Multi-level Fgf4- and apoptosis-dependent regulatory mechanism ensures the plasticity of ESC-chimaeric mouse embryo Free

Mammalian (including mouse and human) embryonic development comprises a sequence of cell fate decisions, whereupon cells gradually decrease developmental potential and increase specialisation. As a result, a blastocyst equipped with primary cell lineages, i.e. epiblast (EPI), primitive endoderm (PrE) and trophectoderm (TE), precursors of the embryo body and extraembryonic structures, respectively, is generated. However, in contrast to most non-mammalian species, where development follows a fixed set of instructions, the preimplantation development of a mammalian embryo is highly regulative. It means that during cleavage, the embryo shows outstanding flexibility to adapt in response to the experimental disruption of development and changing circumstances, i.e. environmental cues. Numerous embryological studies have shown that mouse (and human) embryos can withstand even very drastic manipulations, such as loss, addition or rearrangement of cells (Coticchio et al., 2021; Grabarek et al., 2012; Klimczewska et al., 2018; Suwińska et al., 2008; Tarkowski, 1959, 1961; Zhu and Zernicka-Goetz, 2020). Depending on the circumstances, embryonic cells use diverse strategies (including switching their fate) to adapt, allowing them to successfully continue the process of embryogenesis. It is worth noting that the self-organisation ability of the embryo has found clinical applications. It enables human embryos to adjust to alterations caused by extrinsic manipulation, such as cryopreservation or embryo biopsy for preimplantation genetic diagnosis. The existence of chimaeras generated by the combination of cells derived from two or more genetically different embryos is another spectacular manifestation of this developmental plasticity. Mouse chimaeras can be produced also by combining a host embryo with genetically dissimilar, or modified, embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs). Putative pluripotent cells integrate within the embryo and resume normal development, which serves as the most stringent test for their pluripotency. Moreover, the generation of experimentally produced mammalian chimaeras from genetically modified ESCs and iPSCs has been extensively used to create transgenic animals (Thomas and Capecchi, 1987). Despite many lines of evidence for the regulative nature of mammalian development, the mechanisms underlying this phenomenon are not fully understood.

The cleaving mouse embryo appears to possess a regulatory mechanism that controls the proportion of symmetric and asymmetric blastomere division during two cleavage rounds (8→16 and 16→32). The lower number of inner blastomeres created after the cleavage of the eight-cell embryo is compensated by the increased frequency of asymmetric divisions of outer blastomeres at the 16-cell stage, and vice versa. The existence of such a mechanism, based on the pattern of cleavage, has been proposed by our group in aggregates composed of inner and outer blastomeres aggregated in different proportions (Krupa et al., 2014). Interestingly, the introduction of supernumerary inner-like cells (i.e. ESCs) into the eight-cell embryo triggers the regulative mechanisms that rely on precise cell–cell communication between ESCs and host cells. As the introduced ESCs localise in the inner cell mass (ICM) while displacing the host's own ICM cells from the epiblast, the blastomeres of the host embryo must colonise other available niches – extraembryonic TE or PrE. Secreted factors may facilitate intercellular communication between both components and provide a spatial and temporal control mechanism during development. The candidate potentially involved in these interactions is fibroblast growth factor 4 (FGF4), which activates mitogen-activated protein kinase (MAPK) signalling by binding to the FGF receptors and stimulating diverse biological responses. Modulating the FGF4/MAPK cascade greatly influences the balance between the PrE and EPI lineages in the ICM. Treatment with a high level of exogenous FGF4 ligand leads to the upregulation of GATA6 and repression of NANOG, converting all ICM cells to adopt a PrE identity (Yamanaka et al., 2010). Conversely, chemical inhibition of FGF signalling with specific small molecules results in the generation of embryos that lack PrE cells and contain the ICM comprising exclusively Nanog-expressing cells (Bessonnard et al., 2017; Hamilton and Brickman, 2014; Nichols et al., 2009; Saiz et al., 2016; Yamanaka et al., 2010). Bearing in mind that ESCs are the source of FGF4 (Wilder et al., 1997) and that the cell specification within the embryo depends on the activation of this pathway (Kang et al., 2013, 2017; Krawchuk et al., 2013; Molotkov et al., 2017; Morris et al., 2012; Saiz et al., 2016, 2020; Soszyńska et al., 2019), we hypothesised that interactions between ESCs and host blastomeres, which are involved in the regulation of plasticity of chimaeric mouse embryos, rely on the FGF4/MAPK pathway. Moreover, having known that ESCs express pro-apoptotic proteins, such as BAX (Xiang et al., 2018), NOXA (also known as phorbol-12-myristate-13-acetate-induced protein 1; PMAIP1) and PUMA (BBC3; LeBlanc et al., 2018), and that apoptosis has been reported in the ICM during the sorting of the EPI and PrE (Morris et al., 2010; Plusa et al., 2008), we decided to check whether selective cell death serves as an additional mechanism regulating the cell lineage composition during chimaera formation.

We revealed that ESCs introduced into the cleaving embryo modify the cell division pattern and the length of the cell cycle of the embryo blastomeres, thus gaining a competitive advantage over host cells within the emerging EPI. Moreover, using Fgf4^−/− ESCs as donor cells we proved that the establishment of the correct ratio of PrE:EPI cells in the chimaeric mouse embryos is due to paracrine interactions between ESCs and the blastomeres of the host embryo, mediated by the FGF4/MAPK pathway. We also showed that apoptosis regulates the size of the ESC component rather than the proportion of blastomeres in the embryo, which serves as a control mechanism ensuring the generation of EPI of correct size and quality.

