Loss of Cubilin, the intrinsic factor-vitamin B12 receptor, impairs visceral endoderm endocytosis and endodermal patterning in the mouse

The visceral endoderm is a polarized epithelial monolayer necessary for early embryonic development in rodents. A key feature of this epithelium is an active endocytosis and degradation of maternal nutrients, in addition to being the source of various signaling molecules or inhibitors required for the differentiation and patterning of adjacent embryonic tissues. Endocytosis across the visceral endoderm epithelium involves specific cell surface receptors and an extensive sub-membrane vesicular system with numerous apical vacuoles/lysosomes. We previously reported that Cubilin, the endocytic receptor for intrinsic factor-vitamin B12, albumin and apolipoproteinA-I/HDL allows maternal nutrient uptake by the visceral endoderm. In the present study, we show that the germline ablation of Cubilin impairs endodermal and mesodermal patterning, and results in developmental arrest at gastrulation. Notably, visceral endoderm dispersal is impeded in Cubilin null embryos. We further confirm the essential role of Cubilin in nutrient internalization by the early visceral endoderm and highlight its involvement in the formation of apical vacuoles. Our results reveal essential roles for Cubilin in early embryonic development, and suggest that in addition to its nutritive function, Cubilin sustains signaling pathways involved in embryonic differentiation and patterning.


Results
Cubilin expression in the primitive endoderm and its derivatives between E3.5 and E8.5. To detail early Cubilin expression we combined whole-mount in situ hybridization and fluorescent whole-mount immunostaining. In E3.5 blastocysts Cubn mRNA and protein are detected in ICM cells adjacent to the blastocoelic surface, the final position of PrE cells. (Fig. 1A,B). In these cells Cubilin is not localized at the plasma membrane but is distributed around the nucleus in a punctate pattern (Fig. 1B). Additionally, at the same stage Cubilin immunostaining reveals a scattered intracellular and membrane distribution in the trophectodermal layer, as previously reported (Fig. 1B and ref. 18 ). Around one day later corresponding to E4.5, PrE cells start to migrate along the trophectoderm as they form the parietal endoderm (Fig. 1C,D). At this stage Cubn mRNA is abundantly expressed in the PrE and Cubilin immunostaining is clearly associated with the apical plasma membrane reflecting the polarization of PrE cells (Fig. 1C,D). Between E5.25 and E5.5, the PrE cells that remain in contact with the epiblast and the extraembryonic ectoderm form the embryonic (em) and extraembryonic (ex) VE respectively (Fig. 1E,F). From this stage onward the Cubn mRNA signal appears to progressively decline in the emVE (Suppl. Fig. 1 and ref. 34 ). In sharp contrast, the protein signal is readily detected throughout the VE (Figs 1F and 2). The reason for this discrepancy is unclear but may be due to the perdurance of the protein.
Around E5.5, emVE cells begin to express alpha-fetoprotein (AFP), a typical VE marker 35 . To clearly define the distribution of Cubilin in the emVE at this stage we used the transgenic mouse line Afp-GFP 36 where GFP is expressed under the regulatory sequences of Afp providing a useful tool to visualize VE and its derivatives. Cubn is required for morphogenesis during gastrulation. To obtain complete Cubn inactivation, Cubn Lox/+ mice 22,30 were first crossed with a line ubiquitously expressing Cre under the control of the PGK promoter, active during oogenesis 37 , to generate a Cubn null allele named Cubn 0 . Intercrosses of Cubn 0/+ animals then gave rise to Cubn 0/0 embryos where no residual expression of Cubilin was observed (Fig. 3A). Cubn 0/0 embryos develop normally until the onset of gastrulation at E6.5 (data not shown and Suppl Fig. 2A-C'). At E7.5 they are smaller than their control littermates and have a shorter PS (Fig. 3B,C and ref. 21 ). Additionally, in the mutants the proximally-located extra-embryonic region of the conceptus appears more developed than the distal embryonic region (Fig. 3B,C). Histological analysis confirms that gastrulation had occurred in E7.75 Cubn 0/0 embryos as evidenced by the presence of mesoderm cells (Fig. 3D). However, unlike the wild-type or heterozygous littermate (control) the mutants do not show head-folds or foregut structures (Fig. 3D). At E8.5, while in control embryos the first somites are detectable and the heart starts to form, in the mutant embryos the embryonic region remains small and disorganized without any signs of somites or other trunk structures (Fig. 3E). The extraembryonic tissues, including exVE, amnion and allantois are formed, but the allantois remains short and large, and never attaches to the chorion (Fig. 3E). Abnormal enlarged blood islands are seen in the mutant VYS ( Fig. 3E and ref. 21 ). These observations are consistent with a previous report on Cubilin loss of function 21 and indicate that Cubilin is required for proper patterning and morphogenesis of the gastrulating embryo.
