ETV4 and ETV5 orchestrate FGF-mediated lineage specification and epiblast maturation during early mouse development

FGF signalling is essential for mouse pre-implantation development where it acts in multiple successive steps. At the blastocyst stage, within the inner cell mass FGF4 drives the specification of primitive endoderm progenitors, an extra-embryonic endoderm lineage that later forms the yolk sac. Culture in exogenous FGF ligands directs all inner cell mass progenitors to differentiate into primitive endoderm (Nichols et al., 2009; Yamanaka et al., 2010). Conversely, inactivation of FGF signalling through genetic or biochemical means generates embryos with an inner cell mass containing all epiblast cells (Chazaud et al., 2006; Kang et al., 2017, 2013; Krawchuk et al., 2013; Molotkov et al., 2017; Nichols et al., 2009; Yamanaka et al., 2010). However, these epiblast cells have sustained elevated NANOG levels, indicating an inability to exit naïve pluripotency (Kang et al., 2017, 2013; Molotkov et al., 2017; Nichols et al., 2009). FGF signalling is well known to regulate naïve pluripotency exit and later germ-layer differentiation in vivo (Lanner and Rossant, 2010; Nichols et al., 2009), and in mouse embryonic stem cells (Kunath et al., 2007; Ying et al., 2008). However, the direct targets of FGF/ERK and the molecular mechanism underlying how FGF/ERK signalling executes cell fate decisions are poorly understood.

Etv4 and Etv5, are ETS transcription factors and downstream transcriptional activators of the FGF signalling pathway during embryonic development, with cooperative roles in the morphogenesis of the lung, limb bud and kidney (Herriges et al., 2015; Zhang et al., 2009). ETS transcription factor expression is induced by RTK/MAPK signalling, and post-translationally activated by MAPK phosphorylation (Charlot et al., 2010; Janknecht et al., 1996; O'Hagan et al., 1996). Etv4 homozygous mutant males are viable, but sterile (Laing et al., 2000), or have neuronal defects (Livet et al., 2002). Whereas, Etv5 homozygous mutants die soon after birth (Zhang et al., 2009) or have mid-gestation (E8.5) lethality (Lu et al., 2009), the causes of which have not been established.

Single-cell transcriptomics of mouse (Ohnishi et al., 2014) and human (Blakeley et al., 2015) blastocysts identified expression of Etv4 (epiblast and primitive endoderm) and Etv5 (inner cell mass progenitors and epiblast), which are downregulated in Fgfr1 mutant mouse blastocysts (Kang et al., 2017), suggesting these ETS variant (ETV) proteins may govern the transcriptional output downstream of FGF signalling in the inner cell mass. However, despite being implicated in regulation of the pluripotent state in vitro (Akagi et al., 2015; Kalkan et al., 2019; Zhang et al., 2018), the role of these transcription factors during epiblast development in vivo has not been assessed.

Here, we investigated the role of Etv4 and Etv5 (Etv4/5) as potential downstream effectors of FGF signalling in the establishment and maturation of the epiblast lineage. Analysing compound mutant mouse embryos, we find that loss of Etv5 compromises the formation of the primitive endoderm at pre-implantation stages. At peri-implantation, the loss of Etv4/5 causes a delay in the progression of pluripotency and epiblast maturation, leading to developmental delay and AVE migration defects at early post-implantation stages. Together, our work sheds light on the successive roles of FGF signalling effectors in mediating inner cell mass fate decisions, and later epiblast maturation required for establishing the embryonic body plan at gastrulation.

Given the role of PEA3 family members in the regulation of the pluripotent state in stem cells, we hypothesised that PEA3 family member ETS transcription factors are involved in the establishment of the epiblast lineage. We first characterised where ETV transcripts were expressed during early mouse embryo development using our scRNA-seq dataset from mouse pre- to early post-implantation embryo development (Nowotschin et al., 2019). Etv4 transcripts were expressed in late-blastocyst stage primitive endoderm (PrE) and epiblast (EPI), whereas Etv5 transcripts were expressed in early-blastocyst stage inner cell mass (ICM) progenitors, and later in epiblast cells (Fig. 1A).

