The Oct4 homologue PouV and Nanog regulate pluripotency in chicken embryonic stem cells

Embryonic stem cells (ESC) are self-renewing pluripotent cells that can be maintained in culture for an indefinite period. In mammals, pluripotency is under the control of key transcription factors, including Oct4 (also known as Pou5f1 - Mouse Genome Informatics)(Nichols et al., 1998), Nanog(Mitsui et al., 2003; Chambers et al., 2003), Sox2(Avilion et al., 2003) and FoxD3 (Hanna et al., 2002). Oct4 is found in oocytes and is expressed in cleavage stage cells up to the morula stage (Kirchhof et al.,2000), and subsequently in the epiblast of the pre-primitive streak stage embryos. Oct4 expression is downregulated in trophectodermal cells but maintained in the inner cell mass, becoming restricted to primordial germ cells and oocytes(Kehler et al., 2004; Boiani et al., 2002). In vitro, Oct4 is expressed in proliferating murine and primate (including human) ESC, as well as in tumourigenic cells such as embryonal carcinoma (EC)(Ben-Shushan et al., 1995) and germ cell tumour (GCT) cells (Looijenga et al., 2003).

Oct4 appears to control cell-fate decisions of ESC in vitro. Inhibition of Oct4 expression in mouse ESC (mESC) causes a loss of proliferation and the induction of trophectodermal and endodermal markers(Velkey and O'Shea, 2003; Hay et al., 2004). By contrast, overexpression of Oct4 leads to primitive endoderm differentiation (Niwa et al.,2000) and it appears that a fine balance between Oct4 and Cdx2 expression controls the choice between embryonic and trophectoderm cell fates (Niwa et al.,2005; Tolkunova et al.,2006).

Oct4 contains a POU-specific domain and a POU homeodomain and belongs to the class V POU homeodomain family of transcription factors. A complex of proteins including Oct4 and Sox2 has been found to regulate expression of the growth factor Fgf4 (Dailey et al.,1994; Ambrosetti et al.,1997) and of the transcription factors Utf1(Nishimoto et al., 1999),Zfp42 (Rex1) (Ben-Shushan et al.,1998), Fbx15 (also known as Fbxo15 - Mouse Genome Informatics)(Tokuzawa et al., 2003), Nanog(Kuroda et al., 2005; Rodda et al., 2005) and Sox2 itself (Tomioka et al., 2002). Different nuclear receptors participate in the regulation of Oct4expression including Sf1 (Barnea and Bergman, 2000), Lrh-1 (Nr5a2)(Gu et al., 2005a), Gcnf(Nr6a1) (Fuhrmann et al.,2001; Gu et al.,2005b), CoupTF (Nr2f2)(Ben-Shushan et al., 1995) and Rar/Rxr heterodimers, the latter being responsible for the downregulation of Oct4 expression by retinoic acid(Schoorlemmer et al., 1994; Pikarsky et al., 1994). Oct4 expression is under the control of its own protein(Okumura-Nakanishi et al.,2005; Chew et al.,2005) through specific response elements located in its own promoter (Yeom et al., 1996; Nordhoff et al., 2001; Gu et al., 2005a; Gu et al., 2005b).

Nanog expression is also confined to pluripotent tissues and cell lines and its overexpression is able to maintain mESC in an undifferentiated state, even in the absence of Lifr/gp130 stimulation. Inhibition of Nanog expression in mESC results in their differentiation into primitive endoderm (Chambers et al.,2003; Mitsui et al.,2003).

To date, this relationship between Oct4 and/or Nanog and stem cell pluripotency has only been demonstrated in mammals. Indeed, in zebrafish, it was reported that the pou2 gene (also known as spg and pou5f1 - ZFIN), initially identified by a mutation that caused neural and endoderm defects, is the fish homologue of the mammalian Oct4 gene based on protein similarities, chromosomal syntenic relationship and developmental expression pattern, but not in terms of function (Burgess et al.,2002). No evaluation of a putative role in fish ESC pluripotency was described in an assessment of murine Oct4 activity in medaka ESC(Hong et al., 1998). The X. laevis Pou91 (XlPou91) gene product, encoded by one of three X. laevis PouV genes, has been demonstrated to have a similar activity to the mouse Oct4 gene in mESC and to participate in the maintenance of putative stem cell populations during early development(Morrison and Brickman,2006).

Given that Oct4 appears to be so important in the maintenance of pluripotency, a report suggesting that the chicken genome lacks a homologue of Oct4 (Soodeen-Karamath and Gibbins, 2001) was very surprising. Indeed, no corresponding sequence was identified in the chicken genome annotation, even in the latest release (Ensembl 42, December 2006).

Here we report the isolation of chicken PouV (cPouV) and Nanog (cNanog), homologues of mammalian Oct4 and Nanog. Both genes are expressed in early embryos before gastrulation and thereafter in germ cells. Taking advantage of chick ESC (cESC)(Pain et al., 1996; Petitte et al., 2004), we demonstrate that chicken PouV and Nanog are required for the maintenance of cESC pluripotency and for continued proliferation. Together,these findings show that the mechanisms by which these two genes regulate pluripotency and self-renewal are not exclusive to mammals.

