Oct1 (Pou2f1) is a transcription factor of the POU-homeodomain family that is unique in being ubiquitously expressed in both embryonic and adult mouse tissues. Although its expression profile suggests a crucial role in multiple regions of the developing organism, the only essential function demonstrated so far has been the regulation of cellular response to oxidative and metabolic stress. Here, we describe a loss-of-function mouse model for Oct1 that causes early embryonic lethality, with Oct1-null embryos failing to develop beyond the early streak stage. Molecular and morphological analyses of Oct1 mutant embryos revealed a failure in the establishment of a normal maternal-embryonic interface due to reduced extra-embryonic ectoderm formation and lack of the ectoplacental cone. Oct1–/– blastocysts display proper segregation of trophectoderm and inner cell mass lineages. However, Oct1 loss is not compatible with trophoblast stem cell derivation. Importantly, the early gastrulation defect caused by Oct1 disruption can be rescued in a tetraploid complementation assay. Oct1 is therefore primarily required for the maintenance and differentiation of the trophoblast stem cell compartment during early post-implantation development. We present evidence that Cdx2, which is expressed at high levels in trophoblast stem cells, is a direct transcriptional target of Oct1. Our data also suggest that Oct1 is required in the embryo proper from late gastrulation stages onwards.
Oct1 (Pou2f1) belongs to the POU protein family (Veenstra et al., 1997), which historically included four transcription factors: the mammalian Pit1, Oct1 and Oct2 and C. elegans UNC-86. A common feature of the family is the POU domain, a bipartite DNA-binding motif consisting of two structurally independent subdomains, the POU-specific domain (POUs) and the POU homeodomain (POUH), which are tethered by a linker of variable length ranging from 14 to 26 amino acids (Phillips and Luisi, 2000). The POUS and POUH domains bind independently, but in a cooperative manner, to each half-site of the target consensus. This modular structure operates as a single functional unit while conferring high DNA-binding affinity and specificity (Pomerantz and Sharp, 1994; Verrijzer et al., 1992). POU factors bind to the asymmetrical octamer canonical sequence ATGCAAAT and variants of this motif, and this has been shown to drive the expression of both ubiquitous and tissue-specific genes (Schöler, 1991). Most of the known POU proteins are temporally and spatially restricted during development. In accordance with their expression patterns, POU factors play pivotal roles in specific cell fate determination events. Oct6 (Pou3f1), for example, regulates Schwann cell differentiation (Jaegle et al., 1996), Brn3.2 (Pou4f2) controls retinal ganglion cell survival and differentiation (Erkman et al., 1996) and Oct2 (Pou2f2) has been implicated in the transcription of octamer-containing promoters, such as those of immunoglobulin genes in B cells (Muller et al., 1988; Scheidereit et al., 1987). To our knowledge, Oct4 (Pou5f1) is the only POU factor that has a role during early embryogenesis, when it is essential for the specification of a pluripotent inner cell mass (ICM) (Nichols et al., 1998) and for primordial germ cell survival (Kehler et al., 2004). Being ubiquitously expressed, Oct1 is an exception to the POU factor tissue-specific functionality. Oct1 activates the housekeeping genes encoding histone H2B and the U6 and U2 snRNAs (Hinkley and Perry, 1992; Segil et al., 1991; Yang et al., 1991), but it can also control transcription of Pax6, Cdx2, immunoglobulins and other tissue-specific genes, usually via interaction with cell-specific binding partners (Donner et al., 2007; Jin and Li, 2001; Mason et al., 1985; Strubin et al., 1995).
Development of an organism requires that cell fate is specified at the correct place and time in the embryo. The blastocyst is the first embryonic landmark in which lineage segregation is apparent, as it comprises cells of two different lineages. The ICM gives rise to the embryo proper and to the primitive endoderm, whereas the trophectoderm (TE) contributes only to extra-embryonic tissues. The proliferation and differentiation of extra-embryonic tissues is an absolute requirement for ensuring intra-uterine growth and survival of the embryo. The TE cells lining the blastocoel cavity (mural TE) differentiate after implantation into a layer of primary trophoblast giant cells, which are essential for promoting the exchange of nutrients and oxygen with the maternal uterine environment before the placenta has developed (Hemberger et al., 2003). The TE cells overlying the ICM (polar TE, or pTE) have a high proliferative potential, and form the extra-embryonic ectoderm (ExE) and the ectoplacental cone (EPC) of the post-implantation embryo, which are inherently different with respect to growth potential. Only the pTE and the ExE harbor trophoblast stem (TS) cells, which depend on Fgf4 to proliferate, provided either by the adjacent ICM or epiblast (EPI). TS cells can also be cultured and expanded ex vivo in the presence of recombinant Fgf4 (Tanaka et al., 1998; Uy et al., 2002). By contrast, the EPC contains only differentiated diploid precursors that give rise to secondary giant cells and later on to the spongiotrophoblast layer of the placenta. Genetic studies have revealed that TE development and TS proliferation also depend on the endogenous expression of the transcription factors Cdx2, Eomes and Elf5 (Donnison et al., 2005; Russ et al., 2000; Strumpf et al., 2005).