Next, we assessed protein expression at pre-implantation stages. We were unable to detect the ETV4 protein at pre-implantation stages, likely due to the lack of a good commercially available antibody; however, ETV5 protein was readily detectable. We performed immunofluorescence staining for ETV5, NANOG, GATA6 and quantified nuclear fluorescence intensity to categorise cell lineages (Fig. S1A,B), as described previously (Saiz et al., 2016). In early blastocyst (32-64 cells) stage embryos, ETV5 protein was not expressed. At mid-blastocyst (64-128 cells) stage, low levels of ETV5 protein were detected in uncommitted ICM progenitor cells (NANOG+GATA6+) and higher levels in epiblast cells (NANOG+GATA6-) (Fig. 1B,C). The highest levels of ETV5 protein were detected in the epiblast cells (NANOG+GATA6- or NANOG-GATA6-) of late blastocyst stage embryos (Fig. 1B,C). ETV5 protein was not present in the primitive endoderm (NANOG-GATA6+) or outer trophectoderm cells.

Further to this, we assessed early post-implantation embryos for ETV5 protein expression. We performed immunofluorescence analysis for ETV5, GATA6 (visceral endoderm marker) and SOX2 (epiblast and extra-embryonic ectoderm marker). At the egg cylinder stage (E5.5), ETV5 protein is expressed at low levels within the nuclei of the epiblast and in the distal-most part of the extra-embryonic ectoderm adjacent to the epiblast, and is absent in the visceral endoderm (Fig. 1D). At mid-gastrulation stage (E6.5), ETV5 was expressed heterogeneously in the epiblast and in the nascent mesoderm (Fig. 1D).

Taken together, the transcript and protein profiling reveal that Etv5 is first expressed in ICM progenitors, and then specifically upregulated in epiblast cells, where expression is maintained in these pluripotent cells through peri- and post-implantation stages.

To determine whether ETVs might act as downstream effectors of FGF signalling in pre-implantation development, we next wanted to assess the role of ETVs in blastocyst formation. Therefore, we analysed cell lineages upon loss of one or both Etv4 (Livet et al., 2002) and Etv5 (Zhang et al., 2009) genes . We collected embryos from Etv4^+/−;Etv5^+/− inter-cross matings at mid- to late blastocyst stage (64- to >128-cell stage), and assessed the number of cells in each lineage by immunofluorescence staining for CDX2 (trophectoderm), NANOG (epiblast) and GATA6 (PrE, Fig. 2A). ICM lineages were assigned by quantification of NANOG/GATA6 levels (Fig. S2A), as described earlier.

From the Etv4^+/−;Etv5^+/− inter-cross progeny, wild-type embryos and embryos with at least one intact copy of Etv5, specified the correct ratio of epiblast cells (Fig. 2A-C), roughly 40% to 60% of PrE (Saiz et al., 2016). Therefore, loss of Etv4 alone does not alter the epiblast:PrE ratio. Complete loss of Etv5 (embryos with the genotypes Etv4^+/+;Etv5^−/−, Etv4^+/−;Etv5^−/− and Etv4^−/−;Etv5^−/−), by contrast, had a marked effect on the ratio of epiblast:PrE at both mid-blastocyst and late blastocyst stages (Fig. 2A-C). Etv5^−/− embryos, regardless of Etv4 genotype, had a reduction in the proportion of primitive endoderm (PrE) cells (57-63% epiblast: 37-43% PrE) by the late blastocyst stage (Fig. 2C). Etv5^−/− embryos exhibited a range in the fraction of primitive endoderm cells in the ICM, with some embryos having fewer percentage of primitive endoderm cells, while others exhibit a complete loss (Fig. 2A,B,  Fig. S2B). The Etv5^−/− embryos phenocopy Fgfr1^−/− embryos (Kang et al., 2017; Molotkov et al., 2017), suggesting a hierarchical relationship of Fgfr1 and Etv5 in relaying FGF4 signalling to specify primitive endoderm.

