TEAD4, YAP1 and WWTR1 prevent the premature onset of pluripotency prior to the 16-cell stage

Pluripotency describes the developmental potential to produce all adult cell types. However, in mammals, the establishment of pluripotency takes place in the context of lineage decisions that establish the extra-embryonic lineages such as the placenta and yolk sac (Chazaud and Yamanaka, 2016; Lanner, 2014; Posfai et al., 2014). The mouse embryo has provided an invaluable tool with which to understand the molecular mechanisms that initially create pluripotent cells, which are also the progenitors of embryonic stem cells. Although much progress has been made in understanding how pluripotency is maintained once pluripotent cells are established, the mechanisms driving the initial establishment of pluripotency remain relatively obscure.

In the mouse embryo, pluripotent cells emerge from the inner cell mass of the blastocyst. Establishment of inner cell mass first occurs around the 16-cell stage, when select cells occupy the inside of the morula (Posfai et al., 2014). Later, at embryonic day (E) E3.75 blastocyst stage, the inner cell mass differentiates into either pluripotent epiblast or non-pluripotent primitive endoderm (Chazaud et al., 2006; Morris et al., 2010; Plusa et al., 2008; Saiz et al., 2016; Yamanaka et al., 2010). As the epiblast matures, it gradually acquires a more embryonic stem cell-like transcriptional signature (Boroviak et al., 2014, 2015).

Although studies in mammalian embryos and embryonic stem cells have developed an extensive catalog of transcription factors that promote pluripotency, the only pluripotency-promoting transcription factor known to distinguish inside cells as they form at the 16-cell stage is Sox2 (Guo et al., 2010; Wicklow et al., 2014). At this stage, other pluripotency factors, such as Nanog and Oct4, are detected in both inside and outside cells (Dietrich and Hiiragi, 2007; Palmieri et al., 1994; Strumpf et al., 2005). Therefore, understanding how Sox2 expression is regulated at the 16-cell stage can provide unique insight into how pluripotency is first established.

Here, we use genetic approaches to test mechanistic models of the initial activation of Sox2 expression. We investigate the contribution, at the 16-cell stage and earlier, of factors and pathways that are known to regulate expression of Sox2 at later preimplantation stages and in embryonic stem cells. We show that embryos are competent to express high levels of Sox2 as early as the four-cell stage, although they normally do not do so. Finally, we uncover the molecular mechanisms that ensure that Sox2 expression remains repressed until the appropriate developmental stage.

To identify mechanisms contributing to the onset of Sox2 expression in the embryo, we first focused on the role of transcription factors that are required for Sox2 expression in embryonic stem cells. The core pluripotency genes Nanog and Oct4 (Pou5f1) are required for Sox2 expression in embryonic stem cells (Chambers et al., 2003; Mitsui et al., 2003; Okumura-Nakanishi et al., 2005) and are expressed in embryos at the eight-cell stage (Dietrich and Hiiragi, 2007; Palmieri et al., 1994; Rosner et al., 1990; Strumpf et al., 2005), prior to the onset of Sox2 expression at the 16-cell stage, raising the possibility that NANOG and OCT4 could activate the initial expression of Sox2.

We previously showed that the initiation of Sox2 expression is Oct4 independent, as normal levels of SOX2 are detected in blastocysts at E3.5 in the absence of Oct4 (Frum et al., 2013). We therefore hypothesized that Nanog and Oct4 could act redundantly to initiate Sox2 expression. To test this hypothesis, we bred mice carrying the null allele Nanog-GFP (Maherali et al., 2007) with mice carrying a deleted allele of Oct4 (Kehler et al., 2004) to generate Nanog;Oct4 null embryos (Fig. S1A). In wild-type embryos, Sox2 is first detected in inside cells at the 16-cell stage, with increasing levels in inside cells of the 32-cell stage embryo (Frum et al., 2013; Guo et al., 2010). In Nanog;Oct4 null embryos, SOX2 was detectable at the 16-cell (E3.0) and 32-cell (E3.25) stages (Fig. 1A,B). We observed no difference in the proportions of SOX2-expressing cells at the 16- and 32-cell stages between non-mutant embryos and embryos lacking Nanog or Oct4 or both (Fig. S1B,C), nor did we observe a difference in total cell numbers among the genotypes at any early stage examined (Fig. S1E-G). These observations indicate that Nanog and Oct4 do not regulate initial Sox2 expression.

