Abstract
In mammals, the first few days of development entail segregation of pluripotent and extraembryonic trophectoderm cells. The challenge for the embryo at this time is to override cell plasticity to ensure that cells adopt distinct fates. Here, we identify novel mechanisms restricting expression of Sox2 and Cdx2 to mutually exclusive domains. We show that Sox2 is repressed in the trophectoderm downstream of ROCK1/2 and HIPPO pathway LATS1/2 kinases. LATS1/2 kinases are thought to antagonize YAP1 and WWTR1, the transcriptional partners of TEAD4. However, the combined loss of Yap1; Wwtr1 has not been reported. Using female germ line deletion and null alleles, we show that YAP1/WWTR1/TEAD4 simultaneously antagonize Sox2 and promote Cdx2 expression in trophectoderm. However, Cdx2 is less sensitive to Yap1 and Wwtr1 dosage, which can lead cells to aberrantly coexpress CDX2 and SOX2, reminiscent of conflicted cell fate. We show that HIPPO resolves cell fate conflicts by elevating Sox2 expression and driving cells to the inner cell mass, but Sox2 is not required for HIPPO-mediated cell repositioning. Rather, HIPPO signaling represses cell polarity components PAR and aPKC, facilitating cell internalization upstream of Sox2. We propose that HIPPO engages in negative feedback and PAR-aPKC providing a fail-safe to ensure lineage segregation.
Introduction
During embryogenesis, cells gradually differentiate, adopting distinct gene expression profiles and fates. In mammals, the first cellular differentiation is the segregation of trophectoderm and inner cell mass. The trophectoderm, which comprises the outer surface of the blastocyst, will mainly produce cells of the placenta, while the inner cell mass will produce pluripotent cells, which are progenitors of both fetus and embryonic stem cells. Understanding how pluripotent inner cell mass cells are segregated from non-pluripotent cells will therefore reveal how pluripotency is first induced in the biologically significant setting of the embryo.
The first cell fate decision has been studied mainly from the perspective of trophectoderm specification, because the transcription factor CDX2, which is expressed in and essential for trophectoderm development (Strumpf et al., 2005), has provided a way to distinguish future trophectoderm cells from non-trophectoderm cells (Beck et al., 1995) and identify the mechanisms that break the symmetry in the mouse embryo. For example, knowledge of CDX2 as a marker of trophectoderm cell fate enabled the discovery of mechanisms by which the HIPPO signaling pathway interprets cellular differences in polarity and position to restrict Cdx2 expression to trophectoderm. Briefly, polarization of outer embryonic cells around the 16-cell stage leads to inactivation of HIPPO pathway kinases LATS1/2. LATS1/2 inactivation enables Cdx2 expression in outer cells since LATS1/2 can antagonize activity of the transcriptional complex upstream of Cdx2, which is thought to include YAP1, WWTR1 and TEAD4 (Anani et al., 2014; Cockburn et al., 2013; Hirate et al., 2013; Kono et al., 2014; Korotkevich et al., 2017; Leung and Zernicka-Goetz, 2013; Lorthongpanich et al., 2013; Mihajlović and Bruce, 2016; Nishioka et al., 2009, 2008; Posfai et al., 2017; Rayon et al., 2014; Watanabe et al., 2017; Yagi et al., 2007; Zhu et al., 2017). However, the specific requirements for Yap1 and Wwtr1 in regulation of cell fate has been inferred from overexpression of wild type and dominant-negative variants, neither of which provide the standard of gene expression analysis that null alleles can provide. Moreover, studies of Cdx2 regulation do not provide direct knowledge of how pluripotent cells are first formed, because the absence of Cdx2 expression does not necessarily indicate acquisition of pluripotency. As such, our knowledge of how symmetry is broken in the early mouse embryo is incomplete.
Progenitors of inner cell mass are first morphologically apparent at the 16-cell stage as unpolarized cells residing inside the morula (reviewed in Frum and Ralston, 2018). However, at this stage, pluripotency genes, such as Oct4 and Nanog, are expressed in both trophectoderm and inner cell mass (Dietrich and Hiiragi, 2007; Niwa et al., 2005; Palmieri et al., 1994; Strumpf et al., 2005). Moreover, Oct4 and Nanog only become restricted to inner cell mass in response to CDX2, which represses their expression in the trophectoderm (Strumpf et al., 2005). Therefore, despite the widespread usage of OCT4 and NANOG as pluripotency markers, during the stages that pluripotent progenitors first emerge in the embryo, OCT4 and NANOG do not provide specific readouts of pluripotency.