First, we wanted to validate our experimental model and see whether ESCs affect the fate of the accompanying blastomeres during chimaera formation. To verify our previous observations that ESCs introduced into an eight-cell embryo direct the emerging ICM cells to the PrE lineage (Humięcka et al., 2016), we microinjected eight to ten DsRed ESCs into GFP-expressing eight-cell embryos, cultured them for 48 h until the blastocyst stage and compared them with the corresponding uninjected embryos (Fig. 1A). We revealed that both chimaeric (n=28) and control, non-chimaeric (n=35) embryos displayed no significant difference in the TE:ICM ratio (P>0.01; Fig. S1). This proportion amounted to approximately 3:1, regardless of the presence of supernumerary ESCs within the embryo (Fig. S1). Moreover, we confirmed our previous observations that the mean percentage of PrE and EPI cells in the blastocysts did not differ between chimaeric and control uninjected groups (11.67%±2.06% versus 11.19%±3.06% for PrE and 11.57%±2.56% versus 11.43%±2.77% for EPI, respectively; mean±s.d.; P>0.05; Dunn's post-hoc test; Fig. 1B,C) and the average 1:1 ratio of PrE to EPI was preserved irrespective of the absence or presence of ESCs (49.37%±8.81% to 50.63%±8.81% versus 50.42%±7.08% to 49.58%±7.08%, P>0.05; Dunn's post-hoc test; Fig. 1B,D). However, the presence of exogenous ESCs decreased the proportion of the host embryo cells contributing to EPI (1.83%±2.85 out of all 49.58% EPI cells) within the ICM compared with control, non-chimaeric embryos (50.63%±8.81%; P<0.001; Dunn's post-hoc test, Fig. 1D).

We observed that the presence of ESCs within the embryo biased the embryo-derived inner blastomeres to form PrE rather than EPI, thus leading to sustaining the proportions of PrE and EPI cells similar to those in control uninjected embryos. As ESCs express and secrete FGF4 (Schoorlemmer and Kruijer, 1991; Wilder et al., 1997), and cell lineage specification within the ICM of the embryo depends on the activation of this pathway (Chazaud et al., 2006; Frankenberg et al., 2011; Kang et al., 2013, 2017; Krawchuk et al., 2013; Messerschmidt and Kemler, 2010; Molotkov et al., 2017; Plusa et al., 2008; Yamanaka et al., 2010), we decided to examine whether the final ICM composition of the chimaeric blastocyst is a consequence of paracrine interactions between ESCs and blastomeres of the host embryo, mediated by FGF4 signalling. To that end, we microinjected Fgf4^−/− knockout ESCs to GFP-expressing eight-cell stage embryos, cultured them for 48 h in a standard KSOM medium and compared them with the corresponding blastocysts containing DsRed (Fgf4^+/+) ESCs (Fig. 1A). We revealed that, of the total of 41 Fgf4^−/− ESC-injected embryos, 30 (73.17%) contained chimaeric EPI, composed of both the progeny of ESCs and blastomeres of the host embryo (Fig. 1B,E). EPI of the remaining 11 blastocysts (26.83%) was built exclusively from ESC daughter cells (Fig. 1B,E). Thus, the presence of Fgf4-deficient ESCs resulted in a decrease in the percentage of blastocysts with ‘pure’ (i.e. ESC-derived) EPI compared with chimaeric control embryos after injection with DsRed (Fgf4^+/+) ESCs [9 and 19 blastocysts with chimaeric (32.14%) and ESC-derived EPI (67.86%), respectively; P<0.01; χ² test; Fig. 1B,E].

We observed that experimental blastocysts containing Fgf4^−/− knockout ESCs, control blastocysts containing DsRed (Fgf4^+/+) ESCs and non-chimaeric control blastocysts were comparable in terms of the average percentage of TE (76.41%±4.41%, 76.76%±3.42% and 77.38%±4.18%, respectively; P>0.05; Kruskal–Wallis test) and ICM cells (23.59%±4.41%, 23.24%±3.42% and 22.62%±4.18%, respectively; P>0.05; Kruskal–Wallis test; Fig. S1). In contrast, the proportion of PrE and EPI cells building the experimental blastocysts that had been microinjected with Fgf4^−/− ESCs, was significantly disturbed. These embryos were characterised by a lower percentage of PrE cells (7.39%±2.15%) compared with control, non-chimaeric blastocysts (11.19%±3.06%; P<0.001; Dunn's post-hoc test; Fig. 1C) and chimaeric blastocysts with DsRed (Fgf4^+/+) ESCs (11.67%±2.06; P<0.001; Dunn's post-hoc test; Fig. 1C). Consequently, they showed a higher proportion of EPI cells (16.20%±3.49%) when compared with both control groups (11.43%±2.77% and 11.57%±2.56%; P<0.001; Dunn's post-hoc test; Fig. 1C).

Analysing the ICM composition of the blastocysts obtained, we observed that a lower level of FGF4 in chimaeric blastocysts (due to the presence of Fgf4^−/− ESCs) resulted in a decrease in the contribution to PrE (31.50%±7.68%) in relation to chimaeras with DsRed (Fgf4^+/+) ESCs (50.42%±7.08; P<0.01; Dunn's post-hoc test; Fig. 1D) in favour of the EPI (10.55%±8.01 versus 1.83±2.85; P<0.001; Dunn's post-hoc test; Fig. 1D), disrupting the ultimate 1:1 ratio of PrE to EPI cells observed in both control groups (P<0.01; Dunn's post-hoc test; Fig. 1D). This corroborates the hypothesis that interactions between ESCs and host cells rely on cell–cell communication mediated by FGF4/MAPK signalling.

Having determined that the PrE:EPI ratio is impaired in Fgf4^−/− ESCs-chimaeras (Fig. 1D), we wanted to check whether supplementation of the culture medium with exogenous FGF4 would rescue the observed phenotype of the chimaeric blastocysts by restoring balanced numbers of these two ICM lineages. To that end, chimaeric embryos containing Fgf4^−/− ESCs were cultured until the blastocyst stage in a medium with the addition of FGF4 (Fig. 1A). We revealed that exogenous FGF4 in the medium shifted the proportions of the PrE and EPI building the ICM of these blastocysts compared with the blastocysts containing Fgf4^−/− ESCs, cultured in standard KSOM medium (P<0.001; Dunn's post-hoc test; Fig. 1B). The contribution of PrE and EPI of embryo origin in these blastocysts was also higher and lower, respectively, compared with the control embryos injected with Fgf4^+/+ ESCs (P<0.001; Dunn's post-hoc test; Fig. 1B,D). It is also worth noting that upon FGF4 treatment all host ICM cells acquired the PrE fate and, as a consequence, EPI was composed exclusively of ESC progeny (Fig. 1B,E).

