Nanog Is the Gateway to the Pluripotent Ground State

1Pluripotency is the capacity of a single cell to generate in a flexible manner all cell lineages of the developing and adult organism. This is an essential, albeit transient, attribute of cells in embryos that undergo regulative development. After fertilization, mammalian zygotes follow a program of cleavage divisions and elaborate two extraembryonic lineages, trophoblast and hypoblast. This preparatory phase of development culminates in creation of the embryo founder tissue, a population of unrestricted pluripotent cells known as the epiblast. The epiblast proliferates to provide the substrate for axis formation, germlayer specification, and gastrulation. Naive early epiblast cells can be immortalized in culture in the form of embryonic stem (ES) cells. Pluripotent cells can also be created outside the embryo by reprogramming somatic cells, either by fusion with pre-existing pluripotent cells or, more compellingly, by transfection with regulatory transcription factors. Information is accumulating on the molecular composition and inferred regulatory circuitry of the pluripotent state. Understanding of how pluripotent cells are generated remains rudimentary, however.

2Nanog is a highly divergent homeodomain-containing protein commonly accorded a central position in the transcriptional network of pluripotency. It is essential for early embryonic development. Nanog is expressed in pluripotent embryo cells, derivative ES cells, and the developing germline of mammals and birds. Forced expression of Nanog is sufficient to drive cytokine-independent self-renewal of undifferentiated ES cells.

3Surprisingly, however, Nanog is not one of the canonical quartet of transcription factors employed to reprogram mouse fibroblasts. Moreover, addition of Nanog to this quartet has not been reported to increase efficiencies. However, Nanog is expressed weakly or not at all in incompletely reprogrammed cells that fail to activate properly the endogenous pluripotent transcriptional circuitry. Selection or screening for activation of endogenous Nanog expression facilitates isolation of fully reprogrammed induced pluripotent stem (iPS) cells that can contribute to adult chimeras and give germline transmission. Furthermore, in human cells Nanog does facilitate molecular reprogramming. It has also been shown that Nanog promotes the transfer of pluripotency after ES cell fusion.

4Nanog null embryos do not develop beyond implantation. An inner cell mass (ICM) is evident in mutant blastocysts and the collapse of post-implantation development has been assumed to reflect a requirement for Nanog to maintain and expand the pluripotent epiblast. However, conditional gene deletion in ES cells revealed that Nanog is not essential for propagation of pluripotency ex vivo. Nanog null ES cells are more prone to differentiate but can be maintained indefinitely. Moreover, they contribute extensively to somatic chimeras, presenting a major discrepancy with the embryo deletion analysis.

5In this study, by clarifying the role of Nanog in generation versus maintenance of pluripotency, we seek to resolve paradoxes arising from previous findings. We compare experimental induction of pluripotency from somatic cells with natural development of pluripotency in the blastocyst.

6Transgenic expression of Nanog promotes formation of pluripotent hybrids after fusion of ES cells with somatic cells. We investigated whether upregulation of endogenous Nanog may have a similar effect. Exposure of ES cells to 3 μM MEK inhibitor (PD184352 or PD0325901) in the presence of serum and leukemia inhibitory factor (LIF) results in increased expression of Nanog without altering levels of Oct4 (Figures 1A and 1B). Rex1, a sensitive indicator of undifferentiated ES cell status, is also unchanged suggesting that the increase in Nanog is not secondary to reduced differentiation. Nanog has been shown to fluctuate in ES cells cultured in serum and LIF. MEK inhibition increases the fraction of Nanog-positive cells to over 90% and also increases the mean and maximum levels of expression (Figure 1C and Figure S1 available online). We treated ES cells with 3 μM MEK inhibitor for 3 days prior to polyethylene glycol (PEG) mediated fusion with brain-derived neural stem (NS) cells. The NS cells constitutively express tauGFP and puromycin resistance whereas the ES cells express the dsRed2 and hygromycin resistance, enabling detection and selection of hybrids. Fused cells were purified by flow cytometry 24 hr after PEG treatment, quantitated (Figure 1D), and plated in complete ES cell medium. MEK inhibitor was maintained for 72 hr after sorting, then withdrawn. Puromycin plus hygromycin selection was then applied. Macroscopic colonies of typical ES cell morphology emerged after 5–6 days under selection. All of these expressed GFP and dsRed2 (Figure 1E). Plates were fixed on day 12 and stained for alkaline phosphatase, a marker of ES cells (Figure 1D). MEK inhibitor-treated cultures yielded a greater than 40-fold increase in undifferentiated hybrid colonies, normalized to the number of fused cells plated to eliminate variation due to differences in fusion efficiency (Figure 1F). This dramatic effect of MEK inhibition is likely to be mediated at least in part via upregulation of Nanog since endogenous Nanog is normally limiting for transfer of the pluripotent state.