The reduction in the proportion of primitive endoderm cells in Etv5-null embryos occurred without changes in total cell counts within the ICM (Fig. S2C). The compensatory increase in epiblast versus primitive endoderm cell numbers in Etv5-null embryos suggested that uncommitted cells were preferentially specified to epiblast. The similarity in the number of double-positive cells in single Etv5^−/−, single Etv4^−/−, double Etv4^−/−;Etv5^−/− knockout and wild-type embryos indicated timely specification of uncommitted progenitors for all embryo genotypes (Fig. 2C). These data, combined with the comparable number of ICM cells between all genotypes (Fig. S2D), argue against a delay in specification or a selective loss of primitive endoderm cells.

Although epiblast specification did not appear to be negatively affected in any of the mutant embryos, we wanted to determine if the epiblast lineage was being properly maintained after specification. Declining NANOG levels are indicative of the exit from naïve pluripotency, as the epiblast undergoes maturation prior to implantation, and is dependent on FGF signalling (Kang et al., 2017; Molotkov et al., 2017; Nichols et al., 2009). NANOG levels were elevated in Etv4^−/− and Etv5^−/− single knockout embryos, when compared with wild-type embryos at the late-blastocyst stage (Fig. 2D). This unphysiologically elevated level of NANOG was further compounded in double knockout embryos (Fig. 2D), with NANOG levels significantly higher in Etv4^+/−;Etv5^−/− and Etv4^−/−;Etv5^−/− compared to wild-type embryos, suggesting that both ETV factors play complementary and synergistic roles in the maturation of the epiblast lineage.

All together, these data suggest Etv5, but not Etv4, is required for balancing the specification of uncommitted ICM cells towards primitive endoderm and away from epiblast cells, to achieve the robust and stereotyped tissue proportions of the blastocyst. However, both Etv4 and Etv5 function synergistically to modulate NANOG expression within the epiblast to allow exit from naïve pluripotency.

Given the phenotypic similarity of Etv5 mutants to mutants with reduced FGF signalling activity (e.g. Fgf4^+/− and Fgfr1^−/− embryos) (Kang et al., 2017, 2013), we reasoned that FGF signalling may be disrupted or dampened in Etv5 mutant embryos. Etv5 is expressed at intermediate levels in uncommitted ICM cells and upregulated in epiblast cells, mirroring the expression of Fgf4 (Fig. 1, Fig. S1B; Nowotschin et al., 2019). Hypothesising that Etv5 may regulate Fgf4 transcription, we analysed ChIP-seq data of ETV5 binding (Kalkan et al., 2019) in mouse embryonic stem cells (mESC) cultured in 2i (naïve pluripotent conditions, similar to pre-implantation epiblast) and 16 h after transfer to N2B27 (to induce transition out of the naïve pluripotent state, similar to peri-implantation epiblast). ETV5, a transcriptional activator, is enriched at the upstream promoter region of Fgf4 in naïve and transitioning mESCs (Fig. 3A), also suggesting direct regulation in epiblast cells in vivo.

To test this hypothesis, we analysed Fgf4 expression after loss of Etv5 in embryos. Quantitative real-time PCR (qRT-PCR) of E4.5 whole blastocysts demonstrated that Fgf4 transcripts are elevated in Etv5^−/− embryos compared with wild type (Fig. 3B, fold change=1.59, P=0.01). Therefore, in embryos, Etv5 is dispensable for Fgf4 expression, which is sustained in Etv5^−/− embryos with higher Fgf4 expression in Etv5^−/− embryos correlating with an increased number of epiblast cells. Thus, loss of primitive endoderm in the Etv5 mutants is not caused by the non-cell autonomous effect of limited FGF4 availability.

Given the dampened response to FGF signalling in the Etv5^−/− embryos, we attempted to rescue the reduction in primitive endoderm numbers with excess exogenous FGF ligand. We treated wild-type, Etv5^+/− and Etv5^−/− embryos with a saturating dose of FGF4 (1 μg/ml) and heparin (1μg/ml) from E2.5 (8-16 cell stage) for 48 h (Fig. 3C). In control culture, wild-type, Etv5^+/− and Etv5^−/− embryos comprised 52%, 49% and 20% GATA6+NANOG− primitive endoderm cells, respectively (Fig. 3D,E and Fig. S3). The significant impairment of primitive endoderm specification in Etv5^−/− control cultured embryos is consistent with freshly recovered Etv5^−/− embryos (Fig. 2). After FGF treatment, wild-type and Etv5^+/− embryos had only GATA6+NANOG− primitive endoderm cells throughout their ICM, as did over half of the Etv5^−/− embryos (4/7 embryos) (Fig. 3D,E and Fig. S3). However, some FGF-treated Etv5^−/− embryos retained GATA6−NANOG+ epiblast cells (3/7 embryos), restoring epiblast ratios to 41%, comparable to untreated wild-type embryos (40% epiblast) (Fig. S3B). These data demonstrate that high, non-physiological levels of exogenous FGF4 can partially rescue primitive endoderm specification in Etv5^−/− mutants, but cannot robustly induce a complete switch in cell fate. Therefore, Etv5 plays a crucial cell-autonomous role in tuning the sensitivity of ICM cells to the FGF4 signal during primitive endoderm specification.