Besides being essential for placenta development, the ExE has an instructive role in patterning the embryo proper. The establishment of the proximal-distal (P-D) axis and its conversion into the anterior-posterior (A-P) axis depend on reciprocal and concerted interactions between the EPI, the visceral endoderm and the ExE, which ultimately lead to the formation of the anterior visceral endoderm (AVE) and the primitive streak (PS) at the prospective anterior and posterior sides of the embryo, respectively (Tam et al., 2006). The formation of both AVE and PS is intrinsically regulated by FGF, Bmp4, Nodal and Wnt signaling pathways (Tam et al., 2006; Thisse and Thisse, 2005).
Once the PS has been specified, cell delamination through the streak results in the formation of the mesoderm and the definitive endoderm. Proximal migration of the extra-embryonic mesodermal cells leads to expansion and then coalescence of the anterior and posterior amniotic folds into a single cavity, which is enclosed distally by the amnion and proximally by the chorion. By E8.5, the primitive body plan has been established and the allantois, which emerged as a finger-like structure where the PS first formed, expands upwards to make contact with the chorion. This embryonic stage marks the beginning of chorio-allantoic placenta formation and organogenesis.
As Oct1 is expressed in pre- and post-implantation mouse embryos, we were interested in investigating the biological function of Oct1 during early development. A mouse model with severely hypomorphic Oct1 alleles was generated previously (Wang et al., 2004). In a hybrid genetic background, a low level of Oct1 expression has been reported to cause midgestation lethality due to decreased erythropoiesis and anemia. In primary embryonic fibroblasts derived from wild-type versus hypomorphic Oct1 fetuses, total amounts of U2/U6 snRNAs and H2B transcripts were indistinguishable, although transfected octamer-driven reporter expression was affected by low Oct1 (Wang et al., 2004). In order to identify Oct1 function in fetal tissues other than the liver, expression profiling of Oct1 hypomorphic and wild-type fibroblasts was undertaken. Even though this analysis revealed that Oct1 modulates genes that mediate cellular response to oxidative and metabolic stress (Shakya et al., 2009; Tantin et al., 2005), the question of whether, and how, Oct1 contributes to embryonic development remained unanswered.
Here, we have generated an Oct1 loss-of-function mouse model by gene targeting. We show that Oct1-null mutant embryos display severe growth defects and die in utero at ∼E7.0-8.0. We demonstrate that Oct1 primarily plays a novel and unexpected role in trophoblast development by ensuring TS cell maintenance and differentiation. We also provide evidence that the embryonic function of Oct1 is necessary to ensure development from the late gastrulation stage onwards.
MATERIALS AND METHODS
Generation of Oct1 mutant mice and PCR genotyping of mice and embryos
Murine Oct1 genomic sequences used in the construction of the Oct1 targeting vector were derived from a mouse genomic lambda phage library, mapped and sequenced. A 6.3 kb KpnI-BamHI genomic fragment that includes the two exons encoding the POUS domain was used as the 5′ homology arm, and a 0.89 kb XhoI-BglII genomic fragment derived from the intron sequence between the two exons encoding the POUH domain was used as the 3′ homology arm. The homology arms were cloned on either side of a TK promoter-driven NeoR cassette. The targeting vector was linearized and electroporated into 129/Ola E14 embryonic stem (ES) cells. Correctly targeted G418-resistant clones were identified using a 735 bp BglII probe corresponding to the genomic POUH domain, which detected a 2.3 kb wild-type and a 6.5 kb mutant fragment on Southern blots of HindIII-digested genomic DNA. Two of these ES cell clones were used for aggregation to C57BL/6 morulae, and the resulting chimeric mice were backcrossed to C57BL/6 animals to obtain germline transmission of the targeted allele.
Tail tips or embryos were digested in 100 mM Tris pH 8.0, 0.5% Tween 20, 0.5% NP40 and 0.1 mg/ml proteinase K at 55°C. The Oct1 wild-type allele was detected by amplification of a 930 bp PCR product using primers specific for the POU genomic domain (see Table S1 in the supplementary material). The targeted Oct1 allele was detected by amplification of a 980 bp product using primers specific for the inserted NeoR cassette and the POU genomic domain (see Table S1 in the supplementary material).
Tetraploid embryo aggregation experiments were conducted as previously described (Eakin and Hadjantonakis, 2006).
Cell culture and immunofluorescence
ES and TS cells were derived and grown under standard conditions (Cavaleri et al., 2008; Tanaka et al., 1998). ES and TS cells were processed for immunostaining as previously described (Cavaleri et al., 2008). Antibodies and dilutions were: anti-SSEA-1 (Developmental Studies Hybridoma Bank, MC-480) 1:200; anti-Oct1 (Santa Cruz, C-21) 1:50; and anti-Cdx2 (BioGenex, Cdx2-88) 1:500.