These results suggest that balanced numbers of EPI and PrE cells depend on the delicate system of paracrine intercellular interactions, mediated by FGF4/MAPK, and they cannot be restored in Fgf4^−/− ESC chimaeras solely by the introduction of growth factor to the culture medium.

Next, we wanted to determine the effect of complete inhibition of FGF4/MAPK signalling on the development of ESCs-chimaeric embryos. To that end, DsRed ESC-microinjected embryos were cultured until the blastocyst stage in a medium supplemented with PD173074 and PD0325901, inhibitors of FGF receptor (FGFR) and MEK kinase, respectively (Fig. 2A). First, we confirmed that, in contrast to the uninjected control embryos (n=31) and ESC-injected embryos (n=27) cultured in standard medium, the corresponding uninjected (n=22) and ESC-injected embryos (n=32) cultured in the presence of FGF4/MAPK inhibitors contained ICMs devoid of PrE (Fig. 2B; Fig. S2B,C). In addition, we observed that chimaeric blastocysts cultured in the standard KSOM medium had either chimaeric (29.63% of embryos) or fully ESC-derived (70.37%) EPI, whereas all of those grown after microinjection in the presence of FGF4/MAPK inhibitors were characterised by chimaeric EPI (P<0.001; Fisher's exact test; Fig. 2C). Comparing the contribution of ESCs and host blastomeres derivatives to EPI in both groups of blastocysts, we noted that inhibition of FGF signalling resulted in a lower percentage of ESCs building this lineage (24.79%±22.83%) than the share of host embryo cells (75.21%±22.83%; P<0.001; Mann–Whitney U-test; Fig. 2D).

Moreover, both uninjected and ESC-injected embryos treated with FGFR and MEK inhibitors exhibited a lower percentage of TE cells than the corresponding groups of embryos cultured without additional supplementation (67.25%±5.37% versus 75.93%±2.57% and 67.25%±5.37% versus 76.98%±4.04%; P<0.001; Dunn's post-hoc test; Fig. 2E). Taken together, embryos subjected to FGF inhibitors are unable to receive the FGF signal from donor wild-type (WT) ESCs, and therefore cannot rescue PrE formation. This provides definitive proof that FGF secreted by ESCs enables the dialogue with blastomeres, guaranteeing the proper formation of PrE. Moreover, inhibition of FGF4/MAPK signalling in embryos, irrespective of the presence or absence of ESCs, compromises the TE lineage, and in chimaeras decreases ESC share in the ICM.

Bearing in mind that introduced ESCs correspond to the inner cells of the embryo, competing with the host blastomeres for the contribution to the ICM, and that the TE:ICM ratio remains constant in ESC-chimaeras, we hypothesised that the chimaeric embryo regulates the correct cell lineage proportions by increasing the frequency of symmetric cell divisions, which are the source of outer, future TE cells. To verify this hypothesis, we used time-lapse imaging allowing us to track the cell fate within the embryo. To visualise cell divisions during chimaera formation, we injected zygotes with mRNA for β-tubulin-GFP (to mark microtubules) and mRNA for histone H2B-mCherry (to mark nuclei). After reaching the eight-cell stage, embryos were subjected to microinjection of CAG::H2B-GFP ESCs and their further development up to the blastocyst stage was recorded by time-lapse imaging (Fig. 3A; Movie 1).

During the fourth cell division (eight- to 16-cell stage transition), both uninjected control (n=16) and experimental embryos (containing ESCs, n=20) were characterised by a similar average number of oblique divisions (P>0.05; Fig. 3B,C) dominating over other cleavage types. In both variants, this type of division was more frequent than symmetric and asymmetric ones (P<0.001; χ² tests; Fig. 3B,C). Interestingly, we noted that blastomeres of the ESC-injected embryos underwent symmetric divisions more frequently than blastomeres of control uninjected embryos (P<0.01; χ² tests; Fig. 3C).

Analysing further the location of the cell nuclei of the resulting daughter blastomeres, we noted that the average number of inner and intermediate cells (which occupied the position between the outer and inner blastomeres and did not change localization) was similar for control and experimental embryos (P>0.05; Mann–Whitney U-tests; Fig. 3D). It is worth noting that ESC-injected chimaeric embryos exhibited a higher average number of daughter outer blastomeres than the corresponding control embryos (P<0.01; Mann–Whitney U-test; Fig. 3D; TableS1A).

Tracking the types of divisions occurring during the fifth cleavage division (16- to 32-cell stage transition), we observed that in both control uninjected (n=6) and experimental ESC-injected embryos (n=8) oblique divisions were less frequent than symmetric (P<0.001; χ² test; Fig. 3E) and asymmetric divisions (P<0.05; χ² test; Fig. 3E), whereas there was no significant difference between the frequency of symmetric and asymmetric divisions (P>0.05; χ² test; Fig. 3E). It is of note that the frequencies of the corresponding cell division types, and consequently the number of inner, outer and intermediate blastomeres, were similar at the transition from the 16- to the 32-cell stage in the experimental chimaeric as well as the control group (P>0.05; unpaired two-tailed Student's t-test; Fig. 3E,F; Table S1B). Overall, these observations indicate that, in response to the presence of ESCs, the embryo increases the frequency of symmetric divisions at the transition from eight- to 16-cell stages, forming more outer blastomeres to sustain the correct final TE:ICM ratio.