7We then exploited the availability of Nanog null (Δ) ES cells to evaluate whether Nanog may be necessary to produce pluripotent hybrids. As controls we employed both parental ES cells and NanogΔ cells with restored Nanog expression from a constitutive transgene (Δ+Ng) (Figure 1G). Fusion of parental ES cells with Oct4-GiP NS cells generated around 70 colonies per plate (Figures 1H and 1I). Δ+Ng ES cells yielded increased numbers of hybrid colonies, >400 per plate, consistent with the enhancing effect of overexpression of Nanog. In contrast, fusions using NanogΔ ES cells yielded only 2–3 hybrid colonies in identical conditions (Figures 1H and 1I). This 25-fold reduction in hybrid yield indicates that Nanog expression is a critical determinant for the efficiency of reprogramming. Furthermore the occasional hybrid colonies recovered from NanogΔES×NS cell fusions expressed Nanog from the NS cell genome (Figure S2). Activation of Nanog after fusion could compensate for absence from the ES cell partner. We therefore fused NanogΔ ES cells with NanogΔ NS cells (Figure S3). We introduced into the NanogΔ NS cells a dual puromycin resistance and dsRed expression construct. After fusion, 4 × 106 cells were plated per 10 cm dish and placed under selection the following day. A few hybrid colonies, approximately 10 per plate, were evident after 2 weeks, identified by combined red and green fluorescence. These cells did not have typical compact ES cell colony morphology, however (Figure 1J). To clarify their status we transferred them to the selective ground state culture combination of MEK inhibitor and GSK3 inhibitor (2i) plus LIF. NS cells cannot survive in 2i and puromycin was maintained to eliminate any persisting unfused ES cells. All cells died or differentiated within days. In contrast hybrids generated between null ES cells and null NS cells expressing a Nanog transgene expanded readily without differentiation. Nanog is not required by established pluripotent cells in 2i/LIF because Nanog null ES cell lines can readily be propagated (Figure S4). We conclude that if Nanog is absent from both fusion partners, hybrid cells do not attain pluripotency.

8The hybrid results led us to hypothesize that Nanog may be dispensable in the initial stages of reprogramming the somatic cell genome but essential for instating pluripotency. We therefore examined requirements for Nanog during molecular reprogramming. NS cells undergo rapid reprogramming after transfection with Oct4, Klf4, and c-Myc, without requirement for exogenous Sox2. As a control population we introduced a constitutive Nanog transgene (Ng) into NanogΔ NS cells (Figure 2A). Oct4, Klf4, and c-myc were transduced into NanogΔ and Δ+Ng NS cells. After 3 days, cells were transferred into medium containing serum and LIF. Two days later the plates were near-confluent with compact proliferating cells that had lost the bipolar morphology of NS cells and morphologically resembled undifferentiated ES cells (Figure 2B). These cells downregulated the NS cell marker Olig2 and acquired the intercellular adhesion molecule Ecadherin (Figure S5B). They expressed ES cell markers SSEA-1 and alkaline phosphatase (Figures S5A and S5C). However, endogenous core pluripotency factors were not robustly expressed (Figure 2F). There were no discernible differences between transduced NanogΔ and Δ+Ng cells. These observations suggest that Nanog is not required to produce a dedifferentiated partially reprogrammed state.

