Cdx2 Gene Expression and Trophectoderm Lineage Specification in Mouse Embryos

See allHide authors and affiliations

Science  17 Feb 2006:
Vol. 311, Issue 5763, pp. 992-996
DOI: 10.1126/science.1120925


Controversy exists as to whether individual blastomeres from two-cell-stage mouse embryos have identical developmental properties and fate. We show that the transcription factor Cdx2 is expressed in the nuclei of cells derived from the late-dividing but not the first-dividing blastomere of two-cell embryos and, by lineage tracing and RNA interference knock-down experiments, that this lagging cell is the precursor of trophectoderm. Cdx2 mRNA is localized toward the vegetal pole of oocytes, reorients after fertilization, and becomes concentrated in the late-dividing, two-cell-stage blastomere. The asymmetrical distribution of Cdx2 gene products in the oocyte and embryo defines the lineage to trophectoderm.

In most animals, the proper development of the embryo depends on the asymmetrical distribution of maternal transcripts and protein in the egg. In Drosophila, gradients of transcription factors are established that provide spatially restricted, cis-regulatory control over downstream zygotic genes (1). The Xenopus oocyte, although radially symmetrical, also possesses an asymmetrical distribution of mRNA that reflects the animal and vegetal poles of the egg and subsequent patterning during embryogenesis (2).

Early mammalian embryos have long been considered to lack the polarity so evident in frogs and insects because of the ease with which individual blastomeres can be manipulated and used to regenerate entire embryos and form chimeras (39). Nonetheless, considerable debate rages about whether the mature mouse oocyte contains factors localized in such a manner that they direct future cell differentiation (1013). This argument has recently focused on the individual fates of blastomeres of the two-cell-stage embryo. Whereas some argue that these two cells are equivalent in their developmental properties (1416), others have reported that one blastomere contributes predominantly to the embryonic part of the blastocyst [polar trophectoderm and inner cell mass (ICM)] and the other to the abembryonic portion (1720). Here, we provide evidence that trophectoderm lineage specification occurs early in development, certainly by the two-cell stage and most likely before the zygote first divides, in agreement with studies indicating the existence of some patterning at the early stages of mouse development (2123).

The caudal-type homeodomain transcription factor Cdx2 is required for proper development of trophectoderm in mice (24, 25). We noted that blastomeres of the two-cell mouse embryo could usually be distinguished according to the distribution of Cdx2 protein (Fig. 1). In 71% (37 of 52) of the cases, both blastomeres showed positive fluorescence, but nuclear localization of Cdx2 was observed in only one of them (Fig. 1B and fig. S1). Of the remainder, Cdx2 expression was either more or less confined to one blastomere (10 of 52, 19%) (Fig. 1A) or distributed relatively uniformly between the two blastomeres (5 of 52) without any obvious nuclear localization (26). The second polar body was also Cdx2-positive in most two-cell-stage embryos. The nuclear staining for Cdx2 (Fig. 2, G and I) matched the 4′,6′-diamidino-2-phenylindole (DAPI) staining pattern and nuclear architecture, which was often dominated by a large, centrally placed nucleolus. The peripheral ring of fluorescence was not a consequence of the focal plane selected, as is evident from examination of a series of optical sections (Fig. 2, G to I, and fig. S1). Localization of mEomesodermin (mEomes), another transcription factor associated with trophectoderm differentiation (27) and used here as a control, did not show any asymmetry in its distribution between blastomeres (Fig. 3 and fig. S2). Figure 2, A to C, shows that oocytes themselves (n = 45) expressed little or no Cdx2 protein, whereas zygotes (n = 56) consistently exhibited clumps of antigen throughout their cytoplasm but no signs of asymmetry in terms of protein distribution (Fig. 2, D to F), which is quite unlike that seen at the two-cell stage (Fig. 2, G to L). One possibility is that these images of zygotes represent protein either being sorted in some manner or in the process of degradation.

Fig. 1.

