Smad4 is essential for epiblast scaling and morphogenesis after implantation, but nonessential before implantation

In animal embryos, including mice, frogs, fish and flies, the bone morphogenic protein (BMP) signaling pathway oversees crucial patterning events early in development. In non-mammalian species, BMP signaling is essential for specification of the dorsal/ventral axis of the early embryo (De Robertis and Sasai, 1996; O'Connor et al., 2006; Zinski et al., 2018). However, the mammalian embryo has an additional developmental task immediately following fertilization: specification of the extra-embryonic lineages that will give rise to placenta and yolk sac and enable development within the mother. Published studies support roles for BMP signaling in extra-embryonic lineage specification before implantation and in subsequent axial patterning, which occurs after implantation (Chu et al., 2004; Graham et al., 2014; Reyes de Mochel et al., 2015; Senft et al., 2019; Sirard et al., 1998; Yamamoto et al., 2009). However, differences in technical approaches used, as well as challenges intrinsic to mouse, have limited origination of a universally accepted model of the role of BMP signaling in mouse embryos throughout pre- and post-implantation stages.

BMP is one of several related and highly conserved molecular signaling pathways belonging to the transforming growth factor beta (TGFβ) superfamily of cytokines. The molecular mechanisms of TGFβ signaling have been carefully studied (Chang, 2016; Massagué and Sheppard, 2023). BMP proteins, like other members of the TGFβ pathway, are secreted ligands that elicit cellular responses by binding to heterodimeric, transmembrane serine-threonine kinase receptors. The activated receptor complex then phosphorylates members of a family of intracellular effectors known as receptor-associated SMADs (R-SMADs). Phosphorylation of R-SMADs allows their association with a co-factor SMAD (co-SMAD) and accumulation in the nucleus, where they impact chromatin and transcription (Hill, 2016). In mammals, R-SMAD activity is encoded by several Smad paralogues, with SMAD1, SMAD5 and SMAD9 (also known as SMAD8) primarily transducing BMP signals, and SMAD2 and SMAD3 primarily transducing Nodal, activin and TGFβ. Remarkably, the mammalian genome encodes a single co-SMAD, SMAD4, which is shared by BMP, Nodal, Activin, and TGFβ signaling pathways.

Across species, BMP signaling has been visualized in embryos using antibodies that specifically recognize the phosphorylated form of the BMP-responsive R-SMADs. This approach has been used to observe gradients of BMP signaling activity that correspond with the dorsal/ventral axis in fly, fish and frog embryos (Dorfman and Shilo, 2001; Plouhinec and De Robertis, 2009; Schohl and Fagotto, 2002; Tucker et al., 2008). In mouse, no graded pSMAD1/5/9 pattern has been reported. Rather, pSMAD1/5/9 is reportedly detected in all cell types of the embryo throughout preimplantation (Graham et al., 2014; Reyes de Mochel et al., 2015). After implantation, pSMAD1/5/9 is detected within a subdomain of extra-embryonic cells, and not within the embryo itself, until it is detected in primordial germ cells and emerging mesoderm during gastrulation (Senft et al., 2019). These observations suggest fundamental differences in the roles of BMP signaling between mammalian and non-mammalian animal embryos, but raise the need for additional, functional lines of evidence.

In mice, individual members of the BMP signaling pathway have been knocked out, but all appear to be dispensable before embryonic day (E) 6.5. Knockout of genes encoding the predominant ligand Bmp4 (Lawson et al., 1999; Winnier et al., 1995), the receptors Bmpr2 (Beppu et al., 2000), Bmpr1a (Mishina et al., 1995), Actr1a (Gu et al., 1999), the R-SMADs encoded by Smad1 (Tremblay et al., 2001) and Smad5 (Chang et al., 1999), and the co-SMAD Smad4 (Sirard et al., 1998; Yang et al., 1998, 2002) all point to essential roles for BMP signaling in extra-embryonic mesoderm, extra-embryonic endoderm and germ cell development. Mechanistically, BMP also interacts with Nodal to pattern the visceral endoderm (VE) and specify distal, and then anterior, VE – extra-embryonic cell types that pattern the embryo by specifying the location of the primitive streak (Robertson, 2014; Waldrip et al., 1998; Yamamoto et al., 2009). These events define gastrulation and anterior/posterior axial patterning, processes which therefore rely on BMP signaling. None of these studies reported that BMP signaling loss of function had any effect on development before E5.5. However, the presence of maternal gene products, provided within the oocyte, could complicate interpretation of zygotic knockout phenotypes. Indeed, evidence exists in other species that maternally supplied BMP pathway members are functional (Das et al., 1998; Faure et al., 2000; Kramer et al., 2002; Miyanaga et al., 2002; Zhang et al., 2020). An additional complication is that mouse embryos are particularly challenging to recover between E4.5 and E6.5, and an in vitro protocol that robustly recapitulates in vivo development during these stages is lacking. For all of these reasons, the roles for BMP signaling during the peri-implantation period are still unclear.

