In rodents, the naïve early epiblast undergoes profound morphogenetic, transcriptional and epigenetic changes after implantation. These differences are maintained between blastocyst-derived embryonic stem (ES) cells and egg cylinder-derived epiblast stem cells (EpiSCs). Notably, ES cells robustly colonise chimaeras, whereas EpiSCs show little or no contribution. ES cells self-renew independently of mitogenic growth factors, whereas EpiSCs require fibroblast growth factor. However, EpiSCs retain the core pluripotency factors Oct4 and Sox2 and the developmental barrier dividing them from unrestricted pluripotency can be surmounted by a single reprogramming factor. This provides an opportunity to identify molecules that can reset the naïve state. We undertook a forward genetic screen for effectors of EpiSC reprogramming, employing piggyBac transposition to activate endogenous gene expression at random and selecting for undifferentiated colonies in the absence of growth factor signalling. Three recovered clones harboured integrations that activate the closely related orphan nuclear receptor genes Nr5a1 and Nr5a2. Activity of Nr5a1 and Nr5a2 was confirmed by direct transfection. Reprogrammed colonies were obtained without transgene integration and at 10-fold higher frequency than with other single factors. Converted cells exhibited the diagnostic self-renewal characteristics, gene expression profile and X chromosome activation signature of ground state pluripotency. They efficiently produced adult chimaeras and gave germline transmission. Nr5a receptors regulate Oct4 transcription but this is insufficient for reprogramming. Intriguingly, unlike previously identified reprogramming molecules, Nr5a receptors play no evident role in ES cell self-renewal. This implies a different foundation for their capacity to reset pluripotency and suggests that further factors remain to be identified.
Mouse embryonic stem (ES) cells and post-implantation epiblast stem cells (EpiSCs) are considered to represent two phases in the ontogeny of the pluripotent epiblast (Rossant, 2008). ES cells are obtained from the initial naïve population that emerges in the mature blastocyst (Brook and Gardner, 1997; Evans and Kaufman, 1981; Nichols et al., 2009). They contribute extensively to chimaeras, demonstrating retention of early epiblast identity and associated unbiased developmental potential. EpiSCs are derived by culturing post-implantation egg cylinder epiblast (Brons et al., 2007; Tesar et al., 2007). They can differentiate into some cell types in culture, including germ cells (Hayashi and Surani, 2009), and can produce teratomas. However, they colonise blastocyst chimaeras poorly, if at all (Guo et al., 2009; Tesar et al., 2007). Intriguingly, stem cells derived from human embryos show features more in common with rodent EpiSCs than with ES cells (Nichols and Smith, 2009; Rossant, 2008).
ES cells and EpiSCs express the central pluripotency factors Oct4 (Pou5f1 – Mouse Genome Informatics) and Sox2 at similar levels, but diverge in other important respects. EpiSCs do not express, or have lower levels than ES cells, of many pluripotency-associated transcription factors, and they show upregulation of early lineage specification markers (Guo et al., 2009; Hanna et al., 2009; Tesar et al., 2007). They also exhibit a differentiated epigenetic feature in female cells: the decoration of one X chromosome with the silencing mark H3K27me3 (Guo et al., 2009; Silva et al., 2003). EpiSCs do not respond productively to the ES cell self-renewal cytokine leukaemia inhibitory factor (LIF) (Brons et al., 2007) and they cannot be propagated in fully defined ES cell culture conditions, known as 2i, in which the mitogen-activated protein kinase (Erk; Mapk1) cascade and glycogen synthase kinase 3 (Gsk3) are selectively inhibited (Guo et al., 2009; Ying et al., 2008). ES cells are maintained in a relatively homogenous ‘ground state’ by