Oligonucleotides (Proligo) were designed using Primer 3 software(http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi)and are listed in Table 1. The coding sequences of the different genes were identified using the chicken genome assembly(http://www.ensembl.org/Gallus_gallus/),GenBank(http://www.ncbi.nlm.nih.gov)or sequenced directly from newly isolated clones.

Total RNA from cESC and from 2-day-old chicken embryoid bodies (cEB)obtained as previously described (Pain et al., 1996) were reverse transcribed. The cDNAs were subject to a subtractive liquid hybridisation procedure(http://www.genome-express.com/). The transcripts enriched in cESC were subcloned, sequenced and filtered sequences assembled using PHRAP software. Target sequences delivered by the assembly process were subject to BLAST analysis.

The cDNA library from chicken embryonic stem cell mRNA(Acloque et al., 2001) was screened using T7 or T3 vector primers and internal sequences P06(381)S(5′-GTTGTCCGGGTCTGGTTCT-3′) or P06(382)AS(5′-GTGGAAAGGTGGCATGTAGAC-3′) derived from the 1P06g01 initial clone. A 5′-RACE strategy was developed with the P06RAAS2(5′-TGAGTGAAGCCCAGCATGAT-3′) primer followed by a second amplification with P06RAAS1 (5′-AACATCTTCCCATAGAGCGTGC-3′) and AnchPS (5′-GACCACGCGTATCGATGTCGACTTTTTTTTTTTTTTT-3′) primers. A second round of amplification using P06(pL7-2)AS(5′-TGCTTGAGGTCCTTGGCAAA-3′) and PCRprimseq primers led to the isolation of 300 bp upstream of the 1P06g01 clone, including an in-frame ATG. A full-length cDNA was cloned into pGEM-T-easy (Promega) using primers P06EcoRIS (5′-ATGAATTCATGCATGTAAAAGCCAAA-3′) and P06EcoRIAS(5′-ATGAATTCTCAGTGGCTGCTGTTGTT-3′).

Real-time RT-PCR was performed using the MXP-3000P PCR-system (Stratagene)using Mix-Quantitect SYBR Green (Qiagen). Samples were run in duplicate and gene expression levels were calculated using the ΔΔCt method(http://www.gene-quantification.info)with the chicken ribosomal gene RS17 (X07257) as reference. The number of independent experiments performed is indicated in each figure legend.

Hens' eggs were incubated for 0-36 hours and embryos staged according to Eyal-Giladi and Kochav (Eyal-Giladi and Kochav, 1976) for pre-primitive streak stages and according to Hamburger and Hamilton (Hamburger and Hamilton, 1951) for later stages. Embryos were subjected to whole-mount in situ hybridisation (Streit and Stern, 2001). Fluorescent Vasa and cPouVprobes were labelled with digoxigenin and fluorescein, respectively, and successively revealed using an HRP-coupled anti-digoxigenin and an HRP-coupled anti-fluorescein antibody and the TSA-Plus Cyanine3/Fluorescein system (Perkin Elmer). SSEA-1 labelling (DSHB, Iowa) was performed on frozen sections (15μm) and revealed with an anti-mouse IgM conjugated to Texas Red (Abcam). Adjacent sections were processed for in situ hybridisation as previously described (Strähle et al.,1994).

Reverse-transcribed chicken embryonic stem cell mRNA was used with P06GFPEcoRIS (5′-GGGAATTCGCATGTAAAAGCCAAA-3′) and P06GFPKpnIAS(5′-ATGGTACCTCAGTGGCTGCTGTTGT-3′) primers to amplify the cPouV coding region. The product was subcloned into pEGFP-C1(Clontech) to produce the pGFP-cPouV expression vector. cPouV cDNA was amplified with P06EcoRIS (5′-ATGAATTCATGCATGTAAAAGCCAAA-3′)and P06EcoRIAS (5′-ATGAATTCTCAGTGGCTGCTGTTGTT-3′) primers and cloned into pCAGIP (Niwa et al.,2000). The 1.8 kb pou2 zebrafish coding sequence was amplified from a pCSL2-Pou2 template using Pou2EcoRIS(5′-ATAGAATTCTATGACGGAGAGAGCGCAG-3′) and Pou2EcoRIAS(5′-GTAGAATTCTTAGCTGGTGAGATGACCC-3′) primers and cloned into pCAGIP. Murine Oct4 and Nanog coding sequences were reverse transcribed from mESC total RNA with primer pairs mOct4EcoRIS(5′-ATGAATTCTGCTGGACACCTGGCTTC-3′) with mOct4EcorIAS(5′-ATGAATTCTTAACCCCAAAGCTCCAG-3′) and mNanogXhoIS(5′-GTCTCGAGATGAGTGTGGGTCTTCC-3′) with mNanogNotIAS(5′-ATGCGGCCGCTCATATTTCACCTGGT-3′), respectively, then inserted into pCAGIP. The cNanog coding sequence was obtained from reverse-transcribed cESC total RNA using cNanogEcoRIS(5′-ATGAATTCATGAGCGCTCACCTGGCC-3′) and cNanogEcoRIAS(5′-ATGAATTCCTAAGTCTCATAACCATT-3′) primers and cloned into pCAGIP.

