Preface
The molecular mechanisms and signalling pathways that regulate the in vitro preservation of distinct pluripotent stem cell configurations, and their induction in somatic cells via direct reprogramming approaches, continue to constitute a highly exciting area of research. In this Review, we provide an integrative synthesis on recent discoveries related to isolating unique naïve and primed pluripotent stem cell states with altered functional and molecular characteristics, and from different species. We overview pathways underlying pluripotent state transitions and interconversion in vitro and in vivo. We conclude by highlighting unresolved key questions, future directions and potential novel applications of such pluripotent cell states.
Introduction
Pluripotency describes cells that have the potential to give rise to cells from all three embryonic germ-layers and primordial germ cells (PGCs, but not extra-embryonic tissues1. While pluripotency is a transient cell state in vivo, pluripotent cells can be derived from different stages of early embryonic development and indefinitely maintained in an artificially induced self-renewal state in vitro, by supplementing exogenous cues2. Thus, it is important to stress that self-renewal is not a defining feature of pluripotency and is only transiently assembled during early development. Pluripotency is highly dynamic and evolves at different stages of pre- and post-implantation stages3. However, the self-renewal aspect is a highly useful in vitro artificial “engineering trick”4 that has brought pluripotent stem cells to the front stage as a potential tool for tissue replacement, disease modelling and animal engineering technologies5.
There are multiple pluripotent stem cell types that can be isolated from vertebrates, including rodents and human, typically annotated based on their donor cell of origin (Fig. 1). Embryonic stem cells (ESCs) are isolated from the inner cell mass (ICM) of developing pre-implantation mouse or human blastocysts6,7. Epiblast stem cells (EpiSCs) are isolated from mouse post-implantation epiblasts8,9, however equivalent derivations have not been attempted with human embryos due to ethical complexities. Early rodent migrating PGCs can be converted in vitro into pluripotent ESC-like cells termed embryonic germ cells10. Mouse neonatal and adult spermatogonial stem cells can be reverted toward pluripotency and generate male germ stem cells (GSCs)11-14. The latter have the disadvantage of retaining only male imprint signature, which compromises their safety for clinical use13. Intriguingly, indefinitely stable and validated EGs and GSCs have not been isolated from primates thus far15,16 (Fig. 1).
Somatic cell reprogramming provides alternative routes for isolating pluripotent cell types. Human and rodent somatic cells can be artificially reprogrammed into ESC-like cells following reprogramming via nuclear transfer, termed NT-ESCs17-19. Ten years ago, Yamanaka established direct in vitro reprogramming of somatic cells to pluripotency via ectopic expression of defined factors20, that yield induced pluripotent stem cells (iPSCs) without the need for oocytes or embryos21-24 (Fig. 1). NT-ESCs and iPSCs offer the advantage for being able to generate patient specific pluripotent cells with genomic DNA identical to the donor somatic cell, however mitochondrial DNA in NT-ESCs is non-isogenic and provided by the a-nucleated donor oocytes25. The latter can be an advantageous in settings aiming at correcting maternally inherited mitochondrial diseases26-28.
While the above overview pertains to classify different pluripotent cell types based on their tissue derivation source, the growth conditions used to expand such cells dictate the pluripotent state they attain (ICM-like, ESC-like, EpiSC-like state etc.)29,30. iPSCs generated in classical mouse ESC growth conditions yield ESC-like iPSCs; while those reprogrammed in EpiSC growth conditions yield EpiSC-like iPSCs29,31. The same analogy applies to explanting rodent ICM cells in ESC or EpiSC growth conditions29,32. In comparison to developmentally restricted mouse EpiSCs, ESCs are highly competent in generating high-contribution chimeric mice after microinjection into host-blastocysts, retain a pre x-inactivation state in female cell lines and reduced expression of lineage commitment factors29,30. Such attributes influence the utility of pluripotent cells in cell differentiation, animal transgenics and molecular disease modelling. Thus it is of critical importance to understand and define different pluripotent states and configurations, and their implications on pluripotent cell differences across different species4,33.
In this review we provide an integrative perspective on recent breakthroughs in understanding the diversity and complexity of pluripotent state regulation in vitro. This includes advances on preserving naïve pluripotency from non-rodent species and alternative EpiSC pluripotent states. We highlight unresolved issues, key questions and future directions in this exciting front of stem cell research.
Murine pluripotent states
Mouse ESCs were thoroughly shown to reside in an ICM-like state29, now referred to as the naïve state of pluripotency30, since they retain many of the molecular characteristics of ICM. In 2007, Mckay and Vallier groups derived a novel type of pluripotent cells termed EpiSCs from post-implantation rodent epiblasts8,9. In comparison to naïve ESCs, EpiSCs retain an alternative pluripotency configuration, commonly referred to a primed pluripotency30. Dramatic molecular and functional differences exist between different pluripotent cells, which subsequently influence their characteristics, function and safety.
Growth conditions for naïve pluripotency
To fully grasp the biology of mouse naïve ESCs and their developmental context, it is of relevance to review the evolution of growth conditions devised to isolate such cells over the course of the last thirty years. Martin and Evans derived ESCs from 129-mouse strain7,34, by utilizing mitotically inactive embryonic fibroblast cells (MEFs) as feeder cells and foetal bovine serum (FBS). Leukaemia Inhibitory Factor (LIF) that activates the STAT3/JAK pathway was later identified as a key ingredient that allowed expansion of mouse ESCs in FBS/LIF conditions without MEFs35,36 (Fig. 2). Such naïve mouse ESCs express hallmark pluripotency factors (e.g. Oct4, Nanog, Esrrb) and retain a pre-x inactivation state in female cell lines37 (Fig. 3). Functionally, ESCs can populate the host pre-implantation mouse ICM upon microinjection, and generate high-contribution chimeras with colonization of the germ-line niche38,39.