Localization of Cdx2 (green) in mouse embryos by using an indirect immunofluorescence technique. White arrows indicate blastomeres with perinuclear localization of Cdx2. (A) Two-cell embryo with strong cytoplasmic localization of Cdx2 in one blastomere. The red arrow indicates a polar body (PB). (B) Typical pattern of Cdx2 expression in a two-cell embryo. (C to E) Typical pattern of Cdx2 expression in three-, four-, and six-cell embryos. Scale bar indicates 25 μM.

Fig. 2.

Cdx2 expression in mouse oocytes and preimplantation stage embryos. (A to C) Typical unfertilized oocyte with no detectable expression of Cdx2. (D to F) Zygote with typical punctate staining. (G to I) Typical two-cell embryo. (J to L) Optical section from a two-cell embryo showing the intense Cdx2 expression (indicated by white arrows) in the nucleus. (M to R) Embryos at the four- and eight-cell stages of development with half the blastomeres showing nuclear localization of Cdx2. (S to U) Typical morula-stage embryo showing nuclear localization of Cdx2 in outer cells. (V to X) Typical blastocyst with nuclear localization of Cdx2 in trophectoderm. (a to h) Control blastocyst stained with Cdx2 antibody preadsorbed with recombinant Cdx2 protein. Tr, trophectoderm. Scale bars, 25 μM.

Fig. 3.

Distribution of Eomes at two-cell, four-cell, and blastocyst stages of preimplantation development in mouse shown by using immunofluorescence staining of Eomes (red) and nuclei (blue). (A to C) Two-cell-stage embryos with nuclear and some cytoplasmic staining for Eomes. Scale bars, 25 μM. (D to F) Four-cell-stage embryos showing nuclear localization of Eomes. (G to I) Blastocyst showing nuclear localization of Eomes in the trophectoderm (Tr) as well as the ICM cells. (J to L) Blastocyst stained without the use of the primary antibody for Eomes.

At the three-cell stage, Cdx2 nuclear staining was confined to the largest of the three blastomeres in 88% (53 of 60) of the embryos examined (Fig. 1C). By the four-cell stage, two of the four blastomeres demonstrated nuclear staining (Figs. 1D and 2, M to O, and fig. S3). Therefore, it is the lagging blastomere of the two-cell embryo that expresses Cdx2 in its nucleus. In 47 of 49 embryos examined at the six-cell stage, two blastomeres showed positive Cdx2 nuclear staining, whereas the other four were negative (Fig. 1E). At the eight-cell stage, 52 of 55 embryos exhibited four cells with and four cells without nuclear Cdx2 expression (Fig. 2, P to R, and fig. S4). As expected (24), Cdx2 at the late morula stage (∼16 cells; n 30) was nuclear and confined to exterior cells of the embryo (Fig. 2, S to U, and fig. S5). At blastocyst (n = 35), the protein was localized in nuclei of the trophectoderm and absent from the ICM (Fig. 2, V to X). The asymmetrical staining pattern for Cdx2 shown in Figs. 1 and 2 has been confirmed with a second antibody from a different source. These results suggest that Cdx2 expression patterns can be used to trace the trophectoderm lineage.

To determine whether the late-dividing blastomere from two-cell embryos was the predominant lineage precursor of trophectoderm, we injected neutral Texas red–conjugated dextran (TRD) (28), molecular weight of 40,000, randomly into blastomeres of two-cell embryos. A majority (37 of 40, 92%) of the embryos survived injection. About half (19 of 37) showed red fluorescence in the lagging blastomere at the three-cell stage (Fig. 3E) and progressed through the four-, eight-cell, and morula stages to the blastocyst (Fig. 3, F to H). All 19 blastocysts that formed showed red fluorescence in trophectoderm. The ICM was generally unlabeled except for the region adjoining the mural trophectoderm. To prove that injection of TRD had not caused this effect, we followed the development of embryos (n = 10) in which the TRD-injected blastomere had divided first. All 10 such embryos progressed normally to blastocyst, and nine showed red fluorescence localized to the ICM (Fig. 3, I to L). Therefore, the lagging blastomere at the two-cell stage contributes mainly to trophectoderm, whereas the leading blastomere is the precursor of the ICM. These observations support the previous conclusions of Piotrowska et al. (17), except our data indicate that the progeny of the lagging blastomere contribute to the polar as well as the mural part of the trophectoderm. The reason for this discrepancy is not clear but could be due to the method used to mark the blastomeres or, more likely, to mouse strain differences.

