Research Article

Blastocyst Axis Is Specified Independently of Early Cell Lineage But Aligns with the ZP Shape

See allHide authors and affiliations

Science  04 May 2007:
Vol. 316, Issue 5825, pp. 719-723
DOI: 10.1126/science.1138591


The mechanisms controlling the establishment of the embryonic-abembryonic (E-Ab) axis of the mammalian blastocyst are controversial. We used in vitro time-lapse imaging and in vivo lineage labeling to provide evidence that the E-Ab axis of the mouse blastocyst is generated independently of early cell lineage. Rather, both the boundary between two-cell blastomeres and the E-Ab axis of the blastocyst align relative to the ellipsoidal shape of the zona pellucida (ZP), an extraembryonic structure. Lack of correlation between cell lineage and the E-Ab axis can be explained by the rotation of the embryo within the ZP.

During vertebrate development, the fertilized egg gives rise to an embryo with an asymmetric body axis. In species such as the frog, information relating to the specification of the future body axis is asymmetrically localized as stored maternal factors in the fertilized egg (1). By contrast, in mammalian embryos, no proteins or other cellular components are known to be localized asymmetrically in the fertilized egg. Mammalian embryos possess a highly regulative character that is also a basis for recent regenerative medicine, and the pattern of cell division and allocation of cells within an embryo during the early stages vary between embryos. There are reports suggesting that intrinsic information relating the future embryonic axis is already present in early cleaving mouse embryos (27). However, several groups have presented data contradicting this hypothesis, suggesting instead that the bias relating to the embryonic axis is found relatively late in development (813). We have designed a series of experiments to directly address this controversial issue.

The blastocyst consist of two types of cells, those of the inner cell mass (ICM) and those of the trophectoderm (TE). After the fifth cell division, the mouse embryo starts to form a cavity (blastocoel or blastocyst cavity). The localization of the ICM on one side of this cavity marks the first appearance of an embryonic axis, known as the embryonic-abembryonic (E-Ab) axis. Thus, the differentiation of cells into ICM and TE is closely linked to the specification of the blastocyst E-Ab axis. There has been progress in understanding the molecular basis for the differentiation of ICM and TE (14, 15). However, the onset of specification of these two cell types has still not been defined. One hypothesis is that the cells located inside the morula mainly contribute to the ICM whereas outside cells differentiate into TE cells (1619). This “inside-outside” hypothesis has its basis in analyses of separated blastomeres, and continuous information from intact embryos is currently very limited. In this study, we focused on the preimplantation stage of mouse development to reveal factor(s) that function in specification of the blastocyst axis. Embryo shape [with and without the zona pellucida (ZP), the membrane surrounding the embryo proper] and the correlation between early cell lineages and cell fate were examined.

