and others inhibiting it (as in AloXR-PSCs and fPSCs), in combination with F/A.

Expression of pluripotency state-specific marker genes.

Various genes help distinguish naïve (e.g. Klf2, Klf4, Tfcp2l1), intermediate/formative (e.g. Otx2, Sox3, Oct6), and primed (e.g. Fgf5, Lef1, Nodal) PSCs^5,35,102. However, care should be taken in defining pluripotency states based on gene expressions due to inherent heterogeneity. In addition, significant species-specific differences exist. Certain mouse naïve markers, such as Klf2, Nrob1, Bmp4, and Esrrb, are not expressed in human or marmoset ICMs. Conversely, the human preimplantation epiblast expresses unique genes, such as KLF17, and members of the TGF signaling pathway (e.g., NODAL, GDF3, and LEFTY1^131,135), which are absent in mice. Furthermore, transcriptome profiling reveals continuous expression of NANOG and PRDM14 in post-implantation epiblasts until gastrulation in NHPs¹²⁰.

Transcriptomic similarity to in vivo epiblasts.

Comparing transcripotomes of cultured PSCs with those of in vivo epiblasts at various developmental stages, particularly at the single cell level, can provide a more accurate assessment of their pluripotency states. Recent advances in transcriptomic analysis of mammalian embryos before, during, and after implantation have yielded a wealth of data crucial for such comparisons^{5,120,131,135–141}. Ideally, the expression patterns of genes in naïve, intermediate/formative, and primed PSCs should mirror those in pre-implantation, pre-gastrulation, and peri-gastrulation epiblasts, respectively. However, it is important to account for differences between in vivo and in vitro conditions, which can cause variations in gene expression.

Global DNA methylation status.

Mouse and human preimplantation epiblasts feature widespread DNA hypomethylation^142–146, a key trait of naïve pluripotency, except in imprinted regions. 2i/L-mESCs exhibit low DNA methylation^17,147, likely attributable to JMJD2C-dependent TET1 activation and DNMT3A/B protein degradation mediated by PRDM14/G9a¹⁴⁸, a feature not maintained under the S/L condition. Naïve hPSCs grown in 5i/L/A⁴⁹ and t2iLGö/PXGL^50,149 conditions show DNA hypomethylation similar to the human ICM but lose genomic imprinting^150–152. Naïve hPSC cultures with slightly higher DNA methylation levels (~42%) than the human ICM but lower than primed hPSCs (~75%) show improved genomic stability and imprint preservation^59–61,63. These hPSCs are in more advanced pluripotency states than 5i/L/A and t2iLGö/PXGL cells, reminiscent of mouse ASCs (~70% CpG methylation). This level is higher than 2i/L-mESCs (~20%) but lower than EpiSCs (~90%)⁹⁸. Unlike 2i/L-mESCs, where imprints are erased^98,153, most imprints are retained in ASCs⁹⁸. The methylation status of most intermediate hPSCs is unreported, but those cultured on laminin-111 and in lipid-deprived conditions show methylation levels similar to or slightly lower than primed hPSCs^113,115.

From E4.0 to E6.5, mouse epiblasts experience a genome-wide increase in DNA methylation, coinciding with upregulation of de novo methyltransferase 3 (DNMT3) enzymes¹⁵⁴. In mammals, three types of DNMT3s have been identified: DNMT3a and DNMT3b are catalytically active, whereas DNMT3L acts as a cofactor. Dnmt3b expression starts at E3.5, followed by Dnmt3a at E4.5, with Dnmt3l expressed briefly between E4.5 and E6.5¹⁴¹. Human embryos show a similar increase in median DNA methylation from 26.1% on day 6 to 60.0% by day 10¹⁴¹.

DNA methylation patterns in various in vitro PSCs largely reflect in vivo patterns. 2iL-mESCs show global hypomethylation similar to ICM cells, whereas S/L-mESCs have higher DNA methylation, resembling the early post-implantation epiblast^153,155. Interestingly, the intermediate pluripotent state regulated by ISY1/miRNAs, despite having a similar transcriptomic signature to S/L-mESCs, has a hypomethylated genome as in 2iL-mESCs⁹⁹. ASCs, further along in development, exhibit high methylation levels across various genomic regions⁹⁸. Contrasting with 2iL-mESCs, which exhibit lower expression of DNA methyltransferases^{17,147,153,155}, RSCs express Dnmt3b and Dnmt3l, showing elevated 5mC levels intermediate between naïve and primed cells¹⁰¹. Globally, fPSCs exhibit high DNA methylation similar to EpiLCs and E6.5 epiblast cells¹⁰³. EpiSCs and rsEpiSCs, at the far end of the pluripotency spectrum, have a hypermethylated epigenome (~87%)⁷⁹. rsEpiSCs show hypermethylation in regions linked to regionalization and anterior/posterior pattern specification⁷⁹. Notably, the global level of 5-methylcytosine in EpiSCs is similar to S/L-mESCs^153,156. However, EpiSCs show a distinct bias towards hypermethylation at promoter regions, especially those with high CpG island promoters. This DNA methylation signature is unique to EpiSCs and differs from their in vivo counterparts, the E6.5 and E7 gastrulating epiblasts, indicating potential culture artifacts¹⁵⁷.

X chromosome inactivation status in female cells.