In contrast to Oct4 and Nanog, Sox2 is expressed specifically in inside cells at the 16- cell stage and is therefore a specific marker of pluripotency (Guo et al., 2010; Wicklow et al., 2014). Importantly, expression of Sox2 is not regulated by CDX2 (Wicklow et al., 2014), consistent with the suggestion that loss of Cdx2 does not necessarily indicate acquisition of pluripotency. Thus, identifying how Sox2 expression is regulated addresses the question of how symmetry is broken in the early mouse embryo from the unique perspective of when and how pluripotency is first established in vivo.
Results
Patterning of Sox2 is ROCK-dependent
We sought to identify the mechanisms by which the patterned expression of SOX2 is achieved in the blastocyst. We therefore focused on how Sox2 expression is normally repressed in the trophectoderm, to achieve the inside cell-specific SOX2 expression pattern. We previously showed that SOX2 patterning is normal in the absence of the trophectoderm factor CDX2 (Wicklow et al., 2014), suggesting that mechanisms that repress SOX2 in the trophectoderm act upstream of CDX2. Rho-associated, coiled-coil containing protein kinases (ROCK1 and 2) are thought to act upstream of Cdx2 because embryos developing in the presence of a ROCK-inhibitor (Y-27632) exhibit greatly reduced Cdx2 expression (Kono et al., 2014). Additionally, quantitative RT-PCR showed that Sox2 mRNA levels are elevated in ROCKi-treated embryos (Kono et al., 2014), suggesting that ROCK1/2 repress expression of SOX2 in the trophectoderm. However, the role of ROCK1/2 in regulating the patterned expression of Sox2 has not been investigated.
To evaluate the roles of ROCK1/2 in repressing Sox2 expression in the trophectoderm, we collected 8-cell stage embryos prior to compaction (E2.5), and then cultured these either in control medium or in the presence of ROCK-inhibitor (Y-27632) for 24 hours (Fig. 1A). Embryos cultured in control medium exhibited normal cell polarity, evidenced by the apical localization of PARD6B in outside cells, and the basolateral localization of E-cadherin (CDH1) in both outside and inside cells (Fig. 1B, C) as expected (Vestweber et al., 1987; Vinot et al., 2005). Additionally, SOX2 was detected only in inside cells in control embryos (Fig. 1C, D), as expected (Wicklow et al., 2014). By contrast, embryos cultured in ROCK inhibitor exhibited defects in cell polarity (Fig. 1B’, C’), consistent with prior studies (Kono et al., 2014). Moreover, ROCK-inhibited embryos exhibited ectopic Sox2 expression in outside cells (Fig. 1C’, D). These observations strongly suggest that ROCK1/2 regulate expression of Cdx2 and Sox2, promoting expression of the former, and repressing the latter, in the trophectoderm.
YAP1 is sufficient to repress expression of SOX2 in the inner cell mass
We next investigated how ROCK1/2 repress expression of SOX2 in outside cells. ROCK1/2 kinases are required for nuclear localization of YAP1 (Kono et al., 2014), a transcriptional activator that partners with TEAD4 in the blastocyst (Nishioka et al., 2009). Thus, the gain of Sox2 in outside cells of ROCKi-treated embryos could be due to loss of nuclear YAP1. Moreover, we previously showed that TEAD4 represses expression of Sox2 in the trophectoderm (Wicklow et al., 2014), raising the possibility that YAP1 partners with TEAD4 to repress Sox2 expression in the trophectoderm. To test this hypothesis, we utilized a constitutively active variant of YAP1 (YAP1CA). Normally, LATS-dependent phosphorylation of YAP1 on serine 112 leads YAP1 to interact with 14-3-3 protein and become tethered in the cytoplasm. Therefore substitution of alanine at serine 112 (YAP1S112A or YAP1CA hereafter) leads YAP1 to be constitutively nuclear and constitutively active (Dong et al., 2007; Nishioka et al., 2009; Zhao et al., 2007). We injected mRNAs encoding YAP1CA and GFP into one of two blastomeres at the 2-cell stage, and then cultured these to morula or blastocyst stages (Fig. 1E). Mosaic overexpression permitted comparison of YAP1CA-overexpressing and non-injected cells, which served as internal negative controls. We first examined localization of YAP1 in these embryos at the morula stage, with the expectation that YAP1 would be detected in nuclei of both inside and outside cells in YAP1CA- overexpressing cells (Nishioka et al., 2009). As expected, YAP1 was observed in nuclei of all YAP1CA-overexpressing cells (Fig. 1F, G). By contrast, YAP1 was detected in nuclei only in outside cells. Additionally, pYAP1, which is normally cytoplasmic in inside cells, was also observed in nuclei of YAP1CA-overexpressing cells, suggesting that YAP1CA can still be phosphorylated on intact residues. These observations confirmed that S112 is essential for cytoplasmic retention of YAP1.