We hypothesised that ETV factors may play a role in epiblast maturation, as suggested by the elevated NANOG levels observed in mutant embryos (Fig. 2D). To probe the role of Etv4 and Etv5 in pluripotency exit further, we assessed additional markers of pluripotency at the late blastocyst stage. Embryos from Etv4^+/−;Etv5^+/− inter-crosses were stained for a core pluripotency marker (SOX2), naïve marker (KLF4) and formative/primed marker (OTX2) (Fig. 4A). While SOX2 and KLF4 in the epiblast were expressed at similar levels across all genotypes (Fig. S4A,B), epiblast OTX2 expression was significantly reduced in embryos lacking Etv5 (Fig. 4B,C). Analysis of published ETV5 chromatin binding in mESC (Kalkan et al., 2019) shows strong binding at a downstream Otx2 enhancer region (Fig. 4D). Together, these data indicate that during pre-implantation development, Etv5 is required for timely exit of pluripotency, likely through direct regulation of Otx2 and indirect regulation of Nanog by an as yet unknown mechanism.

We then assessed pluripotency at early post-implantation stages at E5.5. In wild-type, Etv4 and Etv5 null mutants embryos, the epiblast cavitated and retained expression of the core pluripotency factor SOX2 (Fig. S4C). Interestingly, OTX2 was expressed in both the visceral endoderm and the epiblast at this stage, indicating that ETV factors are not necessary for upregulation of primed pluripotency factors at early post-implantation stage.

We then tested the requirements of Etv4 and Etv5 for pluripotency in ESCs. We differentiated double knockout Etv4^−/−;Etv5^−/− ESCs (Lu et al., 2009) to epiblast-like cells (EpiLC) that are representative of the early post-implantation epiblast, which required exogenous FGF2 and activin. After 2 days of EpiLC differentiation, both wild-type and double knockout cells maintained core pluripotency transcription factors (OCT4 and SOX2), downregulated naïve markers (NANOG and KLF4) and upregulated primed markers (Fig. S4D,E). This indicated that loss of Etv4/5 is not sufficient to completely block naïve pluripotency exit, in agreement with in vivo early post-implantation embryo data (Fig. S4C). However, given the failure in downregulation of NANOG and upregulation of OTX2 at pre-implantation stages, ETVs likely control the timely exit of naïve pluripotency in the embryo; this delay has been similarly shown in ESCs (Kalkan et al., 2019).

We next looked at later embryonic time-points to determine if the delayed naïve pluripotency exit impacted later stages of development. At E6.5, at gastrulation, wild-type and Etv4 mutant embryos expressed high levels of OTX2 in the anterior visceral endoderm (AVE), and NANOG expression was confined to the proximal posterior region of the epiblast, marking the primitive streak (Fig. 5A). In Etv5-null and compound mutants, the distal visceral endoderm (DVE) marked by high OTX2 failed to migrate anteriorly (Fig. 5A, left panel arrowheads). In single Etv5 mutants, this resulted in a failure to properly position the anterior-posterior axis, as evidenced by the ∼45° rotation of the NANOG expression domain, marking the primitive streak. Double knockout embryos at E6.5 were overall smaller compared to wild-type embryos at the same stage (Fig. 5), and were instead comparable in size to E5.5 wild-type embryos (Fig. S4C).