The p(ATGCAAAT)×3-luc reporter gene was constructed by inserting double-stranded oligonucleotides Oct4BS(5′-CTAGCATGCAAATAACAGCGCGCATGCAAATAACAGCGCATGCAAATAACAGCGCCCC-3′)and Oct4BAS(5′-GGGGCGCTGTTATTTGCATGCGCTGTTATTTGCATGCGCGCTGTTATTTGCATG-3′)into the pGL3 vector (Promega). To construct the pΔPE-luc reporter gene,a 1.4 kb fragment from the mOct4 distal enhancer was amplified from pGOF18ΔPE-GFP using ODES(5′-GTACGCGTGAATTCAGACAGGACTGCTGGGC-3′) and SVAS(5′-AGCATCACAAATTTCACAAATAAAGAATTCACGGCTTT-3′) primers(Hong et al., 2004) and subcloned into pGL3. For luciferase assays, ZHBTc4 cells were plated at 1×10⁵ cells per well with 2 μg/ml tetracyclin. Twenty-four hours later, 75 ng of reporter plasmid, 150 ng of the test plasmid and 10 ng of the Renilla reporter plasmid were co-transfected using 600 ng FuGENE 6 (Roche) and incubated overnight before fresh medium was added with 2μg/ml tetracyclin. Cell lysates were analysed 48 hours after transfection as described by the manufacturer (Promega).

pFLΔNeo was obtained by inserting into pBSK the 2 kb PCR-amplified product mU6ΔNeoΔ, derived from the mU6ΔNeo-ΔApaIDXhoI template (Coumoul et al.,2004) using mU6SmaIS(5′-ATCCCGGGGTATATCCGACGCCGCCAT-3′) and mU6HindIIIAS(5′-ATAAGCTTAACAAGGCTTTTCTCC-3′) primers. Double-stranded short hairpin (sh) RNA was cloned into pFLΔNeo, generating pFLΔNeo-XshRNA vectors for each gene to be targeted. The oligonucleotides containing the HindIII and XhoI sites used for generating the 21 bp shRNA sequence were: cPouV-shRNA-2S(5′-AGCTTAAGATGTTCAGCCAGACCACCTTCAAGAGAGGTGGTCTGGCTGAACATCTTTTTTTTC-3′)and cPouV-shRNA-2AS(5′-TCGAGAAAAAAAAGATGTTCAGCCAGACCACCTCTCTTGAAGGTGGTCTGGCTGAACATCTTA-3′)against cPouV; cNanog-shRNA-1S(5′-AGCTTAACAGAAACCTTCAGGCTGTGTTCAAGAGACACAGCCTGAAGGTTTCTGTTTTTTTTC-3′)and cNanog-shRNA-1AS(5′-TCGAGAAAAAAAACAGAAACCTTCAGGCTGTGTCTCTTGAACACAGCCTGAAGGTTTCTGTTA-3′)against cNanog; as well as cNanog-shRNA-3S(5′-AGCTTAAGGCCAAGAGCCGCACAGCTTTCAAGAGAAGCTGTGCGCTCTTGGCCTTTTTTTTC-3′)and cNanog-shRNA-3AS(5′-TCGAGAAAAAAAGGCCAAGAGCCGCACAGCTTCTCTTGAAAGCTGTGCGGCTCTTGGCCTTA-3′)and cOct6-shRNA-3S(5′-AGCTTAAGCAGCGGCGGATCAAGCTGTTCAAGAGACAGCTTGATCCGCCGCTGCTTTTTTTTC-3′)and cOct6-shRNA-3AS(5′-TCGAGAAAAAAAAGCAGCGGCGGATCAAGCTGTCTCTTGAACAGCTTGATCCGCCGCTGCTTA-3′)against cOct6. The Cre-ERT2 coding sequence(Feil et al., 1997) was cloned into the pCIFL-Hygro vector, derived from pCINeo (Promega), by replacing the neomycin cassette with a hygromycin cassette to produce pCre-ERT2-Hygro.

cESC were maintained and transfected as previously described(Pain et al., 1996; Pain et al., 1999). For kinetic experiments, formation of cEB was achieved by allowing dissociated proliferating cESC to float in bacterial dishes. When used, retinoic acid was added at 10^-7 M 24 hours after plating and considered as T=0. Cycloheximide and actinomycin D were added to the culture medium at 10μg/ml for various times as indicated.

ZHBTc4 cells were maintained as described(Niwa et al., 2000). Expression of the endogenous murine Oct4 can be downregulated by addition of 1 μM doxycyclin (Sigma). For transfection,5×10⁶ cells were electroporated (BioPulser, BioRad) at 575μF with 25 μg of the various linearised vectors. From twenty-four hours after electroporation, doxycyclin was added at daily intervals, and puromycin was added 72 hours after electroporation at 1 μg/ml and administered daily for 6 days.