Having confirmed that YAP1CA exhibits constitutively nuclear localization, we next evaluated the consequences of ectopic nuclear YAP1 on expression of SOX2 in inside cells. We observed that the majority of YAP1CA-overexpressing cells (92%) exhibited no detectable SOX2 (Fig. 1G, H). Therefore, nuclear localized YAP1 is sufficient to repress Sox2 expression in the inner cell mass. These observations, together with our previous work (Wicklow et al., 2014) suggest that, YAP1/TEAD4 normally repress Sox2 in trophectoderm and that a key determinant in Sox2 repression is the nuclear localization of YAP1.
YAP1 and WWTR1 restrict Sox2 expression to the inner cell mass
Our observation that nuclear YAP1 is sufficient to repress expression of Sox2 in the inner cell mass raised the possibility that Yap1 is required to repress expression of Sox2 in in outside cells. However, it was previously reported that Yap1 null embryos survive until E9.0 (Morin-Kensicki et al., 2006) and exhibit normal Cdx2 expression at the blastocyst stage (Nishioka et al., 2009), suggesting that other factors compensate for loss of Yap1 during preimplantation development. Accordingly, the closely related gene Wwtr1, which is coexpressed with Yap1 in the blastocyst (Varelas et al., 2010), has been proposed to compensate for the loss of Yap1 during preimplantation development (Nishioka et al., 2009). However, the Yap1; Wwtr1 double knockout phenotype has not been reported.
To generate Yap1; Wwtr1 double knockout embryos, we mated mice carrying null alleles of Yap1 and Wwtr1, generated by CRE-mediated deletion of each conditional allele (Xin et al., 2013, 2011) (see Methods), and then examined embryos at E3.25. Surprisingly, CDX2 expression appeared normal in Yap1; Wwtr1 double knockout embryos (Fig. 2A-D). In contrast, SOX2 was detected in most cells of Yap1; Wwtr1 double knockout embryos, indicating that Yap1 and Wwtr1 are required to repress expression of SOX2 in outside cells. Yap1 null embryos carrying only a single functional copy of Wwtr1 (or vice versa, Wwtr1 -/-; Yap1 +/-) exhibited minor defects in SOX2 expression, with ectopic SOX2 detected in a few outside cells (Fig. 2B, C, arrowheads). These observations indicate that SOX2 expression is normally repressed in outside cells in a Yap1/Wwtr1-dependent manner and indicate that regulation of Sox2 expression is more sensitive to Yap1/Wwtr1 dose than is the regulation of Cdx2 expression.
Oocyte-expressed Yap1 has been shown to be required for Cdx2 expression in preimplantation embryos (Yu et al., 2016). We therefore hypothesized that maternally provided Yap1 and/or Wwtr1 contribute to regulation of Cdx2 and Sox2. To test this hypothesis, we deleted Yap and Wwtr1 from the female germ line by generating mice carrying conditional alleles of Yap and Wwtr1 in the presence of the female germ line-specific Zp3Cre (de Vries et al., 2000). We then crossed these females to males carrying null alleles of Yap1 and Wwtr1. From these crosses, we obtained embryos lacking maternally provided Yap1 and Wwtr1 and carrying all expected combinations of maternal and zygotic alleles (Table 1). Interestingly, embryos lacking maternal Yap1 and Wwtr1 and carrying wild type alleles of both zygotic Yap1 and Wwtr1 exhibited normal CDX2 and SOX2 expression patterns (Fig. 2E), arguing that maternal Yap1 and Wwtr1 are not required in the presence of zygotically expressed Yap1 and Wwtr1. This observation contrasts with the published observation that expression of CDX2 is lost in the absence of maternal Yap1 (Yu et al., 2016). However, we observed that loss of maternally provided Yap1 and Wwtr1 worsened the Yap1 zygotic null SOX2 phenotype (Fig. 2F compare to B). Conversely, loss of maternal Yap1 and Wwtr1 worsened the Wwtr1 zygotic null SOX2 phenotype (Fig. 2G compare to C). Meanwhile, CDX2 expression appeared normal in embryos of these genotypes. These results are consistent with our proposal that regulation of SOX2 is more sensitive to the dose of Yap1/Wwtr1 than is CDX2. In fact, CDX2 expression was disrupted only by loss of both maternal and zygotic Yap1 and Wwtr1 (Fig. 2H). We therefore conclude that Yap1 and Wwtr1 are functionally equivalent and that maternal Yap1 and Wwtr1 are genetically redundant with zygotic Yap1 and Wwtr1. Additionally, these observations provide the first loss of function evidence using null alleles to show that YAP and WWTR1 promote expression of Cdx2 while repressing expression of Sox2 in the trophectoderm.