Morphological and molecular defects of the Cubn-deficient VE. The main site of Cubn expression during gatrulation is the VE. In Cubn 0/0 embryos the overall morphology of the VE is altered showing intermixing www.nature.com/scientificreports www.nature.com/scientificreports/ of tall columnar and more cuboidal cells (Fig. 4A,B). To investigate the molecular defects of the mutant VE, we compared the expression of factors involved in the nutritive and/or hematopoiesis and vasculogenesis functions of the exVE by qRT-PCR and whole mount mRNA in situ hybridization analyses. The early (vHnf1, Hnf4, Gata4) and late (Afp and transthyretin) VE differentiation markers and the exVE markers Amn and Pem are expressed in mutant embryos, indicating that VE differentiation and molecular regionalization occur in the absence of Cubilin function (Fig. 4C-E). However, the expression of VE markers appears significantly increased in E7.5 mutant embryos compared to control littermates (Fig. 4C). During gastrulation, VE cells in the embryonic region are dispersed and displaced proximally towards the extraembryonic region as definitive endoderm is formed and  www.nature.com/scientificreports www.nature.com/scientificreports/ progressively covers the distal embryonic region of the conceptus. The increased expression of VE markers in Cubn mutants as compared to controls might be linked to the relative small size of the embryonic region in these mutants, developmental delay and/or to defects in the generation of the definitive endoderm germ layer.
Despite these endodermal defects, the expression of Bmp4, a known inducer of visceral endoderm involved in the proximodistal patterning of the embryo 38 , is not altered in the extra-embryonic ectoderm of the mutants (Fig. 4F).
Mesodermal and endodermal defects in the Cubn null mutants. In order to characterize the patterning defects observed in gastrulating Cubn null embryos we analyzed the distribution of established mesodermal, endodermal and prospective neuroectodermal markers. Anterior-posterior pattern evidenced by the expression of Hex or Lefty1 in the anterior visceral endoderm (AVE) and Brachyury (T) in the nascent mesoderm, is established in Cubn 0/0 around E6.5 (Suppl. Fig. 2A,B'). However, the expression domain of T remains restricted to the proximal most part of the mutant PS (Suppl. Fig. 2A'). Additionally, Lefty2 has not yet been induced posteriorly (Suppl. Fig. 2B') and Nodal, involved in the formation and maintenance of the PS remains present throughout the epiblast as in control embryos at earlier stages (Suppl. Fig. 2C,C'). We further analyzed Cubn null embryos at E7.5. Fgf8 is required for normal gastrulation and specification of the paraxial mesoderm 39 . In Cubn mutants, Fgf8 expression is upregulated (Suppl. Fig. 2D) and Fgf8 transcripts are homogenously distributed along the entire PS. This is in contrast to the proximo-distal gradient of Fgf8 expression observed in E7.5 control littermates (Fig. 5A,A'). Snail1, a marker of the embryonic mesoderm is detected in the mutant PS (Fig. 5B,B' and Suppl. Fig. 2D). These observations indicate that PS and mesoderm formation occur in the absence of Cubilin, however proximo-distal PS extension and/or patterning might be deficient.
In the mouse, the proximal PS produces extra-embryonic mesoderm, the middle streak produces lateral, paraxial and intermediate components of the trunk mesoderm, whereas the anterior-most region of the PS produces the anterior mesendoderm (AME), the axial mesoderm and definitive endoderm. The expression of the paraxial and presomitic marker Twist is significantly reduced and mainly marks the mutant extra-embryonic mesoderm as well as mesodermal cells close to the proximal PS in E7.5 Cubn 0/0 embryos (Fig. 5C,C' and Suppl. www.nature.com/scientificreports www.nature.com/scientificreports/ Fig. 2D). One day later, the expression of Wnt3a, a WNT ligand essential for caudal somite development 40 is strongly reduced in the mutants (Fig. 5D,D'). These observations indicate that paraxial mesoderm formation is impaired in Cubilin mutants.