To investigate a role for Nanog and Oct4 in maintaining expression of Sox2, we evaluated double knockout embryos at a later time point. By E3.5, SOX2 appeared weak or undetectable in most cells of Nanog;Oct4 null embryos (Fig. 1C). Moreover, the proportion of cells expressing a wild-type level of SOX2 was significantly lower in Nanog;Oct4 null embryos (Fig. 1D), but not in embryos lacking Nanog or Oct4 only (Fig. S1D). We therefore conclude that Nanog and Oct4 redundantly maintain Sox2 expression up to E3.5.

To evaluate whether Nanog and Oct4 redundantly maintain Sox2 expression later, we examined SOX2 in embryos lacking either Nanog or Oct4 at E3.75 and E4.25. At E3.75, SOX2 levels were similar among non-mutant, Nanog null and Oct4 null embryos (Fig. 1E,F). Notably, Nanog-GFP was detected in all inner cell mass cells in the Nanog null embryos (Fig. 1E,G), compared with non-mutants and Oct4 null embryos, in which NANOG was downregulated in non-epiblast cells. Therefore, Nanog is required for repression of Nanog expression in primitive endoderm. This observation is consistent with a non cell-autonomous requirement for Nanog in promoting primitive endoderm fate (Frankenberg et al., 2011; Messerschmidt and Kemler, 2010).

By contrast, both Nanog null and Oct4 null embryos exhibited defects in SOX2 by E4.25. Nanog null embryos exhibited the more severe SOX2 expression phenotype, with almost undetectable SOX2 (Fig. 1G). Oct4 null embryos exhibited a less severe SOX2 expression phenotype, with reduced, but detectable SOX2 (Fig. 1H), possibly owing to developmental delay in Oct4 null mutants at E4.25 (Frum et al., 2013). These observations indicate that, although the initial phase of Sox2 expression is independent of Nanog and Oct4, this is followed by a period during which Nanog and Oct4 act redundantly to maintain Sox2 expression, which then gives way to a phase during which Nanog and Oct4 are individually required to achieve maximal Sox2 expression.

Having observed that the initiation of Sox2 expression is Nanog and Oct4 independent, we next examined the role of other factors in regulating initial Sox2 expression. TEAD4 and its co-factors WWTR1 and YAP1 repress Sox2 expression in outside cells, starting around the 16-cell stage (Frum et al., 2018; Wicklow et al., 2014). However, YAP1 is detected within nuclei as early as the four-cell stage (Nishioka et al., 2009), suggesting that the TEAD4/WWTR1/YAP1 complex is active prior to the 16-cell stage. Recent studies have highlighted the roles and regulation of TEAD4/WWTR1/YAP1 in promoting expression of CDX2 during outside cell maturation to trophectoderm during blastocyst formation (Anani et al., 2014; Cao et al., 2015; Cockburn et al., 2013; Hirate et al., 2013; Kono et al., 2014; Leung and Zernicka-Goetz, 2013; Lorthongpanich et al., 2013; Menchero et al., 2019; Nishioka et al., 2009; Posfai et al., 2017; Rayon et al., 2014; Shi et al., 2017; Yagi et al., 2007; Yu et al., 2016). Yet the developmental requirement for TEAD4/WWTR1/YAP1 prior to the 16-cell stage has not been investigated. We therefore hypothesized that TEAD4/WWTR1/YAP1 repress Sox2 expression prior to the 16-cell stage.