In contrast to peri-implantation, preimplantation embryos are relatively easy to isolate and culture in vitro. Accordingly, several studies have examined BMP signaling in preimplantation development. Culturing preimplantation embryos in the presence of small-molecule BMP inhibitors led to decreased numbers and cell cycle rate of extra-embryonic trophectoderm (TE) and primitive endoderm (PrE) cells, as well as changes in expression of lineage-specific transcription factors, including markers of PrE (SOX17, GATA6), TE (CDX2) and inner cell mass/epiblast (ICM/EPI; OCT4) (Graham et al., 2014; Reyes de Mochel et al., 2015; Stuart et al., 2019). Some of these observations were recapitulated following microinjection of siRNA against Bmp4 or overexpression of dominant-negative forms of Bmpr2 (Graham et al., 2014). Overexpression of dominant-negative Smad4 reportedly phenocopied loss of the upstream signaling components. In principle, these approaches could interfere with the activities of both maternally and zygotically expressed signaling components and thereby achieve more complete loss of function than zygotic null embryos. However, in these manipulated embryos, pSMAD1/5/9 was not examined, and so the extent to which these manipulations disrupted BMP signaling was not directly tested. Moreover, inhibitors are prone to off-target effects, which could further confound interpretation of results (Lowery et al., 2016).

In the present study, we visualize pSMAD1/5/9 in wild-type embryos and in embryos in which Bmp4 has been maternally and zygotically deleted. We evaluate their lineage specification and morphogenesis throughout preimplantation, peri-implantation and early post-implantation stages. We report that, in contrast to previous studies, BMP signaling is apparently dispensable during mouse preimplantation development. However, we describe a novel role for SMAD4-mediated signaling in limiting FGF/ERK signaling to enable the timely execution of EPI morphogenetic events shortly after implantation.

To determine when BMP signaling becomes active in the mouse embryo, we first optimized a method to examine the localization of transcription factors SMAD1, 5 and 9, which are phosphorylated in response to ligand/receptor binding (Dijke and Hill, 2004). To achieve this, we used immunofluorescence and an antibody that recognizes phosphorylated SMAD1/5/9 (pSMAD1/5/9; Fig. 1) (Senft et al., 2019; Xu et al., 2019; Yuan et al., 2015) and performed a time course analysis of mouse early embryos. We did not detect pSMAD1/5/9 in preimplantation embryos flushed from uteri between E3.75 and E4.25 (Fig. 1A; Fig. S1A), but pSMAD1/5/9 first became detectable in E4.5 peri-implantation embryos (Fig. 1A). We acknowledge that the assay did not allow us to reliably distinguish possible low levels of pSMAD1/5/9 from background, so we quantified only those cells which were highly pSMAD1/5/9-positive. At E4.5, we detected pSMAD1/5/9 in nuclei of a few ICM cells in 29% of embryos examined (Fig. 1B,C). By E4.75, when embryos have undergone implantation, we observed pSMAD1/5/9-positive cells in 87.5% of embryos evaluated (Fig. 1A-C). Starting at E5.0, we observed pSMAD1/5/9-positive cells within 100% of embryos examined (Fig. 1A-C; Fig. S1A). The observed pSMAD1/5/9 overlapped with a subset of GATA6-expressing PrE (E4.5 and E4.75) and VE (E5.5) cells (Fig. 1A).

To test the specificity of the pSMAD1/5/9 signal, we cultured E5.5 wild-type embryos for 6 h in the presence of LDN-193189 (LDN hereafter), which has been used to disrupt BMP signaling in mouse embryos (Graham et al., 2014; Reyes de Mochel et al., 2015). A concentration of 1 µM LDN was reported as sufficient to inhibit BMP signaling in preimplantation mouse embryos (Reyes de Mochel et al., 2015). However, we found that treatment with 1 µM LDN was highly toxic to embryos at this stage (Fig. S1B). Treatment with 0.25 µM LDN, however, led to complete loss of pSMAD1/5/9 signal in E5.5 embryos (Fig. S1B). Altogether, these observations suggest that BMP signaling becomes active around the time of embryo implantation but is not active during preimplantation stages.