XCI is a crucial developmental process in female mammals to balance the dosage of X-chromosomal genes between the sexes. It occurs dynamically during mammalian development and has two forms: imprinted and random. Imprinted XCI, happening at the preimplantation stage in mice, involves selective silencing of the paternal X chromosome at the 4-cell stage^122,158, with the cis-acting, non-coding RNA Xist (X-inactive specific transcript) playing a key role¹⁵⁸. Imprinted XCI begins to be erased in the ICM of early blastocysts, leading to a temporary reactivation of both X chromosomes. Subsequently, random XCI begins in the post-implantation epiblast^136,158. A study tracking Xist expression during peri-implantation development found that Xist erasure starts around E3.75 and completes by E4.5 in mice. Notably, Xist re-expression was observed as early as E4.75, indicating the early initiation of random XCI in these cells¹⁵⁹. This study also uncovered that a notable proportion of epiblast cells displayed biallelic Xist expression at E5.25 (13.9%) and E5.5 (9.3%)¹⁵⁹. Interestingly, similar expression patterns were noted in male epiblast cells at comparable stages¹⁶⁰. These findings suggest substantial epigenomic changes during this phase of developmental transition. By E6.5, nearly all cells exhibited a single Xist cloud, indicating that random XCI is likely completed by this stage¹⁵⁹.

Pre-implantation human embryos exhibit a unique XCI pattern compared to mice, lacking paternal XIST imprints and showing random XCI in both embryonic and extraembryonic tissues^136,158,161. Notably, ~85% of cells in day 6 and 7 female blastocysts display bi-allelic XIST coating, indicating a distinct mechanism of XCI in humans compared to mice^136,158. Interestingly, despite XIST accumulation, bi-allelic expression of X-linked genes is evident at the chromosome-wide level throughout pre-implantation development, starting as early as E3^136,158. These studies establish that human embryos are in a pre-XCI state before implantation (Xa^XIST+Xa^XIST+). Information on human random XCI timing is limited due to restricted access to peri-implantation embryos. XCI involves the accumulation of histone 3 lysine 27 trimethylation (H3K27me3) marks on the condensed X-chromosome. In a study modeling human implantation, day 5 human blastocysts were cultured for an additional three days on decidualized endometrial stromal cells. Intriguingly, at the end of this co-culture, H3K27me3 foci were absent in epiblast cells, suggesting that random XCI did not initiate at this stage¹⁶². Advances in extending human blastocyst culture beyond implantation stages^163,164 have provided insights into XCI during early post-implantation. Studies examining allele-specific expression of X-linked genes from days 6 to 12 in these cultured embryos suggest that random XCI begins at peri-implantation stages without being complete by day 12¹⁴¹.

In female mouse PSCs, XCI status is another key indicator of their pluripotency state and resemblance to in vivo development stages. Female mESCs have two active X chromosomes (XaXa), mirroring pre-implantation epiblast cells. Intermediate PSCs transition from this XaXa status to having one active and one inactive chromosome (XaXi). Both RSCs and FTW-ESCs retain the XaXa status, consistent with E5.0-E5.5 epiblast stages^35,101. In contrast, primed F/A-EpiSCs and rsEpiSCs⁷⁹, intermediate CX/CR1-EpiSCs¹⁰⁷, IWP2-EpiSCs⁸⁵, fPSCs¹⁰³ and FS cells¹⁰² exhibit XaXi status, corresponding to later epiblast stages (E6.5-E7.75).

XCI in female hPSCs is more complex than in mice. Initially, conventional hPSCs were thought to have two active X chromosomes without XIST expression (Xa^XIST−Xa^XIST−)¹⁶⁵. Later studies revealed significant variations in XCI status across different hPSC lines and even within subcultures of the same line^166–170. Additionally, factors like oxygen concentration during derivation and culture can influence the XCI status in primed hPSCs¹⁷¹. Primed hPSCs show XCI instability, often losing XCI characteristics over multiple passages, unlike differentiated cells. This instability is characterized by the loss of XIST expression and partial reactivation of the inactive X chromosome^172,173. Similarly, naïve hPSCs exhibit a varying XCI status under different culture conditions, with some (5i/L/A^49,151 and t2iLGö⁵⁰) aligning with a pre-XCI state indicated by biallelic expression of X-linked genes and XIST accumulation on active X chromosomes^174,175. However, the extent to which these cells mirror in vivo pre-implantation epiblasts is debatable, as most show monoallelic XIST expression^174,175. There is also a lack of consensus regarding the XCI marker H3K27me3. Some studies reported an enrichment of H3K27me3 on active X chromosomes that are coated with XIST¹⁷⁴. Others did not observe this accumulation¹⁷⁵, aligning instead with findings in embryos¹⁵⁸. Moreover, during differentiation, naïve hPSCs undergo de novo XCI, but this process differs from the XCI observed in embryos as it follows a non-random pattern^151,174. In another study insufficient inhibition of autocrine FGF2 led to varied X chromosome status in naïve hPSCs. The authors successfully isolated a uniform population of naïve hPSCs with an Xa^XIST+Xa^XIST+ status, which is closely related to the XCI status of human pre-implantation epiblast. In this study, it was also observed that when these cells transitioned from a naïve to primed state, random XCI ensued, mimicking the natural embryonic process¹⁷⁶.

Metabolic features.