To test this hypothesis, we examined SOX2 in embryos lacking Tead4. Consistent with our hypothesis, Tead4 null embryos exhibited precocious SOX2 at the eight-cell stage (Fig. 2A and Fig. S2C). Notably, this phenotype was not exacerbated by elimination of maternal Tead4 (Fig. 2A and Fig. S2A,C), consistent with the absence of detectable Tead4 in oocytes (Yagi et al., 2007). By contrast, deletion of maternal Wwtr1 and Yap1 (Fig. S2B) led to precocious SOX2 at the eight-cell stage (Fig. 2B and Fig. S2C). The presence of wild-type, paternal alleles of Wwtr1 and/or Yap1 did not rescue precocious SOX2 in the maternally null embryos. Therefore, maternally provided WWTR1/YAP1 and zygotically expressed TEAD4 repress Sox2 expression at the eight-cell stage.

We next evaluated SOX2 in embryos lacking maternal (m) and/or zygotic (z) Tead4 or Wwtr1;Yap1 at the four-cell stage. We observed that SOX2 was never detected in four-cell Tead4 z null or Tead4 mz null embryos (Fig. 2C and Fig. S2D). However, four-cell embryos lacking maternal Wwtr1 and Yap1 occasionally exhibited weak ectopic SOX2 (Fig. 2D and Fig. S2D). These observations suggest that Wwtr1 and Yap1 partner with factors other than TEAD4 to repress Sox2 expression at the four-cell stage. As TEAD1 and TEAD2 are also detected during the two- to eight-cell stages (Nishioka et al., 2008), we predict that these factors may partner with YAP1/WWTR1 to repress SOX2 during early embryogenesis.

The premature onset of Sox2 expression in embryos lacking Tead4 or Wwtr1 and Yap1 demonstrates that preimplantation mouse embryos are capable of expressing markers of inside cell identity as early as the four-cell stage and reveals an earlier than expected role for TEAD4/WWTR1/YAP1 in repressing the expression of Sox2 until the formation of inside cells, thus permitting the establishment of discrete trophectoderm and inner cell mass domains of gene expression. Notably, expression of OCT4 and NANOG is unchanged in embryos lacking Tead4 (Nishioka et al., 2008), highlighting the unique regulation of SOX2 in defining initial inner cell mass identity. Whether other pluripotency factors exist that are co-regulated with Sox2, remains an unresolved issue. Our results suggest that the mechanism regulating the onset of Sox2 expression is that constitutive repression of Sox2 by TEAD4/WWTR1/YAP1 is relieved once cells are positioned inside the embryo at the 16-cell stage. The mechanisms that initiate expression of TEAD4, WWTR1 and YAP1 prior to compaction are currently unknown.

In many contexts, TEAD4/WWTR1/YAP1 activity is repressed by the HIPPO pathway LATS1 and LATS2 kinases, which repress nuclear localization of WWTR1/YAP1 (Zhao et al., 2007, 2010). For example, during blastocyst formation, LATS1 and LATS2 repress nuclear localization of WWTR1/YAP1 in inside cells (Nishioka et al., 2009). To evaluate the role of the HIPPO pathway in regulating initial Sox2 expression, we examined whether Sox2 expression is LATS1/2-sensitive prior to the 16-cell stage.

We injected mRNA encoding Lats2 into both blastomeres of two-cell stage embryos, which is sufficient to inactivate the TEAD4/WWTR1/YAP1 complex during blastocyst formation (Nishioka et al., 2009; Wicklow et al., 2014), and then evaluated SOX2 at the four- and eight-cell stages (Fig. 3A). As anticipated, Lats2 mRNA injection, but not injection of green fluorescent protein (GFP) mRNA, greatly reduced YAP1 nuclear localization at four- and eight-cell stages (Fig. 3B,C). In addition, we observed precocious SOX2 in embryos overexpressing Lats2 (Fig. 3B-D). Therefore, LATS kinases can repress TEAD4/WWTR1/YAP1 nuclear activity and induce Sox2 expression prior to the 16-cell stage, but must not normally do so, as SOX2 is not detected prior to the 16-cell stage. After the 16-cell stage, LATS1/2 kinases are thought to be active specifically in inside cells, owing to their unpolarized state (Hirate et al., 2013; Kono et al., 2014; Leung and Zernicka-Goetz, 2013). Therefore, the polarization of all blastomeres of the eight-cell stage embryo (Frum and Ralston, 2018), or other polarity-independent mechanisms, could limit LATS1/LATS2 activation prior to the 16-cell stage.