9Incompletely reprogrammed cells may be promoted to pluripotency by exposure to 2i/LIF. This allowed us to test whether partially reprogrammed NanogΔ cells could transit to pluripotency. NanogΔ cells transduced with the three factors were passaged once on feeders in the presence of serum and LIF, then switched to 2i/LIF. All cells died within 7 days (Figure 2C). No colonies formed from multiple wells plated (Figures 2C and 2D). In marked contrast Δ+Ng cells treated in the same way generated around 100 colonies per plate of 6 × 104 cells, corresponding to a conversion frequency of 0.16% ± 0.02% (Figures 2C and 2D). These colonies were expanded in 2i/LIF and stable lines generated (Figure 2E). Resistance to G418 and hygromycin indicated activation of the endogenous targeted Nanog alleles. They expressed Fgf4 and Rex1 unlike the partially reprogrammed cells (Figure 2F).

10We deleted the floxed Nanog transgene after conversion in 2i/LIF (Figure S6A) by Cre recombination. The deletion is conveniently monitored by gain of dsRed expression. DsRed-positive iPS cells continued to proliferate with undifferentiated morphology and retention of mRNA markers of pluripotency (Figure S6B). To confirm ground state identity and developmental potential, we examined ability to colonize the mouse embryo after morula aggregation. Chimeric fetuses analyzed at mid-gestation exhibited widespread contribution of GFP-positive cells (Figure S6D). Therefore, in line with results in ES cells, Nanog is no longer necessary in iPS cells once pluripotency has been attained by reprogramming. We conclude that Nanog is specifically required for partially reprogrammed pre-iPS cells to reach ground state pluripotency.

11We then investigated whether partially reprogrammed NanogΔ cells that appear incapable of progression to the ground state are indeed pre-iPS cells. To do this we restored Nanog expression using PiggyBac transposition to introduce a floxed Nanog transgene into the partially reprogrammed cells. Stable transfectants were obtained after hygromycin selection. Cells were then transferred into 2i/LIF as above. In contrast to nontransfected and empty vector transfected cells that died, PB-Nanog transfectants produced over 1400 colonies per well (Figure 2G). This corresponds to a conversion efficiency of more than 0.4%. These cells are G418 resistant, indicating activation of the endogenous targeted Nanog locus. qRT-PCR analysis revealed silencing of retroviral expression and upregulation of ground state pluripotency markers Rex1 and Klf2 to levels comparable to those found in ES cells (Figure 2H). After tamoxifen-induced Cre excision of the Nanog transgene these cells gave liveborn coat color chimeras (Figure 2I).

12We then investigated the timing of gene activation in pre-iPS cells after the switch to 2i/LIF. We exploited a clonal line of pre-iPS cells derived from mouse embryo fibroblasts. These cells do not express Nanog or Rex1 (Figure S7), and when maintained in the presence of serum they do not spontaneously acquire pluripotency even under selection for Oct4 promoter activation. However, they convert to germline-competent pluripotency at high efficiency after transfer to 2i/LIF. Conversion takes place over a period of 10 days. Strikingly, we found that Nanog mRNA is only significantly upregulated from day 7. This timing is coincident with the upregulation of endogenous pluripotency markers and of an Oct4-GFP transgene reporter (Figures 2J and S7). These data indicate that Nanog is not the immediate target of 2i in pre-iPS cells. Consistent with this interpretation, transgenic expression of Nanog in pre-iPS cells is not sufficient to enable efficient escape from the pre-iPS cell state without exposure to 2i/LIF. Together, these data indicate that Nanog acts at the final stage of the molecular reprogramming process to instate pluripotency.