Analysis of time-lapse movies allowed us to compare the length of the cell cycle of blastomeres in embryos depending on the presence or absence of ESCs. As the development of embryos was recorded from the onset of ESC microinjection, we were unable to assess the cell cycle duration at the transition from the eight- to 16-cell stage. For this reason, we decided to compare the average length of the cell cycle of blastomeres, defined as the time interval from the completion of the fourth cleavage division (eight- to 16-cell stage) until the completion of the fifth cleavage division (from 16- to 32-cell stage). We revealed that in ESC-containing embryos, the average duration of the cell cycle of blastomeres was longer than in uninjected control embryos (12.44 h±2.22 h versus 11.84 h±2.30 h; P<0.05; Mann–Whitney U-test; Fig. 4). This result suggests that the increase in the blastomere cell cycle duration may be an additional mechanism activated in the mouse embryo in response to the excess of lineage-restricted cells, thus allowing ESCs to compete with the host blastomeres for the EPI niche.

Next, we wanted to see whether the decrease in the cell cycle length of blastomeres could account for displacing them from the ICM in a process termed ‘cell competition’ involving apoptosis. Especially as, when analysing time-lapse movies, we observed single cases of cell death at the early blastocyst stage both in uninjected control embryos (in 2 out of 6; Movie 2) and in ESC-injected embryos (in 2 out of 8; Movies 3, 4 and 5). To get a deeper insight into the role of apoptosis in the plasticity of chimaeric embryos, we decided to compare the frequency of apoptosis in both groups of embryos fixed at two time points. ESCs express and secrete pro-apoptotic proteins, e.g. BAX (Xiang et al., 2018). In addition, bearing in mind that the presence of ESCs leads to a decrease in the contribution of the host cells to the ICM and the fact that mouse embryos express key genes involved in apoptosis at all stages of development (Exley et al., 1999; Jurisicova et al., 1998; Metcalfe et al., 2004), we hypothesised that ESCs promote apoptosis of the blastomeres to displace them from the EPI and to regulate the proportions of the emerging cell lineages in the ICM of the chimaeric blastocysts. To verify this hypothesis, we microinjected eight-cell embryos with ESCs, fixed them at the early and late blastocyst stages, and scored the incidence of apoptosis using immunostaining of cleaved caspase 3, the apoptosis executive protein (Fig. 5A).

We observed that in ESC-injected embryos at the early blastocyst stage (n=45) apoptosis occurred more frequently than in corresponding control uninjected embryos (n=26; P<0.001; Fisher's exact test; Fig. 5B,C). In contrast, at the late blastocyst stage, the incidence of embryos containing apoptotic cells was similar in the experimental (n=40) and control groups (n=18) (P>0.05; χ² test; Fig. 5B,C). Importantly, apoptosis frequency increased with the development of blastocyst, regardless of the presence (P<0.01; χ² test; Fig. 5C) or absence (P<0.001; χ² test; Fig. 5C) of ESCs within the embryo.

Because we observed that the apoptosis in chimaeric early and late blastocysts affected both the blastomeres and the ESCs, we decided to exclude ESCs undergoing cell death from further analysis and focus on host embryo blastomeres. Comparing uninjected control and chimaeric embryos at the same developmental stage, we found no differences in the frequency of apoptosis in the blastomeres (P>0.05; Fig. 5D). However, when comparing embryos between developmental stages, we observed that apoptosis of the blastomeres was more frequent in late control blastocysts than in early control blastocysts (P<0.001; Fisher's exact test; Fig. 5D) and in late chimaeric blastocysts compared with early chimaeric blastocysts (P<0.001; χ² test; Fig. 5D). This means that the phenomenon of programmed death of the blastomeres occurred more often in late blastocysts than in early blastocysts, regardless of the presence of ESCs. It is of note that apoptosis affected ICM cells more often than TE lineage both in the control and chimaeric embryos. Only ICM cells underwent cell death in both variants (control and chimaeric) at the early blastocyst stage, whereas in the late blastocysts apoptosis occurred within the ICM as well as TE cell population, but with a higher frequency in ICM (Fisher's exact test: P<0.001 for late control blastocysts; χ² test; P<0.001 for late chimaeric blastocysts; Fig. 5E). Collectively, we found no supporting evidence that the regulation of lineage proportions in the preimplantation chimaeric mouse embryo is based on the selective elimination of host embryo blastomeres triggered by the presence of exogenous ESCs.

As the microinjection method does not guarantee the presence of the same number of ESCs and, consequently, comparable levels of factors secreted by them between embryos, we decided to confirm the obtained results using another approach. We therefore cultured embryos from the eight-cell stage in the medium conditioned on ESCs and compared them for incidence of apoptosis with two control groups [embryos cultured in medium conditioned on a mouse embryonic fibroblasts (MEFs) feeder layer and in standard medium; Fig. S3A]. We again did not observe any differences in apoptosis frequency between experimental blastocysts (cultured in the medium conditioned on ESCs) and blastocysts from both control variants at the same developmental stage (P>0.05; Fig. S3B). Moreover, apoptosis occurred much more often in the late blastocyst stage, regardless of the presence of factors secreted by ESCs in the culture medium (P<0.01; Fig. S3B). Comparing the dependence of apoptosis frequency on the cell lineage, we observed increased apoptosis in the ICM compared with TE in blastocysts from all variants. However, there was no difference between experimental and control embryos in this regard (P>0.05; Fig. S3C,D). We thus confirmed that neither the presence of exogenous inner-like cells (ESCs) nor the presence of factors secreted by these cells in the culture medium induced an increase in the apoptosis of blastomeres in the blastocyst.

Having determined that ESC-injected embryos at the early blastocyst stage exhibited a higher frequency of apoptosis than control embryos (Fig. 5C), but not in the case of blastomeres (Fig. 5E), we decided to determine the number of ESCs undergoing programmed cell death following microinjection into an eight-cell host embryo. By calculating the ratio of apoptotic ESCs to the total number of ESCs present in the resulting blastocysts we found that the incidence of apoptosis of ESCs was higher in early blastocysts compared with late blastocysts (P<0.001; Fisher's exact test; Fig. 5F). This suggests that selective apoptosis is exploited by the chimaeric embryo to regulate the number of ESCs that ultimately form the EPI.