Histology and in situ hybridization (ISH)
Pregnant females were dissected at the indicated gestational age, counting noon of the day of the vaginal plug as E0.5. For embedding, deciduae were fixed overnight in 4% paraformaldehyde and processed for routine paraffin histology. Whole-mount RNA ISH was performed as described for high-background probes (Zeller et al., 2001). After signal detection, embryos were photographed and genotyped by PCR. Antisense riboprobes were synthesized using a DIG RNA Labeling Kit (Roche) according to the manufacturer's instructions.
Electrophoretic mobility assay (EMSA) and western blotting
EMSA for spleen, thymus and ES whole-cell extracts was performed as described (Sauter and Matthias, 1998). In brief, radioactively labeled DNA fragments containing an octamer site from the IgH chain enhancer were used as probe. The fragments were labeled with [γ-32P]ATP and polynucleotide kinase. Binding reactions (20 μl) were set up with 2 μg whole-cell extract, 10,000 cpm of probe, 1 μg poly(dI-dC) and 1 μg denatured herring sperm DNA in the binding buffer (4% Ficoll 400, 20 mM HEPES pH 7.9, 50 mM KCl, 1 mM EDTA, 0.25 mg/ml bovine serum albumin). After 10 minutes incubation at room temperature, samples were electrophoresed in 4% polyacrylamide gels in 0.25×TBE. The gel was dried and exposed to a phosphorimager screen for quantification.
For western blotting, cells were lysed in 2× Laemmli buffer, vortexed for 3 seconds, heated at 99°C for 10 minutes and centrifuged for 5 minutes at 16,000 g at 4°C. Then 12.5-25 μl of lysate were electrophoresed on an 18% polyacrylamide minigel under denaturing SDS-PAGE conditions. Proteins were transferred to a PVDF Immobilon membrane (Millipore, Schwalbach, Germany) and processed for immunodetection with ECL Plus reagents (GE Healthcare, Solingen, Germany). Antibodies and dilutions were: anti-Oct1 (Santa Cruz, C-21) 1:1000; and anti-β-actin (Actb) (Abcam, 8226) 1:5000.
Quantitative expression analysis
For real-time analysis of gene expression, embryos (or cells) were harvested and processed as previously described (Boiani et al., 2003). Briefly, single E3.5 or E6.5 embryos were lysed in RLT buffer (Qiagen, Hilden, Germany), and 50% or 20% of the lysate was used for genomic DNA purification; the remaining lysate was used for RNA extraction. Complementary DNA synthesis was performed with the High Capacity cDNA Archive Kit (Applied Biosystems, Darmstadt, Germany) following the manufacturer's instructions. Transcript levels were determined using ABI PRISM Sequence Detection System 7900HT (Applied Biosystems) and the ready-to-use 5′-nuclease Assays-on-Demand as follows: Oct1, Mm00448332_m1; Cdx2, Mm00432449_m1; Esrrb, Mm00442411_m1; Eomes, Mm01351984_m1; Fgfr2, Mm00438941_m1; Oct4, Mm00658129_gH; Hand1, Mm00433931_m1; Hprt1, Mm00446968_m1. Quantification was normalized to the endogenous Hprt1 gene using the ΔΔCt method (ABI Prism 7700 Sequence Detection System User Bulletin #2, relative quantification of gene expression).
Chromatin immunoprecipitation assay (ChIP)
ChIP assays were performed following the manufacturer's recommendations (Agilent mammalian ChIP-on-chip protocol), with a few modifications: the number of cells used for each experiment was reduced to 1×106, and the amount of DNA used for each immunoprecipitation was of 1×105 cell equivalents. Briefly, cells were cross-linked and their nuclei were pelleted and lysed. After sonication, DNA was used to amplify the human/mouse reference gene ACTB/Actb to equalize the input used for the immunoprecipitation step. G protein-conjugated Dynabeads (25 μl; Invitrogen) were coupled with 10 μg of a polyclonal antibody directed against the mammalian Oct1 protein (Santa Cruz, C-21) and mixed with the sonicated DNA. Following overnight incubation, the beads were washed and the DNA eluted accordingly to the Agilent protocol.
Quantitative (q) PCR was used to determine the amount of immunoprecipitated DNA. Normalization and quantification were carried out as previously described (Johnson et al., 2002) using the ΔΔCt method relative to the control gene ACTB/Actb. Input control was used to determine the linear dynamic range and the efficiency of each qPCR reaction. The regions including the putative Oct1 binding consensus (human, –117 bp; mouse, –154 bp; relative to the transcription start site) and two regions more than 600 bp distant therefrom were amplified from human/mouse CDX2/Cdx2 loci with specific primers (see Table S1 in the supplementary material).