PSC states are closely linked to specific metabolic profiles. Although a high glycolytic flux is essential for maintaining pluripotency^177–181, its role varies across different pluripotency states¹⁸². Primed PSCs rely almost entirely on glycolysis. In contrast, naïve PSCs are characterized by a bivalent metabolism, leaning more towards oxidative metabolism^{49,50,183,184}. Intermediate and formative PSCs show mixed metabolic states. EpiLSCs and FTW-ESCs display an intermediate level of maximal mitochondrial activity between mESCs and mEpiSCs. They rely on oxidative phosphorylation during basal respiration but shift to glycolysis during maximal respiration^35,105. FS cells, similar to mEpiSCs, predominantly use glycolysis for energy production¹⁰⁵. These observations mirror the metabolic preference of the early embryos, which initially depend on pyruvate from ovarian follicle cells for oxidative metabolism. As development progresses, increased glucose availability and transporter expression lead to a rise in glycolytic activity^182,185,186. Interestingly, primed PSCs have low mitochondrial activity but elongated mitochondria with defined cristae, in contrast to the spherical but active mitochondria in naïve PSCs^{50,54,183,187,188}. Formative EpiLSCs show an intermediate mitochondrial morphology with a mix of spherical and elongated structures¹⁰⁵.

Lipids are essential in early mammalian development, acting as energy sources, signaling molecules, and membrane components¹⁸⁹. Prior to implantation, lipids come from the oocyte's endogenous reserves and are also synthesized de novo¹⁹⁰. Lipid availability influences acetyl-coenzyme A and α-ketoglutarate levels, impacting ERK signaling and inducing an intermediate pluripotency state with heightened de novo lipogenesis^113,191. This intermediate state thrives in lipid-free culture conditions, promoting active lipid biosynthesis and adding lipids to the medium disrupts this state. Supporting this observation, numerous genes highly expressed in fPSCs are involved in cholesterol/sterol biosynthetic and lipid metabolic processes, underlining the significance of lipid metabolism in maintaining intermediate/formative PSCs¹⁰³.

Amino acids influence pluripotency and stem cell fate, serving as substrates for biosynthesis and facilitating chemical modifications¹⁸². Specific amino acids such as threonine, methionine, and proline are key regulators of pluripotency. mESCs rely heavily on threonine metabolism to sustain pluripotency and self-renewal¹⁹². Threonine supports naive pluripotency by supplying S-adenosyl methionine (SAM) for the trimethylation of histone H3 lysine 4 (H3K4me3)¹⁹³. As an essential amino acid, threonine can be broken down by threonine dehydrogenase (TDH) into glycine and acetyl-coenzyme A. Disruption of threonine metabolism in mESCs, either through inhibiting TDH or depriving cells of threonine, leads to a loss of stemness, diminished proliferation, apoptosis, and cell cycle arrest^192,194. hPSCs do not depend on threonine as the TDH gene in humans is a non-functional pseudogene¹⁹⁵. In primed hPSCs, methionine is essential for synthesizing SAM, with its depletion leading to differentiation and apoptosis¹⁹⁶. Conversely, in naïve hPSCs, low levels of SAM, maintained by nicotinamide N-methyltransferase, are necessary to preserve the naïve state¹⁸⁷. Proline also plays a vital role, particularly in mESCs, where its addition to the culture media enhances proliferation and facilitates the transition to EpiSCs^197,198. The implantation process involves breaking down a proline-rich extracellular matrix, suggesting that the released proline may support the epiblast transition during peri-implantation development^199,200.

3.2. Functional Hallmarks

A range of assays has been developed to assess the developmental potential of cultured PSCs¹²⁸ (Figure 3). These include universal assays, such as in vitro embryoid body differentiation and the teratoma assays, which are suitable for all PSCs. Additionally, there are assays designed to differentiate between various pluripotency states, briefly outlined below.

Blastocyst chimera formation and germline transmission.

Assessing pluripotency in PSCs often involves testing their ability to form chimeras by combining them with preimplantation embryos and transferring these into surrogate mothers¹²⁸. PSCs with high developmental potential not only support normal development but also produce high-grade chimeras capable of generating offspring entirely derived from the donor PSCs via germline transmission. In contrast, PSCs with low developmental potential may result in modest to no chimerism. Mouse and rat ESCs are known for their ability to form blastocyst chimeras and achieve germline transmission. Chimeras are formed not only by naïve PSCs but also by several intermediate and formative PSC types, such as ASCs⁹⁸, RSCs¹⁰¹, FTW-ESCs/INTPSCs^35,100, FS cells¹⁰², fPSCs¹⁰³, Oct4-GFP+ EpiSCs and IWP2-EpiSCs. ASCs, RSCs, and FTW-ESCs/INTPSCs have also demonstrated germline transmission, a trait not observed in fPSCs, Oct4-GFP+ EpiSCs⁷⁵, IWP2-EpiSCs⁸⁵ and not tested in FS cells¹⁰². Generally, the extent of chimera contribution from intermediate/formative PSCs tends to be lower than from mESCs. Blastocyst chimera assays have also been applied to NHPs, with varied success. For instance, cynomolgus monkey PSCs (cyPSCs) cultured under a modified NHSM condition (NHSM supplemented with Vitamin C) showed modest chimeric contribution of less than 20% into two fetuses at around 100 days of gestation⁶⁷. Importantly, cyPSCs cultured in 4CL⁶³ medium resulted in an aborted chimeric fetus and a live neonatal chimeric monkey, with the latter showing a high donor cell contribution of up to 90%⁶⁹. However, this monkey developed severe health problems and was euthanized shortly after birth, preventing assessment of its germline transmission capabilities.