Once transfected and selected with 200 μg/ml neomycin for 7 days,resistant clones of cESC were pooled, transfected with the pCre-ERT2-Hygro vector and selected for 7 days with 0.75 μg/ml hygromycin. Clones were numbered, picked and individually observed during the induction of shRNA expression by adding 1 μM 4-hydroxytamoxifen to the medium. Morphology was assessed by direct microscopic observation and Wright Giemsa staining. For proliferation kinetic assays, clones were picked individually, the cells dispersed and plated in six wells in 250 μl medium. After 24 hours,4-hydroxytamoxifen was added and proliferation was assessed at different times using two wells per time point using Cell Proliferation Kit II (XTT)(Roche).

Subtractive hybridisation of cDNAs from cESC and chicken embryoid bodies(cEB) resulted in identification of a 228 bp cDNA fragment encoding a partial POU domain. A combination of the results of screening a cDNA library and 5′-RACE using mRNA from cESC, allowed us to define an open reading frame(ORF) of 888 bp coding for a 295 amino acid (aa) protein. Comparative analysis and phylogenetic tree construction using maximal parsimony and neighbour-joining methods revealed that this sequence is statistically more closely related to the other PouV proteins than to the other Pou factors(Fig. 1A)(Felsenstein, 1978; Saitou and Nei, 1987). This novel predicted protein is part of the PouV protein subfamily, which contains XlPou91, the D. rerio Pou2 and mammalian Pou5f1 proteins and exhibits high similarity with the other members of the family(Fig. 1B,C).

At the genome level, the novel gene was mapped to chicken chromosome GGA17,specifically between primers SEQ0256 and SEQ0257 described in the ChickRH6 whole-genome radiation hybrid (WGRH) panel(http://chickrh.toulouse.inra.fr/). Syntenic comparison identified a relationship between this chicken gene, XlPou91 and zebrafish pou2. This relationship appears to be absent, either deleted or displaced, in mammalian species, despite the presence of adjacent syntenic loci on mouse chromosome 2 (data not shown).

In conclusion, our data reveal the existence in the chicken genome of a gene belonging to the PouV gene subfamily. We will therefore henceforth refer to this new chicken gene as chicken PouV(cPouV) (GenBank accession DQ867024).

A chicken Nanog gene was predicted in the chicken genome annotation at reference ID ENSGALG00000014319 on chicken chromosome 1 (GGA1). Primers designed using this sequence were used to isolate a clone from the chicken embryonic stem cell library with a 930 bp ORF (GenBank accession DQ867025). Comparative analysis and phylogenetic tree construction revealed that this sequence is closely related to mammalian Nanog genes and that the predicted protein exhibits high similarity with the other Nanog proteins (Fig. 1D). This sequence contains a homeodomain of 57 aa, located between aa 98 and 155, but does not have the WWW repeat in the C-terminus that is characteristic of the mammalian Nanog subfamily (Pan and Pei,2005). In contrast to the recently reported chicken Nanog sequence identified in silico (Canon et al.,2006), our cloned protein does not indicate the existence of a 112 aa segment after aa 50 that could correspond to a putative alternatively spliced form.

To determine the expression profiles of cNanog and cPouV,we performed real-time RT-PCR experiments showing that proliferating cESC express high levels of cPouV and cNanog(Fig. 2A,D, time 0). cESC can be induced to differentiate either as cEB, by preventing cell attachment, or following treatment with chemical inducers such as DMSO or retinoic acid (RA)(Pain et al., 1996). During a 5-day RA treatment, cPouV expression was almost completely abolished in parallel to similar reductions in expression of the markers alkaline phosphatase (AP) and telomerase reverse transcriptase (Tert)(Fig. 2A). Chicken Gcnf expression was also strongly downregulated, as was expression of Sox2 and Nanog, although with a more complex profile. By contrast, Rarγ expression was upregulated following RA treatment (Fig. 2A). Treatment with cycloheximide, known to block de novo protein synthesis, did not affect the downregulation of cPouV and Gcnf transcription, whereas Nanog transcription was no longer responsive to RA, suggesting that downregulation of cPouV and Gcnf transcription are direct transcriptional events following RA treatment(Fig. 2B). Moreover, following actinomycin D treatment, which blocks transcription, a 50% decrease in the expression of cPouV and Gcnf was observed 8 to 12 hours after addition of the drug. A decrease of greater than 50% in NanogmRNA levels was seen as early as 30 minutes after treatment, suggesting that it has a very short half-life (Fig. 2C).

Expression of cPouV and Nanog was also strongly downregulated during formation of cEB (Fig. 2D), as was expression of Sox2 and AP. By contrast, Sox3, Cdx2 and Gata6 were induced, suggesting a complex differentiation process during formation of these three-dimensional bodies (Fig. 2E).