13EpiSCs are a cell type in which many components of pluripotent circuitry are naturally present. They are derived from the epithelialized epiblast of post-implantation embryos or by differentiation of ES cells. EpiSCs are distinct from ES cells in gene expression, growth factor dependency, inactivation of the X chromosome (in XX cells), and inability to contribute to blastocyst chimeras. This is a transcriptionally and epigenetically differentiated cell state compared with naive epiblast or ES cells that does not revert, even in 2i/LIF. However, EpiSCs can be reprogrammed to ground state pluripotency by stable transfection with Klf4 and culture in 2i/LIF. This occurs with an efficiency of 0.1%–1%, similar to somatic cell reprogramming. EpiSCs express Nanog but at lower levels than ES cells. We therefore tested whether elevated expression of Nanog may mediate conversion to the ground state. We used EpiSCs derived from an E5.5 female embryo carrying the Oct4-GiP reporter. Nanog stable transfectants were generated by lipofection with a Piggybac vector containing independent promoters directing expression of Nanog and of dsRed reporter linked to the hygromycin resistance gene (PBNanogdsRed). Transfectants were expanded in EpiSC culture conditions of Fgf2 and activin under hygromycin selection. Under these conditions there is no upregulation of ground state markers. Upon transfer to 2i/LIF, which eliminates parental EpiSCs or control transfectants, multiple undifferentiated colonies form that express Oct4-GFP (Figure 3A). Colonies were obtained at an average of 200–2000 per well of 2 × 104 cells, or 1%–10%. These cells could be stably propagated and exhibited upregulation of ES cell markers (Figure 3D). This frequency of conversion is 10-fold higher than for Klf4 transfectant in side-by-side experiments. Klf4-mediated conversion is dependent on treatment with 2i, which may act in part by inducing Nanog. We asked whether Nanog can mediate reprogramming of EpiSCs without 2i. Nanog-transfected EpiSCs transferred from Fgf2/activin culture to medium supplemented with LIF+BMP4 produced numerous undifferentiated Oct-GFP-positive colonies (Figures 3B and 3C) that could be stably expanded and showed the transcriptional marker profile of ground state pluripotency, including upregulation of Klf4 (Figure 3D). No iPS cell colonies were obtained from parallel cultures of Klf4 transfectants without 2i.

14The Nanog-generated Epi-iPS cell colonies invariably expressed dsRed (Figure 3C) indicating integration of the transgene. Transient transfection of the PB-Nanog construct without PBase did not yield any Epi-iPS cells. We then tested transient cotransfection of Nanog with Klf4. Transfected cells were transferred into 2i/LIF after 48 hr. Epi-iPS cell colonies emerged over the following 7 days at a frequency of 2–6 colonies per well of 2 × 104 starting EpiSCs. Several of these colonies were picked and expanded. They showed uniform expression of Oct4-GFP but no dsRed (Figure 3E). Integration of PB vectors was undetectable by genomic PCR (Figure 3F). qRT-PCR analysis confirmed transcriptional resetting with upregulation of ES cell markers Nr0b1, Klf4, and Klf2 and downregulation of EpiSC markers Fgf5, Brachyury, and Lefty (Figure 3G). Immunostaining for the silencing mark me3H3K27 revealed that the nuclear focus corresponding to the inactive X chromosome was erased in the Epi-iPS cells (Figure 3H). The definitive test of ground state pluripotency is ability to colonize the developing embryo and contribute substantially to term chimeras. Two clones of Nanog/Klf4 Epi-iPS cells were tested by morula aggregation and both colonized the egg cylinder at high frequency (Figure S8), a property never exhibited by EpiSCs. After blastocyst injection of one of these clones, healthy adult mice were obtained with extensive coat color chimerism (Figure 3I).

15These findings confirm that in the presence of other key pluripotency factors, Nanog can orchestrate transition to the ground state, mediating both transcriptional and epigenetic reprogramming. Intriguingly, as previously noted for ES cell self-renewal and hybrid reprogramming, endogenous expression of Nanog is limiting, indicating dosage sensitivity in Nanog action.

