Changes in cell surface and cortical cytoplasmic organization during early embryogenesis in the preimplantation mouse embryo

Membrane topography and organization of cortical cytoskeletal elements and organelles during early embryogenesis of the mouse have been studied by transmission and scanning electron microscopy with improved cellular preservation. At the four- and early eight-cell stages, blastomeres are round, and scanning electron microscopy shows a uniform distribution of microvilli over the cell surface. At the onset of morphogenesis, a reorganization of the blastomere surface is observed in which microvilli becomes restricted to an apical region and the basal zone of intercellular contact. As the blastomeres spread on each other during compaction, many microvilli remain in the basal region of imminent cell-cell contacts, but few are present where the cells have completed spreading on each other. Microvilli on the surface of these embryos contain linear arrays of microfilaments with lateral cross bridges. Microtubules and mitochondria become localized beneath the apposed cell membranes during compaction. Arrays of cortical microtubules are aligned parallel to regions of apposed membranes. During cytokinesis, microtubules become redistributed in the region of the mitotic spindle, and fewer microvilli are present on most of the cell surface. The cell surface and cortical changes initiated during compaction are the first manifestations of cell polarity in embryogenesis. These and previous findings are interpreted as evidence that cell surface changes associated with trophoblast development appear as early as the eight-cell stage. Our observations suggest that morphogenesis involves the activation of a developmental program which coordinately controls cortical cytoplasmic and cell surface organization.     

Spectrin and calmodulin in spreading mouse blastomeres

The role of spectrin and its association with calmodulin in spreading mouse blastomeres was investigated. Embryonic spectrin binds 125I-calmodulin in a calcium-dependent fashion in the blot overlay technique. Double-labeling experiments show coordinate redistribution of spectrin and calmodulin in blastomeres preparing to undergo active spreading movement. At this stage cortical spectrin staining is lost from the region of cell-substrate contact and spectrin and calmodulin become concentrated in two structures closely associated with the contacted region: a group of spherical bodies located on the cytoplasmic side of the cortical layer and a subcortical ring that marks the perimeter of the contacted region. The localization pattern of spectrin and calmodulin is also coordinated with that of actin and myosin. The results suggest that spectrin plays a role in the spreading of blastomeres and that this function may involve linkage of spectrin, calmodulin, and the cortical contractile apparatus.     

Changes in the distribution of vinculin during preimplantation mouse development

It has been proposed that vinculin is a microfilament bundle-membrane linking cytoskeletal protein. We used double-fluorescence microscopy to study the distribution of vinculin and F-actin in mouse oocytes and preimplantation embryos. In oocytes and in the cells of cleavage- and blastocyst-stage embryos, vinculin exhibited a diffuse cytoplasmic distribution and was concentrated in a submembranous layer. The presence of vinculin in oocytes was confirmed by immunoblotting. In oocytes, a distinct concentration of actin was observed above the second metaphase spindle. During the 8-cell stage, compacting blastomeres exhibited partial polarization of cortical vinculin and actin toward their outward-facing surfaces. In precompaction-stage blastomeres, the submembranous layer of vinculin contained a ring-like concentration in the most peripheral region of each intercellular contact area. During later development, the amount of vinculin localized in the areas of intercellular contacts became modified. In embryos ranging from the compacted 8-cell stage to the mid-morula stage, the vinculin-specific fluorescence was only intense in some intercellular contacts, being indistinct in most contact areas. In late morulae, the flattened outer cells increasingly exhibited concentration of vinculin in contact areas. In contrast, actin-specific fluorescence was clearly evident in most intercellular contacts throughout the morula stage. At the early blastocyst stage, all contacts of the trophectoderm (TE) cells again regularly exhibited concentration of both components. At the late blastocyst stage, the staining pattern changed once again: the contact-associated concentration of vinculin-specific fluorescence was not observed in polar TE cells, while remaining clear in mural TE cells. In blastocyst outgrowths, TE cells displayed typical vinculin plaques at the peripheries of the cells. The continuous changes in the distribution of vinculin and actin suggest that these components are involved in the control of cellular relationships during early development. Immunoelectron microscopy and experiments using cytochalasin were performed in an attempt to relate the distribution of vinculin to the ultrastructural features of embryo cells.     