application of these two inhibitors (Silva and Smith, 2008; Wray et al., 2010; Ying et al., 2008). EpiSCs, by contrast, appear to be variable both between and within cell lines. This might reflect both their origin from heterogeneous egg cylinder populations and the use of undefined culture environments, which may either select for, or induce, particular phenotypes. EpiSC-like cells derived on feeders in serum or serum replacement have been reported to acquire ES cell-like features without genetic manipulation (Bao et al., 2009; Greber et al., 2010). This is proposed to represent epigenetic reversion or dedifferentiation. However, the potential heterogeneity and developmental status of these EpiSC cultures are uncertain. Furthermore, EpiSCs on feeders can differentiate into primordial germ cells that may then undergo conversion into embryonic germ (EG) cells (Hayashi and Surani, 2009). Consequently, it is unclear what process underlies such spontaneous conversion. Importantly, for the present study, when EpiSCs are derived and maintained in defined conditions using fibroblast growth factor (FGF) and activin without feeders, they do not spontaneously acquire ES cell properties (Brons et al., 2007) and on transfer to 2i they very rarely form undifferentiated colonies (Guo et al., 2009). Robust conversion can be triggered, however, by transfection with single genes: either Klf4, Klf2 or Nanog (Guo et al., 2009; Hall et al., 2009; Hanna et al., 2009; Silva et al., 2009). Reprogramming is dependent upon withdrawal of the EpiSC self-renewal factors FGF and activin, and is promoted by 2i in combination with LIF (Yang et al., 2010).
Interestingly, despite their closer developmental proximity to naïve pluripotency and their endogenous expression of Oct4 and Sox2, the efficiency of generating induced pluripotent stem (iPS) cells from transfected EpiSCs in defined culture is not demonstrably higher than from fibroblasts. Therefore, in addition to illuminating features of pluripotency and developmental restriction, characterising the transition from EpiSC to iPS cell might also contribute to an understanding of somatic cell reprogramming. To identify factors that can surmount the molecular roadblock between EpiSCs and ground state pluripotency we undertook a genome-wide screen.
MATERIALS AND METHODS
Mouse studies were carried out in a designated facility under licences granted by the UK Home Office.
EpiSCs derived from E5.5 mouse embryos (Guo et al., 2009) were cultured on fibronectin in N2B27 medium (Ying and Smith, 2003) with activin A (20 ng/ml) and FGF2 (12.5 ng/ml) prepared in-house. ES cells and iPS cells were cultured in 2i/LIF medium (Ying et al., 2008) comprising N2B27 with MEK inhibitor (1 μM PD0325901), Gsk3 inhibitor (3 μM Chir99021) and 200 units/ml LIF (Smith, 1991). LIF/BMP4 medium is N2B27 with LIF (100 units/ml) and 5 ng/ml BMP4 (R&D Systems). Immunostaining and blastocyst injection were performed as described (Guo et al., 2009).
EpiSC reprogramming screen
Between 0.5 and 0.7×106 EpiSCs carrying the Oct4-GFPiresPuroR transgene (Guo et al., 2009) were plated per well of a 6-well tissue culture plate in EpiSC culture media. The following day, cells were transfected using Lipofectamine 2000 (Invitrogen) and 1 day later the contents of each well were replated in a 10-cm plate. Hygromycin selection (200 μg/ml) was applied for 3 days in EpiSC culture conditions to enrich for transfectants. Cultures were then transferred into 2i/LIF. Puromycin (1 μg/ml) was applied to eliminate Oct4-negative differentiated cells before colony picking. Rex1-GFPiresPuroR knock in ES cells were generated by gene tarteting.
Vector construction, insertion site identification and PCR
The MSCV 5′LTR with a splice donor site from exon 1 of mouse Foxf2 was amplified by PCR from T2/Onc (Dupuy et al., 2005) and inserted into the NotI/PacI sites of pGG131 (Guo et al., 2009) to produce the pGG134 gene-trap activation vector.