16The preceding evidence that Nanog is crucial for instating pluripotency during somatic cell reprogramming prompted re-examination of its role in embryogenesis. Nanog is essential for development after implantation. This could be because it serves to maintain the epiblast or because it is required to make a pluripotent epiblast. Nanog is present from the late morula throughout formation and expansion of the ICM. The epiblast emerges through partition of the ICM in the mature mouse blastocyst between embryonic day (E) 3.5 and E4.5. Nanog mRNA is downregulated shortly thereafter. We analyzed distribution of Nanog protein at these stages (Figures 4 and S9).

17We compared localization of Nanog with Oct4, generally considered a pluripotency marker but also transiently expressed during hypoblast formation. We contrasted Nanog distribution with Gata6 and Gata4, specification factors for the hypoblast. At E3.5 Nanog was largely confined to the ICM (Figures 4C and S9) and present in most, though not all, of these cells, as previously described. Oct4 was strongly expressed throughout the ICM and also detectable in some trophoblast cells. By E4.5 Nanog immunostaining was confined to a subset of ICM cells overlaid on one surface by trophoblast and on the other by hypoblast (Figure 4A). Oct4 was expressed more broadly and was present in almost all ICM cells (Figure 4). Gata4 and Gata6 were restricted to ICM cells lining or close to the blastocoel, corresponding to the location of the hypoblast (Figures 4A and 4C). Many of those cells coexpressed Oct4. In contrast, we did not observe at E4.5 any instance of a cell that coexpressed Nanog with either Gata4 or Gata6.

18Mouse embryo development can be arrested and synchronized at the mature blastocyst stage if implantation is prevented by ongoing lactation or by experimental deprivation of estrogen. This condition of diapause is a favored stage for the derivation of ES cells. Oct4 is expressed in all ICM cells in diapause. In contrast Nanog is present only in the interior ICM cells and is absent from the hypoblast layer (Figure 4B). These findings establish that Nanog uniquely marks the epiblast in the mature blastocyst and is mutually exclusive with expression of the Gata factors, whereas Oct4 is ubiquitous throughout the ICM.

19In female mouse embryos the paternal X chromosome is silenced during early cleavage. The inactive X (Xi) is then reactivated in the ICM. Reactivation is critical to allow for random X inactivation in the embryo proper and may be considered an essential attribute of ground state pluripotency in females. Indeed a distinctive feature of female ES cells is that both X chromosomes are active. Furthermore, ES cells have the capacity to reactivate a silenced somatic X chromosome during cell fusion-induced reprogramming. We therefore analyzed the relationship between Nanog expression and reactivation of Xi in the ICM. We used immunostaining for Eed, a component of the PRC2 polycomb group complex, to detect epigenetic marking indicative of Xi. We compared the presence or absence of an Eed nuclear body with Nanog expression. All ICM cells of XX E3.5 blastocysts exhibited a prominent Eed focus (Figures 4C and 4D), confirming that Xi is maintained in the early blastocyst. Therefore coexpression of Nanog, Oct4, and Sox2, at least at the levels present at E3.5, is not sufficient for X chromosome reactivation. At E4.5, however, around half of the Oct4-positive ICM cells in XX blastocysts had lost the Eed nuclear body that was readily detectable in neighboring trophoblast cells (Figures 4C and 4E). Absence of the Eed focus correlated precisely with the presence of Nanog (Figures 4C and 4E and Movie S1). ICM cells retaining Xi did not have Nanog but expressed Gata4 (Figures 4C and 4E), indicative of hypoblast differentiation. Similarly in diapause blastocysts an Eed nuclear body was evident in all cells of the hypoblast and trophoblast but absent from Nanog-positive epiblast cells (Figure 5F). From all embryos examined, every cell without an Eed nuclear body also expressed Nanog. These observations indicate that X chromosome silencing is erased in only a subset of the ICM.