Despite many lines of evidence regarding preimplantation embryo plasticity, the mechanisms behind this phenomenon remain largely unknown. We took advantage of the ESC-chimaera model to study this particular aspect of mammalian embryogenesis. The introduction of exogenous cells into the embryo imitates the disturbance of embryogenesis, triggering the regulatory mechanism that would normally not be activated. ESCs derived from ICM after reintroduction into the embryo repopulate ICM and subsequently EPI, and therefore recapitulate some aspects of its specification. Similar to EPI cells, ESCs express pluripotency factors, pro-apoptotic proteins and growth factors, thus serving as a model of inner-like cells, which provide a local source of FGF4 in the ICM of the chimaeric embryo. Therefore, our chimaeric model may reflect, to some degree, what happens in a normal, non-chimaeric embryo.

We showed that the presence of supernumerary inner-like ESCs leads to the emergence of a multi-level mechanism regulating the final proportions of cell lineages within the blastocyst. The greater number of inner-like cells owing to the presence of ESCs must have been accompanied by a greater number of outer cells to ultimately maintain balanced numbers of ICM and TE precursors. Indeed, we observed that, during the fourth cleavage division in embryos injected with ESCs, symmetrical divisions occurred more often than in non-chimaeric embryos, generating more outer cells. As a result, at the late blastocyst stage the TE:ICM ratio remained constant and amounted to approximately 3:1, regardless of the presence of additional inner-like ESCs in the embryos. It suggests that the embryo is endowed with the mechanism controlling the type of division to ensure the formation of balanced proportions of both lineages, which is consistent with previous reports (Humięcka et al., 2016). Humięcka and colleagues analysed the number of inner versus outer and polar versus apolar cells in fixed 16-, 32- and 64-cell embryos microinjected with different numbers of ESCs at the eight-cell stage (Humięcka et al., 2016). They noted that the presence of ESCs reduced the frequency of asymmetric divisions at the expense of symmetric divisions in the fourth and sixth cleavage divisions, i.e. at the transition from the eight- to 16-cell stage and from 32- to 64-cell stage, respectively. In contrast to that approach, in which the origin of inner and outer cells has been inferred post-factum, our live imaging enabled monitoring of the course of cell divisions, including the incidence of oblique divisions. In addition to outer and inner cells, we could distinguish the presence of intermediate cells that occupied a unique position between outer and inner blastomeres and did not undergo internalisation, which contradicts other reports (Anani et al., 2014; Samarage et al., 2015). These discrepancies may result from the method of the analysis – based on the division angle, exposure of the blastomere cell membrane, or inheritance of the apical domain (Anani et al., 2014; Samarage et al., 2015; Watanabe et al., 2014). We cannot exclude that, by using cell membrane labelling, we would be able to classify the intermediate blastomeres as inner or outer ones. Nevertheless, we managed to uncover that, although the proportion of oblique divisions was similar in control and chimaeric embryos and this type of division prevailed over other types, ESC-containing embryos divided symmetrically more frequently than control ones, giving rise to a higher number of outer progeny cells at the transition from the eight- to 16-cell stage.

Whether the orientation and organisation of the spindle in the embryo and, consequently, the pattern of cleavage is determined by the geometry of the cell (Niwayama et al., 2019), the presence of the apical domain of eight-cell blastomeres orienting the spindle along the apical–basal axis (Korotkevich et al., 2017), or is random (Anani et al., 2014; Dietrich and Hiiragi, 2007; Fujimori et al., 2009; McDole et al., 2011; Watanabe et al., 2014) remains an open question. In light of the latest research, the division pattern is not stochastic and depends on the structure of the spindle (commonly occurring anastral spindles promote symmetric divisions and less frequent monoastral spindles trigger asymmetric division patterns) (Pomp et al., 2022).