Blastocyst chimera formation is not suitable for human research due to ethical issues and is challenging for species with long gestation or poorly understood reproductive biology. As an alternative, interspecies chimeras are used, but this approach is often unreliable due to low success rates and xenogeneic barriers between distant species²⁰¹ (Figure 3). Consequently, teratoma formation and in vitro differentiation of embryoid bodies are broadly used for accessing pluripotency in humans. Notably, some naïve hPSCs require an adaptation period in primed media to respond effectively to differentiation signals²⁰², which may explain their limited ability to generate teratomas²⁰³.

Tetraploid complementation.

Tetraploid complementation, the most stringent test for pluripotency, involves introducing donor PSCs into a tetraploid blastocyst⁷, created by fusing two-cell stage embryos. In this process, tetraploid cells predominantly contribute to extraembryonic placenta tissue, whereas the diploid donor PSCs develop the embryo proper. High-quality mESCs and early passages of rESCs⁴³ have demonstrated the ability to complement tetraploid blastocysts, leading to animals entirely derived from donor cells. Additionally, intermediate ASCs have been used to produce full-term offspring in tetraploid hosts⁹⁸. However, tetraploid complementation has yet to be tested for RSCs and FTW-ESCs.

Post-implantation epiblast chimera formation.

EpiSCs show limited ability to generate blastocyst chimeras but can effectively integrate into the post-implantation epiblast, contributing to ex vivo epiblast chimeric embryos^{74,79,204,205}. This outcome differs from mESCs, which are limited in contributing to chimera formation in post-implantation epiblasts²⁰⁴. Notably, primed hPSCs and rhesus monkey PSCs can engraft into E6.5 or E7.5 mouse epiblasts and differentiate into cells from all three germ layers^79,206,207. Grafting to the post-implantation epiblast thus serves as a functional test for primed pluripotency. The capability of intermediate and formative PSCs to generate chimeras through post-implantation epiblast grafting remains to be tested.

Primordial germ cell specification.

Germ cells, vital for species continuity, originate from precursor cells known as PGCs, which arise from competent formative epiblast cells around E5.5-E6.25 in response to signals such as BMPs in mice. These PGCs proliferate and migrate through the developing hindgut before colonizing the nascent gonads, where they differentiate into oocytes or sperm¹⁰⁴. Notably, mESCs require a transient differentiation into formative EpiLCs to generate PGC-LCs in vitro²⁰⁸. This process is also mirrored in rat EpiLCs, which can be induced from rESCs to form functional PGC-LCs in spherical aggregates²⁰⁹. Although previously considered inefficient, we now know that mEpiSCs can produce PGC-LCs and fully functional germ cells²¹⁰. Rat EpiSCs cultured in a WNT-inhibited condition AFI2XY (Activin-A, FGF2, IWP2, XAV and Y27632) demonstrate germline competency, producing functional PGC-LCs in vitro and showing molecular traits akin to pre-gastrulating pluripotent epiblasts, thus representing an intermediate pluripotency state⁸⁹. Consistently, WNT-inhibited cynomolgus monkey and rabbit PSCs also exhibit robust PGC-LC differentiation^94,95. Contrary to mESCs, formative PSCs such as FTW-ESCs, EpiLSCs, FS cells, and fPSCs, directly specify PGC-LCs^{35,102,103,105}. Intermediate or formative hPSCs also show similar PGC responsiveness, especially when grown on laminin-111¹¹⁵, cultured in FTW medium³⁵, or propagated in a simplified NHSM condition, termed '4i'²¹¹. Intriguingly, direct PGC-LC induction from primed hPSCs is achievable either through incipient mesoderm-like cells⁸³ or using a microfluidic device⁸⁴. However, species-specific differences in PGC specification mechanisms exist between primates and rodents, including varied developmental stages and origins from distinct progenitor cells (epiblast and/or amnion)²¹².

4. Dynamic Totipotent Stem Cell States

Zygotes and early blastomeres, capable of forming the entire organism and producing all embryonic and extraembryonic tissues, exhibit totipotency, the highest level of developmental potency. Unlike stem cells, early blastomeres, derived from zygote cleavage, lack self-renewal ability²¹³. Consequently, there's interest in whether totipotent stem cells, recapitulating some features of zygotes/early blastomeres, can be effectively derived and maintained in vitro. Totipotent or totipotent-like stem cells, though present in low proportions in pluripotent cultures, can be isolated and even stably maintained under specific conditions, sparking significant research interest. This summary highlights recent advancements in generating and characterizing these so-called totipotent stem cells, acknowledging their currently limited understanding compared to PSCs.

4.1. Metastable Mouse 2-Cell-Like Cells and Human 8 Cell-Like Cells

Using 2C::tdTomato+ reporter mESCs, generated based on the MuERV-L-LTR retrotransposon expressed in mouse 2-cell blastomeres at the ZGA stage, a study found a small subpopulation (about 0.1–1%) could transiently activate 2-cell specific genes, such as Zscan4 and Zfp352. These cells, known as 2-cell-like cells (2CLCs), quickly revert to tdTomato-, suggesting a metastable state with fluctuating entry into and exit from the 2-cell totipotent-like state²¹⁴. Currently, no study has confirmed that a single 2CLC can generate both embryonic and extra-embryonic tissues in chimeric mice when injected into blastocysts (or earlier stages). More research is needed to fully validate the functional totipotent potency of these cells²¹⁵.