Insertion junctions were amplified using Splinkerette PCR (Guo et al., 2009; Mikkers et al., 2002). PCR products from screen 1 were cloned using the Zero Blunt TOPO PCR Cloning Kit for Sequencing (Invitrogen). Bacterial clones with PCR products were then re-amplified by colony PCR using HMSp2/Sp2 primers for both 5′ terminal repeat (TR) and 3′TR junctions. Colony PCR products were treated with exonuclease I (New England Biolabs) to degrade single-stranded oligonucleotides, followed by ethanol precipitation. PCR products from screen 2 were separated by gel electrophoresis and individual bands were recovered. Sequencing was performed using HMSp3 and Sp3 primers for both 5′TR and 3′TR on an Applied Biosystems 3730xl DNA Analyser. Sequence results were annotated manually by BLAT search.
Open reading frames were amplified by PCR using the primers indicated below from cDNA prepared from Epi-iPS cells in which Nr5a1 or Nr5a2 were activated by the MSCV enhancer insertion. Sequence-verified products were inserted into PB expression vectors by Gateway cloning. For transient expression, open reading frames were sub-cloned into pPyCAG expression constructs (Chambers et al., 2003).
Nr5a1 transgene was detected with primers CAG-For and Sf1-ex4-rev by genomic PCR. Ube1 was amplified as a genomic DNA loading control using Ube1XA and Ube1XB primers.
Primers for PCR and RT-PCR (5′ to 3′):
Ube1XA, TGGTCTGGACCCAAACGCTGTCCACA; and Ube1XB, GGCAGCAGCCATCACATAATCCAGATG.
Total RNA was prepared using the RNeasy Mini Kit (Qiagen) with DNaseI treatment. First-strand cDNA was synthesised using Superscript III reverse transcriptase (Invitrogen) with poly(dT) priming. Real-time PCR was performed using TaqMan Gene Expression Assays (Applied Biosystems) in a 7900HT Fast Real-Time PCR system (Applied Biosystems). Gene expression was determined relative to Gapdh using the ΔCt method. Standard deviation was calculated from PCR triplicates. TaqMan assays: Nr5a1, Mm00446826_m1; Nr5a2, Mm00446088_m1; Oct4, Mm00658129_gH; Klf4, Mm00516104_m1; Klf2, Mm01244979_g1; Klf5, Mm00456521_m1; Nanog, Mm02384862_g1; Rex1, Mm03053975_g; Fgf5, Mm00438615_m1; Lefty, Mm00438615_m1; brachyury (T), Mm01318252_m1; Nr0b1, Mm00431729_m1; stella, Mm00836373_g1; Gapdh, 4352339E.
Cells were harvested in SDS buffer (2% SDS, 10% glycerol, 60 mM Tris-HCl, pH 6.8) with Complete Protease Inhibitor Cocktail tablets (Roche). Samples were sonicated and protein concentration was determined by BCA assay (Pierce, Thermo Scientific). Total protein (11.5 μg) was fractionated on a 4-12% Bis-Tris Novex gel, electroblotted onto nitrocellulose membranes, and the membranes probed sequentially either anti-Nr5a1 (R&D, PP-N1665-00) or anti-Nr5a2 (R&D, PP-H2325-00) and then anti-α-tubulin (Abcam). Blots were incubated with horseradish peroxidase-coupled anti-mouse IgG and developed with ECL Plus (Amersham). Membranes were stripped between probing by incubation at 70°C for 20 minutes in 50 mM Tris, pH 6.8, 2% SDS.