The introduction of lineage-restricted ESCs into the embryo always requires ESCs to compete with the host blastomeres for the epiblast. ESCs force them to occupy other available niches – TE and PrE. Our data indicate that ESC-secreted FGF4 directs ICM progenitor cells of the host embryo into PrE fate and is responsible for maintaining the correct proportions of the PrE:EPI in the resulting chimaeric mouse embryo. Chimaeric blastocysts with a reduced level of FGF4 (due to the presence of Fgf4^−/− ESCs) exhibited an impaired proportion of PrE:EPI cells compared with the control embryos containing ESCs synthesising FGF4 and non-chimaeric blastocysts. This result is in line with the results obtained by Krawchuk and colleagues who studied PrE specification in non-chimaeric mutant mouse embryos and found that Fgf4^+/− embryos had a lower proportion of PrE cells than the WT Fgf4^+/+ embryos (Krawchuk et al., 2013). We revealed that a reduction in the percentage of PrE cells in embryos microinjected with Fgf4^−/− ESCs was accompanied by an increase in the proportion of host embryo cells in the EPI compared with blastocysts containing Fgf4^+/+ ESCs with unaffected FGF4/MAPK signalling. As a result, in embryos injected with Fgf4^−/− ESCs there was an increase in the frequency of chimaeric epiblast compared with embryos injected with Fgf4^+/+ cells in which ‘pure’ ESC-derived epiblast predominated. The reduced level of FGF4 found locally in the ICM of these blastocysts resulted in targeted differentiation of blastomeres into EPI and thus PrE was reduced compared with embryos characterised by a normal level of FGF4/MAPK signalling. Importantly, supplementation of chimaeric embryos containing Fgf4^−/− ESCs with exogenous FGF4 resulted in a reversal of the phenotype of the resulting blastocysts. This treatment yielded 100% of blastocysts with EPI colonised only by ESC progeny cells, without the participation of host embryo blastomeres. Fgf4^−/− ESCs, unable to respond to FGF4 by differentiation into PrE, remained the only cells capable of acquiring EPI fate. Moreover, as all of the host embryo ICM cells differentiated into the PrE, the percentage of cells of this lineage was significantly increased at the expense of EPI cells, compared with embryos injected with Fgf4^−/− ESCs and cultured in standard medium and blastocysts containing Fgf4^+/+ ESCs. This is in line with the results obtained by Krawchuk and colleagues for non-chimaeric embryos. Treatment of Fgf4^+/+, Fgf4^+/− and Fgf4^−/− non-chimaeric embryos with exogenous FGF4 at a concentration higher than 250 ng/ml resulted in the conversion of all ICM cells into PrE cells, regardless of the genotype (Krawchuk et al., 2013). In the light of a model proposed by Kang et al. (2017) and Molotkov et al. (2017), it can therefore be assumed that the ESCs present within the chimaeric mouse embryo are equivalent to those ICM cells of the non-chimaeric embryo, which are characterised by a high level of FGF4 expression. They secrete FGF4, which acts on the FGFR1 receptors on the embryo-derived ICM cells of the early blastocyst. By activating the MAPK pathway, it inhibits the expression of Nanog and maintains the expression of Gata6. Subsequently, other PrE markers such as SOX17 and GATA4 promote the maintenance of the resulting PrE precursor cells. As a consequence, in the chimaeric embryo containing ESCs capable of biosynthesis and secretion of FGF4, the majority of embryo-derived ICM cells differentiate into the PrE lineage, whereas the EPI is inhabited mainly by ESC daughter cells, and the share of host embryo cells in chimaeric EPIs is low. However, in the absence of the FGF4 signal from the introduced ESCs, the level of FGF4 in the embryo is locally reduced, failing to establish the proper, physiological level of FGF4/MAPK activation within the ICM and, as a consequence, impairing the balance between PrE and EPI cells. Our results, showing the paracrine nature of communication between ESCs and blastomeres of the host embryo mediated by the FGF4/MAPK pathway, are consistent with our previous results (Krawczyk et al., 2022). We studied the preimplantation development of ICM aggregated with the eight-cell embryo subjected to double knockdown of Fgfr1 and Fgfr2 genes. We demonstrated that disruption of the FGF4/MAPK pathway signalling between these two components of the chimaeric embryos resulted in their abnormal development (Krawczyk et al., 2022). Using a different model, Saiz et al. also confirmed the involvement of FGF signalling in coupling lineage size with cell fate decisions (Saiz et al., 2020). They constructed chimaeras consisting of eight-cell Fgf4^−/− embryos and a different proportion of WT ESCs and observed that ESCs in the number corresponding to the number of EPI cells found in the blastocyst were able to restore the formation of the PrE in the embryo. They also found that only uncommitted progenitor cells may respond to the FGF4 signal and adopt a PrE fate (Saiz et al., 2016, 2020). In contrast, we showed that ESCs introduced into the embryo previously treated with FGF inhibitors, and therefore unable to receive the FGF signal from the donor ESCs, cannot rescue PrE formation, thus providing definitive proof that FGF secreted by ESCs enables the dialogue with blastomeres, guaranteeing the proper formation of PrE. Our research, in conjunction with the results of Saiz et al. (2020) and Krawczyk et al. (2022) indicates the universality of this signalling pathway as a regulatory strategy in the chimaeric mouse embryo, regardless of the model used.

The predominance of ESCs in the EPI of the chimaeric embryo could also result from selective apoptosis of the host embryo blastomeres competing for EPI. For a long time, apoptosis has been regarded as a cell-autonomous, physiological process aiming at the elimination of cells with genetic abnormalities and cells improperly sorted during PrE and EPI differentiation (Bolton et al., 2016; Brison and Schultz, 1997; Handyside and Hunter, 1986; Hashimoto and Sasaki, 2019; Plusa et al., 2008). However, there is more and more evidence from research on various model organisms that apoptosis can be also non-autonomously controlled by the neighbouring cells in a cell-to-cell mechanism termed ‘cell competition’ in response to enhanced cell proliferation or mechanical forces generated from cell crowding (Di Gregorio et al., 2016; Kawamoto et al., 2016). However, we did not find evidence that supernumerary ESCs induce cell death of the neighbouring blastomeres. We showed that the apoptosis occurs more often in the ICM than TE lineage; however, it depends neither on the presence of ESCs in the host embryo nor on the presence of factors secreted by these cells. It rather constitutes a part of the natural process of blastocyst maturation, as evidenced by the intensification of this process at the late blastocyst stage. A similar result was obtained by Winiarczyk and colleagues, who studied the occurrence of apoptosis in preimplantation mouse embryos, inducing the changes in the number of cells, in the nuclear-cytoplasmic ratio of its blastomeres and their ploidy (Winiarczyk et al., 2021). Despite such serious changes in the organisation of the structure of the mouse embryo, cell death did not occur before reaching the early blastocyst stage, similar to the control embryos, which have not undergone any manipulations. However, in contrast to our research, Winiarczyk et al., who used as a model aggregation chimaera composed of two embryos at the same stage (without any reporter protein to distinguish between the two components), were unable to analyse the influence of one component of the chimaera on the other one in the context of apoptosis.