YAP1 and WWWTR1 maintain outside cell positioning
To examine the roles of Yap1 and Wwtr1 in blastocyst development we examined embryos lacking Yap1 or Wwtr1 at E3.75. We again observed a range of phenotypes, correlating with dose of Yap1 and Wwtr1. Embryos with intact maternal Yap1 and Wwtr1 and at least two zygotic copies of Yap1 or Wwtr1 appeared normal, with CDX2-positive trophectoderm encapsulating a cavity containing SOX2 positive ICM cells (Fig. 3A). By contrast, embryos with intact maternal Yap1 and Wwtr1, but with only one functional copy of zygotically-expressed Wwtr1 exhibited low levels of ectopic SOX2 in trophectoderm cells (Fig. 3B, white arrowheads). Ectopic SOX2 was not detected in this genotype at E3.25, suggestive of rescue by maternal Yap1 and Wwtr1 at the earlier time point. This genotype was phenocopied in embryos that were lacking maternal Yap1 and Wwtr1 and were heterozygous for both Yap1 and Wwtr1 (Fig. 3C, white arrowheads). A more severe phenotyped was observed in embryos lacking maternal Yap1 and Wwtr1 with only one functional zygotic allele of Yap1 or Wwtr1, which did not form a blastocyst and, in addition to exhibiting ectopic SOX2, contained CDX2 negative cells in the outside position (Fig. 3D, E, yellow arrowheads). Strikingly, the number of outside cells appeared severely reduced in embryos lacking maternal Yap1 and Wwtr1 with only one functional zygotic allele of Yap1 or Wwtr1, suggesting that Yap1 and Wwtr1 are essential for maintaining outside cell position.
To quantify these phenotypes, we grouped embryos according to the number of functional doses of Yap1 and Wwtr1, regardless of maternal or zygotic source, counting the maternal contribution of Yap1 and Wwtr1 as a single functional dose, and then compared the numbers of outside and inside cells among embryos with one, two, or more than two functional doses of Yap1 and Wwtr1. To identify inside versus outside cell position, we used the cell polarity marker CDH1 and defined inside cells as both appearing internal and as having uniform CDH1 staining around their entire cell membrane. In some cases, cells appearing internal, but lacking uniform CDH1 staining were present (Fig. 3D, yellow arrowheads) and were therefore counted as outside positioned. This criterion may have led us to overestimate the number of outside cells, but in spite of this, we observed a major decrease in the number of outside cells and an increase in the number of inside cells in embryos with as Yap1 and Wwtr1 dosage decreased (Fig. 3F, G). Finally, we quantified expression of SOX2 and CDX2 among embryos of these genotypic categories. Ectopic SOX2 was detected in the outside cells of embryos with two or fewer function copies of Yap1 and Wwtr1. However, CDX2 expression followed a slightly different trend. While CDX2 was present in all outside cells in embryos with 2 or more functional alleles of Yap1 and Wwtr1, some outside cells were also CDX2-negative in embryos with only 1 functional allele of Yap1 and Wwtr1 (Fig 3H). These observations are consistent with the differential sensitivities of Cdx2 and Sox2 to the dose of Yap1 and Wwtr1 observed at earlier stages and demonstrate that Yap1 and Wwtr1 are required to maintain cell position on the surface of the embryo.