Vector construction, lentiviral particle production and TS cell infection
DNA constructs designed to produce short hairpin (sh) RNAs targeting Oct1 (5′-GCATCTAGCCCAAGTGCTTTGTTCAAGAGACAAAGCACTTGGGCTAGATGC-3′) or lacZ (5′-GTGGATCAGTCGCTGATTAAATTCAAGAGATTTAATCAGCGACTGATCCAC-3′) were cloned in front of the H1 promoter in the pLVTHM vector to produce pLVTHM-shOct1 and pLVTHM-shlacZ, respectively. pLVTHM-Td-tomato was constructed from pLVTHM by replacing the GFP with the Td-tomato coding sequence. pLVTHM-Oct1-2A-tomato was generated by introducing the Oct1 coding sequence and the 2A peptide in frame with Td-tomato. PLVTHM-wtCdx2 and pLVTHM-mutCdx2 were constructed by replacing the EF1α promoter in pLVTHM with the wild-type Cdx2 promoter (–154 to +126) or the Cdx2 promoter containing a mutated Oct binding site (CTGCAGAT) (Jin and Li, 2001), respectively.
The recombinant lentiviral particles were produced by transient transfection of 293T cells with 12 μg of each viral vector, 8.5 μg psPax2 and 3 μg pMD2.G using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. The supernatant was collected after 48 hours and concentrated by ultracentrifugation at 26,000 rpm (80,000 g) for 2 hours at 4°C using an SW41 rotor (Beckman Coulter). After ultracentrifugation, the supernatant was decanted and the viral pellet was resuspended in 200 μl Dulbecco's Modified Eagle Medium. The suspension was stored at –80°C until use. Eight thousand TS cells were plated on gelatin in 4-well plates and 24 hours later 20 μl of the concentrated virus was added to the medium. Cells were washed after 16 hours of incubation and transferred onto mouse embryonic fibroblasts (MEFs).
Oct1 knockdown and transactivation assay
Feeder-free TS cells were infected with pLVTHM-shOct1 or pLVTHM-shlacZ and sorted for GFP expression 48 hours after being plated on MEFs. GFP+ TS cells were lysed in RLT buffer (Qiagen) and reverse-transcribed using the High-Capacity cDNA Archive Kit (Applied Biosystems) following the manufacturer's instructions. The expression of Oct1 and Cdx2 was determined using the ABI PRISM Sequence Detection System 7900HT (Applied Biosystems) and the ready-to-use 5′-nuclease Assays-on-Demand (see above).
TS cells, co-infected with pLVTHM-Oct1-2A-tomato and either PLVTHM-wtCdx2 or pLVTHM-mutCdx2, were sorted for Td-tomato and GFP expression 48 hours after being plated on MEFs. Td-tomato and GFP mean fluorescence intensities were measured using FACSdiva software (BD Biosciences).
Inactivation of the mouse Oct1 gene
To assess the role that Oct1 plays during embryogenesis we inactivated the mouse Oct1 locus by homologous recombination in ES cells. At least four (multiple) alternatively spliced isoforms of Oct1 have been identified in mice (Zhao et al., 2004). These isoforms have a unique 3′ terminus and a common 5′ moiety with an intact POU binding domain. Wang and colleagues previously attempted to disrupt Oct1 by replacement of exon 3 with a neomycin cassette (Wang et al., 2004). This gene targeting strategy was intended to lead to the production of transcripts harboring a frameshift mutation that upon translation would result in the deletion of two-thirds of the Oct1 polypeptide. However, owing to the utilization of translation initiation sites downstream of exon 3, residual Oct1 binding activity was detected with nuclear extracts isolated from Oct1 double-targeted MEF cells. Therefore, the engineered Oct1 mutation was considered to constitute a severely hypomorphic allele. We thus decided to inactivate the Oct1 locus by replacing exon 11, which encodes the linker domain and the 5′ terminus of the POUH domain, with the neomycin resistance gene (Fig. 1A). Properly targeted ES clones were identified by Southern blotting of HindIII-digested genomic DNA (Fig. 1B).
Transcripts from the targeted allele were expected to contain exon 10 spliced to exon 12 and hence a nonsense mutation that would lead to the translation of a C-terminally deleted Oct1 protein unable to bind DNA. In order to characterize the splicing events occurring at the targeted allele, reverse transcription PCR was performed with RNA extracted from wild-type, heterozygous and null Oct1 embryos. A DNA fragment corresponding to the expected length for an exon 10 to exon 12 splicing event was obtained from cDNA samples of heterozygous and null genotypes with an oligo pair specific to exons 9 to 13, whereas no amplification product was obtained with cDNA isolated from null embryos when an oligo pair specific to exons 9 to 11 was used (Fig. 1C). Sequence analysis confirmed the amplification product of the mutant Oct1 cDNA.
We then investigated the binding activity of the mutant Oct1 protein by performing EMSA experiments with nuclear extracts isolated from both spleen and thymus cells from mice heterozygous for the Oct1 mutation. Besides the endogenous Oct protein complexes formed with wild-type extracts, no additional complex that could indicate a residual binding capability of the truncated Oct1 protein was detected in the presence of heterozygous spleen or thymus cell extracts (Fig. 1D). Furthermore, quantification of the EMSA fluorograms showed that the spleen or thymus cells of heterozygous mice contain approximately half of the wild-type Oct1 binding activity (Fig. 1E). Taken together, these results indicate that our Oct1 targeted allele results in the production of a non-functional, truncated Oct1 protein that is unable to bind DNA, and can thus be regarded as a null allele.