Smith group described the first serum- and feeder-free defined conditions for expanding mouse ESCs by combining low dose BMP4 with LIF40. Addition of small molecule inhibitors for MEK signalling, increased ESC derivation efficiency and stability41. Developing different defined conditions, all involving MEK inhibitors, which can be used to isolate murine ESCs, extended the latter observations. A cocktail combing 3 inhibitors termed ‘3i conditions’, was shown to stabilize pluripotent cells without LIF, indicating existence of redundant pathways to isolate ESCs in vitro that can compensate for lack of LIF/Stat3 signalling42. Notably, such cell configuration was labelled as “ground state pluripotency” since the cells in 3i were reported to grow independent of any exogenous signalling stimuli. However, this term is challenged by the fact that growth in these conditions heavily relies GSK3 inhibition that mimics WNT stimulation; and on exogenous insulin that activates PI3K/AKT signalling42. Further, 2i/LIF conditions were adopted as an enhanced mean to expand murine ESCs, and the reduced proliferation in 3i conditions indicates a role for autocrine secreted FGF4 in promoting naïve ESC growth43 (Fig. 2).
“Alternative 2i” conditions, involving small molecule inhibitors for Gsk3 and Src pathways, have been shown to yield germ-line competent ESCs44 (Fig. 2). Paul and colleagues identified atypical PKC small molecule inhibitor Go6983 (aPKCi) as another stimulator for isolating murine ESCs together with LIF and/or MEK inhibitors45. Single cell RNA-seq analysis has proven equivalent global heterogeneity between different naïve conditions, however the difference exists in the identity of genes that underlie heterogeneity in each condition46.
Enriched conditions were important for deriving ESCs from other mouse strains that have until recently been considered “non-permissive” for deriving naïve ESCs. While 129-mouse strain derived ESCs could be expanded in FBS/LIF conditions, for other mouse strains like non-obese-diabetic ICR mice (NOD), supplementation of 2i or GSK3 inhibitors is essential for both derivation and maintenance29. 3i and 2i/LIF conditions yielded rat ESCs, however these conditions are suboptimal47,48. LIF/MEKi/aPKCi have been recently described as a more robust method to support rat ESCs in feeder free conditions49 (Figure 2,3).
The above discoveries underscore the relevance of analysing rodent ESCs expanded in different naïve conditions and from different genetic backgrounds. Further they emphasize the importance of other signalling pathways that remain to be deeply characterized in the context of pluripotency (Fig. 2). NF-kB inhibition has been implicated as a downstream effector underlying aPKC inhibition mediated support of naïve pluripotency45, however other pathways, like Mbd3/NuRD repressor complex, can be neutralized by aPKCi (unpublished observations, J.H.H). SRC functions as a downstream target of ERK and Calcineurin-NFAT signalling to promote ESC differentiation50, and its inhibition promotes naïve pluripotency44,50.
The Hippo signalling pathway has been uncovered as cardinal regulator acting at epiblast vs. trophoblast segregation in late mouse morulas. The Hippo pathway is highly active in pluripotent epiblast cells, leading to exclusion of Yap/Taz effectors from the nucleus51,52. Piccolo and colleagues indicated that Yap/Taz depletion in mouse naïve ESCs expanded in 2i/LIF prevents their differentiation52, while Lian et al. have indicated Yap/Taz as essential regulators for stability of naïve ESCs expanded in FBS/LIF conditions 53. Systematic analysis of these findings in different defined conditions might resolve these seemingly opposing results (Fig. 3).
It should be noted that signalling pathways are often pleiotropic and may simultaneously have positive and negative effects on naïve pluripotency. For instance, nuclear β-catenin stabilization following Gsk3 inhibition, promotes naïve pluripotency via neutralizing the repressive activity of Tcf3 on its bound target genes in the nucleus54. Cytoplasmic β-catenin promotes naïve pluripotency via increasing E-cadherin membrane stability55 (Fig. 2). However, nuclear β-catenin can induce mesodermal gene expression through its Lef co-effectors47. Yet, such differentiation priming effects are outweighed by naïve pluripotency promoting functions of β-catenin under optimized conditions. LIF has also been shown to promote primitive endoderm specification in naïve pluripotency growth conditions56. Such “non-purist” effects should be kept in mind when dissecting the role of signalling pathways on pluripotency.
The ample conditions to grow naïve murine ESCs have been important for better understanding and revisiting the roles of several classical pluripotency regulators. While Nanog was first purported to be absolutely essential and irreplaceable for establishing naive pluripotency through iPSC reprogramming, cell fusion or EpiSC reversion57, multiple conditions enable reprogramming of Nanog null donor cells in vitro58,59. Still however, Nanog null ESCs cannot be derived from mouse ICMs, indicating that while Nanog is indispensible for establishing pluripotency in vivo, it is dispensable during in vitro induction and maintenance by alternative pluripotency regulators like Esrrb and enriched growth conditions60. The latter example highlights that in vitro pluripotency maintenance and induction can not be considered “authentic”, as some in vitro conditions can clearly potentiate the robustness of the naïve pluripotency program and compensate for deficiencies that are not sustainable in vivo. Similarly, Klf2 knockout embryos do not present lethality at the pre-implantation stage, and naïve ESC in FBS/LIF conditions can tolerate Klf2 ablation61. However ESCs in 2i only conditions, can no longer tolerate loss of Klf261, and LIF can compensate for the lack of Klf2.
Another emerging regulatory principle is that not all factors expressed in the ICM or ESCs necessarily promote naïve pluripotency, and some of them in fact are promoting its dissolution. However, they are tolerated by ESCs in vitro due to the optimized and enriched in vitro growth conditions used. For instance, Tcf3 binding represses the expression of its naïve pluripotency promoting target genes, leading to their partial repression, is tolerated in serum/LIF naïve conditions. However, Tcf3 neutralization by adding GSK3i boosts naïve pluripotency54. In a similar manner, mouse ESCs tolerate Mbd3/NuRD complex expression although it partially represses naïve pluripotency targets like Esrrb and Tfcp2l162. However, genetic ablation of Mbd3 leads to upregulation of master regulators of naive pluripotency and allows LIF independent growth62. Consistently, derivation of Mbd3 KO ESCs from null ICMs is uncompromised in 2i/LIF conditions63. In summary, both Tcf3 and Mbd3 are expressed in the ICM and ESCs likely to set the stage for terminating the naïve pluripotency program. Thus the molecular characteristics of a pluripotent state fixated in a certain condition represent the net outcome of conflicting stabilizing and destabilizing factors simultaneously residing and conflicting in that state64.
Collectively, it is important when analysing function of pluripotency regulators to systematically compare different naïve growth conditions, genetic backgrounds and in vivo context65. Such integrative analysis will likely unravel additional layers of underappreciated complexity and may resolve some conflicting results that stem for conducting analysis on pluripotent states with different molecular flavours52,53.