A previous report has shown that BMP4 is sufficient to influence gene expression in preimplantation mouse embryos (Goissis et al., 2023), suggesting that preimplantation embryos can respond to exogenous BMP signals. We therefore examined expression dynamics of genes encoding BMP pathway members during preimplantation stages. We analyzed published single-cell RNA-seq (scRNA-seq) data from mouse embryos at stages E3.5-E6.5 (Nowotschin et al., 2019). Many core components of canonical BMP signaling were detectable as early as E3.5, including the ligand Bmp4, Type I receptor Bmpr1a, Type II receptors Bmpr2 and Acvr2b, receptor-associated SMAD Smad5 and co-factor SMAD Smad4 (Fig. 1D; Fig. S2), consistent with blastocyst competence to respond to BMP signals.

Next, we tested whether pSMAD1/5/9 could be induced in preimplantation embryos treated with exogenous BMP4. We cultured compacted eight-cell-stage embryos (E2.75) in 300 ng/ml BMP4 for 36 h to the blastocyst stage (equivalent in cell number to E3.75, as confirmed by cell counts). Although we did not observe pSMAD1/5/9 in any control embryos cultured in unsupplemented medium, we observed low, but detectable, levels of pSMAD1/5/9 in 82% (n=14/17) of embryos treated with exogenous BMP4 (Fig. 1E,F). Notably, pSMAD1/5/9 was detected only in the ICM but did not preferentially colocalize with either SOX2-positive (EPI) or SOX2-negative (PrE) cells. The presence of both SOX2-positive and SOX2-negative cells in the ICM is consistent with normal ICM differentiation (Wicklow et al., 2014), in spite of elevated pSMAD1/5/9. Therefore, we conclude that BMP signaling is not normally highly active during preimplantation development, but ICM cells are competent to respond to exogenous BMP signals at these stages, consistent with published investigations (Goissis et al., 2023; Graham et al., 2014; Reyes de Mochel et al., 2015).

We next evaluated pSMAD1/5/9 in embryos shortly after implantation. Consistent with previous reports (Senft et al., 2019; Yamamoto et al., 2009), we detected pSMAD1/5/9 within a zone of the VE that flanks the extra-embryonic ectoderm (EXE) at E5.5 and E5.75 (Fig. 1A; Fig. S1A). This observation is also consistent with evidence that several key components, including Bmp2, Smad1, Smad5 and Bmpr2, are substantially upregulated around the time of implantation (E4.5-E5.5), particularly in the PrE/VE lineage (Fig. 1D; Fig. S2A,B). Notably, culturing E5.5 embryos in the presence of exogenous BMP4 for 6 h was sufficient to expand the zone of pSMAD1/5/9 within the VE in a dose-dependent manner (Fig. S1C). Thus, the availability of ligand could limit the extent of pathway activation, during both pre- and post-implantation stages.

Finally, we evaluated pSMAD1/5/9 in Bmp4 null embryos at E5.5. We were unable to detect pSMAD1/5/9 in Bmp4 null embryos, although it was observed at wild-type levels and localization in homozygous wild-type littermate controls (Fig. 1G,H). In Bmp4 heterozygous embryos, we observed an intermediate phenotype, where pSMAD1/5/9 was detectable in an intermediate number of cells (Fig. 1H). This suggests that at E5.5, BMP4 plays a major role in initiating BMP signaling activity in the mouse, and that this function of BMP4 is dose-dependent.

Previous knockout studies of BMP signaling components did not report preimplantation phenotypes (Beppu et al., 2000; Mishina et al., 1995; Sirard et al., 1998; Winnier et al., 1995). However, other groups reported defects in preimplantation lineage specification using pathway inhibitors or microinjection of RNAi or mRNA for dominant-negative overexpression (Graham et al., 2014; Reyes de Mochel et al., 2015; Stuart et al., 2019). These observations raise the hypothesis that components of the BMP pathway are maternally imparted to the oocyte and functionally complement zygotically expressed components during early development. To test this hypothesis, we examined cell fate specification in embryos lacking both maternal (m) and zygotic (z) Bmp4 or Smad4 using the female germ line-expressed Zp3-Cre (De Vries et al., 2000) in combination with floxed alleles of either Bmp4 or Smad4 (see Fig. S3A for breeding scheme). RT-qPCR analysis confirmed the absence of detectable Smad4 transcript in Smad4 mz null oocytes (Fig. S3B), as we have observed for many other loci deleted in this manner (Blij et al., 2012; Frum et al., 2013, 2018; Wicklow et al., 2014).