Loss of Oct1 causes early embryonic lethality
Heterozygous Oct1 mice derived from two independent ES cell clones appeared normal and fertile. However, no viable Oct1–/– pups were recovered from intercrossing heterozygous animals. To assess the time frame of embryonic lethality, deciduae from heterozygote matings were dissected at various stages of gestation and isolated embryos were genotyped. Oct1–/– embryos were recovered at E6.5 at the expected Mendelian ratio (Table 1). However, the frequency with which viable mutant embryos were recovered decreased from E7.5, and at E9.5 only resorbed null embryos were found, suggesting that mutant embryos die between E7.5 and E9.0. Morphological examination of embryos at ∼E7.5 revealed that Oct1–/– embryos were oval or asymmetric in shape and showed no evident demarcation between the embryonic and the extra-embryonic regions (Fig. 2A). They appeared developmentally arrested and resembled egg cylinders with a pro-amniotic cavity, whereas their littermates had developed up to late-streak stage and consequently displayed proper separation of the amniotic and exocoelomic cavities, as marked by chorion and amnion formation (Fig. 2A).
In order to identify possible defects at the tissue and cellular levels we performed in utero histological analysis of Oct1 littermates. Since we were unable to genotype concepti from sectioned deciduae, we attributed a null genotype to concepti that showed no obvious P-D polarity. At E7.0, presumptive Oct1-null mutant embryos appeared reduced in size and lacked a structured EPC, which had invaded the maternal endometrium, although primary giant cells were visible around the conceptus (Fig. 2C). By contrast, the outer epithelial layers of parietal and visceral endoderm did not display any obvious defect (Fig. 2B,C). These results suggest that Oct1 might be crucial for the establishment of a proper embryonic-maternal interface that ensures embryo development before placenta formation takes place.
Oct1-null embryos are defective in trophoblast development
In order to investigate the molecular basis of the early post-implantation lethality caused by Oct1 deficiency, we examined the expression of lineage-specific molecular markers by whole-mount ISH.
Nodal, the Nodal co-receptor Cripto (Tdgf1 – Mouse Genome Informatics), Otx2 and Oct4 are expressed in the EPI of pregastrulating embryos. Nodal and Cripto become restricted to the posterior EPI as gastrulation commences at E6.5 (Brennan et al., 2001; Ding et al., 1998). In Oct1 mutant embryos, as in the wild type, Nodal transcripts were found in a proximal-to-distal distribution and predominantly on the posterior side of the embryo (Fig. 3A). Expression of Cripto was also unaltered in Oct1-null embryos (Fig. 3B). At E6.5 and E7.75, Oct1 mutant embryos displayed a strong and almost ubiquitous pattern of Otx2 and Oct4 expression, respectively, whereas wild-type and heterozygous littermates showed the correct distal pattern (Fig. 3C,D). These results suggest that Oct1-null embryos are mainly composed of embryonic ectoderm surrounded by visceral endoderm and might lack proximal extra-embryonic tissues.
We then probed embryos for extra-embryonic markers. Bmp4 and the caudal-related homeobox gene Cdx2 are expressed within the proximal region of the ExE (pExE) immediately adjacent to the EPI of E6.5 embryos (Beck et al., 1995; Lawson et al., 1999). Eomes (Russ et al., 2000), the nuclear orphan receptor Esrrb (Luo et al., 1997), the fibroblast growth factor receptor Fgfr2 (Haffner-Krausz et al., 1999), Pace4 (Pcsk6 – Mouse Genome Informatics) (Donnison et al., 2005) and Bmp8b (Ying and Zhao, 2000) mark the whole ExE of early gastrulating embryos, with Fgfr2 and Bmpb8b also being expressed in the EPC.
Bmp4 expression was consistently found at the proximal pole of Oct1 mutant embryos (Fig. 3E). By contrast, Cdx2 transcripts could not be detected in null embryos. However, because the intensity of the signal obtained with the Cdx2 probe used was relatively weak, we cannot exclude the possibility that the level of Cdx2 expression was just below our limit of detection (Fig. 3F). qRT-PCR analysis of wild-type, heterozygous and null Oct1 E6.75 embryos confirmed the above interpretation of the Cdx2 ISH data, and showed downregulation of Cdx2, Esrrb and the pan-trophoblastic marker Hand1 (Cross et al., 1995) (see Fig. S1 in the supplementary material). Eomes, Fgfr2 and Pace4 were transcribed in Oct1-null embryos, although in a significantly reduced domain compared with their wild-type or heterozygous counterparts (Fig. 3G-I). Surprisingly, Bmp8b could not be detected in Oct1-null embryos (Fig. 3I). These results indicate that only an ExE-like compartment that is severely compromised in size can form in the absence of Oct1. We then investigated the expression of the Achaete-scute homolog Mash2 (Ascl2 – Mouse Genome Informatics), which is a marker of the diploid precursors of the ExE/EPC transition tissue and of the EPC at early gastrulation stages (Guillemot et al., 1994). Mash2 transcripts were absent in Oct1-null embryos (Fig. 3L). Collectively, these results suggest that loss of Oct1 activity affects ExE development and impedes EPC formation.