Our study confirmed the greater susceptibility of the ICM to apoptosis observed by other researchers (Adiga et al., 2007a,b; Byrne et al., 2002; Handyside and Hunter, 1986; Hardy and Handyside, 1996; Kamjoo et al., 2002; Plusa et al., 2008; Winiarczyk et al., 2021). The sensitivity of the ICM to apoptosis-promoting signals may be related to the differential expression of key factors involved in this process. This hypothesis is supported by the fact that although the anti-apoptotic factor TGFβ (transforming growth factor beta) is already expressed in the blastomeres of a four-cell mouse embryo, at the blastocyst stage its expression is limited to TE (Slager et al., 1991) and the expression of the pro-apoptotic p53 protein is 2.5-fold higher in the ICM cells than in the whole blastocysts (Jurisicova et al., 1998). Apoptosis is also used by the mouse embryo to eliminate cells with genetic abnormalities (Bolton et al., 2016) and cells located in a destination inconsistent with their expression profile during the segregation of the PrE and EPI precursors (Plusa et al., 2008; Wigger et al., 2017). Importantly, recent studies shed new light on the role of apoptosis in the blastocyst as a mechanism of cellular competition for participation in the emerging EPI (Hashimoto and Sasaki, 2019; Virnicchi et al., 2020). It has been shown that levels of HIPPO pathway transcription factor TEAD and its coactivator YAP are highly variable in blastomeres and correlate with the expression levels of pluripotency factors. Studies have shown that unspecified cells showing low TEAD activity are eliminated from the epiblast through cell competition (Hashimoto and Sasaki, 2019), thus providing a mechanism controlling the quality of the EPI cells. The recent work on the WWC2, one of the regulators of HIPPO signalling, confirmed the key role of this pathway in the selection of cells with naive pluripotency that populate the emerging epiblast (Virnicchi et al., 2020). We cannot rule out that a similar HIPPO-based mechanism may function also in the chimaeric embryo.

Apart from the physiological process of cell death, which is part of a normal developmental programme, we observed enhanced apoptosis of ESCs shortly after incorporation into the embryo. Using a live imaging approach to study chimaera formation, Alexandrova et al. revealed that eliminated cells are those in the process of exiting naive pluripotency (Alexandrova et al., 2015). Thus, we suspect that control of ESC size and quality through both apoptosis and lengthening of the blastomere cell cycle at the 32-cell blastocyst stage contribute to a competitive advantage of ESCs over the host embryo cells when it comes to colonising the EPI niche in the chimaeric mouse embryo.

In conclusion, our observations led us to propose a model of a multi-level regulatory mechanism underlying the plasticity of the chimaeric embryo (Fig. 6). According to this model, the mouse chimaeric embryo is subjected to various successive mechanisms regulating the pattern and duration of cell divisions, frequency of cell death and cell-cell communication mediated by FGF4/MAPK, which must be precisely coordinated to generate an embryo of the appropriate size and cellular composition.

The study was approved by the Local Ethics Committee for Experimentation on Animals No. 1 (permission numbers 792/2015, 509/2018 and 929/2019, Warsaw, Poland) designated by the National Ethics Committee for Experimentation on Animals (Poland) and was performed in compliance with national and EU regulations. Mice at the age of 8-12 weeks were kept at a 14 h light/10 h dark regime. F1(C57BL/6/Tar×CBA/Tar) females and F1(C57BL/6/Tar×CBA/Tar) and Swiss-Tg(UBC-GFP)30Scha/J males were used in this study.

To obtain zygotes and eight-cell embryos, F1(C57BL/6/Tar×CBA/Tar) females were induced to superovulation using 10 IU of pregnant mare serum gonadotropin (Intervet) followed after 48 h with 10 IU of human chorionic gonadotropin (hCG; Intervet) and were mated with F1(C57BL/6/Tar×CBA/Tar) or Tg(CAG-DsRed*MST)1Nagy/J males. Females with detectable vaginal plugs were sacrificed 22 h (zygotes) or 63-64 h (eight-cell stage) after hCG injection. Zygotes were recovered from the ampullae of oviducts and cleared of follicular cells using hyaluronidase (300 μg/ml, Sigma-Aldrich). Eight-cell embryos were flushed from the oviducts and uteri of females using a standard medium supplemented with 4 mg/ml bovine serum albumin (BSA; Sigma-Aldrich) and collected in small droplets of this medium under mineral oil at 37.5°C and 5% CO₂.

To visualise microtubules and nuclei, zygotes were microinjected with mRNAs for β-tubulin-GFP and histone H2B-mCherry (at final concentrations of 63 ng/μl and 95 ng/μl, respectively). mRNA for β-tubulin-GFP was transcribed in vitro from pRN3P (T3 promoter) using the mMessage mMachine T3 kit (Ambion). mRNA for histone H2B-mCherry was transcribed in vitro from pCS2-Cherry vectors (T3 promoter) using the mMessage mMachine T3 kit (Ambion).

The microinjections were performed under an inverted microscope (Axiovert 200, Zeiss) equipped with TransferMan (Eppendorf) micromanipulators connected to CellTram (Eppendorf) and FemtoJet (Eppendorf) pumps using in-house prepared injection pipettes. The procedure was carried out at room temperature in an M2+BSA medium under mineral oil. Microinjected zygotes were cultured in KSOM medium for 48 h, until reaching the eight-cell stage. At this stage of development, they were microinjected with eight to ten ESCs.

The following ESC lines were used: G4 ESC line expressing DsRed under the control of the chicken β-actin promoter (Vintersten et al., 2004), CAG::H2B-GFP ESC line expressing constitutively a human histone H2B gene fused with GFP gene under the control of a CAG (CMV-actin-globin hybrid) promoter (Hadjantonakis and Papaioannou, 2004) and an FD6 (Fgf4^−/−) ESC line (Wilder et al., 1997). Cell lines were validated by fluorescence microscopy and karyotyping. All cell lines have been routinely tested for contamination.

ESCs were grown on gelatine-coated (0.2%; Sigma-Aldrich) plastic dishes (Ø 35×10 mm; Falcon, Becton Dickinson), on a feeder layer of MEFs inactivated with mitomycin C (10 µg/ml; Sigma-Aldrich). ESCs were cultured in Knockout DMEM (Dulbecco Modified Eagle Medium; Thermo Fisher Scientific) supplemented with 15% foetal bovine serum (Thermo Fisher Scientific), streptomycin (5 mg/ml; Thermo Fisher Scientific), penicillin (5000 units/ml; Thermo Fisher Scientific), non-essential amino acids (0.1 mM; Thermo Fisher Scientific), L-glutamine (2 mM; Thermo Fisher Scientific), mouse LIF (50 units/ml; Chemicon International) and β-mercaptoethanol (0.1 mM; Sigma-Aldrich). Before microinjection, the ESCs were subjected to 5 min of trypsinization (0.25% Trypsin-EDTA; Thermo Fisher Scientific), collected in a 15 ml conical centrifuge tube (Falcon, Becton Dickinson), centrifuged at 1500 rpm (39 g) for 5 min, resuspended in ESC medium and kept at 4°C until the microinjection.