To determine the likely sites at which cPouV and cNanogfunction during normal development, transcripts were identified in embryos during progressive stages of development using whole-mount in situ hybridisation. cPouV mRNA was found to be ubiquitously expressed in the epiblast of pre-primitive streak stage embryos and in a salt-and-pepper fashion in the forming hypoblast (Fig. 3A,B,B′). As the primitive streak started to form,transcripts were strongly localised in the epiblast of the streak itself(Fig. 3C,C′) and in the mesoderm emerging from it, whereas expression in the lower layer tended to decrease (Fig. 3D-F,D′,F′). Expression in the area opaca was lost by stage 3⁺ (Fig. 3F). At later stages, cPouV continued to be expressed in the mesoderm, but was undetectable in the endoderm (Fig. 3G-I). At stage 8 and subsequently, cPouV was strongly expressed in the neural plate and neural tube with particularly strong expression in the anterior hindbrain/posterior midbrain(Fig. 3I). Later, at stage 9 and subsequently, cPouV was still expressed in neural tissue and expression appeared in primordial germ cells(Fig. 3J).

cNanog showed a different pattern. In pre-streak embryos,transcripts were detected in the whole epiblast but not in the forming hypoblast (Fig. 4A,B,B′). As the primitive streak started to form, transcripts disappeared from the primitive streak epiblast but were still expressed throughout the area pellucida epiblast (Fig. 4C-E,C′,D′). At the end of gastrulation (stage 4^-4⁺), cNanog mRNA was quickly downregulated in the epiblast and persisted in a crescent anterior to the emerging head process(Fig. 4F-H). As the neural plate formed (stages 6-8), expression in the epiblast was restricted to the anterior neural plate (Fig. 4H-J,H′).

In conclusion, following initial, high levels of expression in early pluripotent epiblast cells, cPouV and cNanog present a very restricted in vivo pattern of expression during early embryonic development.

In order to determine the expression profile of cPouV and cNanog during late embryonic development, quantitative (Q) RT-PCR analysis was performed on chicken embryo tissues at day 16 to 17, including intestine, muscle, kidney, spleen, lung, brain, liver, heart and gonads. Expression was detected in gonads (male and female, data not shown) but at a level 270-fold lower than in proliferative cESC, and also in spleen and brain but 530-fold and 1100-fold lower, respectively, than in cESC (data not shown). In situ hybridisations confirmed that cPouV is expressed in gonads,with expression restricted to germ cells. At stage 33 (7 days of incubation), cPouV expression was detected in a salt-and-pepper fashion in the forming gonads (Fig. 5A,B). The cPouV-positive cells were found to also express the germ-cell-specific markers Cvh (by mRNA detection, Fig. 5C-F) and SSEA-1 (by immunostaining, Fig. 5G,H). Sox2 and Cvh, the expression of which is high in embryonic brain and gonads, respectively, were used as control gene markers for tissue specificity (data not shown).

cNanog expression was also detected by QRT-PCR in heart, brain,kidney and gonads, but at levels 20-, 25-, 90- and 100-fold lower,respectively, than in cESC (data not shown). In early embryos, Nanogwas also expressed in scattered cells in the germinal crescent: these cells are likely to correspond to the future germ cells(Fig. 4I and arrowhead in H′). Later in development, cNanog was still expressed in germ cells at stage 33 (Fig. 5I-K)identified as SSEA-1-positive, in a similar manner to the cPouV-expressing cells (Fig. 5L,M). However, cNanog expression became weaker at this stage compared to the previous stages. This expression profile was observed in both male and female embryos (data not shown).

In conclusion, expression of cPouV and Nanog becomes restricted to germ cells at later stages of embryonic development.

In order to compare cPouV function with its orthologues, coding sequences of cPouV, murine Oct4 (mOct4), XlPou91 and zebrafish pou2 were transfected into cESC and mESC. In cESC, overexpression of cPouV using the pCAGIP vector impaired the isolation of proliferating clones. Using pCMV-based vectors, gene expression analysis revealed a 4-fold induction of cPouV, but a strong decrease in expression of Nanog and Tert. By contrast, strong upregulation of Gata4, Gata6 and Cdx2,associated with differentiation, was observed(Fig. 6A). Ectopic expression of XlPou91 induced a similar expression profile, with an increase in endogenous cPouV and of differentiation markers Gata4, Gata6and Cdx2, and a strong decrease in Nanog and Tertexpression (Fig. 6B). By contrast, overexpression of mOct4 did not modify cPouV or Nanog endogenous expression levels and induced only a slight increase in endogenous chicken Cdx2 gene expression(Fig. 6B).

As previously described (Niwa et al.,2002), it was not possible to isolate clones of cells overexpressing mOct4 after transfection of mESC with the neomycin resistance overexpression plasmid. The same pCAGIP vector was used to overexpress cPouV and XlPou91 in mESC, but clones could only be isolated of mESC expressing XlPou91. Endogenous expression of Oct4 was maintained and expression of Gata4, Gata6 and Cdx2, as well as of mesendodermal markers including Hnf1,brachyury, Sox17 and laminin B1, was observed(Fig. 6C). Using a pCMV-based expression vector, cPouV expression enabled identification of clones presenting a similar expression profile, i.e. with a maintenance of endogenous Oct4 expression but only a slight increase in Gata4, Gata6, Cdx2,Hnf1, brachyury, Sox17 and laminin B1 expression(Fig. 6D).