The extra-embryonic ectoderm region was noticeably reduced in size, with a disordered morphology of the adjacent visceral endoderm epithelium (Fig. 5A, right panel arrowhead). NANOG expression was elevated throughout the epiblast, indicating that either the entire epiblast had failed to exit naïve pluripotency, or the presence of an expanded primitive streak region.

To determine if mutant embryos were specifying the AVE correctly and able to initiate gastrulation, we stained for an AVE marker, CER1, at E6.25 (pre-streak stage) and for a mesoderm marker, T (Brachyury) at E6.5 (mid-streak stage). In wild-type and Etv4-null embryos at E6.25 and E6.5, CER1 was localised to the anterior region of the embryo, extending from the embryonic-extraembryonic junction to the distal tip (Fig. 5B, Fig. S5A). At E6.5, T was localised to the posterior epiblast marking the nascent mesoderm (Fig. 5B). However, in single Etv5 and double Etv4;Etv5 homozygous mutant embryos at E6.25 and E6.5, while CER1 expression indicated that the AVE had been specified, the migration of the AVE appeared to be disrupted, delayed or arrested at the distal tip of the embryo (Fig. 5B, Fig. S5A, arrowheads). The occasional extension of CER1 expression beyond the embryonic-extraembryonic junction shows an over-migration of the AVE. The CER1 domain size was noticeably reduced relative to the size of the embryos. Again, abnormal thickening and disordered DVE/AVE epithelial morphology were observed in these mutants (Fig. 5B, arrowhead). Finally, at E6.5, mutants were devoid of T+ cells, indicating a failure in gastrulation and primitive streak formation, and/or specification of mesoderm cell types.

The AVE migration phenotype and the reduced embryo size were more severe and pronounced in compound mutant embryos Etv4^+/−;Etv5^−/− and Etv4^−/−;Etv5^−/− when compared with single homozygous Etv4^+/+;Etv5^−/− embryos (Fig. 5A,B, Fig. S5A). Thus, although Etv4 single mutants did not have an overt phenotype, partial or complete loss of Etv4 did increase the severity of the Etv5 phenotype, suggesting that these factors may partially compensate, or have overlapping roles, in the early post-implantation epiblast.

By late-gastrulation (E7.5), Etv5 and double homozygous null embryos were severely developmentally retarded with a failure in gastrulation, as indicated by the absence of the primitive streak marker T (Fig. S5B). Epiblast morphology was grossly abnormal, with ruffles and folds in the epithelium. As a consequence, Etv5 homozygous mutant pups were not recovered from Etv4;Etv5 heterozygous intercrosses, and were drastically under-represented in Etv5 heterozygous intercrosses (Fig. S5C,D). Etv4^−/−;Etv5^+/− pups were also not recovered, further suggesting a combinatorial role for ETV factors during embryonic development. All together, these data demonstrate that embryos of Etv4;Etv5 allelic series have increasingly severe phenotypes with loss of ETV gene dosage at early post-implantation stages exhibiting developmental delay and AVE migration defects that compromise embryonic development.

Here, we explored how transcription factors Etv4 and Etv5 function downstream of FGF signalling to regulate early mouse development. We discovered that Etv5 is essential for primitive endoderm formation before implantation, while the combined loss of Etv4/5 impairs epiblast maturation and proper anterior-posterior patterning after implantation. These findings reveal how FGF signalling effectors orchestrate both early cell fate decisions and subsequent epiblast development necessary for proper gastrulation.

The reduction in the number of primitive endoderm cells in Etv5^−/− blastocyst bears a striking similarity to that of Fgfr1^−/− and Fgf4^+/− phenotypes (Brewer et al., 2015; Kang et al., 2017, 2013). Given that Fgf4 transcripts are not diminished in Etv5 mutants, the phenocopying of low dosage FGF mutants cannot be due to a decreased ligand availability. Exposure to high levels of exogenous FGF can partially restore primitive endoderm cell numbers in Etv5-deficient embryos, suggesting that related ETS family members such as Etv1 and Etv4 may compensate for loss of Etv5 function under these non-physiological conditions, although a complete rescue is not achieved. Therefore, our findings support the hypothesis that Etv5 primarily relays FGF signalling activity in uncommitted ICM cells to specify primitive endoderm. Here, Etv5 acts as a FGF signalling effector, similar to other developmental contexts (Herriges et al., 2015; Zhang et al., 2009). The direct involvement of Etv5 in initiating the primitive endoderm program is further supported by its ability to upregulate multiple endoderm genes (including Sox7 and Sox17) when overexpressed in mouse ESCs (Correa-Cerro et al., 2011).