Remarkably, we were able to recover Bmp4 mz null and Smad4 mz null blastocysts at wild-type rates (Fig. 2), indicating that neither Bmp4 nor Smad4 is required for either fertilization or preimplantation embryo development. Moreover, both Bmp4 and Smad4 mz null embryos exhibited normal morphology, total cell number, and ratio of trophectoderm and ICM cells (Fig. 2; Fig. S3C-E). Moreover, expression of the ICM marker OCT4 and polarized distribution of CDH1 in the TE strongly suggested normal lineage specification (Fig. S4). Similarly, the expression of EPI and PrE cell fate markers at E3.75, E4.25 and E4.5 was also unaffected in either Bmp4 or Smad4 mz null embryos at these stages (Fig. 2; Fig. S3C-E), consistent with normal ICM differentiation. Our observations support the conclusion that canonical BMP signaling does not play a major role in preimplantation development. In a parallel set of experiments, we allowed Smad4 m null embryos to develop to term. Mice lacking maternal Smad4 were born and developed apparently normally to at least 4 months old (15/15 mice, three litters). We conclude that maternal Bmp4 and Smad4 are dispensable for development and that neither zygotic gene plays any obvious role before implantation.

Previous studies have mainly focused on characterization of BMP signaling loss-of-function phenotypes at later stages (>E5.5) (Sirard et al., 1998; Winnier et al., 1995; Yang et al., 1998). However, we first observed BMP signaling activity in most embryos just after implantation at E4.75, prompting us to scrutinize embryos lacking Bmp4 or Smad4 beginning at E4.75 for phenotypes. At E4.75, Smad4 null embryos were grossly normal (Fig. 3A). However, upon close examination, Smad4 null embryos displayed a significant decrease in total cell number (Fig. 3B). We quantified the number of EPI, PrE and TE cells in these embryos and discovered that the decreased cell number was most pronounced in EPI cells at this stage (Fig. 3B,C).

By E5.5, Smad4 null embryos were visibly reduced in size and all displayed disorganization in EPI, VE and EXE compartments, as expected from studies performed at E6.5 (Sirard et al., 1998; Yang et al., 1998) (Fig. 3D). Strikingly, the EPI was greatly reduced in cell number (Fig. 3D-F) relative to controls, and had not yet cavitated, as the proamniotic cavity was not present among most (>80%, n=7/9) Smad4 null embryos examined (Fig. 3D,L,M). Notably, relative to extra-embryonic lineages, the size of the EPI was disproportionately decreased in Smad4 null embryos; the EPI length was decreased even when normalized to proximal-distal embryo length (Fig. 3G-I; Fig. S5A). By contrast, the size of the EXE was appropriately scaled to the reduced size of Smad4 null embryos at E5.5 (Fig. 3J,K). This suggests that Smad4 is not only required for general embryonic growth, but also specifically required for EPI growth or scaling relative to total embryo size. These phenotypes were not observed in E5.5 Bmp4 null embryos, which did not differ from wild-type in morphology, embryo size, or EPI or PrE cell number (Fig. S5B-E), suggesting that additional signaling pathways act upstream of SMAD4 to promote EPI growth and morphogenesis at this stage.

Having discovered that Smad4 is required for EPI cavitation and growth at E5.5, we began to investigate the mechanism. We were struck by the observation that the EXE appeared to be oversized, relative to the size of the EPI, in E5.5 Smad4 null embryos. To confirm the identity of the EXE cells, we examined markers of EXE, including phosphorylated ERK (pERK) (Corson et al., 2003). In Smad4 null embryos, we observed pERK throughout the EXE region (Fig. 4A), consistent with its identity as EXE. Notably, however, pERK levels were dramatically increased within the EXE relative to wild-type embryos (Fig. 4A). As pERK in the EXE is dependent on signaling by the fibroblast growth factor (FGF) pathway (Corson et al., 2003), we hypothesized that increased pERK was because of elevated FGF signaling in Smad4 null embryos.

