In a regulative system, where it is important to avoid invasive experimentation, technical differences between groups can be significant. We have tried but failed to confirm Hiiragi and Solter's rule that first cleavage always occurs perpendicular to the plane of pronuclei alignment3 and now respond to the points raised by Hiiragi et al.1.

Our time-lapse studies, which used multiple focal planes to monitor position carefully, indicated that cleavage tends to correlate with polar body and sperm entry sites before any flow of membrane to the cleavage furrow2. This would be difficult to find by studying just a single focal plane as others have done3,9. We note that their studies of cleavage did not use 40–80 sections, as the reference to a separate study of blastocysts10 might imply, but just a single section. Also, our findings were similar for wild-type and H2B–EGFP strains and the conditions used for culture and microscopy allowed normal development2. In contrast to Hiiragi and Solter3, we applied no selection or manipulation/alignment, but observed cleavage in all embryos in multiple planes2.

We observed cleavage between pronuclei, as described by Hiiragi3, only when we interfered with egg shape or disrupted the actin cytoskeleton2. Thus, it is possible that Hiiragi and co-workers inadvertently preselected embryos of a particular shape when aligning them because, in our experience, only some remain positioned on the culture dish with the polar body and the two pronuclei in the same plane. This could bias their data set as cell shape has an overriding effect11. To guard against such bias, we monitored all embryos and our findings2 did not support their conclusions3. Furthermore, Hiiragi and Solter's model of cleavage states that female and male chromatin mix on the first metaphase plate3, which, to the best of our knowledge, disagrees with the experimental observations of all other groups (see ref. 12, for example).

Our findings that progeny of early blastomeres make a biased contribution to different blastocyst parts are consistent with independent findings4,5,7. In assessing this, we found that it was critical to count the proportion of cells contributing to different blastocyst parts: to be significant, this had to be more than 70%13, and not simply three or fewer cells1.

Most important, if cleavage were random and blastomere fate could not be predicted, as Hiiragi et al. suggest, we would not have been able to isolate four-cell-stage blastomeres of like identity from different embryos and combine them to make chimaeras with different developmental properties8. Further weaknesses in the case against prepatterning are discussed elsewhere14.

Hiiragi et al.1 restate a model to account for how lineages of cells are established. Our findings that both two-cell blastomeres contribute to the inner cell mass and the trophectoderm6,13, and that these lineage decisions are taken from the eight-cell stage onwards, support this long-standing model. Thus, we contend that lineage establishment is not the topic under discussion but that the issue is whether the cavitation site, and therefore the embryonic–abembryonic axis orientation, is defined stochastically or not. Our results indicate that early cleavage patterns can bias the site of cavity formation (Fig. 1).

Figure 1: Differences between blastomeres arise as a result of different cleavage order and orientation; differences accumulate and bias the positioning of cells and hence their fate.
figure 1

In this model, progeny of one cell (dark blue) tend to undertake more symmetric than asymmetric divisions from the eight-cell stage onwards. If the outside cells resulting from symmetric divisions are more weakly connected to inside cells, this would create a preferred site at which fluid could accumulate when the cavity forms. Inside cells, yellow; outside cells, blue; red or green links connect sister cells after symmetric or asymmetric division of outside cells, respectively; orange links connect sister inside cells; orange arrows indicate expansion of the blastocyst as the cavity expands. Not all sister-cell relationships are shown, as sisters can occupy different planes of the spherical embryo. Inside cells give rise to pluripotent inner cell mass (progenitors for future body and some extra-embryonic tissues); outside cells give rise to trophectoderm (progenitors for extra-embryonic tissues).¤

The controversy seems in part to be semantic, resting on the interpretation of prepatterning and determination. By prepatterning, we mean some developmental bias. It does not connote determination: indeed, our own experiments to remove parts of the egg surgically indicate that determinants do not exist15.