20We conclude that Nanog expression at E4.5 defines the embryo founder compartment and is coincident with the domain of X chromosome reactivation. Nanog-negative ICM cells still express Oct4 but are Gata6/Gata4 positive. These hypoblast cells do not erase X chromosome silencing. Thus paternal X chromosome inactivation in the yolk sac endoderm does not entail a second cycle of imprinted inactivation but is inherited from early cleavage as occurs in the trophoblast.

21We then analyzed ICM status in Nanog−/− blastocysts. At E3.5 Nanog protein was undetectable in one quarter of intercross blastocysts, but these were not distinguishable in morphology, size, or ICM cell number (Figures 5A and S9). In contrast, an overt phenotype was evident in embryos harvested at E4.5. Of 57 E4.5 intercross blastocysts, 12 lacked detectable Nanog protein. The Nanog-negative specimens contained substantially fewer ICM cells, and these showed reduced intensity of Oct4 staining (Figures 5A and 5B). Strong Gata4 staining was lacking, though faint immunoreactivity was apparent in a few cells (Figures 5B and 5C). In five Nanog-negative blastocysts with Eed foci in trophoblast nuclei, the nuclear body was also prominent in all intact ICM nuclei (Figure 5D). Persistence of Xi in Oct4-positive ICM cells in mutant embryos at E4.5 is consistent with a putative direct role for Nanog in X chromosome reactivation. It also points to a failure of null ICM cells to develop into epiblast.

22To clarify the cellular basis of the phenotype of Nanog-deficient blastocysts, we examined diapause blastocysts. Diapause effectively extends the preimplantation period and allows the possibility for retarded cells to catch up in development. Eight of thirty-one intercross diapause blastocysts lacked Nanog-positive cells. None of these contained an overt ICM compartment. Four were completely Oct4 negative, while the others contained a few Oct4-stained cells dispersed around the trophectoderm layer (Figure 5E). In Nanog-negative diapause embryos that showed Eed foci, this was present in every nucleus including those staining for Oct4 (Figure 5E), indicating that X reactivation had not occurred. These observations indicate that the reduced size of the ICM at E4.5 and the absence of hypoblast are not due to retarded growth or differentation but are a result of developmental failure. We surmised that Nanog may be required to achieve a pluripotent ground state in the early embryo.

23To test further whether pluripotent cells could be produced in the absence of Nanog we examined responsiveness to 2i. Culture of early embryos in 2i maximizes expansion of the epiblast and facilitates subsequent derivation of ES cells, including from otherwise recalcitrant strains. Early blastocysts cultured in 2i or 2i/LIF for 48 hr develop expanded Oct4-positive ICMs with a large component of Nanog-expressing cells along with Gata4-positive hypoblast (Figure 6A). Nanog null ES cells expand more efficiently in 2i or 2i/LIF than in serum and LIF (Figure S4). Therefore if Nanog−/− embryos have the capacity to produce any epiblast cells, their recovery should be maximized in 2i/LIF. We harvested intercross blastocysts at E3.5 when the newly formed ICM is indistinguishable between mutant and wild-type. After culture in 2i or in 2i/LIF for 48 hr, all 43 Nanog-positive embryos had an ICM containing a large compartment of cells double positive for Oct4 and Nanog, with overlying Gata4-positive cells. In 10 Nanog-negative embryos the ICM had completely degenerated with neither Oct4 nor Gata4 detectable. Five of these displayed Eed foci in all cells (Figures 6A–6C). Lack of Gata4 staining in both freshly harvested and cultured Nanog-deficient blastocysts indicates that the loss of ICM does not arise by unscheduled differentiation into hypoblast, as we previously speculated. Instead, these findings suggest that without Nanog, ICM cells are unable to progress into a correctly specified and viable epiblast and that hypoblast either does not form or rapidly degenerates.