NANOG, a hallmark marker of mouse naïve pluripotency, is downregulated in the epiblast prior to implantation in an FGF/ERK-dependent manner (Nichols et al., 2009). In both Etv4^−/− and Etv5^−/− blastocysts, elevated NANOG levels reveal a failure to exit naïve pluripotency. Etv4 and Etv5 have been implicated in regulating pluripotency in vitro. For example, Etv5 was shown to promote MET during iPSC reprogramming (Zhang et al., 2018). In addition, these ETV factors appear to regulate ESC proliferation, but there is conflicting evidence as to whether they promote or repress epiblast-like fate during differentiation (Akagi et al., 2015; Zhang et al., 2018). Triple knock-out of Etv5, Rbpj and Tcf3 in ESCs can maintain naïve pluripotent state in absence of 2i (Kalkan et al., 2019), implicating these factors in dissolution of the naïve pluripotency transcription factor network. Given that ETV5 does not directly bind to the Nanog locus in ESCs (Kalkan et al., 2019), it seems likely that this regulation by ETV factors in vivo is indirect. Instead, ETVs may activate the primed pluripotency network, which then, in turn, repress the naïve state. Consistent with such a model, the primed pluripotency marker OTX2 fails to be upregulated in Etv5 mutant blastocysts. ETV5 has been shown to directly bind an Otx2 enhancer in mESCs (Kalkan et al., 2019), suggesting it is also a direct target in vivo. As Otx2 and Nanog are antagonistic (Acampora et al., 2017, 2016), the role of ETV in epiblast cells may be to turn on Otx2 and consequently reduce Nanog levels, thereby promoting the exit from naïve pluripotency.

OTX2, a marker of primed pluripotency, is normally expressed in the epiblast from the late blastocyst stage (E4.5) through early post-implantation. In Etv5^−/− embryos, however, OTX2 is absent at E4.5, but is eventually expressed by E5.5, indicating a delayed exit from naïve pluripotency. This delay is consistent with our observation of elevated NANOG levels in Etv5 mutant blastocysts. Furthermore, this is supported by time-course analysis of Etv5^−/− ESC differentiation showing delayed naïve pluripotency exit (Kalkan et al., 2019). Together, these findings highlight a role for the ETVs in regulating the robustness and timely exit of naïve pluripotency.

Timely Nodal-dependent maturation of the epiblast at early post-implantation stages is required for the coordinated and spatial patterning of the embryo and initiation of gastrulation (Huang et al., 2017; Zang et al., 2022). Given that Etv4 and Etv5 are not expressed in the visceral endoderm, we hypothesize that the delayed epiblast maturation observed in Etv5 and Etv4;Etv5 compound mutants leads to the later AVE defects. In the Etv5 and Etv4;Etv5 compound mutants, there is a reduction in the number of CER1-expressing cells, disordered discontinuous AVE and ectopic protrusions. These phenotypes are reminiscent of mutants affecting AVE development, including mutants in Nodal signalling pathway components (Stower and Srinivas, 2014). The most severely affected Etv4^−/−;Etv5^−/− embryos either fail to migrate the AVE, or the AVE over-migrates beyond the embryonic-extraembryonic boundary, and exhibit a disordered multi-layered epithelium, reminiscent of the two classes of Lefty1 mutants (a Nodal antagonist) (Trichas et al., 2011). Our research suggests that FGF, in conjunction with Nodal, plays a role in epiblast development and its interaction with the visceral endoderm, which is crucial for establishing the anterior-posterior axis of the embryo. All together, our results demonstrate the successive roles of ETS factors Etv4 and Etv5 as FGF signalling effectors in early lineage specification and embryonic body plan establishment, increasing our understanding of the molecular mechanisms of mammalian development.