The microinjections were performed under an inverted microscope (Nikon Diaphot 300) equipped with Leitz micromanipulators using an injection pipette (BMIOOt-15, size 42; BioMedical Instruments). The procedure was carried out at room temperature in M2+BSA medium under mineral oil. Approximately ten ESCs were injected under the zona pellucida of eight-cell embryos before compaction. In the case of compacted embryos, treatment with Ca²⁺- and Mg²⁺-free M2 medium supplemented with EGTA for 5 min was applied in advance.

After microinjection, chimaeric embryos were cultured in KSOM medium or KSOM supplemented with exogenous FGF4 (500 ng/ml; R&D Systems) and heparin (1 μg/ml; Sigma-Aldrich) or KSOM medium with the addition of PD173074 (100 nM; Stemgent) and PD035901 (500 nM, Stemgent) (at 37.5°C, in 5% CO₂ for 48 h until reaching the expanded blastocyst stage.

Time-lapse imaging of chimaeric embryos was performed using an inverted fluorescence microscope (Axio Observer Z1, Carl Zeiss) equipped with a camera (AxioCam HR R3, Carl Zeiss) and an incubation chamber ensuring a temperature of 37.5°C and 5% CO₂ (Incubator XL multi S1, Carl Zeiss). Embryos were placed in drops of KSOM medium under mineral oil on a glass-bottomed dish, and their development was recorded using a 40× objective (EC Plan-Neofluar 40×/0.75 M27) for 48 h. The wavelength commands for GFP fluorescence were 488 nm excitation and 509 nm emission. For mCherry, the settings were adjusted to a 590 nm excitation wavelength and a 612 nm emission wavelength. Three-channel (GFP, mCherry and transmitted light) images were acquired every 15 min, every 2.5 μm. ZEN 2012 software (blue edition, Carl Zeiss) and Imaris software (Bitplane) were used for analysis.

ESCs were seeded at a density of 5.0×10⁵ cells per well of a six-well plate on feeder cells. The original medium dedicated to ESCs was replaced with serum-free fresh ESC medium 24 h later. After another 24 h, the supernatant (conditioned medium) was collected and then centrifuged at 4000 rpm (278 g) for 60 min at 4°C. The conditioned medium was stored at −80°C before further processing. Eight-cell embryos were cultured in the medium containing 50% KSOM and 50% conditioned medium. Because ESC cultures were carried out routinely on the MEF feeder layer we used two controls: embryos at the eight-cell stage cultured either in the medium containing 50% KSOM and 50% medium conditioned for 24 h on MEFs (control for MEFs) or in 50% KSOM and 50% ESC medium.

Embryos (deprived of zonae pellucidae by brief exposure to acidic Tyrode's solution) were fixed in 4% paraformaldehyde in Ca²⁺- and Mg²⁺-free PBS (30 min, room temperature), permeabilised in 0.5% Triton X-100 in Ca²⁺- and Mg²⁺-free PBS (30 min, room temperature) and blocked with 3% BSA and 0.05% Tween 20 (Bio-Rad Laboratories) in PBS (4°C, overnight). The blastocysts were incubated overnight at 4°C with primary antibodies diluted in PBS with 3% BSA and 0.05% Tween 20. Primary antibodies used were: anti-Sox17 goat polyclonal (R&D Systems, AF1924, 1:100; Krawczyk et al., 2022), anti-Sox2 rabbit polyclonal (Abcam, ab97959, 1:100; Krawczyk et al., 2022), anti-Cdx2 mouse monoclonal (BioGenex, MU392A-UC, 1:50; Krawczyk et al., 2022), anti-Caspase 3 rabbit monoclonal (Cell Signaling Technology, D175, 1:100). Subsequently, the blastocysts were rinsed with PBS and incubated for 2 h at room temperature with the corresponding secondary antibodies diluted 1:200 in PBS with 3% BSA and 0.05% Tween 20. Secondary antibodies used were: Alexa-fluor 594 donkey anti-goat IgG (A11058), Alexa-fluor 594 donkey anti-mouse IgG (A31573), Alexa-fluor 488 donkey anti-goat IgG (A11055), Alexa-fluor 647 donkey anti-rabbit IgG (A31573), Alexa-fluor 633 rabbit anti-mouse IgG (A21063) (all from Thermo Fisher Scientific). Omitting the primary antibodies was used as a negative control for their validation. Nuclei were stained with chromomycin A₃ (0.01 mg/ml in PBS; Sigma-Aldrich) at 37.5°C, for 30 min. Stained embryos were analysed using a laser scanning confocal microscope (LSM 510, Zeiss). Z-stacks of 50-60 optical sections per blastocyst were collected. Images were analysed using LSM Image Browser software.

Statistical analysis was conducted using IBM SPSS Statistics 23. Quantitative data was shown as a mean±standard deviation (s.d.). The Kruskal–Wallis with Dunn's post-hoc tests, Fisher's exact test, Chi-squared test, and unpaired two-tailed Student's t-test for independent samples were used for the statistical analysis. P-values of less than 0.05 were considered statistically significant.

We thank Kat Hadjantonakis for H2B-GFP ESC, Angie Rizzino for FD6 Fgf4^−/− ESCs, Anna Ajduk for β-tubulin-GFP construct, Berenika Plusa for helpful discussion on the manuscript and Darek Maluchnik for technical assistance. We also thank the staff of the Animal Facility at the Faculty of Biology for excellent care of animals.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201756.reviewer-comments.pdf.