Foxa2 and Goosecoid (Gsc), mark anterior PS and AME cells during gastrulation. Both genes are expressed in the mutants although their domains of expression are restricted to the distal part of the PS, and do not extend anteriorly ( Fig. 5E-F'). Dkk1 encodes a Wnt antagonist expressed first in the AVE and then in the PS derived AME. In control E7.5 embryos, Dkk1 transcripts are restricted to the prospective foregut domain in the anterior definitive endoderm. In the mutants, Dkk1 expression in the outer endoderm layer exhibits a horse-shoe pattern expanding posteriorly around the girth of the embryo (Fig. 5G-H'). This expression is reminiscent of that observed in the VE of E6.5 control embryos. This analysis suggests that AME morphogenesis and definitive endoderm formation are impaired in Cubn mutants.
In the ectoderm layer molecular regionalization also takes place during gastrulation with genes like Nodal being down-regulated from the anterior epiblast to become restricted to the PS region, and others like Otx2 being down-regulated from the posterior epiblast to be maintained in the prospective anterior neurectoderm. In Cubn mutants, the expression of both Nodal and Otx2 is abnormally maintained in the entire epiblast ( Fig. 5I-J' and Suppl. Fig. 2D). The expression of Hesx1, a marker of AVE, AME, and anterior neurectoderm, is strongly reduced in the mutants (Fig. 5K,K').
Together our results indicate that paraxial mesoderm and anterior mesendoderm formation are defective in Cubn mutants, which may in turn affect ectoderm regionalization.
Visceral endoderm dispersal is deficient in Cubn null embryos. We next investigated whether the above gastrulation defects originating at the PS impaired the dispersal of emVE cells known to take place at these stages 6 . The ingression of endoderm progenitors through the distal PS is necessary for definitive endoderm (DE) formation. Between E7.0 and E7.5 epiblast-derived cells intermingle with and disperse the emVE cells www.nature.com/scientificreports www.nature.com/scientificreports/ without displacing them to the extraembryonic regions 5 . To further examine the role of Cubilin in DE formation and concomitant emVE cell dispersal, we used live imaging and fluorescent cell labeling. The distribution of the definitive endoderm progenitor markers Sox17 and Foxa2 6,41 was analyzed in Cubn 0/0 -Afp-GFP transgenic embryos around E7.25, a time when emVE dispersal is extensive (Fig. 6). Sox17 is required for endoderm specification and is expressed in the DE prior to and after intercalation with the emVE. As previously reported Sox17 is strongly expressed in egressing DE progenitors and at lower levels in the emVE (Fig. 6A-A") 6 . Sox17-positive DE precursors are readily identified in surface renderings and transverse epiblast sections either surrounded by GFP-positive emVE cells or traveling alongside the mesoderm (Fig. 6A-A",a). In Cubn mutants most of the cells on the surface epithelium express low levels of Sox17 (Fig. 6B-B",b). A strong Sox17 signal is observed in some isolated cells (Fig. 6B), and the distribution of GFP is almost uniform (Fig. 6B') indicating a failure of emVE dispersal and DE formation. This is confirmed in transverse sections also showing an accumulation of mesodermal cells in the proximal epiblast close to the PS (Fig. 6b1). www.nature.com/scientificreports www.nature.com/scientificreports/ Foxa2-positive cell recruitment from the epiblast, transit through the PS and integration/expansion into the endoderm layer is also important for the establishment of the DE 41,42 . Foxa2 is co-expressed with Sox17 showing likewise a low expression levels in emVE cells and high expression levels in DE progenitors 6 . Foxa2-positive cells are either surrounded by GFP-expressing cells, or co-expressed with GFP in emVE (Fig. 6C-C" and c). In the emVE of Cubn mutants the distribution of Foxa2 and GFP is generally overlapping (Fig. 6D-D"), suggesting that Foxa2 expressing DE progenitors fail to reach the outer layer and disperse the emVE in these mutants. This is particularly evident in transverse sections across the proximal and distal epiblast of Cubn mutants (Fig. 6d1,d2), also showing that the bulk of Foxa2-positive cells accumulate in the mutant PS (Fig. 6d2). Impaired gastrulation and defective emVE dispersal thus appear to be associated, and indeed successive, events in Cubn mutants. Impaired endocytosis in the Cubn-deficient VE. Between E5.0 and E6.5 the VE, and after gastrulation the VYS and its extensive vesicular system are responsible for maternal nutrient uptake and waste exchange between the fetus and its mother. The VE, especially its extraembryonic portion, and the VYS are characterized by the presence of large apical vacuoles/lysosomes in which endogenous and exogenous endocytic markers accumulate, including the lysosome-associated membrane proteins LAMP1/2 and the lysosomal proteinases cathepsins 4,43 . The implication of Cubilin in exogenous/maternal apolipoproteinA-I/HDL endocytosis and degradation, as well as normal blood vessel formation in the VYS was previously reported 18,21 .