Pre-implantation embryo immunofluorescence was carried out as previously described (Saiz et al., 2016). Briefly, embryos were fixed for 10 min at room temperature in 4% PFA. Fixed blastocysts were washed in PBX [0.1% Triton X-100 (Sigma-Aldrich) in PBS], permeabilised for 5 min in a solution of 0.5% Triton X-100 and 100 mM glycine in PBS, and then washed in PBX for 5 min. Embryos were blocked in blocking buffer (2% horse serum in PBS) for 40 min at room temperature, followed by incubation overnight at 4°C with primary antibodies diluted in blocking buffer (Table S1). The next day, embryos were washed three times in PBX, incubated in blocking buffer for 40 min at room temperature before a 1 h incubation with secondary antibodies at 4°C. Embryos were then washed in PBX and incubated in 5 μg/ml Hoechst in PBS for at least 30 min prior to imaging.

For post-implantation stages, embryos were fixed for 30 min at room temperature in 4% PFA. Then, fixed embryos were washed in PBX, permeabilised in 0.5% Triton-X in PBS for 30 min and then washed three times in PBX. Embryos were then incubated in blocking buffer, 5% donkey serum and 0.2% BSA in PBX for 2 h at room temperature, followed by incubation overnight at 4°C with primary antibodies diluted in blocking buffer. The next day, embryos were washed in PBX, followed by a second blocking step for at least 2 h at room temperature, and incubation with secondary antibodies in blocking buffer overnight at 4°C. After antibody staining, embryos were washed in PBX, and incubated with 5 μg/ml Hoechst in PBX for a minimum of 2 h to visualise DNA prior to imaging.

Embryos were imaged on a Zeiss LSM880 laser scanning confocal microscope, in glass bottomed dishes (MakTek) in PBS. Pre-implantation embryos were imaged using a Plan-Neofluar 40×/1.30 oil immersion objective. Post-implantation embryos were imaged at 20× using a Pan-Apo 20×/0.8 air objective. Nuclear segmentation for quantification of fluorescence intensity was carried out using MINS, as described previously (Lou et al., 2014). Correction for fluorescence decay along the z-axis was performed by linear regression and empirical Bayes method, as detailed by Saiz et al. (2016).

ESC lines used in the study were wild-type and Etv4^−/−;Etv5^−/− cells (Lu et al., 2009). ESCs were maintained on 0.1% gelatin (Millipore, 104070) coated tissue-culture grade plates in a humidified 37°C incubator with 5% CO₂. ESCs were grown in DMEM (Life Technologies, 11995073), supplemented with 2 mM L-glutamine (Life Technologies, 25030164), 0.1 mM MEM NEAA (Life Technologies, 11140-050), 1 mM sodium pyruvate (Life Technologies, 11360070), 100 U/ml penicillin and 100 mg/ml streptomycin (Life Technologies, 15140163), 0.1 mM 2-mercaptoethanol (Life Technologies, 21985023), 15% FBS (VWR, 97068-085) and 1000 U/ml LIF (prepared in house). ESCs were differentiated to EpiLCs as previously described (Hayashi et al., 2011). Briefly, ESCs were seeded 2.5×10⁴ cells/cm² in EpiLC medium onto fibronectin-coated (16 μg/ml, Millipore, FC010) 8-well IBIDI plates. EpiLC medium comprised N2B27 medium containing 20 ng/ml activin A (Peprotech, 120-14P), 12 ng/ml FGF2 (R&D Systems, 233-FB-025) and 1% KSR (ThermoFisher Scientific, 10828028).

Animal work was approved by the MSKCC Institutional Animal Care and Use Committee (IACUC). Etv4^tm1Arbr (Livet et al., 2002) and Etv5^tm1.1Xsun (Zhang et al., 2009) knockout alleles (abbreviated to Etv4⁻ or Etv5⁻) were maintained as double and single heterozygous mouse lines on a CD1 background. Mice, whole blastocysts, whole embryos or trophectoderm cells were genotyped by PCR. Three primer PCRs were used to genotype Etv5 samples, as follows: F, 5′-CTCGCAGAGGACAAGGTAGTGAC-3′; R_WT, 5′-GTGTGCACGACATGTTCAAGG-3′; and R_KO, 5′- CCAGCATCGTACAAAACAAGAG-3′. These generated a wild-type band at 270 bp and a knockout band at 374 bp. For genotyping Etv4 samples, the following primers were used: Etv4_WT_F (5′-TCTGGACCCTCTCCAGGTGATG-3′) and Etv4_WT_R (5′-CCACCAGAAACTGCCACAGTTG-3′), generating a wild-type band of 501 bp; and LacZ_F (5′-CATCCACGCGCGCGTACATC-3′) and LacZ_R (5′-CCGAACCATCCGCTGTGGTAC-3′), generating a knockout band, amplifying the LacZ cassette of 360 bp.