Analysis of cell behavior in developing preimplantation mouse embryos using a cell tracing system. One way to determine the correlation between cell lineage and axis formation in the mammalian embryo is to label and continuously observe all of the cells in live, undisrupted embryos during the entire period of interest. We generated a transgenic mouse line (R26H2BEGFP) that ubiquitously expresses EGFP (enhanced green fluorescent protein) fused to human histone H2B (20) under the control of the Rosa26 locus (21). This allowed continuous labeling of chromosomes in the nuclei of all the cells with fluorescent signals. To record development from the two-cell to the blastocyst stage, we captured bright-field and fluorescence images every 10 min (Fig. 1, A to J). Fluorescence images with 5-μm steps were used to reconstruct images that can be viewed three-dimensionally with glasses colored in red and green (Fig. 1, G to J, and movie S1). The position of each nucleus was plotted on the images, and the x-y position recorded simultaneously (Fig. 1, K to O). A total of 124 embryos were recorded in 15 repetitions of the experiment. Of these embryos, 108 (87%) developed into blastocysts, and in most cases the initial formation of the blastocoel was seen after the fifth cell division. The number of nuclei identified and traced varied between embryos. We traced as many cells as possible and the average number of traced cells was 31.8 per embryo. The embryos attached weakly to the culture plate, and the position of the ZP relative to the bottom of the culture plate remained fixed, judging from its shape and pigmentation pattern. By contrast, the position of the embryo proper within the ZP was not fixed, but embryos rotated within the ZP, as reported previously (8). A typical result of tracing the nuclei of an embryo from the two-cell stage is shown in Figs. 1 and 2A. When the traced points were plotted against time, a cell lineage map could be drawn (Fig. 2B). Because of technical limitations, we could not apply molecular markers to distinguish the ICM and TE in embryos used for time-lapse recording. However, whether a cell was in the embryonic region of the blastocyst (ICM and polar TE) or in the abembryonic region (mural TE) was judged from bright-field images, and the H2BEGFP-labeled nuclei was scored appropriately (blue or red, respectively, in Fig. 2, D and E) and traced back to the two- or four-cell stage. In 66 embryos analyzed from the two-cell stage, all of the descendants of two-cell blastomeres contributed to both embryonic and abembryonic regions. When the analysis was carried out from the four-cell stage, a wide range of ratios was observed for cells contributing to the embryonic or abembryonic regions (fig. S1). Out of 264 four-cell blastomeres analyzed, 222 (84%) contributed to both embryonic and abembryonic regions. No contribution to the embryonic region was observed for 20 (7.6%) four-cell blastomeres, whereas the descendants of 22 (8.3%) four-cell blastomeres were found only in the embryonic region (table S1). No correlation was found between the division order of the blastomeres and their future fates. We tried to classify the division patterns in the same way as applied previously (3, 22). However, we were unable to determine the type of the pattern which portion of cytoplasm was divided during the cleavage, because the relative position of the blastomeres changed and the second polar body moved on the surface of blastomeres during cleavage. Thus, we could not reproduce previous results indicating differences in future fates between blastomeres of the four-cell stage after a specific type of cleavage. By contrast, our results suggest that the blastocyst axis is formed independently of the early cell lineage up to the four-cell stage. When cell lineage was examined in later stages, such as the 16-cell stage, cells located relatively close to each other had common fates in the blastocyst. This suggests that future cell fate does relate to the position of the cells in later stages. It was also noticed from the time lapse movies that the speed of embryonic rotation within, and relative to, the ZP is reduced in later morula stages compared with the speed of earlier stages. This observation might explain why a cell labeled in the late morula remains in a similar position at the blastocyst stage, as reported by Gardner and Davies (23).

Fig. 1.

Time-lapse recording and tracing of nuclei in developing preimplantation R26H2BEGFP homozygous embryos. Bright field images (A to F) and images of EGFP (G to J) were captured simultaneously. The position of each nucleus was plotted and numbered (K to O). In this embryo, 35 cells were identified in the blastocyst, and the behavior of all of them was analyzed from the two-cell stage. The four colors (red, green, blue, and purple) correspond to each of the four-cell blastomeres. The second polar body is yellow. This analysis was carried out in reverse order from the blastocyst stage. Thus, several points are overlapping in nuclei of early stages. Scale bar indicates 20 μm.

Fig. 2.

Positions and velocity of movement of nuclei plotted against time from traced images. After the nuclei were traced (A), their positions on the y axis were plotted against time (t)(B). (C) The velocity (v) of each nucleus was plotted against time. Because the nuclei moved a long distance during cleavage, the peaks of the graphs correspond to the period of each cleavage (shown by bars and arrows). (D and E) The same plot points as (A) and (B) are shown in red and blue, corresponding to cells in the abembryonic (mural TE) and embryonic (the ICM and polar TE) regions at the blastocyst stage, respectively. In this embryo, both colors were seen in all of the four-cell blastomeres.

Lineage tracing in blastocysts developed in vivo. The above experiments and those examining the correlation between the two-cell lineage and the E-ab axis of the blastocyst (2, 3, 5, 6, 812, 23, 24) were carried out with embryos developing in vitro. However, these conditions might not be optimal for embryonic development. We therefore carried out lineage tracing in vivo with a labeling method that is noninvasive and milder compared with conventional labeling methods using microinjection of lineage tracers (25). A transgenic mouse strain was established that ubiquitously expresses Kikume Green-Red (MBL, Nagoya, Japan), a fluorescent protein that emits green fluorescence that can be converted to red after exposure to weak ultraviolet (UV) light (26, 27). For cell labeling, a part of a single two-cell blastomere was exposed to weak UV light (Fig. 3, A and B). The converted red proteins were observed throughout the whole blastomere (Fig. 3B). A total of 74 embryos labeled in this way at the two-cell stage were transferred immediately to the oviducts of pseudo-pregnant females. Of these embryos, 70 were recovered: 53 were at the blastocyst stage, and the remaining 17 embryos were at the morula stage. Two colors were distinguishable in 41 embryos. In 29 (71%) embryos examined by confocal microscopy, cells with red and green fluorescence formed two distinct clusters, as seen in Fig. 3, D to S. In the remaining 12 (29%) embryos, cells of different colors were intermingled, at least partially, and two or more clusters of cells with the same color were observed. In most of the embryos with only two clusters of red and green cells, the boundary between these clusters was not straight. It was therefore not easy to determine the angle between the boundary and the E-Ab axis of the blastocyst. For example, in the embryo shown in Fig. 3, the boundary seemed approximately parallel to the E-Ab axis at the focal plane shown in Fig. 3I around the embryonic region. However, this was not the case at the focal plane shown in Fig. 3L, and the boundary between the two colors was roughly perpendicular to the E-Ab axis. Because the distribution of labeled cells differed along the z positions in this embryo (compare Fig. 3, E to I, and Fig. 3, N to S), it was categorized as an example of an embryo in which the boundary was neither parallel nor perpendicular to the E-Ab axis. In two embryos (5%), the boundary of the two colors was close to perpendicular to E-Ab axis. On the other hand, in another four embryos (10%) the angle was close to parallel. In the rest of the embryos (23, 56%), the boundary varied along the z axis. These results suggest that there is no such tendency that the one of two-cell blastomeres contribute mainly to either embryonic or abembryonic region of the blastocyst. These in vivo experiments produced the same results as the in vitro time-lapse analysis, namely, that the axis of the blastocyst is formed independently of cell lineage from the two-cell stage.