Localization of nonerythroid spectrin and actin in mouse oocytes and preimplantation embryos

Mouse oocytes, cleavage-stage embryos, and blastocyst-stage embryos were studied to show the distribution of both an immunoanalog to nonerythroid spectrin (p 230) and F-actin. Using antibodies to nonerythroid spectrin, diffuse, positive cytoplasmic fluorescence was regularly seen in oocytes and embryo cells. The presence of nonerythroid spectrin in oocytes was confirmed by immunoblotting. Oocytes usually exhibited an inconspicuous submembranous layer of nonerythroid spectrin, which was more pronounced in the area of the polar body. Oocytes regularly exhibited a peripheral concentration of actin. Throughout the cleavage and blastocyst stages, a cortical layer of nonerythroid spectrin and actin was usually observed in embryo cells. These submembranous layers on the outer surface of the embryo were relatively thin as compared to those in areas of intercellular contact. The contact areas regularly showed distinct positive staining, including a concentration of label at the most peripheral region of each contact area. This resulted in the presence of ring-like fluorescence around each blastomere. Nonerythroid spectrin and actin showed concentration to the contact area between the oocyte and the polar body. Although the general localization patterns of nonerythroid spectrin and actin were similar, double-staining experiments revealed that slightly different planes of focus were necessary to obtain sharp definition of the fluorescence of these components in areas of intercellular contact: the ring-like concentration of nonerythroid spectrin appeared to be localized more peripherally than that of actin. The cells of preimplantation embryos show motile features that include actual cell movements and striking changes in cell shape (e.g., during compaction). The submembraneous layers of nonerythroid spectrin and actin may contribute to the regulation of the deformability and thus the shape of embryo cells.(ABSTRACT TRUNCATED AT 250 WORDS)     

Membrane-cytoskeletal interactions in the early mouse embryo

Differentiation in the early mouse embryo begins at the 8-cell stage when the blastomeres flatten against each other by active spreading movements and surface and cytoplasmic elements become concentrated in the apical (uncontacted) region of the cells. A ring of cortical myosin marks the demarcation between the contacted and the uncontacted cellular domains. The organization of the cortical contractile apparatus in the blastomeres bears a formal resemblance to that of other cells that are engaged in similar motile activities. It has been proposed that a flow of cortical filaments could provide the motor that powers these movements. The applicability of such a cortical flow model to the early embryo and the implications for cell flattening and cell polarization are discussed in this review.     

Expression of ZO-1 and occludin at mRNA and protein level during preimplantation development of the pig parthenogenetic diploids

Expression of mRNAs and proteins of ZO-1 and occludin was analyzed in pig oocytes and parthenogenetic diploid embryos during preimplantation development using real-time RT-PCR, western blotting and immunocytochemistry. All germinal vesicle (GV) and metaphase (M)II oocytes and preimplantation embryos expressed mRNAs and proteins of ZO-1 and occludin. mRNA levels of both ZO-1 and occludin decreased significantly from GV to MII, but increased at the 2-cell stage followed by temporal decrease during the early and late 4-cell stages. Then, both mRNAs increased after compaction. Relative concentration of zo1α- was highest in 2-cell embryos, while zo1α+ was expressed from the morula stage. Occludin expression greatly increased after the morula stage and was highest in expanded blastocysts. Western blotting analysis showed constant expression of ZO-1α- throughout preimplantation development and limited translation of ZO-1α+ from the blastocysts, and species-specific expression pattern of occludin. Immunocytochemistry analysis revealed homogeneous distribution of ZO-1 and occludin in the cytoplasm with moderately strong fluorescence in the vicinity of the contact region between blastomeres, around the nuclei in the 2-cell to late 4-cell embryos, and clear network localization along the cell-boundary region in embryos after the morula stage. Present results show that major TJ proteins, ZO-1 and occludin are expressed in oocytes and preimplantation embryos, and that ZO-1α+ is transcribed by zygotic gene activation and translated from early blastocysts with prominent increase of occludin at the blastocyst stage.     