24We investigated in more detail the fate and potency of Nanog null ICM cells. We stained intercross E4.5 embryos with Troma1 to detect cytokeratin 8 expressed by trophoblast. We costained for Oct4 and activated caspase 3 to monitor apoptosis. Nanog immunostaining could not be used in this combination with available antibodies. Therefore mutant embryos were identified by low or absent Gata4. This correlated with reduced ICM cell numbers and increased caspase immunoreactivity (Figure 6D). In three out of four of these embryos some Oct4-positive cells costained withTroma1, which was never seen in the wild-type or heterozygous embryos at this stage (Figures 6E and 6F). These cells appeared to be on the surface of the embryo, suggesting incorporation into the trophoblast.

25We isolated 56 ICMs from E3.5–E3.75 intercross blastocysts and cultured them in medium supplemented with LIF and serum. Individual genotypes were determined from the trophoblast lysates. Wild-type and heterozygous ICMs sustained a central cell mass and produced outgrowths of cells with refractile and migratory features characteristic of parietal endoderm (Figures 6G and 6H). In 3 out of 48 cases, outgrowths also contained large flattened cells of trophoblast morphology, suggesting that those ICMs had been harvested from early stage blastocysts and retained the capacity for trophoblast production. Eight out of eight null ICMs did not maintain any undifferentiated cell mass and either failed to attach and degenerated or produced a few trophoblast cells with spread morphology and large nuclei (Figure 6G). In no case did we observe either persistent ICM or hypoblast in a null embryo culture.

26We conclude that Nanog mutant ICM cells are blocked in a transition stage of ICM development. Null cells cannot progress to pluripotency and therefore have only two options, differentiation into trophoblast or death.

27Nanog can command constitutive self-renewal of ES cells and appears able to reverse precommitment perturbations of the pluripotent state. Nanog also greatly increases the efficiency of nuclear reprogramming by ES cell fusion. It is therefore somewhat surprising that Nanog is not represented among the minimal combinations of exogenous factors found to convert mouse somatic cells into iPS cells. This study offers an explanation (Figure 7). Our results indicate that Nanog is in fact decisive for attaining this pluripotent ground state. However, this requirement is during the final phase of reprogramming when other key factors are already present and may be fulfilled by activation of endogenous Nanog. A role in the culmination of somatic cell reprogramming is mirrored by the pivotal function revealed for Nanog in establishing pluripotency in the embryo.

28Analyses of reprogramming in Nanog-deficient cells show that it is fully dispensable for the initial steps: loss of differentiated characteristics and creation of a pre-pluripotent state. Indeed pre-iPS cells, which have silenced somatic genes, exhibit expression of some markers of pluripotency but not Nanog. The failure to respond to 2i/LIF, which efficiently converts wild-type pre-iPS cells to pluripotency, indicates that Nanog−/− pre-iPS cells cannot attain pluripotent status. This block is overcome, however, by introduction of Nanog. Therefore, Nanog mediates acquisition of pluripotency by dedifferentiated partially reprogrammed cells.

29The in vivo phenotype of Nanog deletion shows that it is critical for early ICM cells to mature into pluripotent epiblast. This explains why mutant embryos cannot give rise to ES cells, even though Nanog-deficient ES cells are viable. Cells that are allocated to the ICM but unable to upregulate Nanog progressively degenerate between E3.5 and E4.5. This occurs by a combination of differentiation into trophoblast and apoptosis (Figures 6D to 6F). The failure of Nanog null ICMs to form hypoblast is unexpected. It is possible that Nanog plays a role in potentiating hypoblast specification. Alternatively, hypoblast may be specified but unable to survive in the absence of paracrine support from a nascent epiblast. This would be consistent with the detection of occasional mutant cells showing weak Gata4 staining.

30Oct4 and its partner Sox2 are expressed ubiquitously in morulae and early blastocysts and are present throughout the ICM until after segregation of the hypoblast. Expression of these factors is therefore too wide to define the epiblast. Nanog, in contrast, is upregulated at the time of compaction. It may fluctuate in early ICM cells but subsequently is uniformly and exclusively expressed in the nascent epiblast. This restricted expression of Nanog is associated with and essential for prosecuting the transition to pluripotency in a subset of cells that already express Oct4 and Sox2. We propose that the presence of Nanog harnesses Oct4 and Sox2 to create ground state pluripotency. In this context, Nanog may be considered the specification factor for epiblast.