Fig. 3.

Lineage tracing of embryos developed in vivo after labeling a blastomere by photoconversion at the two-cell stage. A two-cell blastomere of CAG-KikGR-1 mouse embryo (A) was labeled by photoconversion (B). Labeled embryos developed in vivo to the blastocyst stage (C to S) were collected. Embryonic regions are observed on the right of the dashed line (C). The distribution of labeled cells was observed by using laser scanning microscopy [(E) to (S)] at several focal planes.

Correlation between the shape of the ZP and the blastocyst axis. In most embryos, the first cleavage plane and the E-Ab axis of the blastocyst are close to perpendicular (2, 5, 6, 12, 22, 28). There are reports that suggest that the shape of the ZP is not a sphere but sometimes ellipsoidal in preimplantation stages (6, 8, 23, 24, 2830). It has also been proposed that this ellipsoidal shape provides asymmetrical physical pressure to orient the blastocyst axis (8). We analyzed the three-dimensional shape of the ZP [Supporting Online Material (SOM) text] and noticed that it was ellipsoidal and that two-cell blastomeres settled along the longest diameter of the ZP (fig. S2). The shape of the ZP was maintained from the one-cell (fig. S3) to early blastocyst stage (movie S1) before the blastocoel expands. We found that the embryo proper rotates within the ZP. Consequently, the positions of the first cleavage plane and the boundary between the two-cell blastomeres at the end of the two-cell stage (defined as the two-cell boundary, 2CB) were not identical in most embryos, although they were relatively close (fig. S3F). Thus, it is possible that in previous studies the 2CB was used instead of the first cleavage plane to determine the position of the two-cell embryo relative to the blastocyst axis. Two independent methods were therefore used to test the correlation between the 2CB and the E-Ab axis. First, we measured the angle between the E-Ab axis and the 2CB from time-lapse images (Fig. 4, A to C). In 64% of embryos (47 out of 74), this angle was more than 70°. Second, a localized region of the ZP over the polar body was labeled with the lectin WGA-Alexa594 (Invitrogen, Carlsbad, CA) at the late two-cell stage (Fig. 4, D and E). Then, labeled embryos were either cultured in a conventional CO2 incubator or allowed to develop in the oviducts of pseudo-pregnant females. In both cases, a similar distribution of angles between the labeled point and the E-Ab axis was observed (Fig. 4, H and I). In 52% (65 of 124) and 67% (25 of 36) of embryos developed in vitro and in vivo, respectively, the angle was more than 70°. These results suggest that the shorter diameter of the ZP and the E-Ab axis of the blastocyst become nearly perpendicular, not only under cultured conditions but also during normal development in vivo. The ellipsoidal shape of the ZP might affect both the position of the two-cell blastomeres and the orientation of the blastocyst axis, as discussed previously (8, 9). Because a blastocyst can be formed without the ZP, it is not required for the formation of the blastocyst axis, but it might function as a kind of restriction for axis orientation during normal development. This conclusion was also supported by the results of experiments in which embryos were cultured in alginate gel after removal of the ZP (6). In this case, the two-cell stage embryo was used as a template to make a mold in the alginate gel. Under these conditions, the configuration of the alginate gel restricted embryonic shape instead of the ZP.