Oct1 loss leads to ectopic AVE formation
Embryo patterning has been shown to be largely dependent on the reciprocal interaction between ExE and EPI. Since Oct1 deficiency compromised ExE development, the next step was to determine whether the embryo body plan was correctly established in Oct1-null embryos. To this end, we examined AVE and PS formation.
AVE cells express both Lim1 and Hex (Lhx1 and Hhex, respectively – Mouse Genome Informatics) at early to mid-gastrulation stages (Thomas et al., 1998; Tsang et al., 2000). Additionally, Lim1 marks the PS and nascent mesodermal cells migrating away from the streak, whereas Hex marks cells of nascent definitive endoderm. The AVE-specific expression of Lim1 and Hex1 appeared enlarged in Oct1-null embryos as compared with control littermates (Fig. 4A,B). By contrast, expression of the nascent mesoderm markers brachyury (T) and Fgf8 was normal in Oct1-null embryos, except for the fact that it was displaced towards the proximal end of the embryo (Fig. 4C,D). These data indicate that Oct1 deficiency does not impair specification of the A-P axis, but rather affects the induction magnitude of anterior polarity in the embryo.
Oct1 is required for the maintenance of TS cells
Since Oct1-null embryos contain a severely reduced ExE and lack the EPC, we reasoned that Oct1 might not be needed for EPI development but is necessary for proliferation and differentiation of the pTE-derived stem cell pool, both in vivo and in vitro (Chawengsaksophak et al., 1997; Russ et al., 2000).
TS cells can be isolated from E3.5 or E6.5 embryos cultured ex vivo in the presence of Fgf4 plus heparin and can be induced to differentiate by growth factor removal (Tanaka et al., 1998). We investigated Oct1 expression in proliferating and differentiating TS cells. Oct1 and Cdx2 were co-expressed in undifferentiated TS cells (Fig. 5A). During trophoblast cell differentiation, the expression of the stem markers Cdx2, Eomes and Essrb was barely detectable as early as 3 days after Fgf4 withdrawal. However, only a minor downregulation of Hand1 and Oct1 transcripts was found in differentiated trophoblast cells even after 5 days of differentiation, indicating that Oct1 might be a pan-trophoblastic marker (Fig. 5B).
To test whether ICM-TE segregation occurs properly in pre-implantation embryos in the absence of Oct1, we performed qRT-PCR analysis of TE-specific and ICM-specific genes in blastocysts derived from Oct1 heterozygous intercrossing. No significant changes in the expression of Cdx2, Eomes, Fgfr2, Nanog and Oct4 were detected in any of the examined blastocysts, regardless of their genotype (see Fig. S2 in the supplementary material). This result suggests that either pTE specification does not require Oct1 or that the pTE is properly specified in the presence of maternal Oct1. We therefore attempted to derive both ES and TS cells from pre-implantation Oct1-null embryos. Oct1-null ES lines were derived at the expected Mendelian ratio. These cells do not contain any Oct1 protein or binding activity, but continue to express SSEA-1 (Fut4 – Mouse Genome Informatics) and the pluripotency-associated markers Nanog and Oct4 (see Fig. S3 in the supplementary material). By contrast, of the 20 TS lines we derived from blastocysts obtained from Oct1-heterozygous intercrossing, 13 were heterozygous and the remaining seven were wild type. These results indicate that although Oct1 is dispensable for ES cell derivation, it is essential for the establishment or maintenance of pTE-derived stem cells in vitro. Moreover, they support the notion that Oct1 is required in vivo for the proliferation of ExE stem cells. However, it cannot be ruled out that Oct1 might be necessary for the differentiation of ExE TS cells as well.
We then tested the hypothesis that Oct1 might be involved in the regulation of Cdx2 by binding to its promoter region, as previously shown for human cells of the intestinal epithelium (Jin and Li, 2001) and of the Caco2 colorectal adenocarcinoma line (Almeida et al., 2005). Sequence analysis of the mouse and human proximal Cdx2 promoter revealed conservation of both the octamer canonical consensus and flanking regions (data not shown). ChIP assay performed with anti-Oct1 antibody showed a clear enrichment in the amplification of the region containing the octamer-binding consensus in comparison to the mock immunoprecipitation for both the human Caco2 and mouse TS samples (–116 and –154 bp relative to the transcription start site, respectively), but not for the MEF sample (P=0.0013) (Fig. 5C). In order to test whether Oct1 regulates Cdx2 transcription, we knocked down Oct1 in TS cells using Oct1-specific shRNA. We observed downregulation of Cdx2 expression following Oct1 downregulation (P<0.05; Fig. 5D). Additionally, mutation of the octamer binding site in the Cdx2 promoter led to decreased expression of a GFP reporter gene in both Caco2 (P=0.0216) and TS (P=0.005) cells, as compared with expression from an unmutated Cdx2 promoter (Fig. 5E). These results indicate that Oct1 specifically binds and regulates the Cdx2 promoter region in murine TS cells as well as in human Caco2 cells.