A piggyBac insertional activation screen identifies EpiSC reprogramming factors
We prepared a piggyBac (PB) gene-trap vector containing the murine stem cell virus (MSCV) enhancer/promoter (Dupuy et al., 2005) (Fig. 1A). PBase-mediated transposition (Guo et al., 2009; Wang et al., 2008) was used to integrate the vector throughout the genome of Oct4GiP reporter (O4G) EpiSCs (Guo et al., 2009). Hygromycin selection for PB vector integration was maintained for 7 days. We speculated that transient co-expression of Klf4 might facilitate reprogramming. Therefore, we co-transfected a pool of EpiSCs with PB gene-trap construct plus circular Klf4 plasmid. After 3 days in FGF plus activin, transfectants were replated in 2i/LIF, which is both non-permissive for continued EpiSC propagation (Guo et al., 2009) and inductive/selective for iPS cells (Silva et al., 2008; Yang et al., 2010). Puromycin was added after 6 days in 2i/LIF to select for Oct4-positive cells. We picked two colonies on day 13 in 2i/LIF and 7 colonies on day 18. These colonies all expanded with ES cell-like morphology and stable GFP expression (Fig. 1B and Table S1 in the supplementary material).
Integration sites were analysed using Splinkerette PCR (see Fig. S1A and Table S2 in the supplementary material). Among these, a potentially productive insertion was identified before the first coding exon of the Nr5a1 gene on chromosome 2 (Fig. 1C). Unexpectedly, however, we detected integration of the Klf4 plasmid in several clones recovered from the screen (see Fig. S1B in the supplementary material). The Klf4 plasmid does not contain PB terminal repeats and is therefore not a substrate for transposition. We surmised that genomic double-strand breaks created by PBase might facilitate plasmid integration. Indeed, we found that co-transfection of circular Klf4 plasmid with PBase and a control PB vector was sufficient for low-frequency integration of the Klf4 transgene and subsequent generation of EpiSC-derived iPS (Epi-iPS) cell clones in 2i/LIF.
We therefore performed a second screen without inclusion of Klf4 plasmid. We transfected 2.5×106 EpiSCs. Conservative assumptions of 40% transfection efficiency and 30% plating efficiency with an average of four integrations per cell gave an estimate of 1×106 insertions, or 15 times gene density. Nine colonies were picked and expanded. Splinkerette PCR indicated that these represent three independent clones, each with two to five integrations (see Fig. S1C in the supplementary material). One clone contained a vector insertion upstream of Nr5a1 at a location distinct from that isolated in the first screen (Fig. 1C). A second clone harboured the gene-trap vector at the Nr5a2 locus on chromosome 1. Nr5a1 and Nr5a2 are closely related genes that encode nuclear receptors also known as steroidogenic factor 1 (SF1) and liver receptor homologue 1 (Lrh1), respectively (Fayard et al., 2004; Galarneau et al., 1996; Hoivik et al., 2009; Lala et al., 1992; Morohashi et al., 1992).
We examined expression of Nr5a1 and Nr5a2 in EpiSCs, ES cells and reprogrammed clones. Neither Nr5a1 nor Nr5a2 transcripts were detected in EpiSCs (Fig. 1D). Nr5a1 mRNA was undetectable in ES cells, but was present at low levels in the reprogrammed IPS2-1 clone with insertion at the Nr5a1 locus. Nr5a2 was expressed in ES cells and in both clones. Increased expression was apparent in the IPS2-2 clone with the Nr5a2 insertion (Fig. 1D). These data indicate that the insertions activate endogenous Nr5a genes.