Recent advancements have led to the generation of metastable human 8 cell-like cells (8CLCs) under naïve hPSC culture conditions. Similar to mouse 2CLCs, human 8CLCs activate a range of ZGA genes, such as ZSCAN4/5B, DUX4, TPRX1, and LEUTX, and constitute about 1.6% of all cells cultured in PXGL medium²¹⁶. In a second study using the e4CL (enhanced 4 chemicals + LIF) medium, 8CLCs were transiently induced to make up 11.9% of the cell population for less than a week, though they could not be passaged. These isolated 8CLCs, identified using TPRX1 as a reporter, demonstrated both embryonic and extraembryonic developmental potential in teratomas and interspecies chimeras in mice⁶³. In a third study, 8CLCs were identified from prEpiSCs and PXGL cultured naïve hPSCs using single-cell RNA-sequencing (scRNA-seq). The researchers created a LEUTX reporter to aid in isolating 8CLCs and developed a culture condition to increase the proportion of 8CLCs from prEpiSCs. In this culture, 8CLCs were able to be sustained at a relatively high percentage for 2–3 weeks⁶⁵.

The discovery of metastable 2CLCs and 8CLCs^63,65,216, which mirror the ZGA stage blastomeres in mice and humans, respectively, hints at the potential to culture totipotent-like cells in vitro. Subsequent research, following these initial findings, has identified several regulatory factors, including DUX4/Dux and Dppa2/4, that promote the shift from pluripotent to totipotent states^217–220. Despite these advances, none of these efforts have succeeded in achieving the stable maintenance and long-term passaging of totipotent stem cells.

4.2. Stable Culture of Mouse Totipotent Stem Cells

Notably, a type of so-called mouse and human expanded/extended pluripotent stem cells (EPSCs) has been cultured and maintained in specific culture media for long-term passages by two separate research groups^32,33. Interestingly, these cells don’t express any typical totipotency related genes, but instead still highly express pluripotency markers, such as Oct4, Nanog, Sox2 and Klf4/5. Surprisingly, functional assays have suggested that they are able to generate extraembryonic and embryonic tissues in vivo and in vitro. However, the totipotent features of these EPSCs have been challenged at molecular and functional levels³⁴. EPSCs are more appropriately considered as a type of intermediate PSCs rather than totipotent cells, given their closer transcriptional resemblance to early post-implantation epiblasts in mice and humans.

In a surprising development, a mouse totipotent stem cell line was successfully isolated and stably maintained by inhibiting spliceosome activity²²¹. Spliceosomes, essential for mRNA splicing and maturation, are typically expressed at low levels in zygotes and totipotent blastomeres, with their activation occurring around the ZGA stage, a pattern observed across various mammals. The success of this method suggests that splicing might influence the transition from totipotency to pluripotency. Suppression of splicing factors in mESCs triggers the activation of totipotent genes while repressing pluripotent genes, steering cells towards totipotency. Remarkably, adding Pladienolide B (PlaB), a splicing inhibitor, to the S/L medium, reprograms mESCs into totipotent blastomere-like cells (TBLCs) and enables their stable maintenance over long-term passages. Molecular analyses indicate that these TBLCs closely resemble 2-/4-cell blastomeres at the transcriptomic and epigenomic levels. In chimeric mice, single mouse TBLCs contribute to both embryonic and extraembryonic tissues, including 6 trophoblast and 1 yolk sac lineage²²¹.

Mouse TBLCs are different from 2CLCs. TBLCs show a wider and more pronounced activation of totipotent genes, such as Zfp365, Ddit4l, Plk2 and Btg2, which are predominantly present in zygotes and early to middle 2-cell blastomeres at the pre-ZGA stage. In contrast, key pluripotent genes, such as Sox2, Nanog, Pou5f1, Zfp42, Esrrb, Klf2, and Tdgf1, which remain highly expressed in 2CLCs, are completely silenced in TBLCs. Epigenetically, 2CLCs are similar to mESCs, including chromatin accessibility, H3K4me3, H3K27me3, and H3K27ac marks, especially at pluripotent and totipotent loci, unlike in vivo 2-cell blastomeres²²². Conversely, TBLCs, induced through spliceosome repression, show epigenetic features closely mirroring 2-/4-cell blastomeres, including DNA methylation, chromatin accessibility, and histone modifications, both at specific loci and globally. Therefore, although 2CLCs exhibit both pluripotent and totipotent characteristics, fitting their metastable nature, TBLCs are stable totipotent stem cells, having shed all pluripotent regulatory networks. scRNA-seq analysis indicates that TBLCs uniformly activate totipotent genes and suppress pluripotent genes, suggesting a relatively homogenous cell population.