In conclusion, high ectopic expression of cPouV impairs the proliferation of both cESC and mESC, but a moderate level of expression of exogenous cPouV is tolerated by cESC and mESC with an associated modification of the observed gene expression profile.

A good test of whether cPouV is functionally equivalent to its murine counterpart Oct4, is to assay the ability of the chick gene to rescue the ZHBTc4-inducible cells in which endogenous Oct4 expression is downregulated by addition of doxycyclin. Transfection of ZHBTc4 cells with expression vectors for mOct4 or XlPou91 allowed isolation of proliferating clones in the presence or absence of doxycyclin,(Fig. 7A,B) as predicted(Niwa et al., 2000; Morrison and Brickman, 2006). In the presence of doxycyclin, expression of cPouV was able to support the growth of slowly proliferating AP-positive colonies(Fig. 7C) with a rescue index(the ratio between the number of clones in the presence versus the absence of doxycyclin) of 0.5 (Fig. 7D),as compared with 1.0 for expression of mOct4 and 3.5 for XlPou91. However, the colonies recovered after cPouVexpression were limited in their capacity to be passaged or amplified and exhibited a differentiated morphology. No clones were obtained after expression of zebrafish pou2 in the presence of doxycyclin.

Real-time RT-PCR analysis performed on RNA from the clones generated by cPouV complementation revealed a complete loss of endogenous mOct4 mRNA, but high expression of the exogenous cPouV mRNA(data not shown). Expression of pluripotency-associated markers such as Nanog, Sox2, Utf1 and Zfp42 (Rex1) was maintained at the same level in cells complemented by XlPou91 as in cells complemented by mOct4 (and expression was even higher for Tert and Fgf4). Expression of these markers was reduced, but detectable, in cells complemented by cPouV, with the exception of Sox2 and Fgf4 for which no expression could be detected in the presence of cPouV (Fig. 7E).

To test the ability of this gene to transactivate specific Oct4-responsive elements, promoters containing either the mOct4 consensus binding site (ATGCAAAT), or the 1.4 kb ΔPE fragment from the mOct4 promoter(Yeom et al., 1996; Hong et al., 2004), linked to a luciferase reporter, were transfected into ZHBTc4 cells. These promoters were activated in ZHBTc4 cells treated with doxycyclin in the presence of the expression vectors coding for mOct4, XlPou91 or cPouV, as measured by luciferase activity (Fig. 7F). This interesting result suggests that the cPouV protein is able to recognise mOct4-response elements and activate transcription.

In conclusion, these experiments suggest that the cPouV gene is able to partially rescue the loss of mOct4 function in mESC, and does interact with and activate mOct4-dependent regulatory elements.

Overexpression of mouse Nanog protein in mESC results in growth factor-independent maintenance of the pluripotent cell phenotype. To test whether overexpression of cNanog can confer the same growth-factor independence on mESC, proliferation of mESC was assessed in the absence of LIF, after transfection of a cNanog expression plasmid. Colonies did form in the absence of LIF, indicating that cNanog is able to confer growth factor independence (Fig. 8A). In the absence of LIF, the transfected cells were undistinguishable from the parental cells, on the basis of morphology, AP staining and growth rate (Fig. 8B-G). Real-time RT-PCR analysis of these proliferating clones indicated that expression of pluripotent factors, including mOct4 and Sox2, was maintained, but with the notable exception of Fgf4, the expression of which was almost completely abolished(Fig. 8H).

In contrast to mESC, cESC are not dependent on a single cytokine for their proliferation and survival (Pain et al.,1996) (our unpublished results). Overexpression of cNanogconferred the ability of the cESC to grow in a low-serum medium in the absence of growth factors and cytokines that are usually required for proliferation(Fig. 8I). The clones obtained proliferated actively and were easily passaged and amplified (data not shown). It was particularly surprising to obtain proliferative avian primary stem cells in the presence of only 1% foetal bovine serum. Interestingly, under these drastic conditions, mNanog expression did not have any pronounced effect on the chicken cells. Real-time RT-PCR analysis confirmed overexpression of cNanog, the maintenance of Tert and reduced, but detectable, expression of cPouV and AP(Fig. 8J-K).

In conclusion, we have shown that cNanog functions in a very similar way to mNanog in mESC and has a more dramatic effect in cESC,where overexpression permits maintenance of the stem cell phenotype in the absence of growth factors and in low serum.

To assess cPouV and cNanog function, constructs expressing shRNAs were designed to knockdown transcripts of these genes, using a tamoxifen-inducible Cre system to activate the expression of the shRNAs. Following induction of Cre recombinase activity by tamoxifen addition, a rapid and dramatic morphological change was observed, involving changes associated with differentiation (Fig. 9A-D). These changes were seen in ∼60% of the clones when specific shRNAs were used against cPouV and cNanog(Fig. 9G) This morphological change was observed even in the presence of growth factors and was accompanied by a loss of AP activity and of SSEA-1 antibody staining(Fig. 9H), plus a growth rate alteration 48 and 96 hours after Cre induction(Fig. 9I). Comparison of the gene expression profiles between differentiated clones and clones that continued to proliferate revealed strong inhibition of endogenous cPouV expression as well as of cNanog and Gata4 and strong induction of Gata6 (Fig. 9J). No upregulation of Cdx2 was detected, nor of other mesendodermal markers such brachyury or Hnf3β. Similar experiments involving inhibition of another POU-domain gene, Oct6(Levavasseur et al., 1998),did not change the endogenous level of cPouV and proliferating clones were obtained (Fig. 9G,J).