LATS kinase is sufficient to induce inside cell positioning
The observation that Yap1 and Wwtr1 are essential for maintaining cell positioning on the surface of the embryo led us to hypothesize that HIPPO signaling, which represses YAP1/WWTR1, maintain inside positioning within the embryo. We therefore aimed to hyperactivate HIPPO signaling in the embryo, which is typically achieved by overexpression of LATS2 kinase (Nishioka et al., 2009). We previously showed that by injection of Lats2 mRNA into both blastomeres of the 2-cell stage embryos leads all cells to induce expression of Sox2 (Wicklow et al., 2014), consistent with the notion that HIPPO signaling induces inner cell mass fate. However, since all cells are converted to inner cell mass following Lats2 overexpression, these experiments do not enable us to know whether Lats2-overexpressing cells would adopt an inside position were they in the context of a wild type embryo. We therefore injected only one of two blastomeres with Lats2 mRNA, and then evaluated their gene expression and positioning at later stages (Fig. 4A). We observed that almost all Lats2-overexpressing cells ended up contributing to the inner cell mass (Fig. 4B, C, G). Notably, SOX2 was detected in all Lats2-overexpressing cells observed within the inner cell mass (Fig. 4D), suggesting that Lats2-overexpressing cells were not only localized to the inner cell mass but had also acquired an inner cell mass fate.
The increased prevalence of Lats2-overexpressing cells in the inner cell mass was also associated with a striking underrepresentation of Lats2-overexpressing cells within the trophectoderm and a decrease in the number of outside cells compared to embryos injected with GFP mRNA alone (Fig. 4B, C, E), suggesting that Lats2 overexpression led cells to move to the inside or to die on the outside. Consistent with this latter possibility, a smaller proportion of Lats2-overexpressing cells were observed in the total embryo population than in embryos in which one blastomere was injected with GFP mRNA only (Fig. 4C). Moreover, Lats2 overexpression led to formation of cellular fragments within the trophectoderm (Fig. 4B, yellow arrowheads), suggestive of cell death among Lats2-overexpressing outside cells. Interestingly, SOX2 was detected in rare outside Lats2-overexpressing cells (Fig. 4D). Therefore, LATS2 is sufficient to induce expression of SOX2, regardless of cell position within the embryo. Notably, the kinase-dead variant of LATS2, which functions as a dominant-negative (Nishioka et al., 2009), did not alter the positioning, survival, or Sox2 expression status of injected cells (Fig. 4F,G), consistent with a previous report (Posfai et al., 2017). Therefore, Lats2- overexpression acts on cell position and cell fate through target proteins normally phosphorylated by LATS2 in response to HIPPO signaling.
To pinpoint when Lats2-overexpressing cells come to occupy the inside of the embryo, we performed a time course, examining the position of injected and non-injected cells from the 16-cell to the blastocyst stage (up to 80 cells). Interestingly, between the 16 and 32-cell stages, the proportion of injected and non-injected cells in the total, outside, and inside cell populations were comparable whether embryos had been injected with Lats2 and GFP or GFP mRNA alone (Fig. 4H-J). In embryos injected with GFP mRNA alone, the proportion of injected and non-injected cells in the total, outside and inside cell populations remained constant throughout the time course. In contrast, starting after the 32-cell stage, the average proportion of injected Lats2-overexpressing cells occupying the inside position increased dramatically. The Lats2-overexpressing cells increased the total number of inside cells in the ICM of embryos injected with Lats2/GFP mRNA compared to embryos injected with GFP mRNA only, without a significant decrease in the number of non-injected cells in the ICM, arguing that the non cell-autonomous effects of Lats2-overexpression on the positioning of non-injected cells are minimal (Fig 4K). The increase in the proportion of Lats2-overexpressing cells in the inside position was concurrent with a severe decrease in the proportion of Lats2- overexpressing cells occupying the outside position, consistent with internalization of Lats2-overexpressing cells after the 32-cell stage. Additionally, Lats2-injected cells became underrepresented as a proportion of the total embryo population after the 32- cell stage, lending further support to the idea that Lats2-overexpressing cells that fail to internalize undergo cell death. We therefore conclude that Lats2 overexpression acts on cell position and survival around the time of blastocyst formation.
LATS2 acts through YAP1 to induce inner cell mass fate
To examine the mechanism by which LATS2 overexpression drives cells into the inner cell mass, we first asked whether LATS regulation of YAP1/WWTR1 mediates cell positioning. To evaluate this possibility, we co-overexpressed mRNAs encoding Lats2 and YAP1CA. We predicted that, if Lats2-overexpression drives cells to adopt inner cell mass fate by repressing YAP1/WWTR1, then co-overexpression of Yap1CA would enable Lats2-overexpressing cells to contribute to trophectoderm.