Subsequently, several research groups have reported the successful isolation of other types of mouse totipotent stem cells, primarily through epigenetic modulation. These cell types include totipotent potential stem cells (TPSCs), totipotent-like stem cells (TLSCs), and chemically induced totipotent stem cells (ciTotiSCs) (Figure 4A)^222–224, which are derived using chemical compounds that target epigenetic regulators, such as KDM5B, DOT1L, and HDAC1/2, influencing histone modifications, such as H3 acetylation and H3K79 methylation, as well as activation of RARγ signaling. Similar to mouse TBLCs, TPSCs, TLSCs, and ciTotiSCs show extensive activation of totipotent genes and repression of pluripotent genes at the molecular level. Functionally, these cells contribute to a range of embryonic and extraembryonic lineages (Table 1).

Figure 4. Summary of in vitro cultured totipotent stem cells.

(A) Developmental Timeline of In vitro Totipotent Stem Cell Cultures. The reference to 'similar to (embryo stage)' relates specifically to similarities at the transcriptomic level.

(B) Heatmap showing the comparative expression of representative totipotent genes across different cell types.

(C) PCA analysis of transcriptomes of TBLC²²¹, TLSC²²², TPSC²²³, EPSC^32,33,223, ciTotiSC²²⁴, 2CLC ²²²and their corresponding mESCs, as well as mouse pre-implantation embryos based on the global transcriptome. Stars represent TBLC data, triangles represent TLSC data, solid squares represent TPSC data, hollow squares represent EPSC data, rounded rectangles represent 2CLC data, and hexagons represent ciTotiSC data.

Table 1.

Dynamic Totipotent Stem Cells

Feature	Developmental potentials
	Chimera assay (in vivo)			Teratoma (in vivo)		Blastoid-formation assay (in vitro)		Trophoblast stem cell (TSC) reprogramming
	Contribute to embryonic tissue	Contribute to extra-embryonic tissue	Chimeric mice	Embryonic lineages	Extra-embryonic lineages
mESC (for comparison)	Yes	No	Yes	Yes	No	No		No
D-EPSC32	Yes	Conflicting results: Yes32 and No34	Yes	Three germ layers	No	Conflicting results: Yes230 and No231		Conflicting results: Yes32 and No34
L-EPSC33	Yes	Conflicting results: Yes33 and No34	Yes	Three germ layers	No	Conflicting results: Yes230 and No231		Conflicting results: Yes33 and No34
2CLC214	No single cell chimera tested	No single cell chimera tested	Not tested	Not tested	Not tested	Not tested		Not tested
TBLC221	Yes	Yes	Yes	Yes	Yes (scRNA-seq)	Yes246		Yes
TLSC222	Yes	Yes	Not tested	Not tested	Not tested	Yes		Not tested
TPSC223	Yes	Yes	Yes	Mesoderm, ectoderm	Yes (scRNA-seq)	Yes		Yes
TotiSC224	Yes	Yes	Yes	Three germ layers	TGCs (IHC)	Not tested		Yes
Feature	Transcriptomic profiles
	Global comparison with mouse early embryonic states		Totipotency or 2C marker		Pluripotency marker	Transposon element (MERVL, MT2-Mm)		Maternal
mESC	Similar to epiblast		+		+++	+		+
D-EPSC	Similar to E5.534		+		+++	+		+
L-EPSC	Similar to E3.5-E4.534		+		+++	+		+
2CLC	Similar to blastocyst		++		+++	++		+
TBLC	Similar to 2- and 4-cell		+++		+	+++		+++
TLSC	Similar to 2- and 4-cell		+++		+	+++		+++
TPSC	Similar to 2-cell stages		+++		+	+++		Not tested
TotiSC	Similar to 2-cell stages		+++		+	+++		+++
Feature	Genomic landscape
	Chromatin accessibility	DNA methylome	Histone modification
			bdH3K4me3	H3K4me3	H3K27me3	H3K9me3	acetylated H3	acetylated H4
mESC	Low	High	-
D-EPSC	Not tested	Not tested	Not tested	similar to mESC	increasing compared with mESC	Not tested	Not tested	Not tested
L-EPSC	Not tested	Not tested	Not tested	similar to mESC	No clear conclusion	Not tested	No clear conclusion	Not tested
2CLC	Middle, resembled mESCs, but not 2-cell	Middle247	Not tested	Higher level	Not tested	Not tested	similar to mESCs	similar to mESCs
TBLC	High, open peaks in 2-,4-cell	Low	Not tested
TLSC	High, open peaks in 2-,4-cell	Not tested	retain many oocyte/2C-specific bdH3K4me3 domains	similar to 2-cell	No clear conclusion	No clear conclusion	Not tested	Not tested
TPSCs	High, open peaks in 2-cell	Low	Not tested	Not tested	Not tested	Not tested	Higher levels	Not tested
TotiSC	High, open peaks in 2-,8-cell and ICM	Low	Not tested
Feature	In vitro culture					Medium Target
	Cell culture passages# in vitro	Resource	Karyotype	Self-renewal	Cell cycle	chemical compounds
mESC	Yes	ICM	Normal	Yes	G2	LIF
D-EPSC	P10	mESC blastocysts	Normal (P10, female)	Yes	Not tested	hLIF, CHIR99021 (Wnt agonist), DiM, MiH (PARP1 inhibitor)
L-EPSC	P10	mESC, iPSC, 4-/8-cell	Normal (P10, female)	Yes	Not tested	hLIF, CHIR99021(Wnt agonist), PD0325901(Mek1 inhibitor), VIII (JNK Inhibitor), SB203580(p38 inhibitor), A-419259(Src kinase), XAV939(TNKS1/2 inhibitor and stabilize AXIN)
2CLC	Transient	mESC	Not tested	No	Longer G2	No need
TBLC	P18	mESC	Normal (P12, male)	Yes	Not tested	Pladienolide B (splicing inhibitor)
TLSC	P21	mESC,2-/4-/8-cell	Normal (P21, male; P16, female)	Yes	Longer G2	SGC09469 (DOT1L inhibition), AS8351 (KDM5B inhibitor), A366 (G9a inhibitor), sIL-6R, IL6
TPSCs	P15	EPS, 2-cell	Normal (P15, male)	Yes	Not tested	CD1530 (RARγ agonist), VPA (HDAC inhibitor), EPZ004777 (Dot1L inhibitor), CHIR99021 (Wnt agonist)
TotiSC	P8	mESC	Normal (passage not indicated, male)	Yes	Longer G2	TTNPB (RARγ agonist), 1-azakenpaullone (GSK-3β inhibitor), WS6