Characterization of intercellular junctions in the preimplantation mouse embryo by freeze-fracture and thin-section electron microscopy

Intercellular junction formation in preimplantation mouse embryos was investigated with thin-section and freeze-fracture electron microscopy. At the four-cell stage, regions of close membrane apposition with focal points of membrane contact and occasional underlying cytoplasmic densities were observed between blastomeres of thin-sectioned embryos. Corresponding intramembrane specializations were not, however, observed in freeze-fractured embryos. At the 8- to 16-cell stage, small gap and macula occludens junctions and complexes of these junctions were observed at all levels between blastomeres of freeze-fractured embryos. As development progressed from the early to mid 8- to 16-cell stage, the size of the occludens/gap junction complexes increased, forming fascia occludens/gap junction complexes. At the morula stage, gap junctions and occludens/gap junction complexes were observed on both presumptive trophoblast and inner cell-mass cells. Zonula occludens junctions were first observed at the morula stage on presumptive trophoblast cells of freeze-fractured embryos. The number of embryos possessing zonula occludens junctions increased at the mid compared to the early morula stage. At the blastocyst stage, junctional complexes consisting of zonula occludens, macula adherens, and gap junctions were observed between trophoblast cells of freeze-fractured and thin-sectioned embryos. Isolated gap and occludens junctions, adherens junctions, and occludens/gap junction complexes were observed on trophoblast and inner cell-mass cells.     

Expression of syndecan, a putative low affinity fibroblast growth factor receptor, in the early mouse embryo.

Syndecan is an integral membrane proteoglycan that binds cells to several interstitial extracellular matrix components and binds to basic fibroblast-growth factor (bFGF) thus promoting bFGF association with its high-affinity receptor. We find that syndecan expression undergoes striking spatial and temporal changes during the period from the early cleavage through the late gastrula stages in the mouse embryo. Syndecan is detected initially at the 4-cell stage. Between the 4-cell and late morula stages, syndecan is present intracellularly and on the external surfaces of the blastomeres but is absent from regions of cell-cell contact. At the blastocyst stage, syndecan is first detected at cell-cell boundaries throughout the embryo and then, at the time of endoderm segregation, becomes restricted to the first site of matrix accumulation within the embryo, the interface between the primitive ectoderm and primitive endoderm. During gastrulation, syndecan is distributed uniformly on the basolateral cell surfaces of the embryonic ectoderm and definitive embryonic endoderm, but is expressed with an anteroposterior asymmetry on the surface of embryonic mesoderm cells, suggesting that it contributes to the process of mesoderm specification. In the extraembryonic region, syndecan is not detectable on most cells of the central core of the ectoplacental cone, but is strongly expressed by cells undergoing trophoblast giant cell differentiation and remains prominent on differentiated giant cells, suggesting a role in placental development. Immunoprecipitation studies indicate that the size of the syndecan core protein, although larger than that found in adult tissues (75 versus 69 x 10(3) Mr), does not change during peri-implantation development. The size distribution of the intact proteoglycan does change, however, indicating developmental alterations in its glycosaminoglycan composition. These results indicate potential roles for syndecan in epithelial organization of the embryonic ectoderm, in differential axial patterning of the embryonic mesoderm and in trophoblast giant cell function.     

Cell surface interaction induces polarization of mouse 8-cell blastomeres at compaction

The development of the polarized surface binding of the fluoresceinated ligand concanavalin A (FITC-Con A) was studied in blastomeres of the early mouse embryo. Single 8-cell blastomeres, natural 8-cell couplets derived from the in vitro division of individual 4-cell blastomeres, and reaggregated couplets made from dissociated 8-cells were cultured for varying periods of time and on a variety of substrata. The development of surface polarity was found to be highly dependent upon cell contact. Over 50% of the cells in couplets were polarized after 4–5 hr in culture, with the smaller cell in the couplet usually more advanced in its polarization than the larger cell. The orientation of the poles of FITC-Con A binding was opposite the point of contact between cells in the couplets regardless of their previous orientation within the embryo or the plane of cleavage.     