Next, we determined whether silencing of Cdx2 in the lagging blastomere would disrupt blastocyst formation by injecting one of the two-cell-stage blastomeres with TRD along with a predesigned Cdx2 small interfering RNA (siRNA) molecule. Most (54 of 62) of the embryos survived the injection. At the three-cell stage, we separated the embryos according to whether the labeled blastomere had divided first (Fig. 4, R and S) or last (Fig. 4, M and N) and examined the two groups at the morula (72 hours of development) (Fig. 4, O and T) and blastocyst (86 hours) stages of development (Fig. 4, P and U), as well as at several intermediary stages in a separate experiment (26) to track labeled cells. All of the embryos (24 of 24) in which the unlabeled blastomere had been the first to divide either became arrested at the morula stage or gave rise to abnormal structures with no expanded blastocoel cavity (Fig. 4P). In addition, the outer cells of these developmentally compromised embryos failed to express either Cdx2 itself or cytokeratin Endo-A (29) (fig. S6, I to L), both markers of trophectoderm. Oct4 was expressed in all but a few outer cells of these embryos that lacked apparent trophectoderm (fig. S6K). In the group of embryos where the RNA interference (RNAi) and TRD-injected blastomere had divided first, 93% (27 of 30) of the embryos developed into normal-appearing blastocysts (Fig. 4U) with label largely confined to the ICM. The expected pattern of Cdx2, cytokeratin Endo-A, and Oct4 localization occurred in these blastocysts (Fig. 4X and fig. S6, B to E). Therefore, injection of RNAi targeted to Cdx2 in the lagging blastomere leads to a failure of trophectoderm formation.

Fig. 4.

The lagging two-cell-stage blastomere contributes to the formation of trophectoderm. In (A) to (D), (W), and (X), Cdx2 (green), nuclei (blue), and the merge (cyan) shows Cdx2-positive nuclei. Scale bars, 25 μM. (A to D) Localization of Cdx2 in the lagging blastomere and its progeny at the three-, four-, eight-cell, and blastocyst stages. (E to H) Trophectoderm lineage tracing from two-cell-stage embryos injected with TRD in the lagging blastomere (red indicates TRD). (I to L) ICM lineage tracing from two-cell stage embryos injected with TRD in the leading blastomere. (M to P) Development of a two-cell-stage embryo coinjected with TRD and siRNA for Cdx2 in the lagging blastomere (at 34 hours) to an abnormal blastocyst without cavity (at 86 hours). (R to U) Development of a normal blastocyst from a two-cell-stage embryo coinjected with TRD and Cdx2 RNAi into the leading blastomere. (Q and V) Normal blastocyst development after coinjection of negative control RNAi and TRD in the lagging and leading two-cell blastomeres, respectively. (W) Blastocyst showing retention of trophectoderm Cdx2 expression after injection of negative control siRNA in the lagging two-cell blastomere. (X) Retention of trophectoderm Cdx2 expression in a blastocyst after injection of Cdx2 RNAi and TRD in the leading two-cell blastomere.

We next examined the presence of Cdx2 transcripts in oocytes, zygotes, and developing embryos by in situ hybridization. The results were surprising. Of 30 oocytes examined, 25 had Cdx2 mRNA concentrated in one-half of their cytoplasm (Fig. 5, A to C; for controls, Fig. 5a). In at least 21 of this group of 25 oocytes, the hybridization signal for Cdx2 mRNA was located on the side of the egg opposite the position of the first polar body, i.e., toward the vegetal pole (fig. S7). The remaining oocytes showed a less pronounced polarity (26). We also found that the expression of Cdx2 mRNA and its localization to one hemisphere of the oocyte in preovulatory oocytes in metaphase II arrest was not evident at metaphase I (fig. S8).