Fig. 4.

E-Ab axis was close to perpendicular to the 2CB both in vitro and in vivo, and it is regulated by the ZP. By using time-lapse images (A and B), we measured the angle between the 2CB and the E-Ab axis of the blastocyst. The distribution of the angles is shown in (C). (D to I) A point on the ZP over the 2CB was labeled [(D) and (E)]. Embryos developed to the blastocyst stage were then analyzed [(F) and (G)]. The angle between the fluorescent label (G, arrowhead) and the E-Ab axis was measured in blastocysts developed in vitro (H) or in vivo (I). (J to L) Development of embryos without the ZP was recorded with the time-lapse system, and the angles between the 2CB and E-Ab axis were analyzed.

To test whether the ZP has an effect on the blastocyst axis, we examined the development of two-cell stage embryos after removal of the ZP with acid tyrode solution (Fig. 4, J to L). Embryos developed to the blastocyst stage without the ZP at a relatively high rate (43/50), and the shape of the blastocoel was closer to spherical in these embryos than in controls (Fig. 4K). This result suggests that physical pressure from the blastocoel overcomes the tension of the embryonic surface. The correlation between the 2CB and blastocyst E-Ab axis was lost when the ZP was removed (Fig. 4L). This finding suggests that the shape of the ZP is an important factor for the positioning of the blastocyst axis, although not predictive of the E-Ab direction. These results also support the idea that the major factor determining the correlation between the 2CB planeand the E-Ab axis of the blastocyst lies not within the embryo proper but in the shape of the ZP.

If this hypothesis is correct, movement of the embryo within the ZP might be the major reason for the lack of correlation between cell lineage and the spatial orientation of the blastocyst axis. In other words, although there is a geographical relation between the position of the two two-cell blastomeres and the future blastocyst axis, a consistent correlation between cell lineage and axis specification is not observed. These two observations seem inconsistent and have resulted in a major point of controversy in the field. In previous reports, lineage dependency was concluded on the basis of analyzing the spatial relation between the position of the two-cell blastomeres and the axis of the blastocyst (6, 23). We noticed that embryos rotated within the ZP (movie S1) and that the relative position of each cell changed during cleavage. This might explain the inconsistency mentioned above. When embryos were embedded within alginate gel, the movement of embryos within the ZP was also restricted (movie S2). Under these conditions, none of the 21 embryos examined rotated or changed their position relative to the ZP. The E-Ab axis formed nearly perpendicular to the short diameter of the ZP. Cells derived from a two-cell blastomere (colored in green and red in movie S2) localized mainly in the embryonic region, whereas cells derived from the other two-cell blastomere were found mainly in the abembryonic region of the blastocyst. This shows that a correlation between cell lineage and the axis of the blastocyst can be observed when the movement of the embryos within the ZP is restricted. In a previous study, we reported that descendants of a two-cell blastomere localized along the E-Ab axis in 73% of blastocysts (12). These embryos were also cultured in alginate gel, and thus the movement of the embryos within the ZP might have been restricted. This may explain why we observed lineage dependency in these experiments. This argument might also apply to other experiments carried out under the similar conditions restricting embryonic movements (2, 7).

In addition to the rotation of embryos within the ZP, other factors may also promote intermingling of cells within the embryo during normal development. For example, the velocity of nuclear movement (i.e., the distance the nucleus moves over time) becomes highest when a cell divides (Fig. 2C and SOM text), reflecting the fact that nuclei move a long distance during cell division. Thus, both complex movements of the embryo within the ZP and the rearrangement of cells within the embryo might contribute to the fact that blastocyst axis specification is not strictly dependent on early cell lineage in the mammalian embryo.

The molecular mechanisms controlling the formation and specification of the blastocyst axis have not been studied in this paper. In the future, the localization of functional molecules such as Cdx-2, Nanog, and Oct4 (14, 15) must be examined in developing live embryos utilizing imaging techniques similar to those described here to reveal the mechanism of differentiation. In addition, examination of ZP shape in other mammalian embryos may provide information about whether our findings with the mouse embryo are applicable to other species.

Conclusion. We have shown that the specification of the mouse blastocyst E-Ab axis is not dependent on early cell lineage. By contrast, the position of the blastocyst axis coincides with the ellipsoidal shape of the ZP, which is maintained during preimplantation stages. Thus, the bias for axis positioning in early-stage mouse embryos is imposed by factors outside the embryo, namely the ZP and its shape, and not by intrinsic factors residing within the blastomeres.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3

Movies S1 and S2

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article