Embryos for this study were obtained from natural matings between Etv5 male and female heterozygotes. The sex of embryos was not determined. E2.5 morulae were flushed from oviducts with flushing holding medium (FHM, Millipore) as described previously (Behringer et al., 2014). Embryos within litters were randomly assigned in even-sized groups for control and exogenous FGF treatment: control, KSOM (MR-121-D, Sigma); FGF stimulation, 1 μg/ml FGF4 (R&D Systems) and 1 μg/ml heparin (Sigma) in KSOM. Medium was equilibrated 30 min prior to culture to reach the correct temperature and pH. Embryos were cultured in groups in droplets of medium in 35 mm dishes (∼1 μl/embryo) overlaid with mineral oil (Sigma) for 48 h in total in a humidified incubator at 37°C with 5% CO₂. After 24 h (E3.5), the zona pellucidae were removed by brief incubation in acid Tyrode's solution (Sigma), washed three times in their respective culture medium and then cultured in fresh droplets of medium. Embryos were assayed for lineage markers by immunofluorescence at the end of a 48 h culture.

E4.5 stage embryos from Etv5 intercrosses were prepared for genotyping and qPCR as previously described (Kang et al., 2017; Morgani et al., 2018), using CellsDirect One-Step kit in accordance with the manufacturer's instructions. Blastocyst were first washed in PBS, then incubated in 0.5% trypsin for 3 min at 37°C. Using a glass capillary, a small number of mural TE cells were removed for PCR genotyping, then each blastocyst was added to 5 µl of 2×Reaction Mix (Invitrogen, CellsDirect One-Step qRT-PCR Kit), snap-frozen on dry ice and stored at −80°C until processing.

For cDNA and target-specific pre-amplification, 5 μl of a reverse transcription/pre-amp mix was added to each blastocyst lysate. For each sample this comprised 0.2 μl SuperScript II RT/Platinum Taq mix (Invitrogen), 2.5 μl TaqMan assay (pooled assay mix with a concentration of 0.2× for each probe, detailed in Table S2), 2.3 μl RNase-free H₂O. To perform combined cell lysis, cDNA synthesis and pre-amplification of specific targets, samples were incubated at 50°C for 20 min, 95°C for 2 min, followed by 18 cycles of 95°C for 15 s then 60°C for 4 min, in a T100 Thermal Cycler (Bio-Rad).

Blastocyst cDNA was diluted 1 in 5 by adding 40 μl H₂O to the 10 μl cDNA to a total of 50 μl. To assay the amount of mRNA, each qPCR reaction was set up in duplicate in 96-well plates (Applied Biosystems, 4306737) overlaid with MicroAmp clear adhesive film (Applied Biosystems, 4306311). For each qPCR reaction, the mix was as follows: 7.5 μl TaqMan Universal PCR Master Mix, 0.75 μl TaqMan Assay and 5.25 μl RNase free H₂O. Sample cDNA (1.5 μl diluted) was then added to each reaction, to a total of 15 μl. Real-time PCR was then carried out in a QuantStudio 7 Flex System (Applied Biosystems). Gene expression was calculated as 2^ΔΔCt. Target gene (FGF) expression was calibrated to the arithmetic mean of the wild-type samples, and normalised to the expression of two reference genes within each sample, Gapdh and Actb, using the geometric mean (Vandesompele et al., 2002).

We thank the members of the Hadjantonakis and Niakan labs for helpful discussions and comments on the manuscript. We thank the MSKCC Mouse Genetics Core Facility for the import and maintenance of mouse strains, the Francis Crick Institute's Genomics Equipment Park for the use of RT-qPCR equipment, and Frank Costantini for providing wild-type and Etv4;Etv5 knockout mouse ESC lines.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.204278.reviewer-comments.pdf