In this table, the classifications of high/middle/low and yes/no are based on comparisons with mESCs.

^(-)means mESC serving as the baseline.

The subjective ratings (+) are employed to visually present the extent of the observed differences.

^#represents the highest passage mentioned in the article.

In summary, various methods, including spliceosome repression, epigenetic manipulation, and RARγ signaling modulation, successfully capture and culture mouse totipotent stem cells in vitro. Despite the diversity of these approaches, they interestingly converge to induce a similar totipotent state in cultured stem cells. This convergence points to the existence of complex regulatory networks crucial for establishing and maintaining a stable homeostatic state in totipotent stem cells. However, it remains unknown whether these approaches can be effectively translated to humans, as stable maintenance of human totipotent stem cells has not yet been achieved.

5. Assessing and Evaluating Emerging Totipotent Stem Cells: Challenges and Criteria

Although various mouse totipotent stem cells have been identified, challenges persist in fully understanding their totipotency and potential applications. A critical step is a thorough comparison and characterization of these distinct totipotent stem cells. For example, all the stably cultured cells closely resemble 2-/4-cell blastomeres and show induction of classical totipotent markers such as Zscan4, Gm8300, and Rab43, compared to mESCs. However, some still lack expression of other key totipotent genes, such as Zfp365, Mmp19, and Ddr2, particularly ciTotiSCs and TPSCs²²⁴ (Figure 4B and 4C). Although scRNA-seq analyses have been conducted on some of these cells, assessing the heterogeneity of different totipotent stem cells across passages remains complex and warrants cautious evaluation.

The growth rate of all reported totipotent stem cells so far appears slow, or at least not as fast as that of PSCs. This is particularly evident in ciTotiSCs, where certain totipotent markers are transiently induced only in early passages (1–4) and not at later stages, such as passage 8 (Figure 4B). This transient induction raises concerns about the stability of these cells over long-term culture, questioning the claimed self-renewal capabilities. Given that zygotes and totipotent blastomeres naturally lack proliferative capacity, the self-renewal of these in vitro totipotent stem cells might depend on artificially activated cell signaling pathways. An important question arises: can we manipulate specific pathways to achieve rapid and stable expansion of totipotent stem cells, akin to PSCs?

Researchers have started to uncover the epigenetic characteristics of totipotent stem cells, such as histone modifications, DNA methylation, parental imprinting, and X-inactivation. For instance, mouse TBLCs, TPSCs and ciTotiSCs exhibit notably lower DNA methylation levels compared to PSCs, contrasting with in vivo blastomeres that have higher methylation than blastocysts^221–225. Moreover, the 3D chromosomal structures in totipotent stem cells are relatively less compact, indicating that their organization is still being established. These distinct epigenetic features could lead to genetic and epigenetic aberrations, impacting the stability and developmental potential of these cells during prolonged in vitro culture.

Based on the stringent definition, a single totipotent stem cell should be capable of generating an entire conceptus. This high standard has not yet been met in the functional verification of reported totipotent stem cells. Currently, the functional proof of totipotency often involves assessing the cells' ability to contribute to extraembryonic tissues in chimeric mice in vivo. In this regard, scRNA-seq analysis emerges as a superior approach, enabling the identification of all embryonic and extraembryonic lineages and elucidating the developmental trajectories of various totipotent stem cells. However, it is important to note that false positives can be introduced during fluorescence-activated cell sorting of cells derived from the chimeric conceptus. Accurate identification of “true” chimeric contribution necessitates verifying the expression of genes (e.g. fluorescent marker genes) that are specifically present in the donor cells, a step sometimes overlooked in some research.

Moreover, the limitations of in vivo chimeric assays, particularly for human totipotent-like cells, make in vitro differentiation systems directed towards extraembryonic tissues both invaluable and essential for assessing totipotency. However, artificial manipulation of signaling pathways and inclusion of undetermined factors in cell culture can inadvertently induce transdifferentiation of the plastic hPSCs, leading to the formation of extraembryonic lineages. Such manipulations can skew the assessment of stem cell potency, potentially leading to incorrect conclusions. For instance, the reported generation of trophoblast stem cells (TSCs) from naïve^66,226–228 and primed²²⁹ hPSCs involves complex signaling manipulations and prolonged culturing, which could be seen as artificial transdifferentiation events that do not align with transient embryonic development in vivo. To effectively develop differentiation systems that can assess the developmental potency of totipotent stem cells, two key principles should be followed: firstly, the formation of extraembryonic trophectoderm or primitive endoderm lineages should occur within a time frame comparable to in vivo development, avoiding long-term culturing; secondly, external factors that could induce cell fate transdifferentiation should be minimized. Adhering to these principles is crucial to ensure that the differentiation potential into extraembryonic lineages truly reflects the inherent totipotent characteristics of these stem cells. In this regard, in vitro embryo-like structures, for instance blastoids, could be potential models to test the developmental potency of a single totipotent stem cell. It is noteworthy that although single EPSCs, aided by helper cells, have been shown to form blastoids²³⁰, there remains conflicting evidence on the capability of mouse EPSCs to generate blastoids on their own^230,231.