Growth conditions for primed pluripotency
Primed EpiSCs were derived from post-implantation epiblasts of a variety of rodent strains in FGF2/Activin A conditions8,9 (Fig. 1). EpiSCs are capable of differentiating into cells of all three germ layers in vitro or in teratoma assay, and thus are pluripotent. However, they are inefficient in yielding chimeric animals once injected in pre-implantation epiblasts (Fig. 3), likely because they have altered molecular characteristics and correspond to a more advanced developmental stage in comparison to the host pre-implantation environment8,9.
While EpiSCs maintain Oct4 and Sox2 expression, they down-regulate most of the other pluripotency factors including Nanog, Esrrb, Klf2 and Klf43. EpiSCs have not undergone differentiation, but they upregulate lineage commitment factors like Otx2, Brachyury and Zic266. Epigenetically, EpiSCs retain distinct characteristics from naïve ESCs: they inactivate X chromosome in females, upregulate global DNA methylation levels and acquire H3K27me3 at developmental regulators67,68. Enhancer landscape and activity is rewired between naive and primed states66, and developmental regulator gene associated “seed enhancers” convert from a dormant to an active state in EpiSCs, thus pre-marking differentiation tendency of primed PSCs69. Summary of divergent signalling and molecule characteristics between murine primed and naïve cells are highlighted in Fig. 2-3.
At the regulatory level, naïve and primed pluripotent cells have been shown to retain opposing dependence on epigenetic repressors65 (Fig. 4). Naïve ESCs tolerate loss of epigenetic repressors like Dnmt1, Dicer, Dgcr8, Eed, Mbd3 and Mettl365, and in fact renders these cells “hyper–naïve” and resistant to differentiation65,70,71 (Fig. 4). On the contrary, murine primed pluripotency maintenance and viability depends on these regulators, and their ablation destabilizes the murine primed pluripotent state65 (Fig. 4). Defining how the depletion of each of these repressors precisely destabilizes the primed configuration is of future interest.
Alternative growth conditions to expand murine EpiSCs have begun to emerge. Elegant work by Ying and colleagues SHOWED THAT simultaneous use of a GSK3 inhibitor (that induces β-catenin stabilization), together with a Tankyrase small molecule inhibitor IWR1 (that up regulates Axin1/2 levels thus leading to retention of β-catenin in the cytoplasm) maintain novel primed EpiSCs without exogenous Fgf2/Activin A supplementation72 (Fig. 3). Removal of IWR1 leads to increased nuclear β-catenin shuttling and EpiSC differentiation72. The mechanisms by which cytoplasmic β-catenin prevents EpiSC differentiation remains to be uncovered72. It is tempting to speculate whether the recently described ability of cytoplasmic APC/Axin/β-catenin destruction complex to act as a sequestration “sink” for Yap/Taz and prevent their nuclear shuttling52, is involved in the ability of GSK3i/IWR1 conditions to maintain EpiSCs. Notably, the latter alternative EpiSC state is different from EpiSCs expanded in classical Fgf2/Activin A conditions, and retains higher expression of naïve markers like Dppa2, 4, 572, and are thus relatively “less primed” (Fig. 3).
Recent studies indicate that different primed conditions used can endow EpiSCs with region specific characteristics of the post-implantation epiblasts. Tam and colleagues showed that EpiSCs expanded in Fgf2/Activin A correspond transcriptionally and functionally to anterior late-gastrula primitive streak cells73. Belmonte and colleagues showed that alternative FGF2/IWR1 conditions generate murine EpiSCs corresponding rather to posterior-proximal epiblasts74. Further, even in classical Fgf2/Activin A conditions distinct subpopulation of EpiSCs can co-exist, each representing different stages of post-implantation embryonic development75.
Finally, the time length at which pluripotent cells are maintained under primed conditions greatly influences their characteristics and functionality76. Counter intuitively; while murine PGCs are specified from the post-implantation epiblast in vivo77, EpiSCs maintained in vitro for more than 7 days in FGF2/Activin A, lose competence to generate PGCs in response to Bmp476. Starting with naïve cells and inducing priming for no longer than 2-4 days, yields distinct primed cells highly competent for generating PGCLCs, termed EpiLCs76. The latter are transcriptionally more similar to in vivo post-implantation epiblast than EpiSCs76. Thus, the above paradigm indicates another aspect of artificial features that can be acquired by pluripotent cells once expanded indefinitely in vitro, in contrast to their in vivo “counterparts” that transiently exist during development.
Studies involving clonal lines and single cell analysis will be key for deeper understanding of features of region specific EpiSCs and shortly after in vitro induction from a naïve state in different priming conditions73. This may help understand how lineage priming is established at the single cell level during these key early developmental transitions73,74 and might be relevant for optimizing other differentiation protocols and predicting PSC behaviour.
Interconversion between naïve and primed states
As somatic cells can be reprogrammed into a naïve ESC-like state via combined overexpression of pluripotency factors together with LIF, primed EpiSCs can also be reverted to naive iPSCs. Overexpression of Klf4 or Myc in EpiSCs, under LIF containing conditions, generates naïve ESCs29,78. FBS/LIF signalling alone can be sufficient to induce such conversion from permissive mouse genetic backgrounds (i.e. 129 strains)79, but not from “non-permissive” ones like NOD, where supplementation of small molecules like 2i is necessary29. Other factors like Nanog, Prdm14 and Esrrb have been shown to synergistically induce and boost the efficiency of this process80,81. Explanting post-implantation E5.5-E7.5 epiblasts in naïve conditions also reverts them into naïve PSCs29,79. The opposite conversion is attainable from in vitro and in vivo isolated naïve cells, as expanding murine naïve PSCs or ICMs in primed conditions leads them to gradually adapt an EpiSC state29,78,82.
Studies focusing on in vitro molecular changes accompanying naïve to primed pluripotency conversion have unravelled key events in understanding mechanisms of reprogramming67. Naïve ESCs expanded in 2i/LIF retain global hypo-methylated levels in both promoters and gene bodies, highly similar to those measured in ICMs83,84. When transferred into LIF/FBS naïve conditions, this is accompanied by an increase in global DNA methylation levels, however promoter and enhancer regulatory regions remain protected from invasion by DNA methylation85. Only after transfer into primed Fgf2/Activin A EpiSC inducing conditions, DNA methylation accumulates over enhancer and promoter regulatory elements85.