We injected mRNAs encoding Yap1CA and Lats2 into one of two blastomeres at the two-cell stage, and then examined cell positions at the blastocyst stage (Fig 5A). Consistent with our hypothesis, Yap1CA-overexpression led to a significant decrease in the proportion of Lats2-overexpressing cells contributing to the inside cell position (Fig 5B- D). Additionally, Yap1CA-overexpression significantly increased the proportion of Lats2- overexpressing cells remaining in the outside position (Fig 5D-F). Therefore, because overexpression of Yap1CA attenuated the effects of Lats2-overexpression on cell positioning and cell survival, we propose that LATS2 alters cell position and cell survival by regulating YAP1, and possibly WWTR1, activity.
LATS2 induces positional changes independent of Sox2
Our observation that Lats2-overexpression induces expression of SOX2 and leads cells to adopt inner cell mass fate prompted us to investigate whether SOX2 helps drive cell position, downstream of Lats2. In support of this hypothesis, SOX2 has been proposed to determine inner cell mass fate (Goolam et al., 2016; White et al., 2016). We therefore investigated whether Sox2 is required for the inner cell mass-inducing activity of LATS2 by overexpressing Lats2 in embryos lacking Sox2 (Fig. 5G). Surprisingly, however, Lats2-overexpressing cells were equally likely to occupy inside position in the presence and absence of Sox2 (Fig. 5H, I). Moreover, Lats2-overexpressing cells were equally unlikely to occupy outside position in the presence and absence of Sox2 (Fig. 5H, I) (Fig. 5H, J). Therefore, although Lats2 overexpression induces expression of SOX2, LATS2 acts on cell positioning/survival upstream of, or in parallel to, Sox2.
LATS2 antagonizes formation of the apical domain
Cell position in the preimplantation mouse embryo has been proposed to be determined by differential inheritance of apically localized of membrane components that pattern myosin activity to either constrict future ICM cells to the inside of the embryo, or to maintain the position of future trophectoderm cells on the embryo surface (Anani et al., 2014; Korotkevich et al., 2017; Maître et al., 2016, 2015; Samarage et al., 2015; Zenker et al., 2018). Indeed, the apical membrane component aPKC is required for maintaining outside cell positon (Dard et al., 2009; Hirate et al., 2015; Plusa et al., 2005) and fate (Alarcon, 2010). We hypothesized that, because Lats2 overexpression led cells to adopt an inside position, that LATS2 antagonizes formation of the apical domain in outside cells before these cells are lost from the outer surface of the embryo.
To determine if the apical domain becomes disrupted by Lats2 overexpression, we examined the localization of PARD6B and aPKCz in embryos at the 16 to 32-cell stages. While apical membrane components PARD6B and aPKCz were detected at the apical membrane of non-injected cells and cells injected with GFP only, most Lats2- overexpressing cells lacked detectable PARD6B and aPKCz (Fig. 6A-D). Notably, PARD6B and aPKCz appeared downregulated, rather than mislocalized, suggesting that LATS2-overexpressing cells could be properly polarized. To evaluate cell polarization, we examined localization of CDH1, which was restricted to the basolateral membrane of both Lats2-overexpressing and non-injected cells (Fig. 6E). Moreover, other apically localized proteins were also properly localized in LATS2-overexpressing cells, including filamentous Actin (Fig. 6F), and phospho-ERM (pERM) (Fig. 6G). These observations demonstrate that Lats2-overexpressing outside cells are properly polarized, but that apical membrane components required for outside cell position and fate are repressed in Lats2-overexpressing cells, providing a mechanism by which HIPPO/LATS drives cell positioning and cell fate assignment.
Discussion
During preimplantation development, many lineage-specific transcription factors, including CDX2, OCT4, NANOG, and GATA6 are initially expressed throughout the embryo before their expression refines to their lineage-appropriate domains (Chazaud et al., 2006; Dietrich and Hiiragi, 2007; McDole and Zheng, 2012; Ralston and Rossant, 2008; Schrode et al., 2014; Strumpf et al., 2005). In striking contrast to these genes, SOX2 is never detected in outside cells (Wicklow et al., 2014), indicating that robust mechanisms must exist to prevent its aberrant expression in trophectoderm. In the present study, we show that the YAP1/WWTR1/TEAD4 complex exerts tight control over Sox2 expression, downstream of cell polarization and HIPPO signaling, to ensure that Sox2 is not expressed in the trophectoderm. This model contrasts with models proposing that SOX2 acts in select embryonic cells as early as the 4-cell stage to dictate inner cell mass positioning and fate (Goolam et al., 2016; White et al., 2016). However, these studies did not evaluate the requirement for Sox2 in regulating cell position, in contrast to the present study.