Distribution of RNA binding protein MOEP19 in the oocyte cortex and early embryo indicates pre-patterning related to blastomere polarity and trophectoderm specification

We report the cloning and characterization of MOEP19, a novel 19 kDa RNA binding protein that marks a defined cortical cytoplasmic domain in oocytes and provides evidence of mammalian oocyte polarity and a form of pre-patterning that persists in zygotes and early embryos through the morula stage. MOEP19 contains a eukaryotic type KH-domain, typical of the KH-domain type I superfamily of RNA binding proteins, and both recombinant and native MOEP19 bind polynucleotides. By immunofluorescence, MOEP19 protein was first detected in primary follicles throughout the ooplasm. As oocytes expanded in size during oogenesis, MOEP19 increased in concentration. MOEP19 localized in the ovulated egg and early zygote as a symmetrical spherical cortical domain underlying the oolemma, deep to the zone of cortical granules. MOEP19 remained restricted to a cortical cytoplasmic crescent in blastomeres of 2-, 4- and 8-cell embryos. The MOEP19 domain was absent in regions underlying cell contacts. In morulae, the MOEP19 domain was found at the apex of outer, polarized blastomeres but was undetectable in blastomeres of the inner cell mass. In early blastocysts, MOEP19 localized in both mural and polar trophectoderm and a subset of embryos showed inner cell mass localization. MOEP19 concentration dramatically declined in late blastocysts. When blastomeres of 4- to 8-cell stages were dissociated, the polarized MOEP19 domain assumed a symmetrically spherical localization, while overnight culture of dissociated blastomeres resulted in formation of re-aggregated embryos in which polarity of the MOEP19 domain was re-established at the blastomere apices. MOEP19 showed no evidence of translation in ovulated eggs, indicating that MOEP19 is a maternal effect gene. The persistence during early development of the MOEP19 cortical oocyte domain as a cortical crescent in blastomers suggests an intrinsic pre-patterning in the egg that is related to the apical-basolateral polarity of the embryo. Although the RNAs bound to MOEP19 are presently unknown, we predict that the MOEP19 domain directs RNAs essential for normal embryonic development to specific locations in the oocyte and early embryo.     

Change in the adhesive properties of blastomeres during early cleavage stages in sea urchin embryo

Blastomeres of sea urchin embryo change their shape from spherical to columnar during the early cleavage stage. It is suspected that this cell shape change might be caused by the increase in the adhesiveness between blastomeres. By cell electrophoresis, it was found that the amount of negative cell surface charges decreased during the early cleavage stages, especially from the 32-cell stage. It was also found that blastomeres formed lobopodium-like protrusions if the embryos were dissociated in the presence of Ca2+. Interestingly, a decrease in negative cell surface charges and pseudopodia formation first occurred in the descendants of micromeres and then in mesomeres, and last in macromeres. By examining the morphology of cell aggregates derived from the isolated blastomeres of the 8-cell stage embryo, it was found that blastomeres derived from the animal hemisphere (mesomere lineage) increased their adhesiveness one cell cycle earlier than those of the vegetal hemisphere (macromere lineage). The timing of the initiation of close cell contact in the descendants of micro-, meso- and macromeres was estimated to be 16-, 32- and 60-cell stage, respectively. Conversely, the nucleus-to-cell-volume ratios, which are calculated from the diameters of the nucleus and cell, were about 0.1 when blastomeres became adhesive, irrespective of the lineage.     

Cell-cell contact modulation of myosin organization in the early mouse embryo

Preimplantation mouse embryos are characterized by a polarized distribution of cortical myosin (J. S. Sobel (1983). Dev. Biol. 95, 227-231.). Myosin was present in the peripheral regions of the blastomers and was not detectable in regions of cell contact. Disaggregation of the embryos yielded blastomeres which had a continuous layer of cortical myosin. Development of new contact relations in aggregates, between daughter cells of divided blastomeres, and in chimaeras resulted in renewed polarization of cortical myosin. The results indicate that continuous cell contact interaction modulates the distribution of myosin throughout the preimplantation stages of development. The loss of detectable myosin from regions of cell contact was correlated with development of cell contacts that remained stable after Triton X-100 extraction.     

Development of cellular polarity of hamster embryos during compaction.

Development of cellular polarity is an important event during early mammalian embryo development and differentiation. Blastomeres of hamster embryos at various stages were examined by scanning electron microscopy (SEM) and immunocytochemical staining. SEM observations revealed that 1- to 7-cell-stage embryos showed a uniform distribution of microvilli throughout the cell surface. Microvillous polarization was initially noted in the blastomeres (10-35%) of 8-cell-stage embryos. The polarized microvilli were observed mostly in the basal region of cell-cell contact and occasionally at the apical, outward-facing surface of the blastomere. Fluorescein-isothiocyanate-conjugated concanavalin A failed to reveal any polarity in the blastomeres regardless of the stages of the embryos. Actin staining showed that microfilaments were present beneath the cell surface, and in addition, areas of cell contact were more heavily stained, indicating a thick microfilament domain. Microtubules were located throughout the cytoplasm and were heavily concentrated near the nucleus during interphase, although they became redistributed in the region of the mitotic spindle during karyokinesis. The position of nucleus changed from the cell center to the apical, outward-facing surface of the cell, and it distanced itself from the basal microvillous pole. It is suggested that the changes in the cell surface and nuclear position are the first manifestations of cell polarity in peri-compacted hamster embryos, which appear as early as the 8-cell stage; furthermore, the outward migration of the nuclei may parallel the redistribution of microtubules in the cytoplasm.     