Transitioning primed EpiSCs or naïve FBS/LIF PSCs into 2i/LIF conditions leads initially to dramatic changes in Oct4, Sox2 and Nanog pluripotency factor occupancy86. Changes in H3K27me3 deposition and enhancer landscape follows only later, likely in response to the rewiring in transcription factor binding86. Downregulation in DNA methylation follows next, which has mainly been attributed to downregulation in de novo DNA methyltransferase enzymes83. It should be noted however, that ablation of Dnmt3a/b in ESCs in FBS/LIF condition does not lead to such rapid loss of DNA methylation87, and other yet to be identified events might be involved in this rapid 2i induced epigenetic response. MEK/ERK inhibition influences polycomb interactions and leads to decreased occupancy of PRC2 and RNA PolII at developmental regulatory genes88, leading to loss of H3K27me3 and increased PolII pausing at bivalent developmental regulatory genes67.
Analysis of other molecular changes in naïve to primed transitions is likely to increase our knowledge of pluripotency regulation and lineage priming. Analysing other defined naïve pluripotency growth conditions (2i/LIF/PKCi, alternative 2i etc.) and in other rodents, will be important for discerning redundancies and specificities of different singling pathways and how they cross-talk with chromatin organization.
Human conventional pluripotent cells
Thomson group first isolated human ESCs in 1998 from bastocysts6. Surprisingly, they were drastically different from murine ESCs in their characteristics and tissue culture requirements. FGF2 and TGFβ1, but not LIF, signalling are at the core signalling modules maintaining such conventional human ESCs derived from the ICM, or iPSCs obtained via direct in vitro reprogramming89.
A primed pluripotent state
Differences between conventional human and mouse ESCs had been initially attributed only to unknown species genetic differences, since human ESCs were also derived from the ICM and not from post-implantation stages. However studies on different mouse strain derived stem cells have discerned a scenario where ICM cells can adapt in vitro into a primed state if naïve conditions that match the requirements of the particular genetic background of donor embryos used, are not devised29. Specifically, NOD mice are relatively “less-permissive” than 129 mice to yield naïve ESCs/iPSCs, as LIF alone is not sufficient to maintain NOD naïve pluripotency and 2i/LIF are permanently required to stabilize and maintain this state in vitro in NOD PSCs29. Further, ICMs from both 129 and NOD strain expanded in primed conditions yield EpiSCs that are indistinguishable from EpiSCs derived from E6.5 embryos29 or in vitro from already established ESCs29.
The relevance of the latter in vitro priming scenario to dictating conventional human ESC identity is supported by the fact that human conventional ESCs/iPSCs retain a great milieu of primed pluripotency features. This includes low expression of naïve pluripotency markers (e.g. KLF5, TFCP2L1, DPPA3), deposition of H3K27me3 over developmental genes, lack of exclusive nuclear localization of TFE3, loss of pluripotency upon inhibition of MEK/ERK pathway, lack of global hypomethylation as seen in ICM cells, lack of a pre x-inactivation state in most conventional female PSCs lines68,90,91. Further, human primed ESCs do not tolerate complete loss of DNMT187, similarly to what has been shown for mouse EpiSCs65 (Fig. 4). Complete human KO ESCs have not been obtained thus far for DICER, MBD3 or METTL363,65 (Fig. 3).
Less primed than murine EpiSCs
In spite of the above, it is of critical importance to realize that human conventional/primed ESCs are not identical to murine EpiSCs, and can be considered relatively “less primed”. For instance, human ESCs do not upregulate FGF5 and N-CADHERIN as seen in murine EpiSCs, and express high-levels of E-CADHERIN as detected in mouse naïve ESCs68. Human ESCs express high levels of some naïve markers like NANOG, PRDM14 and REX1 that are not expressed in murine EpiSCs92. Moreover, human primed ESCs are functionally dependent on NANOG and PRDM14 and their ablation induces differentiation92. DNA methylation distribution in human ESCs shows they rather correspond to murine naïve ESCs expanded in FBS/LIF conditions, rather than mouse FGF2/Activin A expanded mouse EpiSCs, as their promoters are protected from invasion by repressive DNA methylation85,93. Further, while murine EpiSCs demonstrate exclusive TFE3 cytoplasmic localization and naïve 2i/LIF ESCs show exclusive nuclear TFE3 localization94, human primed ESCs show an intermediate configuration where TFE3 is present in both the cytoplasm and the nucleus68.
Human naive pluripotent cells
The metastability of naïve and primed pluripotent state depending on the growth conditions applied29, and the stringency in requirement for exogenous naïve pluripotency promoting factors to isolated naïve PSCs from previously “non-permissive” rodent strains29,30, have underscored a scenario of whether unique and more stringent conditions can be applied to isolated previously unidentified alternative naïve-like pluripotent states in humans.
Transgene-dependent generation
2i/LIF conditions are not sufficient to maintain naïve human ESCs or iPSCs95. However, additional transgene expression can induce an artificial transgene dependent state that may be of considerable interest. Continued exogenous OCT4/KLF4 or KLF2/KLF4 transgene expression can maintain human ESCs/iPSCs in a unique pluripotent state in 2i/LIF conditions95. Recently, these observations were extended by optimizing over-expression of KLF2/NANOG transgenes, allowing expansion of human naïve iPSCs in 2i/LIF96. Smith and colleagues overexpressed KLF2 and NANOG transgenes in primed ESCs and expanded them in 2i/LIF/aPKCi conditions97 These cells exhibited more extensive DNA demethylation, and strong upregulation of naïve markers like TFCP2L1, KLF2, and KLF4. However, as KLF2 is not expressed in the human ICM98, and as 2i/LIF/aPKCi are insufficient to convert primed ESCs without exogenous transgene induction97, and as transgene-free cells remain to be validated in 2i/LIF/aPKCi conditions, it is unclear whether this state is indefinitely stable without retaining leaky transgenes and/or MEFs as claimed by the authors. Further, independent examination of DNA methylation landscape in reset cells indicates aberrant global loss of imprinting, excessive hypomethylation of endogenous retroviral genes90,99. Finally, while these growth conditions do not contain exogenous FGF or TGFβ1/Activin A cytokines, applying short-term inhibition of FGFR/TGFR signalling is not sufficient evidence to validate FGF/TGF/Activin A signalling independence97 (Fig. 2).