Direct transfection with Nr5a1 or Nr5a2 converts EpiSCs to naïve pluripotency
To test directly whether Nr5a receptors can reprogramme EpiSCs we prepared expression constructs in a PB vector. O4G EpiSCs were transfected and selected in hygromycin. In FGF plus activin, stable transfectants showed no overt phenotype shift and the marker profile was unchanged (Fig. 2A). Upon transfer to 2i/LIF most cells died, but within 10 days over 100 GFP-positive undifferentiated colonies emerged per well of 1×104 cells (Fig. 2B). We noted that expression of dsRed from the PB vector was downregulated in these colonies, indicative of transgene silencing. We therefore investigated transient expression by transfecting O4G EpiSCs with PB-Nr5a1 only. Transfected cells were transferred to 2i/LIF after 48 hours. Vector-only transfections yielded no colonies, whereas ~100 colonies per 1×104 plated cells were obtained from Nr5a1 transfection. Expanded colonies lacked detectable transgene integration or expression (Fig. 2C). Similar results were obtained by transient transfection of Nanog-GFP (TNG) EpiSCs. The Nanog reporter is barely detectable in EpiSCs, in line with low levels of Nanog transcript, but is bright in ES cells and iPS cells. Five or six days after transfection with Nr5a1, clusters of GFP-positive cells became visible within emerging colonies (Fig. 2D). Nr5a2 is expressed as a different isoform in ES cells compared with somatic cells owing to the use of an internal promoter (Gao et al., 2006). The gene-trap insertion does not induce the longer, somatic cell transcript (see Fig. S2A,B in the supplementary material). Nonetheless, we tested both isoforms and found that either can reprogramme EpiSCs (see Fig. S2C in the supplementary material).
We expanded cells obtained by transient transfection and interrogated them for expression of ES cell and EpiSC markers. In all cases, we found that the lineage-specific markers Lefty (Lefty1) and Fgf5 were downregulated, and ES cell-specific markers, such as Rex1 (Zfp42), Nr0b1, stella (Dppa3), Klf2 and Klf4, were upregulated (Fig. 2E). Immunostaining for the H3K27me3 silencing mark on the X chromosome showed that the nuclear focus in the XX EpiSCs was erased in derivative Nr5a1-reprogrammed cells (Fig. 2F).
Propagation in 2i/LIF with appropriate marker expression is a good surrogate for ground state pluripotency, but the definitive criterion is contribution to healthy chimaeras. We injected two clones of O4G cells reprogrammed by transient expression of Nr5a1 into C57BL/6 blastocysts. Twenty of 25 live pups showed overt coat colour chimaerism, with 14 from one clone and six from the other. Three female chimaeras mated with C57BL/6 males produced agouti offspring, demonstrating transmission of the EpiSC genome (Fig. 2G).
Nr5a receptors are potent EpiSC reprogramming factors but are not ES cell self-renewal factors
By utilising the MSCV enhancer/promoter, which drives moderate transcription and is subject to silencing, the screen was tailored to isolate factors that do not require high or sustained expression. This could explain why we did not isolate Nanog, Klf2 or Klf4, which have previously been shown to effect EpiSC reprogramming by stable transfection (Guo et al., 2009; Hall et al., 2009; Hanna et al., 2009). To examine directly whether Nr5a receptors are more potent, we prepared a series of transient expression constructs driven by the powerful CAG unit (Chambers et al., 2003; Niwa et al., 1991). In addition to known reprogramming factors, we included the orphan nuclear receptor Nr0b1 (Dax1), which is a well-defined target of Nr5a1 in steroidogenic cells (Hoivik et al., 2009), is regulated by Oct4/Sox2 in ES cells (Sun et al., 2008), and directly interacts with Oct4 (van den Berg et al., 2010). We transiently transfected wells of 2×104 TNG EpiSCs and then cultured cells in 2i/LIF for 9 days (Fig. 3A). An average of over 200 Nanog-GFP-positive Epi-iPS cell colonies were obtained from Nr5a transfection, a frequency of at least 1% (assuming an EpiSC transient transfection efficiency of 50%, which is rather higher than routinely obtained). Slightly fewer colonies were obtained from TNG cells, possibly because the starting EpiSC population is more differentiated in the absence of selection. Nr5a1 transfections consistently produced more colonies than Nr5a2 transfections. An order of magnitude fewer colonies were obtained with Nanog, Klf2 or Klf4, and no colonies were obtained from Nr0b1 transfections.