To test this hypothesis, we used a previously-published protocol to inhibit FGF signaling in embryos (Yamanaka et al., 2010), which effectively eliminated pERK in control embryos (Fig. S6A). Additionally, inhibition of FGF signaling partially rescued EPI defects in E5.5 Smad4 null embryos (Fig. 4A). That is, we observed a significant increase in EPI size, as well as increased rates of cavitation in FGF-inhibited Smad4 null embryos, relative to untreated Smad4 null embryos (Fig. 4B-E). However, FGF inhibitor treatment did not fully rescue the growth restriction of Smad4 null embryos (Fig. 4F), nor was the number of OCT4-positive cells restored (Fig. 4G). These observations are consistent with SMAD4 limiting the level of FGF signaling within the EXE during early post-implantation stages to promote EPI morphogenesis.

To determine whether elevated FGF signaling is sufficient to antagonize cavitation, we treated embryos with exogenous FGF4. In wild-type E5.5 embryos treated with exogenous 1 µg/ml FGF4, we observed elevated levels of pERK within the EXE and ectoplacental cone (Fig. S6A). However, we observed no impact on EPI cavitation or EPI size following this treatment (Fig. S6A-D). Altogether, these data suggest that pERK/FGF signaling antagonizes EPI cavitation, but upregulation of pERK alone is insufficient to induce cavitation defects in wild-type embryos, at least under the conditions tested here.

Here, we have used a combination of immunofluorescence and genetic approaches to elucidate the roles of BMP4, pSMAD1/5/9 and SMAD4 in peri-implantation development. We have considered why our preimplantation pSMAD1/5/9 findings differ from previously published findings (Graham et al., 2014; Reyes de Mochel et al., 2015), and we have identified several differences in experimental design that could be relevant. The first difference is the source and specificity of antibodies. We used the same anti-pSMAD1/5/9 antibody that was used in post-implantation embryos in Senft et al. (2019), producing concordant post-implantation results. However, a different antibody was used to detect pSMAD1/5/9 in preimplantation embryos in Reyes de Model et al. (2015). The source of antibody that was used to detect pSMAD1/5/9 in blastocysts in Graham et al. (2014) was not described in detail, but a more recent study from the same lab reported absence of pSMAD1/5/9 in the blastocyst using the same antibody that we used (Weatherbee et al., 2024). Ours is the only study to confirm pSMAD1/5/9 specificity in Bmp4 null blastocysts, providing strong evidence for antibody specificity.

A second possible difference between our study and previous studies is embryo preparation: we characterized pSMAD1/5/9 localization in embryos immediately after retrieval from pregnant mice, whereas embryos were cultured before analysis in Graham et al. (2014). Moreover, embryos were cultured in the presence of bovine serum albumin (BSA) in Graham et al., which could influence the signaling environment of the embryo. A third possible difference is the loss-of-function approach: previous studies have primarily modulated BMP signaling using either pharmacological inhibitors or overexpression of dominant-negative receptors (Graham et al., 2014; Reyes de Mochel et al., 2015). It is therefore possible that the previously reported preimplantation phenotypes were due to expression of proteins at super-physiological levels and/or off-target effects of chemical inhibitors. Consistent with this latter proposal, the BMP inhibitors dorsomorphin, LDN-193189 and DMH2 have been shown to inhibit dozens of kinases that are not members of the BMP signaling pathway (Boergermann et al., 2010; Lowery et al., 2016; Vogt et al., 2011). This possibility also complicates interpretation of post-implantation embryo phenotypes resulting from treatment with BMP signaling inhibitors (Sozen et al., 2021).

Here, we also showed that pSMAD1/5/9 is first detectable within the PrE lineage. As Bmp4 is expressed in EPI cells at the blastocyst stage, and in EXE during post-implantation stages, and as our data show that SMAD1/5/9 phosphorylation in the PrE lineage is Bmp4-dependent, our observations indicate that BMP4 signals to the PrE lineage non cell-autonomously. Yet, the role for BMP4-SMAD1/5/9-SMAD4 in the PrE lineage is unclear. Intriguingly, BMP4 is sufficient to induce differentiation of PrE-derived extra-embryonic endoderm stem (XEN) cells to VE-like cells (Artus et al., 2012; Paca et al., 2012). Therefore, BMP4 signaling via pSMAD1/5/9 may facilitate PrE maturation in vivo.