Deficient nutrient uptake may affect cell proliferation and/or cell survival. Cell proliferation identified by the M-phase cell marker phospho-histone H3 (H3S28P) is only marginally reduced in Cubn deficient embryos (Suppl. Fig. 4A-D). In contrast, apoptosis followed by TUNEL staining is much higher in the mutant embryos, suggesting that an active cell death process may account for the smaller size of the mutant epiblast (Suppl. Fig. 4A,B,D).
Lysotracker is a product that freely permeates cell membranes and accumulates in intracellular compartments with low pH; namely endolysosomal compartments and/or autophagolysosomal structures 44,45 . Control and mutant embryos were labeled for 15 min or 30 min at 37 °C with LysoTracker Green (LT) and then incubated Additionally, a scarce punctuated staining of the emVE and the control epiblast is evident (Fig. 7C,C"). In Cubn deficient embryos (n = 5) the LT signal is undetectable in VE cells suggesting defective formation of the acidic compartments (Fig. 7D-D"'). In contrast, an intense signal is observed in the anterior epiblast (Fig. 7D,D"), most likely reflecting increased cell death 46 also identified by the TUNEL staining.
Immunomorphological analysis of control and Cubn deficient embryos (Fig. 8) shows that Cubilin, Amn and Lrp2 are expressed and/or co-localize at the apical pole of exVE cells (Fig. 8A,B). Additionally, Amn can also be www.nature.com/scientificreports www.nature.com/scientificreports/ found in large apical vacuoles positive for the lysosomal marker Lamp1 (Fig. 8C). In control embryos the EEA1 expressing endosomes are disc-shaped, of various sizes and are localized closer to the apical plasma membrane than the much larger Lamp1-positive apical vacuoles (Fig. 8D).
In the mutants, Amn loses its polarized distribution and is diffusely expressed throughout the VE cell (Fig. 8E). By contrast, Lrp2 is readily found at the apical pole of the exVE cells (Fig. 8F). The Lamp1-positive structures are smaller, fewer and localized just beneath the apical membrane together with sparse, small and condensed EEA1 vesicles (Fig. 8G,H). Furthermore, the endosomal marker clathrin adaptor complex 1 (AP-1) is exclusively seen in small subapical vesicles and the expression of Lamp2, a marker of the apical vacuoles/lysosomes is barely detectable in exVE extracts (Suppl. Fig. 5A,B). Collectively these results show defective endosome and lysosome formation in Cubn mutants and strongly support a role for Cubilin in endo-lysosomal assembly and/or function in the VE.

Discussion
We have investigated the embryonic and cellular processes that require Cubilin function during gastrulation. Cubn deficiency results in mesodermal, endodermal and ectodermal patterning defects in gastrulation. Furthermore, we show that Cubilin is an early pan-VE marker involved in emVE dispersal, as well as the formation of the VE endocytic apparatus. In keeping with previous findings in other contexts, we establish the importance of Cubilin in VE nutrient endocytosis 18,20,21,47 .
As previously reported 21 , loss of Cubn is not compatible with embryonic survival. Cubn null embryos initiate gastrulation, but do not undergo normal organogenesis and display various mesodermal, endodermal and ectodermal defects. Here we show that anterior-posterior patterning is established and that PS formation and mesoderm migration are initiated in Cubn mutants. However, subsequent PS elongation and patterning are impaired and paraxial mesoderm population is reduced. Additionally, the AME, which is instrumental for the induction and maintenance of anterior neuroectoderm identity and the inhibition of Wnt signaling is formed, but the domains of expression of the AME-specific genes Foxa2 and Gsc are abnormal, as the anterior mesendoderm does not seem to extend anteriorly. Finally, the dispersal of emVE is impaired in Cubn mutants and the ingression of DE precursors is restricted to the posterior epiblast.