We then investigated the combination of transient Nr5a with stable Nanog or Klf4 transgene expression. EpiSCs stably transfected with either Nanog or Klf4 convert at a frequency of 0.5-1% upon transfer to 2i/LIF. This was substantially increased upon transient transfection with Nr5a factors (Fig. 3B), by 4- to 6-fold in the case of Klf4 transfectants and ~2-fold for Nanog. Furthermore, significantly more colonies were obtained than from parallel Nr5a transfections of parental EpiSCs (Fig. 3A). In the case of transient Nr5a1 transfection into Klf4-expressing EpiSCs, the yield of over 1000 colonies corresponds to a frequency of ~10% of transiently transfected cells. These results indicate that Nr5a receptors act combinatorially with Klf4 or Nanog to promote EpiSC reprogramming.
The previously identified EpiSC reprogramming factors Klf4, Klf2 and Nanog are known to inhibit ES cell differentiation and sustain self-renewal without addition of LIF (Chambers et al., 2003; Hall et al., 2009; Niwa et al., 2009). We tested whether forced expression of Nr5a receptors (see Fig. S3 in the supplementary material) could similarly confer reduced dependency on LIF. Upon LIF withdrawal, Nr5a-transfected ES cells maintained Oct4 expression. However, they lost the expression of the ES cell markers Nanog and Rex1 (Fig. 3C) and acquired a spread, differentiated morphology (data not shown). We also observed that Rex1-GFP knock-in reporter ES cells transfected with Nr5a lost GFP expression throughout the entire population after LIF withdrawal (Fig. 3D). Therefore, we find no evidence that Nr5a receptors sustain ES cell self-renewal even though they retard downregulation of Oct4 (Gu et al., 2005).
Sustained expression of Oct4 is not sufficient to reprogramme EpiSCs
Oct4 is downregulated by 24 hours in parental EpiSCs transferred to 2i/LIF, but we noted that expression is maintained in Nr5a transfectants. Nr5a1 and Nr5a2 have been reported to bind to Oct4 upstream elements and promote transcription (Barnea and Bergman, 2000; Gu et al., 2005). We tested whether maintenance of Oct4 expression is sufficient to mediate reprogramming in 2i/LIF. Transient transfection with a PB-Oct4 expression vector did not yield any iPS cells. We therefore generated stable transfectants driving modest overexpression of Oct4 (Fig. 4A). Upon transfer to 2i/LIF these Oct4-EpiSCs maintained Oct4 transgene expression while downregulating endogenous Oct4 (Fig. 4A). They showed reduced cell death compared with parental EpiSCs, but differentiated. Oct4-GFP expression was lost completely and no iPS cell colonies were obtained. A caveat to this experiment is that increased Oct4 can cause ground state cells to differentiate (Niwa et al., 2000), raising the possibility that iPS cells might be lost if Oct4 overexpression continued after reprogramming. To test whether the Oct4 transfectants could produce Epi-iPS cells, we transiently transfected them with Nr5a1. Epi-iPS cell colonies were readily recovered and expanded in 2i/LIF (Fig. 4B). They showed transcriptional resetting to the ground state (Fig. 4C). Interestingly, these cells no longer overexpressed Oct4 (Fig. 4D), indicating either that the transgene is silenced upon reprogramming or that autoregulatory feedback is in operation. These observations indicate that although sustaining the expression of Oct4 may be part of the contribution of Nr5a receptors to reprogramming, it is not sufficient.
We also investigated whether Nr5a receptors directly induce Oct4 enhancer switching. In EpiSCs and ES cells, Oct4 expression is directed by distinct proximal (PE) and distal (DE) enhancers, respectively (Tesar et al., 2007; Yeom et al., 1996). Potential Nr5a binding sites have been identified in both enhancers as well as in the Oct4 proximal promoter (Barnea and Bergman, 2000; Gu et al., 2005). The Oct4ΔPE-GFP transgene is active in ES and iPS cells but not in EpiSCs (Bao et al., 2009; Yoshimizu et al., 1999). Stable transfection with Nr5a1 did not activate ΔPE-GFP in EpiSCs maintained in FGF plus activin (Fig. 4E). Upon transfer to 2i/LIF, GFP became apparent in patches of cells only after 5-6 days (Fig. 4F). Reporter expression then spread throughout the expanding colonies, similar to observations described above for Nanog-GFP. These observations confirm that the distal enhancer is activated upon reprogramming, but also show that it is not an immediate target of Nr5a receptors.