While the field has shifted to study transgene independent conditions as detailed below, it should be noted that such transgene dependent states might be important, since it may be possible that robust naïve pluripotency currently obtained in mouse ESCs is a rodent specific phenomenon. Capturing human naive PSCs identical to those obtained from mice might still involve genetic modifications. Nevertheless, the latter studies provide evidence for the possibility to generate naïve-like pluripotent states in humans and other species29,95.
Transgene independent generation
Our team was the first to describe naïve conditions, designated as NHSM conditions, which entail complete ablation of MEK/ERK signalling and are compatible with indefinitely expanding genetically unmodified human PSCs both in MEF-containing and -free conditions68. These naïve MAPK independent pluripotent cell lines could be derived from human pre-implantation embryos, through de novo iPSC generation, or from previously established primed ESCs/iPSCs68. NHSM conditions contain 2i/LIF together with P38i, JNKi, aPKCi, ROCKi, low doses of FGF2 and TGFβ1 (or Activin A), and render human PSCs more similar, but not identical, to murine naïve PSCs68 (Fig. 2,3). In fact these cells have features of so called naïve 2i/LIF “ground state” pluripotency, which are not found even in naïve mouse ESCs expanded in FBS/LIF. This includes exclusive nuclear localization of TFE3, and cleansing of H3K27me3 over developmental genes67,68,94. Transcriptionally, these cells down regulated expression of lineage commitment markers like OTX2, ZIC2 and CD24 and moderately upregulated pluripotency genes (more prominently on MEFs)68,100. Enhancer rewiring has been attained in these human naive PSCs as seen with mouse cells69. The cells exhibited downregulation in DNMT3B101 and a mild global decrease in global DNA methylation levels, while maintaining imprinting integrity and chromosomal stability68. While conducting chimeric analysis with human PSCs and using human embryos as hosts is ethically and legally forbidden, these human naïve PSCs cells showed better integration upon microinjection into host mouse morulas and were able to contribute at low-grade levels in mouse embryos up to E17.568 (Fig. 3).
Important publications describing alternative conditions that yield human ERK independent naïve pluripotent cells emerged soon after, each producing cells with different enhanced molecular properties (Figure 3). A combination of 2i/LIF, ROCKi, BMPRi, high doses of FGF2 and TGFB1 were able to maintain human PSCs only in the presence of MEFs102. These cells demonstrated transcriptional upregulation of pluripotency markers like STELLA and KLF5. Jaenisch team described conditions96 that adopted most of components found in NHSM68 (i.e. 2i/LIF, ROCKi, Activin A (instead of TGFβ1) - with or without FGF2 and JNKi) and supplemented inhibitors for BRAF and SRC pathways (conditions termed 5i/LA-MEF with optional inclusion of JNKi or FGF2). In comparison to the previous two studies, cells in 5i/LA-MEF conditions demonstrated a more impressive upregulation of naïve pluripotency markers. However, the cells did not down regulate DNMT3B, maintained an inactive X chromosome state in female cell lines, and demonstrated an unusual expression pattern of early pre-ICM genes96. Intriguingly, the process of converting primed cells back to naïve state in 5i/LA-MEF conditions is inefficient, taking two weeks to isolate initial clones that retain a very slow growth rate96. Further, these conditions exclusively yield chromosomally abnormal cell lines96. Thus, it remains to be determined whether such chromosomal abnormalities are in fact inherent to 5i/LA-MEF cells and dictate the properties described for this state96, and are being selected for during this inefficient conversion process. Finally, DNA methylation profiling and whether these cells maintain epigenetic imprinting integrity is another important aspect that remains to be evaluated.
It is clear from the above summaries that none of the many already published conditions generate human naïve PSCs that are identical to mouse ESCs or human icm98,103,104. However, these studies implicate new signalling pathways and pave avenues for further optimization and characterization of such novel PSCs (Fig. 2). Mechanistically it will be interesting to test whether there is a connection between RAF and aPKC inhibition. Further, RHO signalling has been shown to promote YAP/TAZ nuclear localization in primed human ESCs and sustain their pluripotency105. Thus it remains to be defined whether ROCKi influencing naïve pluripotency characteristics68 via YAP/TAZ modulation.
The role of FGF2, ACTIVIN A, TGFβ1 signalling, either autocrine or exogenously provided at low doses, remains to be understood in human naïve PSCs. The latter demonstrate upregulation of Activin like ligand GDF397, and human, but not mouse, ICM cells abundantly express Activin receptors98. Thus, it is tempting to speculate that primed human ESCs are relatively less primed than murine EpiSC due to differences in response to Activin-like ligands, where it might promote some naïve features in human68,96, but not in mouse. To conclude, systematic analysis of the response of pluripotent states from different species to a variety of TGF ligand family members is of importance (Fig. 2), while the possibility to generate human PSCs that are entirely independent of FGF and/or TGF signalling cannot be excluded.
Differences between mouse and human epiblasts
Recent studies focusing on single cell RNA-seq of human pre-implantation embryos are starting to provide answers to some of the questions highlighted above. While human and mouse blastocyst do not display morphological differences, they retain profound molecular differences at the cellular level98 (Fig. 5a,b). Human ICM epiblast cells do not express genes that are considered important pluripotency factors in mouse, such as KLF2 and ESRRB. Instead, KLF17 might have a human-specific role in the ICM98. ERAS, an ESC specific form of RAS that is constitutively active, is not expressed in humans, as it became a pseudo-gene106. Eras null mouse ESCs propagate slowly in FBS/LIF conditions106, and whether they can be expanded in 2i/LIF conditions remains to be defined.
At the post implantation stage, dramatic differences exist between rodent and human embryos (Fig. 5a,b). Rodents are rather unusual as their post-implantation epiblast assumes egg-like cylinder shape, while in humans the post-implantation epiblast assumes a flat disc shape, like in most other mammals107. While it might be impossible to conduct single cell analysis on early human post-implantation epiblasts, non-human primates might provide some relevant insights. Collectively, these species differences might directly influence the distinct pluripotent characteristics observed in PSCs from different species in vitro and their distinct growth requirements. Further, they are of relevance for understanding the developmental context of human in vitro isolated pluripotent cells.