Within the VE, Cubilin is co-expressed with the classical VE marker AFP 36,48,49 . It is worth noticing however that whereas Cubilin protein is readily detected throughout the VE, the Cubn mRNA becomes downregulated in the emVE around E6.5. Although this difference may be due to the lower sensitivity of the in situ hybridization technique, it could alternatively reveal a rapid downregulation of the Cubn mRNA in the emVE, as previously reported for other VE markers concomitant, with protein perdurance in a population which is not actively dividing 5,50 . In the VE Cubilin is also co-expressed with Lrp2 and Amn. Amn in the extra-embryonic tissues is essential for normal gastrulation and its mutant displays molecular defects similar to those observed in Cubn mutants 13 . Although the lack of Lrp2 in the VE does not affect embryonic development, it is possible that the persistence of the three partners is necessary for an optimal Cubilin, Amn and Lrp2 activity in the VE. Supporting this hypothesis loss of apical expression of Lrp2 in the VE is thought to contribute to the gastrulation defects of the Mesd and Dab2 mutants 51,52 .
Inactivation of Cubn exclusively in the epiblast does not interfere with mesoderm formation or the induction of anterior neuroectodermal characters 22 . It is therefore likely that the phenotype of Cubn null mutants at E7.5 is due to the function of Cubilin in the VE. During mouse gastrulation, the VE is an essential component of the embryo-maternal interface required for the exchange of nutrients and waste. Cubn null mutants show defective gastrulation, developmental arrest, increased apoptosis in the epiblast and defective endocytosis of nutrients. These observations indicate that the nutritional function of the VE is severely impaired in Cubn mutants. Cross-interactions between VE and the underlying tissues, the extra-embryonic ectoderm and the epiblast, are also essential for embryo patterning as exemplified by the role of VE secreted Nodal antagonists in the restriction of PS inducing signals to the posterior epiblast 9 . Whether the failure to restrict Nodal expression in Cubn mutants is solely explained by the developmental arrest of these embryos or is also the result of impaired signaling between VE and the adjacent epiblast will require further investigations.
Degradative endocytosis of nutrients and signaling molecules is a key function of the VE. During the early steps of the process extra-cellular molecules bound to cell surface receptors are transported to subapical early endosomal vesicles. Mutations that affect these steps lead to developmental defects [52][53][54][55] . The later stages of endocytosis involve the engulfment of the endosomal vesicles by the apical vacuoles, specialized lysosomal compartments of the VE 4,56,57 . Lysosomal digestion is essential for nutrient delivery to the embryo-proper and signal termination. Disruption of lysosomal formation thus results in gastrulation defects [58][59][60] . Inactivation of the late endosomal adaptor p18, abundantly expressed in the VE, causes a gastrulation phenotype similar to that of Cubn mutants 60 . It is of interest that in the p18 null embryos, the normally large apical vacuoles/lysosomes are not detectable; only small lysosomal structures are evident, and Cubilin as well as Lrp2 are predominantly found within mutant VE cells. The authors propose that the disorganization of the endosomal/lysosomal compartments affects the trafficking and apical insertion of these membrane receptors 60 . Defective membrane expression of Cubilin, Lrp2 and most likely Amn may therefore contribute to the abrogation of nutrient uptake and subsequently affect epiblast viability.
Increased cell death in the epiblast, as the one we report here, was also observed in embryos lacking mVam2, a late endosomal protein, involved in BMP signaling 59 . In these mutants no large apical vacuoles are assembled and the delivery of internalized proteins, such as albumin or transferrin, from the endosomes to lysosomes is deficient despite normal initial uptake 59 . In the Cubn mutant both initial transferrin uptake and formation of apical vacuoles appear to be affected. No transferrin accumulates in subapical vesicles and the EEA1-positive endosomes are small, sparse and locate just beneath the plasma membrane. Furthermore, the morphology and topology of the (2019) 9:10168 | https://doi.org/10.1038/s41598-019-46559-0 www.nature.com/scientificreports www.nature.com/scientificreports/ apical vacuoles are severely perturbed as suggested by the Lysotracker and Lamp1 staining. Thus, both the initial, and to a greater extent the late, steps of endocytosis appear to be affected by the loss of Cubilin.
How Cubilin, a peripheral membrane protein might impair endocytosis is not clear. Either the formation of the Cubilin-Amn-Lrp2 macromolecular complex is an essential step for early endocytosis in the VE or additional Cubilin partners may be involved.