Like activation of the distal Oct4 enhancer, most ES cell-specific markers were not acutely upregulated after transfer into 2i/LIF and their expression rose gradually over 7 days in reprogramming cells (Fig. 4G). An exception is Klf2, which was upregulated within 12 hours (Fig. 4H). Klf2 is an integral component of the core gene regulatory network of ground state pluripotency (Hall et al., 2009; Jiang et al., 2008; Smith, 2010) and can replace Klf4 in reprogramming (Nakagawa et al., 2008). It has recently been suggested that the expression of endogenous Klf2 might mediate conversion of feeder-maintained EpiSCs (Greber et al., 2010). However, acute upregulation of Klf2 is not sufficient in our conditions as it was observed in non-transfected cells that subsequently died or differentiated. Nr5a transfection did increase this initial expression of Klf2 in 2i/LIF (Fig. 4H). It has been suggested that Klf2 could be a direct target of Nr5a receptors (Heng et al., 2010), and transfection with Klf2 can enable EpiSC reprogramming in 2i/LIF (Hall et al., 2009). We therefore tested the possibility that a combination of Klf2 and Oct4 regulation could account for the reprogramming efficacy of Nr5a receptors by transfecting the Oct4-EpiSCs with Klf2. As observed in parental EpiSCs, we found that Klf2 transient transfection yielded only a small number of colonies – an order of magnitude fewer than with Nr5a1 (Fig. 4B).
Collectively, these findings indicate that induction of previously identified reprogramming factors is not sufficient to explain the potent ability of Nr5a receptors to reset the naïve pluripotent state.
Reprogramming factors have primarily been discovered by testing candidate genes that are implicated in the pluripotent phenotype of ES cells (Feng et al., 2009; Silva et al., 2009; Takahashi and Yamanaka, 2006). In this study, a genomic screening approach has identified the potent activity of Nr5a nuclear receptors. Transient expression of these receptors in combination with 2i/LIF culture can reliably reset EpiSCs to ground state pluripotency. Furthermore, when combined with Klf4, reprogramming frequency is ~1 in 10 transfected cells. Although Nr5a receptors are known to regulate Oct4 expression, our findings indicate that neither increasing nor maintaining Oct4 levels is sufficient to reprogramme EpiSCs into naïve pluripotent stem cells.
The Nr5a class of nuclear receptors are thought to be able to function independently of ligand (Fayard et al., 2004). They are well-described regulators of steroid metabolism and endocrine homeostasis (Fayard et al., 2004; Hoivik et al., 2009). Nr5a1 also plays an essential role in sexual differentiation, whereas Nr5a2 has functions in primitive and definitive endoderm. Neither gene is essential in ES cells, nor for formation of the epiblast in vivo (Hoivik et al., 2009). Nr5a1 is not expressed in ES cells and there is no indication of loss of pluripotency or self-renewal capacity in Nr5a2-deficient ES cells (Gu et al., 2005). In the embryo, deletion of Nr5a2 causes disorganisation of the egg cylinder with loss of Oct4 and failed gastrulation (Gu et al., 2005; Pare et al., 2004). However, this phenotype is non-cell-autonomous and appears to be secondary to defective visceral endoderm development because Nr5a2-null cells are capable of progressing through gastrulation in tetraploid aggregation chimaeras (Labelle-Dumais et al., 2006). Consistent with a non-essential role in Oct4 regulation, EpiSCs lack both Nr5a1 and Nr5a2 yet show abundant Oct4.