6. Emerging Concepts and Conclusion

Totipotent and pluripotent stem cells have been acclimated to artificial culture environments that differ significantly from their in vivo developmental niches. These cultures typically include supporting feeder cells, non-physiological biological matrices, excessive amounts of glucose and nutrients, and, notably for PSCs, a lack of interaction with extraembryonic cells. Recognizing these discrepancies, efforts have been made to optimize stem cell culture niches in vitro to better mimic in vivo conditions. These efforts have led to advancements in several areas, such as refining and identifying signaling pathways critical for the maintenance of totipotent and pluripotent stem cells, applying mechanobiological principles (such as substrate topography and stiffness), using 3D encapsulation and microcarriers²³², and co-culturing with extraembryonic stem cells²³³. The potential benefits of these improvements for stem cell differentiation remain largely unexplored. Surprisingly, a study revealed that S/L culture is more effective than other chemical-based mESC cultures (2i/L, a2i/L and LCDM) in producing complete mice through tetraploid complementation, with higher survival rates to adulthood. Mice generated using 2/L culture exhibit no visible abnormalities for up to 2 years, unlike those from extended chemical-based cultures, which develop atypical teratomas or leiomyomas²³⁴. This success may be linked to the downregulation of metabolic pathways in S/L-mESCs compared to those in chemical-based mESC cultures²³⁴. Therefore, optimizing PSC cultures to reduce energy activity could be key to preserving and prolonging their developmental potential in vitro.

Interestingly, primed and intermediate/formative cultures are generally more universally applicable compared to naïve and totipotent conditions, allowing for the maintenance of PSCs across a broader range of species. This universal applicability could be due to stabler environmental conditions experienced by intermediate/formative and primed cells after embryo implantation, as opposed to the uterine environments to which naïve and totipotent cells are exposed, which differ among species. In line with this observation, changes in the transcriptomic profile of primate epiblasts have been reported upon implantation. Following implantation, the epiblast enters a phase of relative transcriptomic stability, which persists for approximately five days^96,120.

Further studies have introduced an intriguing concept: some PSC cultures are effective in deriving and maintaining extraembryonic stem cells. For instance, one study showed that a defined medium containing Activin-A, CHIR, and LIF (ACL medium) enables the establishment of ACL-ESCs and ACL-Extraembryonic Endoderm (XEN) cells from a single blastocyst²³⁵. Additionally, another study found that an EPSC culture medium, LCDM, initially developed for mice and humans³², not only produced a similar type of PSCs in bovines²³⁶, but also supported the derivation and long-term culture of bovine TSCs²³⁷. Strikingly, the FTW medium, known for supporting the derivation of intermediate/formative FTW-ESCs from mouse and horse blastocysts³⁵, enables de novo derivation and culture of ESCs, TSCs, and XENs from mouse and monkey blastocysts²³³. This advancement allows for the co-culture of embryonic and extraembryonic stem cells and reveals a growth inhibition effect exerted by extraembryonic endoderm cells on pluripotent cells, partly mediated by extracellular matrix signaling²³³.

The development of diverse totipotent and pluripotent stem cell cultures has greatly advanced the field of mammalian embryo modeling, especially for humans. It has led to the generation of pre-implantation and post-implantation embryo-like structures that mimic different aspects of early human development^84,238–245. To create human embryo models from cultured hPSCs, two methods are used: either aggregating cells of embryonic and extraembryonic origins^243,244 or inducing differentiation and self-organization from a single PSC type^{84,241,242,245}. The latter approach potentially provides a superior assay for functionally evaluating human pluripotency than the pre-existing embryoid body and teratoma assays. For instance, the generation of human blastoids^238–240, peri-gastruloids²⁴² and gastruloids²⁴¹ from naïve hPSCs, hEPSCs, and primed hPSCs, respectively, serves as a functional assessment of their developmental potency, which facilitates the staging of these hPSCs along the developmental axis. The differentiation of epiblast-like cells in these models occurs in conditions that, while not identical, are more akin to natural embryonic environments than random in vitro differentiation. They offer valuable insights into defining the human epiblast developmental trajectory. The success of this exciting intersection of stem cell biology with embryology requires a deeper understanding of the self-organization capabilities of spatiotemporally distinct early embryonic cells. In-depth insights are now more attainable due to the increasing variety of culture adaptations.

While artificially maintained, the dynamic totipotent and pluripotent stem cells offer an invaluable resource for studying early mammalian development, focusing on cell potency and cell plasticity, environmental and experimental perturbations, and in vitro state transitions. These cells also hold great potential to transform the fields of regenerative medicine and reproductive biology.