Oct1 function is indispensable in the embryo proper
Our morphological and marker analyses showed that Oct1-null embryos are unable to develop beyond early gastrulation stages because of a failure in trophoblast development. However, as Oct1 is expressed in embryonic tissues, the observed phenotype could also result from a combination of effects, i.e. from a lack of Oct1 in both the ExE and EPI. We set out to determine whether the gastrulation defect also occurred in embryos lacking Oct1 function solely in the EPI by performing a tetraploid complementation assay. Single wild-type 4-cell stage tetraploid embryos were aggregated to 8-cell stage diploid embryos derived from Oct1 heterozygous intercrosses and allowed to develop to midgestation. None of the viable 35 chimaeric embryos recovered between E10.5 and E12.5 was Oct1 null. At nominal stage E8.5, wild-type host ↔ Oct1-deficient chimeras were smaller than their littermates, but did not present any obvious morphological or developmental abnormalities. At E9.5, wild-type host ↔ Oct1-null chimeras were consistently growth retarded and developmentally arrested at neurula (headfold) stage with no more than four to five somites (E8.25) (Fig. 6), whereas heterozygous chimeric counterparts had clearly completed the turning process and displayed at least 16 somite pairs and a beating heart. These results suggest that Oct1 function in the embryonic tissues is not required for development from early to late gastrulation stages. They also indicate that Oct1 function is necessary in the embryo proper for normal development and/or growth to proceed from late gastrulation to the early somite stage.
In this study we describe the phenotype resulting from functional inactivation of the mouse Oct1 locus and show that Oct1 plays a novel and unexpected role in TE development. Despite OCT1 transcripts being detected in the human placenta and in some choriocarcinoma cell lines (Jimenez-Mateo et al., 2006), the expression and function of this factor in murine trophoblastic tissues have not been previously reported. Oct1-null embryos are unable to develop past the early streak stage, such that Oct1 deficiency leads to early embryonic death in utero, with all Oct1-null embryos dying by E9.5. Expression of the key molecular determinants of trophoblast fate, Cdx2, Fgfr2 and Eomes, is normal in Oct1-null blastocysts, suggesting that either Oct1 does not take part in TE specification or that maternally inherited Oct1 (protein and/or transcript) is sufficient to sustain pre-implantation development. Additionally, Oct1-null embryos implant normally, indicating that the formation of primary giant cells, which mediate implantation by invading the uterine stroma and remodeling the maternal vasculature, is not affected by Oct1 deficiency. However, in utero histological examination of Oct1 littermates at E6.75 revealed that Oct1-null embryos do not present an obvious embryonic or extra-embryonic polarity and lack an identifiable EPC, suggesting that ExE differentiation into EPC precursors is impaired by Oct1 loss. Our whole-mount ISH analysis of ExE and EPC markers at the early gastrulation stage, combined with the inability to derive Oct1-null TS cells, strongly indicate that both the expansion and differentiation of proliferative ExE TS cells require Oct1. Importantly, Oct1-null embryos contain a small ExE that lacks Bmp8b expression, suggesting not only that the size and molecular identity of the ExE are affected by Oct1 loss, but also that Bmp8b can be considered a direct or indirect Oct1 target in the proliferative ExE.
Previous studies showed that Elf5 and Erk2 (Mapk1 – Mouse Genome Informatics) are essential for TS cell self-renewal, as the ExE is totally absent in embryos deficient of either factor at E6.5. Moreover, TS cell lines could not be derived from Elf5 and Erk2 mutant embryos (Donnison et al., 2005; Saba-El-Leil et al., 2003). However, the presence of a small ExE in Oct1-null embryos suggests that Oct1 has a less stringent role than Elf5 and Erk2 in the maintenance of the proliferative capacity of ExE TS cells. Importantly, our data suggest that Oct1 deficiency is not permissive for the transition of proliferative ExE into ExE cells and EPC diploid precursors (Fig. 7). The absence of Bmp8b expression cannot account per se for these differentiation defects, as disruption of the Bmp8b locus does not have any effect on TE development (Ying and Zhao, 2000).
A candidate target of Oct1 in the ExE could be Sox2, which is expressed in both the EPI and the ExE of the early streak embryo (Avilion et al., 2003). Previous studies have highlighted the presence of multiple Oct binding sites within Sox2 regulatory regions (Catena et al., 2004; Tomioka et al., 2002). Although these studies have elucidated the capacity of the POU factors Oct4 and Brn1/Brn2 (Pou3f3/Pou3f2) to activate Sox2 transcription in ES and neural cells, respectively, the identity of transcription factors regulating Sox2 expression in the ExE has not yet been investigated. As Oct1 is expressed in the ExE, it is possible that Oct1 might be important for driving Sox2 expression in the ExE via the Oct sites.