Our observations additionally suggest that the spatial patterning of BMP signaling is limited in part by the availability of the ligand during early postimplantation. Consistent with this, the range of SMAD1/5/9 phosphorylation could be expanded within the VE by addition of exogenous BMP4. Curiously, however, we did not detect pSMAD1/5/9 within the EXE or proximal EPI of wild-type embryos, suggesting that additional mechanisms exist for limiting BMP signaling to the EXE-adjacent VE. We do not yet know whether this pattern is important for development.

Our study demonstrates a requirement for Smad4 in EPI maintenance and morphogenesis postimplantation (Fig. 5). We have considered whether these findings differ from previous reports. Previous descriptions of Smad4 null embryos report growth restriction and VE disorganization beginning at E5.5, with embryonic lethality by E8.5 (Chu et al., 2004; Sirard et al., 1998; Yang et al., 1998). Our results are consistent with these findings, but also uncover previously unappreciated defects in the EPI. Notably, cavitation defects were not reported for E6.5 embryos homozygous for null or deleted alleles of Smad4 (Sirard et al., 1998; Yang et al., 1998) (exon 8 deleted in both; Yang et al., allele used in the present study). However, E5.5 embryos were not examined in detail in these previous studies. Alternatively, genetic background, which is known to influence the severity of the Bmp4 null phenotype (Winnier et al., 1995) could impact the severity or the penetrance of the Smad4 EPI phenotypes. Both previous studies were performed in a C57BL/6 background, whereas ours were performed in a CD-1 background, where the cavitation phenotype was partially penetrant (80%).

Regardless of the explanation, our observations prompt us to wonder where Smad4 is required for early EPI morphogenesis (E4.75-E5.5). A previous study characterized the spatial requirements for Smad4 in embryogenesis by performing VE-specific Smad4 deletion, which caused gastrulation defects (Chu et al., 2004). These observations led to the conclusion that SMAD4 acts non cell-autonomously within the VE. However, the VE-specific Cre driver used to delete Smad4 in this study is not active until E5.75, which is well after we first observe EPI defects. As our study first observed an EPI defect at E4.75, an earlier knockout of Smad4 within the PrE lineage is needed to determine whether SMAD4 promotes EPI morphogenesis non cell-autonomously via the VE. On the other hand, our study also showed that EPI cavitation relies on SMAD4 attenuation of pERK levels in the EXE, strongly suggesting that SMAD4 promotes EPI cavitation non cell-autonomously via the EXE. Regrettably, distinguishing between these two intriguing possibilities is not experimentally straightforward at the current time.

We note that we have not ruled out a cell-autonomous requirement for Smad4 within the EPI. For example, SMAD4 could promote EPI maturation, defined here as the transition from a naïve to primed state, which normally occurs in EPI cells between preimplantation and post-implantation stages (Boroviak et al., 2015; Nichols and Smith, 2009). EPI maturation has been suggested to be crucial in the formation of the proamniotic cavity (Carbognin et al., 2023; Shahbazi et al., 2017). The possible cell-autonomous requirement for Smad4 is not mutually exclusive with its possible non cell-autonomous roles in the extra-embryonic lineages.

At this point, we understand little about the signals acting upstream and downstream of SMAD4 in EPI cavitation and scaling (E4.75-E5.5). In terms of upstream signals, a previous study has reported that BMP2/4 signaling is necessary and sufficient for cavitation of embryoid bodies (Coucouvanis and Martin, 1999). If BMP4 is functionally redundant with BMP2 or other BMP factors in this context, this could explain why cavitation appeared to be normal in E5.5 Bmp4 null embryos. However, we do not favor this model because phosphorylation of SMAD1/5/9 was dependent on Bmp4, arguing against functional redundancy with other BMP ligands. Deletion of genes encoding all three SMAD1/5/9 factors would be necessary to test the requirements for BMP signaling in early EPI morphogenesis.

Alternatively, another TGFβ pathway may signal through SMAD4 to regulate early EPI morphogenesis. Knockout of Nodal has been shown to decrease embryo size and expression of Oct4 mRNA at early post-implantation stages (Brennan et al., 2001; Mesnard et al., 2006), but Nodal null embryos cavitate normally and expression of OCT4 protein is normal at E5.5 (Senft et al., 2019). As the cavitation defect and loss of OCT4-positive cells are more severe in Smad4 null embryos than either Bmp4 null or Nodal null models alone, multiple TGFβ pathways may regulate EPI morphogenesis cooperatively (Fig. 5).