While the TEAD4/WWTR1/YAP1 complex is widely recognized as a transcriptional activator, it has more recently been shown to act also as a transcriptional repressor (Beyer et al., 2013; Kim et al., 2015). Therefore, we considered two mechanisms by which TEAD4/WWTR1/YAP1 could repress Sox2 expression (Fig. 4A): an indirect model, in which TEAD4/WWTR1/YAP1 promote transcription of a Sox2 repressor; and a direct model, in which TEAD4/WWTR1/YAP1 themselves act as the Sox2 repressor.

To test these models, we employed variants of Tead4 in which the WWTR1/YAP1 interaction domain has been replaced with either the transcriptional activator domain of VP16 (Tead4VP16) or the transcriptional repressor domain of engrailed (Tead4EnR). These variants have previously been used in preimplantation embryos to provide evidence that TEAD4/WWTR1/YAP1 promotes Cdx2 expression through a direct mechanism (Nishioka et al., 2009). We reasoned that if TEAD4/WWTR1/YAP1 represses Sox2 indirectly, then overexpression of Tead4EnR would induce Sox2 expression prematurely. Alternatively, if TEAD4/WWTR1/YAP1 represses Sox2 directly, then Tead4VP16 would induce Sox2 expression prematurely. We injected mRNAs encoding GFP and either Tead4VP16 or Tead4EnR into a single blastomere of four-cell stage embryos to observe the effects on SOX2 prior to the 16-cell stage (Fig. 4B). In these experiments, we commenced overexpression at the four-cell stage in order to achieve maximal expression levels of Tead4VP16 and Tead4EnR by the eight-cell stage. Moreover, we found that these constructs caused lethality at the two-cell stage, which did not enable us to study their effects on SOX2 expression at the eight-cell stage. We observed that overexpression of Tead4VP16, but not Tead4EnR, induced SOX2 at the eight-cell stage (Fig. 4C,D). These observations are consistent with the direct repression of Sox2 by TEAD4/WWTR1/YAP1 prior to the 16-cell stage.

This study highlights distinct phases of Sox2 regulation occurring during the establishment of pluripotency in mouse development. As early as the four-cell stage, blastomeres are competent to express Sox2, but this is overridden by TEAD/WWTR1/YAP1 (Fig. 4E, Box 1). Initiation of Sox2 expression does not require Nanog and Oct4. Instead, LATS1/2 activity in inside cells relieves repression of TEAD4/WWTR1/YAP1 on Sox2 at the 16-cell stage (Fig. 4E, Box 2). After blastocyst formation, the presence of either NANOG or OCT4 ensures that Sox2 expression is maintained (Fig. 4E, Box 3). Finally, as the embryo approaches implantation, Nanog and Oct4 are both required to sustain Sox2 expression (Fig. 4E, Box 4). Given that Sox2 is detectable in preimplantation embryos of many mammalian species (Blakeley et al., 2015; Boroviak et al., 2018; Frankenberg et al., 2013; Goissis and Cibelli, 2014; Petropoulos et al., 2016), examining the functional requirements for HIPPO pathway members in the temporospatial regulation of Sox2 in other species will provide exciting new insight into the evolution of pluripotency.

Animal care and husbandry was performed in accordance with the guidelines established by the Institutional Animal Care and Use Committee at Michigan State University. Wild-type embryos were generated by mating CD-1 mice (Charles River). Female mice used in this study were between 6 weeks and 6 months of age and males were used from 8 weeks to 9 months. Alleles and transgenes used in this study were maintained on a CD-1 background and include: Nanog^tm1.1Hoch (Maherali et al., 2007), Pou5f1^tm1Scho (Kehler et al., 2004), Tead4^tm1Bnno (Yagi et al., 2007), Yap1^tm1.1Eno (Xin et al., 2011), Wwtr1^tm1.1Eno (Xin et al., 2013) and Tg(Zp3-cre)93Knw (de Vries et al., 2000). Conditional, floxed alleles were recombined to generate null alleles by breeding mice carrying conditional alleles to Alpl^tm(cre)Nagy (Lomelí et al., 2000) mice.