A framework for classification of pluripotent states
The advent of different conditions to isolate human naïve PSCs with distinct characteristics, and the limitations in conducting chimeric analysis in humans, simulate discussions and perspectives on classifying pluripotent states. It is often claimed that ability to derive ESCs for human ICMs in a newly devised growth condition constitutes “a gold standard” for proving naivety108. However, it should be kept in mind that the pluripotent state identity is eventually dictated by the derivation growth condition and not by whether their source was from the pre- or post-implantation epiblast29,82. Utilization of OCT4 distal vs. proximal enhancer element as a binary distinguishing marker can be also misinterpreted96. Both OCT4 distal and proximal enhancer elements are active in naïve and primed states, both in humans and mice109,110. The difference rather emerges from their relative activity levels (high/low) and dominance.
Relying on a single attribute marker or functional test is limiting and must be accompanied by systematic analysis of ever increasing characteristics that continue to be uncovered for different pluripotent states (Fig. 3,4). Nevertheless, in our opinion, a molecular and functional characteristic that can be considered as a major divider between naïve vs. primed pluripotent states, is the response of a certain pluripotency circuitry to inhibition of MEK/ERK signalling (Fig. 5c). Human conventional ESCs and mouse EpiSCs rapidly collapse upon MEK/ERK inhibition, while naïve pluripotent cells rather tolerate and consolidate their naivety following this challenge95. Emphasizing this molecular and functional quality is supported by the ability of MER/ERK inhibition to expand murine epiblast in ICMs and signifies consolidation of naïve pluripotency in vivo111.
Within the domains of the naïve and primed ground states of pluripotency, it is clear that if one considers a list of many naïve and primed pluripotency features originally described for mouse naive 2i/LIF and primed FGF2/Activin A, different pluripotency growth conditions can simultaneously endow a mix of primed and naïve properties in the same cell type (Fig. 3). As such, pluripotent states can be classified as “more naïve” or “more primed” by having more of such properties (Fig. 6). Human primed ESCs have a number of naïve pluripotency features (e.g. protection of promoter regions from hypermethylation, dependence on Nanog). Murine naïve ESCs expanded in FBS/LIF conditions can give rise to “all-ESC” chimeric embryos and tolerate Dnmt1/Mettl3 ablation, however, they are globally hypermethylated and acquire H3K27me3 over developmental genes as seen in EpiSCs (Fig. 3,5c).
Other functional tests that can be used to assess the stringency and extent of naivety in different primate naïve PSCs, is whether the cells can tolerate complete ablation of epigenetic repressors like METTL3, DNMT1, DGCR8 and MBD34,65 (Fig. 3,4). Further, such tests might be useful for optimizing conditions that close the gap between mouse and human naive pluripotent cells isolated thus far (Fig. 4, Box 1). Collectively, it will be informative to systematically annotate different naïve and primed sates from many different species isolated thus far according to such criteria (Fig. 3).
Box 1 A ’dark side’ of naïve pluripotency?
With the development of naïve conditions and efforts to endow these cells with more features of naivety, one question that emerges is “how much naivety is needed and is there a dark side for permanently maintaining PSCs under naive conditions?”
Rodent ESCs expanded in 2i/LIF have an increased tendency for acquiring genomic abnormalities43, and it remains unclear whether they occur as a by-product of non-specific activity of small molecule inhibitors used96,117, or as a direct result of intrinsic molecular features of naïve pluripotency (e.g. increased ERV activity, reduction in epigenetic repressive marks). One can envision a scenario where such features can be tolerated in vivo where this configuration exists for only 1-2 days, while prolonged in vitro expansion of this state increases the occurrence frequency of such unwanted damaging events.
The latter concern may also relate to safeguarding the integrity of DNA methylation and imprinting in naïve PSCs expanded in vitro over an extended period of time. Studies focusing on loss of DNA methylation upon transfer of mouse PSCs into 2i/LIF conditions have quantified methylation after 10-24 days of transfer, and have documented rapid global loss of DNA methylation accompanied with relative resistance of retrotransposons and imprinting regions to such demethylation84. However, it is unclear whether that latter described methylation state represents a final plateau that naive cells achieve, or whether carrying out 2i/LIF cultures further eventually leads to erosion in the relative resistance of such regions to demethylation. Indeed, methylation over imprinted genes and retrotransposons are partially, yet significantly, reduced in 2i/LIF conditions84.
If such effects are overwhelmingly frequent, researchers will have to re-evaluate how to optimize applying naive conditions. One scenario might involve decreasing inhibitor levels as a mean to avoid excessive hypomethylation or other unwanted effects. An alternative scenario is to maintain cells in primed and transfer them into naive conditions only for short time before initiation of differentiation.
Implications and future directions
Yamanaka’s breakthrough of reprogramming somatic cells to pluripotency has provided the foundation for deeper sleuthing of pluripotent states and the understanding that pluripotent configurations can be reconfigured. The latter, feeds back and bares direct influence on issues associated with current hurdles and limitations related to human iPSC quality and characteristics (Box 2).
Box 2 Potential implications on iPSC reprogramming.
Recent studies have implicated how certain epigenetic regulators in fact have opposing effects on naïve and primed murine PSC maintenance65. These findings might be relevant when comparing induction of pluripotency mechanisms in human vs. mouse, as human, but not mouse iPSCs, are typically reprogrammed in conventional/primed pluripotency conditions118. Consequently, some of the differences observed between human and mouse iPSC regulators might not be related to species differences but rather to the fact that distinct pluripotent states are being induced. Therefore, it would be imperative to expand human iPSC reprogramming regulator screens, and include different pluripotency conditions, as they might yield different outcomes.
Another implication of pluripotent state characteristics on reprogramming is whether naïve conditions might improve the quality of obtained iPSCs. It should be noted that, residual epigenetic memory involving DNA methylation in human iPSCs appears to be rather permanent even after extended cell passaging, while in naïve murine iPSCs, epigenetic memory was more reminiscent of incomplete reprogramming and faded after a few passages in vitro119. Thus, it should be tested whether the use of adequate naïve conditions might resolve such epigenetic memory in human iPSCs. Similarly, subtle epigenetic differences in DNA methylation found between NT-ESCs and iPSCs generated from the same human donor cells120 might also be neutralized when deriving iPSCs in naïve conditions that mimic more closely the epigenetic features of ICM.