We also do not yet understand the mechanisms regulating early EPI morphogenesis downstream of SMAD4. We showed that SMAD4 promotes cavitation by attenuating pERK levels. Inhibition of FGF/MAPK signaling rescued cavitation in Smad4 null embryos, even though the EPI remained disproportionately small. This observation allows us to propose that cavitation is not dependent on EPI size, but rather the embryo signaling environment. Identification of SMAD4 targets in the EXE using an approach such as CUT&RUN could help reveal the mechanism by which SMAD4 regulates ERK levels in the EXE. Further study will also be needed to discover previously unreported signals from the EXE that act downstream of ERK to promote EPI morphogenesis.

ScRNA-seq data generated by Nowotschin et. al. was used to analyze the expression of TGFβ genes in mouse E3.5, E4.5, E5.5 and E6.5 blastocysts (Nowotschin et al., 2019). The analysis was completed using R v4.1.0 with tools from Seurat v4.3.0 (Hao et al., 2021). We normalized the UMI counts using SCTransform and cells were visualized in 2D space using UMAP performed on the first 30 principal components (Choudhary and Satija, 2022; Hafemeister and Satija, 2019). After excluding TGFβ genes expressed in <10 cells, we used Seurat's FindAllMarkers function with the Wilcoxon rank-sum test to identify TGFβ genes enriched in each cell type versus all other cells. The P-values were corrected for multiple comparisons using the Bonferroni method. Genes with P-adj<0.01 and average log₂ fold change<0.25 were considered cluster enriched. Heatmaps were generated using the pheatmap (v 1.0.12) after averaging the normalized expression for each gene in each cell type.

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 were used in this study and maintained in a CD-1 background (see Table S1): Bmp4^tm1Jfm/J (Liu et al., 2004); Smad4^tm2.1Cxd/J (Yang et al., 2002); Tg(Zp3-cre)93Knw (de Vries et al., 2000). Null alleles were generated by breeding dams carrying homozygous floxed alleles and the Zp3Cre allele to CD-1 males. Mouse genotypes were determined by PCR using genomic DNA extracted using the REDExtract-N-Amp kit (Sigma-Aldrich, XNAT) according to the manufacturer's protocol. Embryo genomic DNA was extracted using the same kit scaled to 10 µl total volume. Genomic extracts (1-2 µl) were then subjected to PCR using allele-specific primers (see Table S3).

Mice were maintained on a 12 h light/12 h dark cycle. Preimplantation (E2.5-E4.5) embryos were collected by flushing the oviduct or uterus with M2 medium (Sigma-Aldrich, M7167). Post-implantation (E4.75-E6.5) embryos were collected by dissecting the embryos from the decidua in ice-cold PBS containing 1% fetal bovine serum (FBS; HyClone, SH30396.02) or BSA (Sigma-Aldrich, A7888). During embryo collection, dissected embryos were held in warm M2 media. For embryo culture, KSOM medium (Millipore, MR-121-D) was equilibrated overnight before embryo collection. Where indicated, the following were included in the culture medium: 1 µM or 0.25 µM LDN-193189 in DMSO (Stemgent, 04-0074-02); 1 µg/ml recombinant FGF4 in PBS with 0.1% BSA (R&D Systems, 235-F4); 1 µg/ml heparin (Sigma-Aldrich, H3149); 100 ng/ml recombinant BMP4 in 4 mM HCl (R&D Systems, 314-BP); 1 µM PD173074 in DMSO (FGFRi, Selleckchem, S1264); 5 µM PD0325901 in DMSO (MEKi, Stemgent, 04-0006); or 0.2% DMSO (New England BioLabs, B0515A) as control. Embryos were cultured at 37°C in a 5% CO₂ incubator under light mineral oil (Millipore, ES-005-C).

Smad4 expression levels in oocytes were assessed by real-time PCR as previously described (Blij et al., 2012). Smad4 levels were assessed in oocytes from three wild-type and three Smad4 maternal null females. Oocytes collected from each female were pooled for mRNA extraction and cDNA synthesis. RT-PCR was performed in quadruplicate technical replicates for each cDNA sample. Primers were (5′-3′): Actb, CTGAACCCTAAGGCCAACC and CCAGAGGCATACAGGGACAG; Smad4 (wild-type allele), CGCGGTCTTTGTACAGAGTTA and ACACTGCCGCAGATCAAAG; Smad4 (deleted allele), CACAGGACAGAAGCGATTGA and CCAAACGTCACCTTCACCTT.