Recent reports have indicated that EpiSCs may revert to ES cell status without genetic manipulation (Bao et al., 2009; Greber et al., 2010). In those studies, however, EpiSC cultures were undefined, being maintained on feeders with serum or serum replacement and without activin. EpiSCs may represent a heterogeneous mixture of cells under these conditions (Chou et al., 2008). Interestingly, these cultures show detectable levels of Klf4 and Klf2 and have similar levels of Nanog to ES cells (Greber et al., 2010; Tesar et al., 2007). By contrast, under defined, feeder-free, serum-free conditions using FGF and activin (Brons et al., 2007), Klf2 and Klf4 are not expressed and Nanog expression is 5- to 10-fold lower than in ES cells, whereas Oct4 and Sox2 are fully maintained (Guo et al., 2009; Hall et al., 2009). From EpiSCs in these defined conditions the incidence of spontaneous generation of cells that can expand in rigorous 2i/LIF ground state culture is less than 1 in 1×10−6. The relevance of EpiSC conversion to direct reprogramming is corroborated by a report, while our study was being finalised, that Nr5a factors can enhance the production of iPS cells from mouse fibroblasts (Heng et al., 2010). These authors assayed a panel of nuclear receptors in combination with the canonical reprogramming factors (Takahashi and Yamanaka, 2006). They also reported that Nr5a receptors can substitute for Oct4. However, in the case of EpiSCs, endogenous Oct4 is already present, yet Nr5a factors have a profound effect. Therefore, the major activity of Nr5a receptors in EpiSCs must be additional to inducing or replacing Oct4. It will be of great interest in future studies to determine which among the many proposed targets of Nr5a receptors (Heng et al., 2010) are functionally relevant to their reprogramming action and whether these are identical in the somatic cell and EpiSC contexts. The high frequency of EpiSC reprogramming triggered by Nr5a receptors, which is complete within 7 days, should also facilitate biochemical dissection.
Unlike other EpiSC reprogramming factors, Nr5a receptor deletion does not compromise ES cells (Gu et al., 2005), nor does their overexpression promote self-renewal or suppress differentiation (Fig. 3C). In fact, the silencing of integrated Nr5a1 constructs and the greater efficiency of transient transfection suggest that continued high expression might even be deleterious in naïve pluripotent cells. The ability of Nr5a factors to induce pluripotency is therefore rather unexpected and likely to have a different origin from previously described factors. From inspection of UniGene EST profiles we observed that both Nr5a1 and Nr5a2 are expressed in mouse oocytes, zygotes and early cleavage stage embryos. It is therefore conceivable that these nuclear receptors might act as maternal reprogramming factors in the zygote and that this underlies their capacity to reset the pluripotent ground state. It will be interesting to examine the effect of oocyte-specific deletion of Nr5a1 and Nr5a2.
In conclusion, these results validate EpiSC conversion in defined culture as an experimental system for interrogating molecular reprogramming and establish an unanticipated function of Nr5a receptors in the induction of pluripotency.
Masayo Fujiwara and Isobel Eyres provided outstanding technical support. We thank Jose Silva for discussion and comment on the manuscript; Jenny Nichols for EpiSC derivations; Bill Mansfield for chimaera generation; Rachel Walker for flow cytometry; Peter Humphreys for imaging; Samuel Jameson and staff for mouse husbandry; Siquin Bao and Azim Surani for ΔPE-GFP EpiSCs; and Marko Hyvonen for providing recombinant proteins. DNA sequencing was performed by the Sequencing Facility of the Department of Biochemistry, University of Cambridge. This study was funded by the Medical Research Council of the UK, The Wellcome Trust, and the European Commission Framework Seven project EuroSyStem. G.G. is a Medical Research Council Stem Cell Career Development Fellow and A.S. is a Medical Research Council Professor. Deposited in PMC for release after 6 months.
Competing interests statement
The authors declare no competing financial interests.
Supplementary material for this article is available at http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.052753/-/DC1
- Accepted July 19, 2010.
- © 2010.