Perhaps one of the most fascinating questions related to the naïve to primed pluripotency continuum is “why do these divergent pluripotent configurations actually exist? The latter is often accompanied by the reductionist question of “Which cells are better to work with - naïve or primed?” In our opinion, as this phenomenon is deeply rooted in early embryonic development in vivo, it is likely that both configurations constitute essential and integral parts for safeguarding optimality and maximizing the benefits of multi-potency and lineage specification simultaneously. We hypothesize that naive pluripotency emerged as an epigenetic erasure state that renders pluripotent cells free from lineage and epigenetic restriction, while simultaneously making these cells relatively less responsive to signalling pathways that might interfere with establishing such a lineage neutral state. After resetting, the induction of specification by morphogens may not be efficiently forced without a short “delay period”. As such, the naïve pluripotency network is gradually resolved and becomes more receptive to inductive cues at the post-implantation stage, and PSCs get differentially patterned/primed according to their spatial localization, before overt somatic differentiation occurs.
At the functional level, it remains to be thoroughly established whether using human naïve PSCs as a starting material, with or without a brief priming, would resolve hurdles currently faced in human in vitro PSC differentiation protocols: 1) Will human naïve PSCs yield increased consistency in differentiation outcome among independent iPSC lines112? 2) Can naïve PSC conditions yield better quality cells in different differentiation protocols when used as a starting material? 3) Can human naïve PSC facilitate the success of differentiation protocols that have not been conductive with conventional human PSCs? Encouraging support for the latter has recently been provided by showing the enhanced ability of human PSCs expanded in NHSM conditions (even in the absence of aPKCi) for undergoing in vitro differentiation into PGCs, a protocol that was inefficient from primed human PSCs113. The molecular rationale for evaluating the potential benefits highlighted above is that naïve pluripotency is more associated with cleansing of epigenetic repressive marks over regulatory regions, compared to primed cells68,97. This might enable more adequate activation of lineage specifiers during differentiation. Further, lineage biases in human primed PSCs are heavily associated with localized accumulation of repressive marks like DNA methylation114.
The recent advances in generating human naïve PSCs will continue to boost attempts to generate naïve-like PSCs from other species, and test same-species and inter-species embryo chimerism assays. Cynomolgus monkey naïve ESCs have been derived in NHSM conditions supplemented with Vitamin C, and gave rise to the first chimera competent ESCs, which following ICM-microinjection, yielded chimeric monkey foetuses115. Developmentally advanced mouse embryos (E10.5-E17.5) with low chimerism levels were obtained following injection of naive human68 or monkey iPSCs116. These observations raise a variety of exciting challenges relating to defining what are the frequency, lineage preference and developmental quality of such integrated primate iPSC derived cells. Systematic efforts will be key to conclude whether humanized animal models might become relevant for disease modelling, studying human development or generate transplantable human organs.
Continued breakthroughs in single cell technologies and applying them on different pluripotent cell types and embryonic samples will facilitate defining properties that are relevant for adequate functionality of PSCs. This will help set standards for desired optimal starting material for stem cell based therapeutics and research (Box 1). It is expected that during this journey aiming at allowing scientists to better control cell fate by deconstructing this previously underappreciated complexity of pluripotency, proposed criteria and standards will likely be debated and revised.
Display items (all will be subject to graphic editing by NRMCB after review)
Copyright permissions
- Figure 2,3,5a-b: Graphic figures are reused from our Nature Reviews Molecular Cell Biology Poster (http://www.nature.com/nrm/posters/pluripotency/index.html) as indicated in figure legends.
- Figure 4 is modified from Geula et al. Science 2015 (Supplementary Figure S34).
- Figure 5c is reused and modified from our Figure 3 in Manor et al. Curr. Opin. In Genetics & development
Further information
Online Web link sites:
Addgene plasmid repository: https://www.addgene.org/
ENCODE project: http://www.nature.com/encode/
Epigenome Roadmap project: http://www.nature.com/collections/vbqgtr
Mouse ES cell ChIP compendium: http://bioinformatics.cscr.cam.ac.uk/ES_Cell_ChIP-seq_compendium.html
CRISPR/CAS9 genome wide screen resource: http://genome-engineering.org/gecko/
Online poster by Nature Reviews Molecular Cell Biology on Pluripotent States:
http://www.nature.com/nrm/posters/pluripotency/index.html
- Glossary
- Primordial germ cells (PGCs)
- embryonic progenitor cells that give rise to germ cells in the gonads (sperm and oocytes).
- Inner cell mass (ICM)
- the mass of cells inside the pre-implantation blastocyst that will subsequently give rise to the definitive structures of the fetus.
- Embryonic stem cells (ESCs)
- in vitro expanded pluripotent cells that originate from the ICM.
- Epiblast stem cells (EpiSCs)
- in vitro expanded pluripotent cells that originate from the rodent post-implantation epiblast.
- Embryonic germ cells
- in vitro expanded pluripotent cells that are derived from embryonic PGCs.
- Germ stem cells (GSCs)
- in vitro expanded pluripotent stem cells that originate from neonatal or adult testis derived spermatogonial stem cells.
- Nuclear transfer
- cloning of somatic cell derived nucleus and its introduction into a-nucleated host oocyte.
- Induced pluripotent stem cells (iPSCs)
- in vitro generated pluripotent cells derived via ectopic expression of defined exogenous factors in somatic cells.
- Naïve pluripotency
- pluripotent state that resembles pre-implantation pluripotent configuration(s).
- Primed pluripotency
- pluripotent state that corresponds to post-implantation embryonic configuration(s).
- Ground state pluripotency
- originally described as a state of pluripotency that is independent of exogenous activator signalling input or stimulation.
- X inactivation
- dosage compensation of X chromosome in female, where one of the X chromosomes gets epigenetically silenced.
- Seed enhancers
- subgroup of enhancers that are dormant in naive cells but become more active in primed pluripotent and somatic cells.
- 3i
- Defined naïve pluripotency growth conditions combing 3 inhibitors (i) for MEK, FGF and GSK3 signalling.