Embryos were collected from naturally mated mice by flushing dissected oviducts or uteri with M2 medium (Millipore-Sigma). All embryos were cultured in 5% CO₂ at 37°C under ES cell grade mineral oil (Millipore-Sigma). Prior to embryo culture, KSOM medium (Millipore-Sigma) was equilibrated overnight in the embryo incubator.

cDNAs encoding Lats2, Tead4VP16 and Tead4EnR (Nishioka et al., 2009) cloned into the pcDNA3.1 poly(A)83 plasmid (Yamagata et al., 2005) were linearized, and then used as a template to generate mRNAs for injection by the mMessage mMachine T7 transcription kit (Invitrogen). NLS-GFP mRNA was synthesized from linearized NLS-GFP plasmid (Ariotti et al., 2015) using the mMessage mMachine Sp6 transcription kit (Invitrogen). Prior to injection, synthesized mRNAs were cleaned and concentrated using the MEGAclear Transcription Clean-up Kit (Invitrogen). Lats2 and NLS-GFP mRNAs were injected into both blastomeres of two-cell stage embryos at a concentration of 500 ng/µl. Tead4VP16 or Tead4EnR mRNAs were injected into a single blastomere of four-cell stage embryos at a concentration of or 150 ng/µl each. NLS-GFP mRNA was included in four-cell stage injections at a concentration of 150 ng/µl to trace the progeny of the injected blastomere. All mRNAs were diluted in 10 mM Tris-HCl (pH 7.4) and 0.1 mM EDTA. Injections were performed using a Harvard Apparatus PL-100A microinjector.

Embryos were fixed in 4% formaldehyde (Polysciences) for 10 min, permeabilized with 0.5% Triton X-100 (Millipore-Sigma) for 30 min and blocked with 10% FBS and 0.1% Triton X-100 for at least 1 h at room temperature or longer at 4°C. Primary antibody incubation was performed at 4°C overnight using the following antibodies: goat anti-SOX2 (Neuromics, GT15098, 1:2000), rabbit anti-NANOG (Reprocell, RCAB002P-F, 1:400), mouse anti-Tead4 (Abcam, ab58310, 1:1000), mouse anti-YAP (Santa Cruz Biotechnology, sc101199, 1:200) and rat anti-ECAD (Millipore-Sigma, U3254, 1:500). Anti-SOX2, anti-TEAD4 and anti-YAP antibodies were validated by the absence of positive staining on embryos homozygous for null alleles encoding antibody target nuclei were labelled with either DRAQ5 (Cell Signaling Technology) or DAPI (Millipore-Sigma). Antibodies raised against IgG and coupled to Dylight 488, Cy3 or Alexa Fluor 647 (Jackson ImmunoResearch) were used to detect primary antibodies. Embryos were imaged on an Olympus FluoView FV1000 Confocal Laser Scanning Microscope using a 20× UPlanFLN objected (0.5 NA) and 5× digital zoom. Each embryo was imaged in entirety using 5 µm optical section thickness.

Confocal sections of entire embryos, collected at 5 µm intervals, were analyzed using ImageJ (Schneider et al., 2012). Each nucleus was identified by DNA stain and then scored for the presence or absence of SOX2. In Fig. 1A,B, cells were classified as inside or outside on the basis of ECAD localization. For analysis of Nanog;Oct4 null embryos in Fig. 1C,D and Fig. S1D, SOX2 staining intensity was categorized as intense or weak. Intense SOX2 staining was defined as the level observed in non-mutant embryos, which was uniform among inside cells. In Fig. 1 and Figs S1, S2, embryo genotypes were not known prior to analysis. In Figs 3 and 4 embryos were grouped according to injection performed, and therefore the researcher was not blind to embryo treatment.

For embryos at the eight-cell stage or older, DNA was extracted from fixed embryos after imaging using the Extract-N-Amp kit (Millipore-Sigma) in a total volume of 10 µl. For embryos at the four-cell stage, DNA was extracted from fixed embryos in a total volume of 5 µl. 1 µl of extracted DNA was used as template, with allele-specific primers (Table S1).

We thank members of the Ralston Lab for thoughtful discussion.