When a similar analysis was performed using shRNA directed against cNanog mRNA, a similar process of differentiation occurred, with thin cytoplasmic protrusions (Fig. 9E,F), a loss of AP activity and SSEA-1 staining(Fig. 9G) and reduced proliferation (Fig. 9H). This phenomenon was observed with two distinct sequences, shRNA-1 and shRNA-3. Real-time RT-PCR expression analysis showed a drastic decrease in the expression of cPouV, Gcnf and Gata4(Fig. 9I) and an induction of Gata6.

In conclusion, inhibition of either cPouV or cNanog leads to a loss of proliferation of cESC and to the induction of differentiation.

Oct4 is established as one of the key factors controlling pluripotency and the unique self-renewing property of mammalian ESC(Chambers and Smith, 2004). Both overexpression and disruption of Oct4 in mESC leads to a loss of pluripotency and induces the cells to differentiate into primitive endoderm,characterised by high Gata6 expression(Li et al., 2004), and into trophectoderm expressing Cdx2(Niwa et al., 2000; Strumpf et al., 2005; Niwa et al., 2005; Tolkunova et al., 2006). In vivo, it is now thought that complex regulatory mechanisms lead to restricted expression in early pregastrulation embryos(Gu et al., 2005a; Boiani et al., 2002) and in the germ line (Kehler et al.,2004; Yeom et al.,1996).

Nanog, a homeodomain transcription factor, was identified as another key factor maintaining the pluripotency of mammalian ESC(Chambers et al., 2003; Mitsui et al., 2003; Hart et al., 2004). In mESC, Nanog overexpression has been shown to substitute for the requirement for growth factors in the maintenance of self-renewal. Disruption of Nanog leads to a loss of pluripotency and to induction of differentiation towards an endoderm-like state(Mitsui et al., 2003).

The existence and equivalent functions of homologues of these genes in non-mammalian vertebrates are still debated. Functional assays were used to identify the zebrafish pou2 gene as the Oct4 homologue(Burgess et al., 2002), but this gene appears to be mainly involved in the endoderm-specification cascade(Reim et al., 2004; Lunde et al., 2004). In Xenopus, XlPou91, a PouV gene, plays a significant role in the maintenance of pluripotent cells during early development and was shown to rescue Oct4 depletion in mESC. XlPou91 knockdown in vivo using morpholinos induces expression of Xcad3, which is considered to be the Xenopus homologue of Cdx2(Morrison and Brickman, 2006). These data are consistent with the idea that PouV family members,including murine Oct4, could act to prevent premature commitment of pluripotent cells present in vertebrate embryos prior to and during gastrulation.

cESC have been isolated and maintained in culture for long periods(Pain et al., 1996; Petitte et al., 2004; Van de Lavoir et al., 2006). These cells were derived from the culture of pre-primitive streak blastodermal cells and are characterised by the presence of typical ESC markers such as AP,Tert activity and reactivity with particular antibodies including ECMA-7,SSEA-1, SSEA-3 and EMA-1 (Pain et al.,1996; Petitte et al.,2004).

In a differential screen from proliferative cESC and cEB, we identified a new coding sequence containing a POU domain. Several strands of evidence support the view that this gene is the chicken homologue of mammalian Oct4. First, comparative analysis and phylogenetic tree construction reveal that this sequence belongs, with high probability, to the PouVsubfamily. Genomic analysis also demonstrates a clear syntenic conservation of the different loci between the non-mammalian species. We therefore refer to this gene as chicken PouV (cPouV). Second, this gene is expressed in vitro only in proliferating ESC. Its expression is rapidly downregulated once differentiation is induced by RA or during formation of cEB. This downregulation is maintained in the presence of cycloheximide,suggesting a direct effect of RA on transcription.

Third, cPouV is expressed in a complex pattern in the embryo,being expressed widely in the early epiblast and later becoming restricted to specific regions, including the mesoderm and nervous system. This initial expression in multipotent epiblast cells, which then becomes restricted once the cells start to be committed, is also shared by the zebrafish, Xenopus and mouse homologues, which have been implicated in regulation of early neural development and patterning(Ramos-Mejia et al., 2005; Burgess et al., 2002; Reim and Brand, 2002; Morrison and Brickman, 2006). During late development, cPouV expression becomes more restricted to migrating and proliferating germ cells, as demonstrated by co-localisation with Cvh-positive cells in the developing gonads. This germ-line-restricted expression is a feature shared with its murine counterpart, in contrast to the zebrafish and Xenopus homologues. We conclude that cPouV plays a similar role to its mammalian and non-mammalian homologues in pregastrulating embryos, but functions more like the mammalian homologue in germ cells.