Fig. 5.

Distribution of Cdx2 mRNA in oocytes and at various stages of mouse preimplantation development. Cdx2 mRNA localization, green; nuclear material, blue. Red arrows indicate PBs. White arrows illustrate the general location of the Cdx2 signal. Scale bars, 25 μM. (A to C) Cdx2 mRNA in unfertilized oocytes. (D to F) Typical zygote with Cdx2 mRNA localized to one-half of the cell. (G to O) Typical embryos at the two-, four-, and eight-cell stages with prominent localization of Cdx2 mRNA in only half of the blastomeres. (P to R) Morula-stage embryo with Cdx2 mRNA localized in the cytoplasm of its outer cells. (S to U) Blastocyst with Cdx2 mRNA localized to the trophectoderm. (a to g) Controls at all stages hybridized to sense probe. (V) RT-PCR analysis for Cdx2 mRNA in oocytes, zygotes, and individual two- and four-cell blastomeres. RNA was extracted from single oocytes, single zygotes, and individual blastomeres from two-cell- and four-cell-stage embryos and amplified as cDNA by RT-PCR. Lane 1, a single ovulated oocyte; lane 2, a single one-cell zygote; lane 3 (2-cell a), one blastomere from a two-cell-stage embryo; and lane 4 (2-cell b), the second blastomere from the same two-cell-stage embryo. Lanes 5 to 8 (four-cell a to d), individual blastomeres from a single two-cell embryo. Lane 9, negative RT control.

After fertilization, there was a reorientation of the Cdx2 mRNA relative to the position of the polar bodies from the vegetal pole toward the animal pole, so that Cdx2 transcripts became concentrated to one side of the axis that bisects the animal and vegetal poles. Presumably, this shift reflects the reorganization of cytoskeletal components after the egg is fertilized (30). As a consequence, Cdx2 mRNA and presumably other transcripts associated with components of the cytoskeleton (31) can then be distributed unequally between the two blastomeres when the zygote divides.

The asymmetry in Cdx2 mRNA distribution observed in oocytes persisted in embryo stages: one-cell (23 of 32), two-cell (25 of 33), four-cell (26 of 36), and eight-cell (30 of 38) (Fig. 5, D to O), although it remains unclear whether the transcripts are entirely of maternal origin or newly transcribed from the embryonic genome. In most instances, the asymmetry was remarkable, with half the blastomeres showing little evidence for the presence of Cdx2 transcripts. The relatively homogeneous distribution of Eomes mRNA in control oocytes and embryos indicated that the asymmetry of Cdx2 was not an artifact (fig. S9). At the blastocyst stage, Cdx2 mRNA was confined to trophectoderm and absent from the ICM (21 of 28) (Fig. 5, S to U). The signal in blastocysts, although outside the nuclei, was generally highly concentrated. We have no explanation for this phenomenon, although it might relate to fixation or permeability artifacts. We confirmed the asymmetric distribution of Cdx2 transcripts in two- and four-cell embryos by performing reverse transcription polymerase chain reaction (RT-PCR) on RNA from individual blastomeres dissected from embryos (Fig. 5V) and likewise proved that Cdx2 mRNA was present in metaphase II oocytes. Together, these data show that the localization of Cdx2 mRNA in the oocyte accurately predicts the lineage of cells destined for trophectoderm.

One puzzle is why Cdx2–/– embryos have been reported to form temporary (24) or even complete (25) blastocoel cavities, whereas our studies indicate that knock-down of Cdx2 mRNA in the lagging blastomere of two-cell-stage embryos leads to complete trophectoderm failure. The most likely explanation is that sufficient maternal oocyte Cdx2 mRNA persists in Cdx2–/– embryos to define the trophectoderm lineage but not to ensure its eventual ability to function.

Our data are at odds with the mainstream view that blastomeres of two-, four- and eight-cell mouse embryos are essentially equivalent (1416). Instead, they support the opposing view that individual blastomeres of early embryos can have dissimilar fates (1723).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S9


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article