- 2i/LIF
- Defined naïve pluripotency growth conditions containing 2 inhibitors (i) for MEK and GSK3 together with LIF cytokine.
- “Alternative 2i”
- Defined naïve pluripotency growth conditions composed of 2 small molecule inhibitors for GSK3 and SRC pathways
- LIF/MEKi/aPKCi
- Defined naïve pluripotency growth conditions containing 2 inhibitors (i) for MEK and atypical PKC signalling, together with LIF cytokine.
- FGF2/ACTIVIN A
- Defined primed pluripotency growth conditions for mouse EpiSCs composed of recombinant FGF2 and ACTIVIN A cytokines.
- FGF2/IWR1
- Defined primed pluripotency growth conditions for mouse EpiSCs composed of recombinant FGF2 and Tankyrase small molecule inhibitor, IWR1.
- GSK3i/IWR1
- Defined primed pluripotency growth conditions for mouse EpiSCs composed of GSK3 pathway inhibitor and Tankyrase small molecule inhibitor, IWR1.
Table of contents summary
Hanna and colleagues review recent advances on molecular underpinnings of alternative primed- and naïve-like pluripotent states isolated in rodents and in man. They highlight potential benefits and identify key unanswered challenges in this rapidly evolving fundamental topic.
Author biographies
Leehee Weinberger M.Sc., has been a PhD student with Jacob H. Hanna at the Department of Molecular Genetics at the Weizmann Institute of Science, Rehovot, Israel for 4 years. She carried out her Masters work with Naama Barkai at the the same department. Her doctoral research focused on regulation and differentiation of human naïve pluripotent cells.
Muneef Ayyash Ph.D., has been a staff researcher in Jacob H. Hanna lab at the Department of Molecular Genetics at the Weizmann Institute of Science Rehovot, Israel for 1 year. He carried docotral work with David Wallach at the Weizmann Institute of Science, Rehovot, Israel and did postdoctoral studies with Yinon Ben-Neriah at the Hebrew University, Jeruslem, Israel. His research currently focuses on comparative analysis of distinct human naïve pluriptoent states.
Noa Novershtern Ph.D., has been a staff scientist and senior bioinformatician in Jacob H. Hanna lab at the Department of Molecular Genetics at the Weizmann Institute of Science, Rehovot, Israel for 5 years. She carried out her doctoral work with Nir Friedman at the Hebrew University, Jerusalem, Israel and Aviv Regev at the Broad Institute, Cambridge, USA. She focuses on dissecting epigenetic dynamics in pluriptoency and reprogramming.
Jacob H. Hanna M.D. Ph.D., has been on the Faculty of the Department of Molecular Genetics at the Weizmann Institute of Science, Rehovot, Israel for 5 years, and is also a Robertson Investigator by the New York Stem Cell Foundation (NYSCF). He carried out his doctoral work with Ofer Mandelobim at the Hebrew University, Jerusalem, Israel and then did his postdoctoral studies with Rudolf Jaenisch at the Whitehead Institute for Biomedical Research - Massachusetts Institute of Technology, Cambridge, USA. His laboratory studies molceular mechanisms of stem cell reprogramming and differentiation. Hanna lab website: http://hannalabweb.weizmann.ac.il/
Acknowledgements
J.H.H is supported by a generous gift from Ilana and Pascal Mantoux; the New York Stem Cell Foundation (NYSCF), Flight Attendant Medical Research Institute (FAMRI), the Kimmel Innovator Research Award, the ERC-StG (StG-2011-281906), Moross Cancer Institute, Israel Science Foundation – NFSC program, Morasha Biomed program, ICORE program, the ICRF Foundation, MINERVA fund, Helen and Martin Kimmel Institute for Stem Cell research, the Benoziyo Endowment fund, David and Fela Shapell Family Foundation INCPM Fund for Preclinical Studies, an HFSPO research grant. J.H.H. is a New York Stem Cell Foundation - Robertson Investigator. We thank W. Greenleaf and members of the Hanna lab for discussions. We apologize to those whose work could not be covered or directly cited due to space limitations.
References
First study to derive primed ESCs from human blastocysts.
One of two studies that derived primed EpiSC lines from rodent post-implantation epiblasts.
- **9.↵
One of two studies that derived primed EpiSC lines from rodent post-implantation epiblasts.
First study to desribe generation of pluripotent ES-like cells from mouse spermatogonial stem cells.
First study to generate validated human NT-ESCs from adult patient specific fibroblasts.
- **18.
First study to generate validated human NT-ESCs from embryonic fibroblasts.
- *19.↵
Study demonstrating feasibility of SCNT in mice.
- **20.↵
First study that invented direct in vitro reprogramming of somati cells into iPSCs with defined transcription factors.
One of two studies first showing interconversion ebtween murine naïve and primed pluriptoent cells, and in different mouse strains.
Study describing defined 3i naïve conditions capable of generating Lif/Stat3 independent mouse ESCs.
First study utulizing aPKCi to boost naïve murine iPSCs and ESCs generation.
First study to show that Src inhibition promotes murine naïvepluriptoency.
- **51.↵
Study identifies Hippo signaling pathway role in epiblast vs. trophoblast specification in pre-implantaioon mouse embryos.
Study establishess Mbd3/NuRD as a repressor of naïve pluriptoency promoting genes in mouse ESCs.
First study proving opposing dependence on epigentic repressors between mouse naïve and primed PSCs.
First study to generate geneticaly unmodified and indefinitely stable human MEK independent naive PSCs, capable of generating advanced cross species mosue-human chimeric embryos.
One offirst two studies descrining region-specific features of in vitro expanded mouse EpiSCs.
- *74.↵
One offirst two studies descrining region-specific features of in vitro expanded mouse EpiSCs.
First study to establish mouse PGCLC induction from murine ERK independent naïve PSCs.
One of two studies first showing interconversion between murine naïve and primed plurippotent cells.
Study provided first evidence for alternative transgene dependent human PSCs that can be expanded in 2i/LIF conditions.
Study connecting RHO and HIPPO signalling pathways in maintenance of primed human pluripotency.
First study to generate human in vitro PGCLCs and demonstrate alteredfunction of human ERK independent naïve PSCs.
Study demonstrating frist ever geenrated chimeric monkey fetuses with naïve monkey ESCs eastablished in NHSM conditions supplemented with Vitamin C.