Another feature of this chicken gene is its ability to induce differentiation when overexpressed in cESC. Expression of cPouV in cESC and mESC alters the morphology and reduces the growth rate of the ESC,and inhibits isolation of clones from cells in which cPouV is expressed from a very strong promoter. Expression using a moderate CMV promoter induces expression of differentiation markers such as Gata4,Gata6 and Cdx2. In mouse, these markers are associated with endodermal and trophectodermal lineages, but their function during early chicken development is still unknown. A similar profile of gene expression is obtained in parallel experiments with XlPou91, suggesting common target genes. Overexpression of cPouV in mESC led to a more moderate phenotype, with a slight induction of differentiation markers such as Gata6 and Cdx2, and also of Hnf1, brachyury and Sox17, which are known in mouse to be strongly expressed in mesendoderm structures (Tada et al.,2005; Yasunaga et al.,2006). cPouV is able to trigger a differentiation programme when overexpressed in both mESC and cESC.

cPouV is only able to partially rescue the phenotype of mOct4-deficient cells. In ZHBTc4 ES cells, cPouV expression can restore limited proliferation, in contrast to the zebrafish pou2gene but in agreement with recent findings regarding XlPou91 function(Morrison and Brickman, 2006). In the presence of cPouV, endogenous mouse Nanog expression in mESC is maintained at a low but detectable level. Expression of Utf1,Zfp42 (Rex1) and Tert is maintained, but expression of Sox2 and Fgf4 is completely abolished. These observations could explain the limited ability of these clones to be passaged and amplified. However, under these same conditions, the various factors are able to transactivate specific promoters containing either the Oct4consensus binding site or the mouse endogenous Oct4 promoter,suggesting that cPouV is able to substitute functionally for mouse Oct4. The N-terminal domains of the mouse and chicken genes are highly diverged. It is probable that stringent interprotein interactions are required for full activity of the chicken protein. A first attempt to test this hypothesis, by constructing molecular chimaeras between the mouse Oct4 N-terminus and the cPouV homeodomain and C-terminus,proved unable to restore complete ZHBTc4 cell proliferation (data not shown),suggesting that other mechanisms and/or partners are likely to be required.

We also report the isolation of the chicken functional orthologue of Nanog (cNanog) from proliferating cESC and demonstrate that cNanog also plays a role in the maintenance of self-renewal and pluripotency of cESC. The sequence we cloned is shorter than the one recently reported by Canon et al. (Canon et al.,2006). The cNanog expression profile differs both in vitro and in vivo from that of cPouV. Specifically, cNanogexpression is downregulated by differentiation but with a different time-course than cPouV in both RA-induced differentiation and cEB formation. In contrast to cPouV, cNanog transcription is maintained and possibly increased in the presence of cycloheximide and the mRNA half-life appears to be reduced in the presence of actinomycin, suggesting a short half-life of the mRNA. In vivo, the expression pattern of cNanog is also different from that of cPouV, with a more rapid disappearance from the epiblast and a subsequent restriction to the anterior neural plate. cNanog expression is detected in migrating germ cells, as was also observed for Nanog in the mouse(Yamaguchi et al., 2005), but also in germ cells of developing gonads.

Following overexpression of either mouse or chicken Nanog in mESC,the resulting proliferating clones are able to grow in the absence of the cytokine LIF, suggesting a functional complementation between the mouse and chicken genes. A notable exception is an almost complete loss of Fgf4expression in clones overexpressing cNanog. In cESC, a drastic reduction in serum in the culture medium revealed an important action of cNanog in maintaining proliferation of undifferentiated cells,suggesting that cNanog is able to stimulate proliferation and cell-cycle machinery in the absence of exogenous growth factors by acting directly on downstream targets.

Finally, inhibition of expression of cPouV and cNanog,using an inducible knockdown approach, promotes rapid growth arrest within 48 hours of shRNA induction. This inhibition of proliferation is accompanied by an induction of differentiation as detected by altered morphology, loss of SSEA-1 labelling and expression of Gata6 and Cdx2. This suggests that these two genes play a key role in the maintenance of the pluripotent character of cESC.

In conclusion, the identification of chicken PouV elucidates some aspects of epiblast proliferation and maintenance of pluripotency in vitro and in vivo, and points the way for a better understanding of germ cell development and proliferation in the chicken embryo. The chicken Nanog gene also plays a role in this process and the functional relationship between these two key genes requires further investigation.

We thank Drs Niwa and Smith for the ZHBTc4 cells, Dr Chambon for the Cre-ERT2 plasmid, Dr Schöler for the pGOF18ΔPE-GFP plasmid, Dr Deng for the mU6 promoter, Dr Brandt for the Spg plasmid, Dr Vignal for the cPouV localisation and Dr Brunet for his help in phylogenetic analysis. F.L. was supported by an INRA fellowship. This work was supported by Region Rhône-Alpes, GenAnimal program and by the Vivalis Company. The work in the laboratory of C.D.S. was supported by the MRC, BBSRC, National Institutes of Health (NIMH) and the European Union Network of Excellence`Cells into Organs' and in the laboratory of H.M.S. by the award of a BBSRC studentship to D.J.M.