We have provided the first analysis of Yap1 and Wwtr1 loss of function using null alleles. Interestingly, we find that the complete absence of both Yap1 and Wwtr1 disrupts expression of Cdx2 completely, in contrast to the Tead4 null phenotype, in which low levels of Cdx2 are still detected (Nishioka et al., 2008). This observation suggests a role for maternal Tead4 or Tead paralogues in promoting expression of Cdx2. We note that YAP1 and WWTR1, which repress Sox2 expression, are not known to act as transcriptional repressors. Therefore, YAP1/WWTR1 may induce expression of an as-yet unidentified transcriptional repressor.
We have shown that Sox2 expression is more sensitive to the dose of functional Yap1 and Wwtr1 alleles than is Cdx2. In other words, moderate decreases in Yap/Wwtr1 cause derepression of Sox2 expression without affecting Cdx2 expression. The differing sensitivities of Sox2 and Cdx2 to Yap1/Wwtr1 dose raise the possibility that intermediate doses of active YAP1/WWTR1 could yield cells that coexpress both SOX2 and CDX2. This proposal is consistent with the observation that CDX2 is detected in inside cells of the embryo during blastocyst formation (Dietrich and Hiiragi, 2007; McDole and Zheng, 2012; Ralston and Rossant, 2008). The coexpression of these lineage markers could be indicative of conflicted trophectoderm/inner cell mass fate (Fig. 7A, B). Therefore, to ensure robust developmental transitions, embryos must have a mechanism for resolving such cell fate conflicts and nudging cells into their correct and final positions.
The timing of HIPPO-induced cell internalization coincides with loss of cell fate plasticity around the 32-cell stage (Posfai et al., 2017). This timing is likely due to formation of tight-junctions form between outside cells (Sheth et al., 1997), which reinforce differences in HIPPO signaling activity between inside and outside compartments of the embryo (Hirate and Sasaki, 2014; Leung and Zernicka-Goetz, 2013). However, cell divisions around the 32-cell stage temporarily disrupt inside and outside regions. Occasionally, mislocalized cells are caught in the act, as unpolarized cells with elevated HIPPO signaling located outside the embryo (Anani et al., 2014; Hirate et al., 2015). We propose two mechanisms by which such cells are eliminated from the trophectoderm (Fig. 7C).
First, a small proportion of conflicted cells undergo cell death. This is consistent with low baseline levels of apoptosis in cells around the 32-cell stage (Copp, 1978). Notably, cell lethality due to elevated HIPPO can be rescued by increasing levels of nuclear YAP1, suggesting that YAP1 activity normally provides a pro-survival signal to trophectoderm cells, consistent with the proposed role of YAP1 in promoting proliferation in non-eutherian mammals (Frankenberg, 2018). Moreover, deletion of Sox2 did not rescue survival of outside cells in which HIPPO signaling was artificially elevated, arguing that HIPPO resolves cell fate conflicts upstream of lineage-specific gene expression.
Second, outside cells with elevated HIPPO signaling drive their own internalization. This is consistent with the observation that cells in which Tead4 has been knocked down become internalized (Mihajlović et al., 2015). However, in contrast to Tead4 knockdown, which preserves cell polarity, we show that LATS2 overexpression leads to repression of PAR/aPKC. Par/aPKC are required for outside cell positioning (Dard et al., 2009; Hirate et al., 2015; Plusa et al., 2005), but HIPPO signaling was not previously known to be capable of regulating the expression of Par/aPKC in the mouse early embryo. Intriguingly, in Drosophila, the Lats orthologue Warts has been shown to regulate the localization of Par/aPKC components during development (Keder et al., 2015; Lucas et al., 2013). Therefore, our observation of reduced Par/aPKC in Lats2-overexpressing cells is consistent with conserved regulation of the Par/aPKC complex by HIPPO signaling.
Our data show that, in addition to moving cells inside the embryo, Lats2 overexpression antagonizes the epithelial character of outside cells by antagonizing PAR-aPKC complex formation. Therefore, the ability of LATS2 to antagonize the PAR-aPKC complex could point to the existence of an exit route for cells to delaminate from the outer trophectoderm epithelium, and then internalize. Alternatively, LATS2 could impact spindle orientation and increase the proportion of cells undergoing asymmetric division, via PAR-aPKC (Korotkevich et al., 2017; Plusa et al., 2005). Identifying the downstream mechanisms by which HIPPO drives cells to inner cell mass will be an exciting topic of future study.