Changes in the organization of the actin cytoskeleton during preimplantation development of the pig embryo

The organization of the actin cytoskeleton was studied in unfertilized porcine oocytes and preimplantation stage embryos from Day 1 through Day 8 of development. Fixed and detergent-extracted oocytes and embryos were analyzed by fluorescence microscopy after staining with either rhodamine-phalloidin to localize filamentous actin or with affinity-purified anti-actin antibodies to localize the total immunodetectable actin. Whereas unfertilized oocytes contain immunoreactive cytoplasmic actin, rhodamine-phalloidin binding is not detected until fertilization when a prominent cortical staining pattern becomes apparent. In early cleavage stage embryos, filamentous actin is concentrated in the cell cortex of blastomeres especially at sites of cell-cell contact. Compacting morulae exhibit a marked accumulation of actin at the margins of blastomeres where numerous interdigitating cell processes are located. The predominantly pericellular distribution of actin becomes a distinguishing feature of trophectodermal cells in the expanding blastocyst at Day 6 of development; these cells form a prominent actin-limited zone circumscribing the inner cell mass. In Day 8 blastocysts, three cell types are present that are readily distinguishable based upon their actin displays among other cytological features. Trophectodermal cells exhibit continuous actin-rich lateral borders and stress fibers along their basal surface. Inner cell mass cells contain a discontinuous actin boundary and prominent foci of actin along their blastocoelic surface. Lining the blastocoel are patches of endodermal cells in which the actin is exclusively cortical. The data are discussed with respect to differences between species and the chronology of actin rearrangements during preimplantation development of the porcine embryo.     

Gap junctional communication in the preimplantation mouse embryo.

In this study, we examined cell-to-cell communication via gap junctional channels between the cells of the early mouse embryo from the 2-cell stage to the preimplantation blastocyst stage. The extent of communication was examined by monitoring for the presence of ionic coupling, the transfer of injected fluorescein (molecular weight 330) and the transfer of injected horseradish peroxidase (molecular weight 40,000). In the 2-cell, 4-cell and precompaction 8-cell embryos, cytoplasmic bridges between sister blastomeres were responsible for ionic coupling and the transfer of injected fluorescein as well as the transfer of injected horseradish peroxidase.In contrast, no communication was observed between blastomeres from different sister pairs. Junction-mediated intercellular communication was unequivocably detected for the first time in the embryo at the early compaction stage (late 8-cell embryo). At that stage, ionic coupling was present and fluorescein injected into one cell spread to all eight cells of the embryo. Injected horseradish peroxidase was passed to only one other cell, however, again indicating the presence of cytoplasmic bridges between sister blastomeres. Junctional communication with respect to both ionic coupling and dye transfer was retained between all the cells throughout compaction. At the blastocyst stage, trophoblast cells of the blastocyst were linked by junctional channels to other trophoblast cells as well as to cells of the inner cell mass, as indicated by the spread of injected fluorescein. In addition, the extent of communication between the cells of the inner cell mass was examined in inner cell masses isolated by immunosurgery; both ionic coupling and the complete spread of injected fluorescein were observed.     

Dorsal and ventral cells of cleavage-stage Xenopus embryos show the same ability to induce notochord and somite formation

Cells in the dorsal marginal zone of the amphibian embryo acquire the potential for mesoderm formation during the first few hours following fertilization. An examination of those early cell interactions may therefore provide insight on the mechanisms important for organization of axial structures. The formation of mesoderm (notochord, somites, and pronephros) was studied by combining blastomeres from the animal pole region of Xenopus embryos (32- to 512-cell stages) with blastomeres from different regions of the vegetal hemisphere. The frequency of notochord and somite development was similar in combinations made with dorsal or ventral blastomeres, or with both. Our results show that during early cleavage stages the ventral half of the vegetal hemisphere has the potential to organize axial structures, a property previously believed to be limited to the dorsal region.     