In summary, the present study uncovers important functions of murine Cubilin during the peri-gastrulation period. Cubilin deficiency affects epiblast patterning and VE morphogenesis. In view of the requirement of Cubilin for the proper assembly and/or maintenance of the apical vacuoles residing within the VE epithelium, we propose that Cubilin-mediated endocytosis in the VE is necessary for both nutrition and morphogenetic signaling in the early embryo. Mice. All mice were maintained under pathogen-free conditions, according to institutional guidelines. The conditional targeting vector for Cubn-deficient mice was described elsewhere 30 . Cubn null (Cubn 0/0 ) embryos were genotyped as described 30 . Afp-GFP TG/+ were genotyped as described 5 . Gestation (E0.0) was considered to have begun at midnight before the morning when a copulation plug was found. A small fragment of the extra-embryonic region of E6.5 and E7.5 embryos or a piece of yolk sac of older embryos was used for genotyping according to standard procedures.

Real time pCR.
Poly-A RNA from single embryos was isolated using Dynabeads mRNA DIRECT kit (Invitrogen, Cergy-Pontoise, France). Reverse transcription of RNA (600 ng/sample) was performed using iScript select cDNA synthesis kit (BioRad, Hercules, CA) according to the manufacturer's protocol. Real-time PCR was performed using a BioRad iCycler and IQ SYBR Green Supermix (BioRad), reactions were performed in triplicate. Transcript levels were calculated using the standard curves generated using serial dilutions of cDNA obtained by reverse transcription of control RNA samples then normalized to HPRT. Primer sequences for the indicated genes are included in Supplementary Table 1. The graphs plot the mean ± s.d. of the fold expression of the control and mutant littermates used. The number of specimens is indicated in the figure legends section. T-test statistical analysis showed significant differences at (**) p < 0.01, and (*) p < 0.1. Amplification specificities were assessed by melting curve analyses and amplicons were sequenced.
Lyso tracker staining of mouse embryos. Embryos were dissected free of decidua and extra-embryonic membranes discarded. LysoTracker green DND-26 (ThermoFisher) was prepared at 5 mM in Hanks BBS. Embryos were incubated at 37 °C for 15 min or 30 min with similar results, and then incubated for 15 min in the absence of the tracer (chase). After washing, embryos were fixed in 4% paraformaldehyde overnight. Actin staining was performed by incubating embryos in Phalloidin (Alexa Fluor Phalloidin 647, ThermoFisher) overnight. Images were collected by confocal microscopy (Zeiss LSM 710) and processed using ImageJ software.
Transferrin uptake. Embryos were removed from decidua and extra-embryonic membranes discarded.
Transferrin alexa Fluor 488 (ThermoFisher) was prepared at 25 μg/ml in DMEM. Embryos were incubated at 37 °C for 5 min. After washing, embryos were fixed in 4% paraformaldehyde overnight. Actin staining was performed by incubating embryos in Phalloidin (Alexa Fluor Phalloidin 647, ThermoFisher) overnight. Images were collected by confocal microscopy (Zeiss LSM 710) and processed using ImageJ software. www.nature.com/scientificreports www.nature.com/scientificreports/ 1:250, Merck Millipore) followed by Alexa 488-conjugated goat anti-rabbit (1:200, Invitrogen) was used. Nuclear staining was achieved by a 20-min incubation in Hoechst 33342 (Invitrogen). Images were collected by confocal microscopy (8 μm/section; LSM710 ConfoCor 3, Carl Zeiss) and processed using ImageJ software. Total numbers of TUNEL-and H3S28P-positive profiles were counted in 7 consecutive sections. TUNEL-and H3S28P-positive profiles were automatically detected using the following set of parameters in Volocity Image Analysis Software (Perkin-Elmer): (1) Find Objects by Intensity; (2) exclude objects smaller than 7 μm 2 ; (3) separate touching objects greater than 25 μm 2 . The apoptotic index (TUNEL) and mitotic index (H3S28P) were calculated as the percentage of cells positive for each marker to the total number of Hoechst 33342 -positive (nuclei marker, blue) cells in ectoderm (epiblast and extraembryonic ectoderm) and embryonic visceral endoderm per embryo. All graphs were generated using GraphPad Prism version 7 and data are shown as mean and SEM. Mann-Whitney U test was used for analysing cell number; TUNEL and H3S28P profiles and ***p < 0.001 was considered highly significant.