The Cdx2 promoter is autoregulated in cells that physiologically express Cdx2 (Xu et al., 1999), and Oct1 is an integral component of this autoregulatory loop, as mutation of the octamer binding site within the Cdx2 proximal promoter interferes with both Cdx2 expression and auto-activation. Autoregulation of the endogenous Cdx2 gene has been observed in TS-like cells produced by inducible, enforced expression of a Cdx2 transgene in ES cells (Tolkunova et al., 2006). Additionally, at stages later than blastocyst, Elf5 has been shown to reinforce Cdx2 transcription via two binding sites located in the –780 to –184 region of the Cdx2 promoter (Ng et al., 2008). In this study we present evidence that Oct1 interacts with the Cdx2 promoter and regulates Cdx2 transcription in mouse TS cells. We then propose that in proximal ExE cells stimulated by Fgf4 and Nodal, Oct1 might contribute to producing the threshold level of Cdx2 transcription that triggers Cdx2 autoregulation and ensures TS maintenance within the pExE (Fig. 7). Additionally, Oct1 is indispensable for pExE differentiation.
An exquisitely regulated network of interactions between the EPI and the ExE guarantees both ExE maintenance and the establishment of A-P polarity in the embryo. Nodal and Fgf signaling emanating from the EPI cooperate in maintaining ExE-specific expression of Cdx2, Eomes and Esrrb and in preventing precocious differentiation of the ExE (Brennan et al., 2001; Guzman-Ayala et al., 2004). Bmp4 signaling emanating from the ExE source mediates mesoderm formation by inducing Wnt3 expression in the EPI, which in turn stimulates T and Nodal transcription (Ben-Haim et al., 2006; Liu et al., 1999). Translation of Nodal transcripts produces a pro-protein that must be processed in order to generate the fully active molecule. However, uncleaved Nodal can sustain extra-embryonic expression of its own convertases, Pace4 and Furin, and of Bmp4 (Ben-Haim et al., 2006).
Oct1 deficiency does not seem to have detrimental effects on Nodal transcription in E6.5-6.75 embryos. This result supports the conclusion that uncleaved-Nodal-dependent signaling acting on the ExE is intact in Oct1-null embryos. Additionally, we detected T and Fgf8 expression on the posterior side of Oct1-null embryos at early gastrulation stage, whereas Elf5- and Erk2-null embryos do not show molecular signs of PS formation. This finding is consistent with the presence in Oct1 mutants of an ExE remnant that still produces Bmp4 and Pace4 at the early streak stage. The expansion in AVE formation observed in the absence of Oct1 could be explained if Oct1-null embryos show lower levels, or a complete lack, of expression of the ExE-derived signaling molecule(s) that restrict DVE induction (Rodriguez et al., 2005). It has been proposed that Bmp8b, which is indeed absent from the Oct1-null ExE, might be such a signaling factor (Ohinata et al., 2009).
Based on our E6.5 analysis of embryonic markers, an Oct1 function that is restricted to the embryo proper is unlikely to contribute to the early streak arrest observed in Oct1-null embryos. Additionally, the gastrulation block of these embryos could be rescued by providing wild-type extra-embryonic tissues. However, because our tetraploid Oct1-null chimeric embryos are developmentally arrested at the early somite stage, we cannot rule out the possibility that the developmental arrest we have observed might have been slightly delayed by a contribution of wild-type cells to the embryo proper. Therefore, a systematic analysis of the differentiation capability of Oct1-null versus wild-type ES cells into all lineages both in vivo and in vitro is required to exactly pinpoint when Oct1 function in the embryonic tissues becomes necessary.
In conclusion, two POU factors play crucial roles during early embryogenesis in distinct tissues of the developing embryo. Oct4 is essential for ICM lineage specification (Nichols et al., 1998), whereas Oct1 primarily regulates ExE development. Further studies are needed to elucidate the modus operandi of Oct1 within the molecular network that underlies the proliferation and differentiation of the trophoblast lineages.
We thank Drs Richard Behringer, Daniel Constam, Jacqueline Deschamps, Brigid Hogan, K. Matsumoto, Janet Rossant and Michael Shen for kind gifts of whole-mount in situ probes; Dr Trono for pLVTHM and viral packaging plasmids; C. Ortmeier and M. Sinn for real-time PCR; and Patrick Kopp and Jean-François Spetz for ES cell work and generation of knockout mice. We are especially grateful to Michele Boiani for fruitful discussions, to Susanne Kölsch for editing the manuscript and to Peter Pfeffer and Tilo Kunath for critical reading of this manuscript. This work was supported by the Max Planck Society and the Novartis Research Foundation.
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.047027/-/DC1
- Accepted August 27, 2010.
- © 2010.