Methods
Mouse strains and genotyping
All animal research was conducted in accordance with the guidelines of the Michigan State University Institutional Animal Care and Use Committee. Wild type embryos were derived from CD-1 mice (Charles River). The following alleles or transgenes were used in this study, and maintained in a CD-1 background: Sox2tm1.1Lan (Smith et al., 2009), Yaptm1.1Eno (Xin et al., 2011), Wwtr1tm1.1Eno (Xin et al., 2013), Tg(Zp3-cre)93Knw (de Vries et al., 2000). Null alleles were generated by breeding mice carrying floxed alleles and mice carrying ubiquitously expressed Cre, 129-Alpltm(cre)Nagy (Lomelí et al., 2000).
Embryo collection and culture
Mice were maintained on a 12-hour light/dark cycle. Embryos were collected by flushing the oviduct or uterus with M2 medium (Millipore). For embryo culture, KSOM medium (Millipore) was equilibrated overnight prior to embryo collection. Y-27632 (Millipore) was included in embryo culture medium at a concentration of 80 μM with 0.4% DMSO, or 0.4% DMSO as control, where indicated. Embryos were cultured at 37°C in a 5% CO2 incubator under light mineral oil.
Embryo microinjection
LATS2 and YAPS112A mRNA was synthesized from cDNAs cloned into the pcDNA3.1- poly(A)83 plasmid (Yamagata et al., 2005) using the mMESSAGE mMACHINE T7 transcription kit (Invitrogen). EGFP or nls-GFP mRNA were synthesized from EGFP cloned into the pCS2 plasmid or the nls-GFP plasmid (Ariotti et al., 2015) using the mMESSAGE mMACHINE SP6 transcription kit (Invitrogen). mRNAs were cleaned and concentrated prior to injection using the MEGAclear Transcription Clean-Up Kit (Invitrogen). Lats2, Lats2KD and YAPCA mRNAs were injected into one blastomere of two-cell stage embryos at a concentration of 500 ng/μl, mixed with 350 ng/μl EGFP or nls-GFP mRNA diluted in 10 mM Tris-HCl (pH 7.4), 0.1 mM EDTA.
Immunofluorescence and Confocal Microscopy
Embryos were fixed with 4% formaldehyde (Polysciences), permeabilized with 0.5% Triton X-100 (Sigma Aldrich), and blocked with 10% FBS, 0.2% Triton X-100. Primary Antibodies used were: mouse anti-CDX2 (Biogenex, CDX2-88), goat anti-SOX2 (Neuromics, GT15098), rabbit anti-PARD6B (Santa Cruz, sc-67393), rabbit anti-PARD6B (Novus Biologicals, NBP1-87337), mouse anti-PKCζ (Santa Cruz Biotechnology, sc-17781), rat anti-CDH1 (Sigma Aldrich, U3254), mouse anti-YAP (Santa Cruz Biotechnology, sc101199), rabbit anti phospho-YAP (Cell Signaling Technologies, 4911), chicken anti-GFP (Aves, GFP-1020). Stains used were: Phallodin-633 (Invitrogen), DRAQ5 (Cell Signaling Technologies) and DAPI (Sigma Aldrich). Secondary antibodies conjugated to DyLight 488, Cy3 or Alexa Flour 647 fluorophores were obtained from Jackson ImmunoResearch. Embryos were imaged using an Olympus FluoView FV1000 Confocal Laser Scanning Microscope system with 20x UPlanFLN objective (0.5 NA) and 5x digital zoom. For each embryo, z-stacks were collected, with 5 μm intervals between optical sections. All embryos were imaged prior to knowledge of their genotypes.
Embryo Genotyping
To determine embryo genotypes, embryos were collected after imaging and genomic DNA extracted using the Extract-N-Amp kit (Sigma) in a final volume of 10 μl. Genomic extracts (1-2 μl) were then subjected to PCR using allele-specific primers (Table 2).
Competing Interests
The authors declare no competing interests.
Acknowledgements
We are grateful to Dr. Hiroshi Sasaki for providing overexpression constructs, to Dr. Randy L. Johnson for providing Yap1 and Wwtr1 conditional alleles, and to Dr. Jason Knott for microinjection training. We also thank Dr. Ripla Arora and members of the Ralston Lab for comments. This work was supported by NIH R01 GM104009.