31Erasure of X chromosome silencing is a unique epigenetic signature of arrival at authentic pluripotency. Presence of a silenced X chromosome in pre-pluripotent female embryo cells along with Oct4 and Sox2 mirrors the circumstances in partially reprogrammed pre-iPS cells. The timing of Nanog action in both somatic cell reprogramming and epiblast formation implies a requirement for coincident expression of other pluripotency factors. Previously, we showed that fusion of ES cells overexpressing Nanog with NS cells results in an equivalent pluripotent hybrid yield as fusion with ES cells, where there is no requirement for reprogramming. Thus, in the appropriate transcription factor context, Nanog levels can modulate reprogramming up to maximal efficiency. Moreover, stable Nanog overexpression in EpiSCs, which express most pluripotent factors, is sufficient to direct reprogramming upon withdrawal of the EpiSC maintainance factors Fgf and activin. Most strikingly, Nanog acts synergistically with the absent factor Klf4, such that transient expression of both is sufficient to convert EpiSCs to ground state pluripotency.

32The reprogramming process can be divided into discrete stages related to the requirement for Nanog activity (Figure 7). The first stage culminates in the generation of dedifferentiated pre-pluripotent cells sustained by Oct4, Sox2, and Klf4 transgenes. In some circumstances, influenced by factors such as character of the reprogramming vector and genetic background, cells are able to transit out of this stage stochastically and reach authentic pluripotency. In many cases, however, cells are blocked in the pre-iPS cell state. Cells in this condition express some markers of pluripotency but Nanog remains silenced. Transfer to serum-free 2i/LIF culture relieves the block and allows generation of pluripotency. Our findings establish that this culminating step is critically dependent on Nanog. How might Nanog achieve this? A recent comparison by chromatin immunoprecipitation of pre-iPS cells and fully reprogrammed iPS cells identified multiple genes that are co-occupied by Oct4, Sox2, and Klf4 in ES cells but not in pre-iPS cells. Many of these genes have been proposed as targets of Nanog in ES cells. Therefore, Nanog may orchestrate transition to the pluripotent ground state by facilitating cooperative binding of the core pluripotency factors to their cognate ES cell targets. Mechanisms could include Nanog-dependent recruitment of protein complexes comprised of the core factors or Nanog-dependent chromatin modulation to render loci accessible, as suggested for intron 1 of the Xist locus.

33In conclusion, our findings suggest that Nanog lies at the heart of a convergent mechanism for attaining the ground state of authentic pluripotency both in embryonic development and in the final phase of somatic cell reprogramming. We surmise that Nanog choreographs the emergent gene regulatory network of pluripotency into a functional and self-sustaining configuration. The next challenge is to elucidate precisely how Nanog brings about this synthesis.

34Nanog−/− NS cells were derived using the established NS cell derivation protocol from E12.5 fetal forebrain of a chimeric fetus after flow cytometric purification of GFP-expressing null cells. Nanog+/+ NS cells were derived from E12.5 fetal forebrain of transgenic lines that constitutively express the fusion protein tauGFP and the puromycin resistance gene or that carry regulatory sequences of the mouse Oct4 gene driving GFP and puromycin resistance. Nanog−/− rescue NS cells were derived by transfection with pPyfloxNanogIPdsRedIB to obtain cells that express constitutively Nanog and puromycin resistance. Nanog−/− pre-iPS cells were transfected using Nucleofection with 1 μg of PBfloxNanogPgkHygro plus 2–3 μg PBase expression vector, pCAGPBase, to obtain cells that express constitutively Nanog and Hygromycin resistance. EpiSCs derived from E5.5 Oct4GiP epiblast were cultured and transfected as described.

35The Nanog null mutation generated by homologous recombination has been described previously. Genotypes of intercross blastocysts were inferred from presence or absence of Nanog immunostaining.