Laser microbeam-induced DNA damage inhibits cell division in fertilized eggs and early embryos

DNA double-strand breaks are caused by both intracellular physiological processes and environmental stress. In this study, we used laser microbeam cut (abbreviated microcut or cut), which allows specific DNA damage in the pronucleus of a fertilized egg and in individual blastomere(s) of an early embryo, to investigate the response of early embryos to DNA double-strand breaks. Line type γH2AX foci were detected in the cut region, while Chk2 phosphorylation staining was observed in the whole nuclear region of the cut pronuclei or blastomeres. Zygotes with cut male or female pronucleus showed poor developmental capability: the percentage of cleavage embryos was significantly decreased, and the embryos failed to complete further development to blastocysts. The cut blastomeres in 2-cell, 4-cell, and 8-cell embryos ceased cleavage, and they failed to incorporate into compacted morulae, but instead underwent apoptosis and cell death at the blastocyst stage; the uncut part of embryos could develop to blastocysts, with a reduced percentage or decreased cell number. When both blastomeres of the 2-cell embryos were cut by laser microbeam, cell death occurred 24 h earlier, suggesting important functions of the uncut blastomere in delaying cell death of the cut blastomere. Taken together, we conclude that microbeam-induced DNA damage in early embryos causes compromised development, and that embryos may have their own mechanisms to exclude DNA-damaged blastomeres from participating in further development.     

Presence and distribution of E-cadherin in the 4-cell golden hamster embryo. Effect of maternal age and parity

Maternal age dependency of gestation time in hamster and in other mammals is a well demonstrated fact. We have recently shown that adult nulliparous and multiparous hamster females show significant asynchrony and retard on early embryo development (from two blastomeres to morula stages) when compared with nulliparous young females. The number of cell-cell adhesions between blastomeres in early embryo development has been reported to be a good indication of the ability of embryos to cleave and develop. In this work we studied, by indirect immunofluorescence, the presence and distribution of E-cadherin in 4-cell embryos obtained from nulliparous young (NYF), nulliparous adult (NAF) and multiparous adult (MAF) hamster females. Distribution and intensity of fluorescence was observed and registered using confocal microscopy. Staining intensities for E-cadherin were quantified by computed densitometry in the free membrane regions, in the cytoplasm region and in the cell-cell adhesion zones of each embryo. E-Cadherin in all the studied zones was significantly higher (p<0.01) in NYF. Cadherin concentration in the intercellular membranes was always statistically higher (p<0.05) than in the free membrane regions. An appreciable concentration of E-cadherin was found in the cytoplasm of the 4-cell embryos obtained from the three groups of females, but was significantly higher in NYF. No statistical differences were observed in any of the parameters studied between NAF and MAF. Our results seem to indicate that changes in the reproductive behavior related to age and/or multiparity may be correlated with changes in the processes related to intercellular adhesions during early cleavage.     

Cell specific loss of polarity-inducing ability by later stage mouse preimplantation embryos

The individual blastomeres of the preimplantation mouse embryo become polarized during the 8-cell stage. Microvilli become restricted to the free surface of the embryo and this region of the membrane shows increased labeling with FITC-Con A and trinitrobenzenesulfonate (TNBS). Previous studies have shown that this polarity develops in response to asymmetric cell-cell contact with stage specific induction competent blastomeres. In the present study, the ability of later stage embryos to induce 8-cell polarization has been investigated. Newly-formed, nonpolar 8-cell stage blastomeres (1/8 cells) were isolated, then aggregated with morulae, inner cell clusters (from morulae), blastocysts, or inner cell masses (ICM) and cultured for 8 hr. Aggregates were then assayed for polarity. The results show a hierarchy of inducing ability, with the ICM and IC cluster possessing greater activity than the morula and polar trophectoderm of the early blastocyst, while the mural trophectoderm shows very little inducing activity. Furthermore, the inducing ability of the polar trophectoderm decreases with complete expansion and hatching of the blastocyst. These results indicate that the ability to induce 8-cell blastomere polarization is retained by the embryo beyond the 8-cell stage and that this ability is lost with further differentiation.