Preimplantation embryos (E2.5-E4.75) were fixed with 4% formaldehyde (Polysciences, 04018) for 10 min, permeabilized with 0.5% Triton X-100 (Sigma-Aldrich, X100) for 30 min, and then blocked with blocking solution (10% FBS, 0.1% Triton X-100) overnight at 4°C. Embryos were incubated with primary antibody overnight at 4°C. The next day, embryos were washed in blocking solution for 30 min, incubated in secondary antibody diluted in blocking solution for 1 h, washed in blocking solution for 30 min, then stained with nuclear stain (DRAQ5) diluted in blocking solution for 10 min or overnight.

Post-implantation embryos (E5.0-E5.75) were fixed with 4% formaldehyde for 1 h, washed three times in 0.1% Tween-20 (Sigma-Aldrich, P9416), permeabilized for 4 h in 0.5% Triton X-100, and then blocked with blocking solution [3% BSA (Sigma-Aldrich, A7888); 0.3% Triton X-100 in PBS] overnight at 4°C. Embryos were incubated with primary antibody overnight at 4°C. The next day, embryos were washed three times in 0.1% Tween-20 for 5 min, then incubated in secondary antibody diluted in blocking solution overnight. The following day embryos were washed three times in 0.1% Tween-20 for 5 min, then stained with nuclear stain diluted in blocking solution for 10 min or overnight.

All embryos (preimplantation or postimplantation) which used antibodies against pSMAD1/5/9 were fixed with 4% formaldehyde for 1 h, methanol dehydration-rehydration series (25%, 50%, 75%, 100%) for 5 min each, washed three times in freshly-made 1% Triton X-100 for 10 min, washed for 20 min in ice-cold acetone at −20°C, washed three times in freshly-made 1% Triton X-100 for 10 min, then then blocked with blocking solution (10% FBS, 0.1% Triton X-100 in PBS) overnight at 4°C. Embryos were incubated with primary antibody overnight at 4°C. The next day, embryos were washed three times in freshly prepared 0.1% Triton X-100 for 10 min, incubated in secondary antibody diluted in blocking solution for 2 h, washed three times in freshly-made 0.1% Triton X-100 for 10 min, then stained with nuclear stain diluted in blocking solution for 10 min or overnight.

All embryos (preimplantation or postimplantation) which used antibodies against pERK were fixed with 4% formaldehyde for 1 h, washed three times for 5 min in PBS, washed for 20 min in ice-cold methanol at −20°C, permeabilized for 30 min in 0.1% Tween-20, then blocked with blocking solution (3% BSA; 0.3% Triton X-100 in PBS) overnight at 4°C. Embryos were incubated with primary antibody overnight at 4°C. The next day, embryos were washed three times in PBS for 5 min, incubated in secondary antibody diluted in blocking solution for 2 h, washed three times in PBS for 5 min, then stained with nuclear stain diluted in blocking solution for 10 min or overnight. All solutions contained HALT protease inhibitor (Thermo Fisher Scientific, 78430) and PhosSTOP phosphatase inhibitor (Roche, 04906837001) diluted 1:500.

Antibodies used are listed in Table S2. Embryos were imaged using an Olympus FluoView FV1000 Confocal Laser Scanning Microscope system with 60× PlanApoN oil (NA 1.42) objective. For each embryo, z-stacks were collected, with 5 µm intervals between optical sections. All embryos were imaged before knowledge of their genotypes.

For each embryo, z-stacks were analyzed using Fiji (ImageJ), which enabled the labeling, based on DNA stain, of all individual cell nuclei. Using this label to identify individual cells, each cell in each embryo was then assigned to relevant phenotypic categories, without knowledge of embryo genotype. Phenotypic categories included marker expression (e.g. OCT4-positive or -negative) and marker localization (e.g. pSMAD1/5/9 nuclear, absent or unlocalized). Proamniotic cavity area was measured using Fiji. Statistical analysis was performed using GraphPad Prism (v. 9.5.1). Comparisons between two groups were performed using unpaired two-tailed Student's t-test, where P<0.05 was considered significant. Comparisons between three or more groups were performed using analysis of variance (ANOVA) followed by Tukey's post-hoc, where P<0.05 was considered significant. Figure images were assembled using Adobe Illustrator.

We thank the lab of Dr David Arnosti for their generous loan of laboratory equipment. We also thank Barbara Makela and Ella Markley for technical support. We thank Dr Melinda Frame and the Center for Advanced Microscopy at Michigan State University for their assistance with confocal microscopy, and thank MSU Campus Animal Resources for maintenance of our laboratory animals.