When and how does cell division order influence cell allocation to the inner cell mass of the mouse blastocyst?

The mouse expanded blastocyst contains two distinct and committed cell subpopulations, the trophectoderm and the inner cell mass (ICM). The allocation of cells to the ICM and trophectoderm lineages depends upon their internal or external position within the embryo at earlier stages (Tarkowski & Wroblewska, 1967; Hillman, Sherman & Graham, 1972). It has been known for some time that the earliest stage at which inside cells can be detected is the 16-cell morula (Barlow, Owen & Graham, 1972) and recent evidence suggests that in the undisturbed embryo this pool of inside cells contributes on average 75 % of the ICM cells. The remaining 25 % of ICM cells derives almost exclusively from a second allocation of cells to the inside in the 32-cell embryo (Balakier & Pedersen, 1982; Pedersen, Wu & Balakier, 1986; Fleming, 1987; Dyce, George, Goodall & Fleming, 1987). The allocation of cells to an inside position is achieved by the division of an asymmetrically organized (polarized) 8- or 16-cell blastomere such that the progeny do not receive equivalent endowments. Such a division is called differentiative, and yields an outer polar cell derived from the apical region of the parent cell and an inner apolar cell derived from the basolateral region (Johnson, 1986). Not all polar 8- or 16-cell blastomeres divide differentiatively, some cleaving along, rather than across, the axis of polarity so yielding two similar progeny both of which are polar and outside. Such a division is called conservative. The number of inside cells within an individual embryo will therefore be determined by the ratio of differentiative to conservative divisions by polarized cells at the 8- to 16-cell and 16- to 32-cell transitions. It has been shown in the intact embryo that once cells are deposited internally as a result of a differentiative division, they rarely if ever re-emerge to assume an outside position and fate (Pedersen et al. 1986; Fleming, 1987; Dyce et al. 1987).

The allocation of cells to the ICM lineage during cleavage is influenced by the division order of the cells, the progeny of early-dividing cells making a preferential contribution (Kelly, Mulnard & Graham, 1978; Surani & Barton, 1984). Such a preferential contribution could be achieved if early-dividing 1/8 or 1/16 polar cells tended to divide differentiatively (yielding one inside and one outside cell) rather than conservatively (yielding two outside cells). Since it is now known that an average 75 % of ICM cells is derived from the inside cell population formed at the 8- to 16-cell transition, we have examined whether those 1/8 blastomeres that are formed first within the embryo might divide differentiatively to yield inside cells more frequently than do later-forming 1/8 blastomeres. We have found (1) that earlier-formed blastomeres do indeed contribute cells preferentially to the inside cell population at the 16-cell stage and (2) that this preferential contribution arises because the cells that are formed early undergo intercellular flattening in advance of the other cells.

Female MFI mice (Central Animal Services, University of Cambridge) were superovulated by an injection of 5i.u. pregnant mares’ serum (PMS) followed 45 – 48 h later by 5i.u. human chorionic gonadotrophin (hCG). The females were paired overnight with HC-CFLP males (Hacking and Churchill Ltd). The presence of a vaginal plug the next morning indicated that a mating had occurred. Embryos were collected at the 2- to 4-cell stage by flushing oviducts 52 h post-hCG with prewarmed (37°C) medium 2 containing 4mg ml⁻¹ bovine serum albumin (M2+BSA; Fulton & Whittingham, 1978). Zona-intact embryos were cultured in pre-equilibrated medium 16 containing 4 mg ml⁻¹ bovine serum albumin (M16+BSA; Whittingham, 1971) in Falcon culture dishes, under oil at 37°C and 5 % CO2 in air. Sterilin culture dishes were used for culture of zona-free embryos and single cells.

Removal of the zona pellucida was achieved by exposure of embryos to prewarmed acid Tyrode’s solution for about 20 s (Nicolson, Yanagimachi & Yanagimachi, 1975). Embryos at the 4-cell stage were disaggregated by first decompacting them by incubation in prewanned (37°C) Ca²⁺-free M2 plus 6 mg ml⁻¹ BSA for 10 – 15 min and then pipetting them gently through a flame-polished micropipette. To disaggregate 16-cell embryos, the Ca²⁺-free M2+BSA was supplemented with a 1 in 10 dilution of 5 % (w/v) trypsin and 2% (w/v) EDTA (Gibco Ltd; Fleming, 1987).

Individual 4-cell blastomeres (1/4 cells) were cultured and examined every hour for evidence of division to 2/8 pairs. Any such pairs were harvested and stored for 0 – 11 h until they were used to make aggregate embryos. Four 2/8 pairs were decompacted by a 5 min incubation in Ca²⁺-free M2+BSA then exposed briefly to phytohaemagglutinin (PHA; Gibco Ltd) diluted 1:20 in M2+BSA to increase their adhesiveness. Aggregation was carried out as described in Kelly et al. (1978). First, the 2/8 pairs were arranged in two sets of four cells with each pair lying parallel to its neighbour so as to form a ‘square’ quartet. When these sets had adhered, one set was placed on top of the other but rotated through 45° so that all the eight cells in each aggregate embryo had similar positions relative to each other. Most aggregate embryos were cultured until the mid-16-cell stage before surface labelling and disaggregation. Some aggregate embryos were fixed during the 8-cell stage for histological analysis.

A stock solution of carboxylated fluorescent latex microparticles (yellow-green latex, 2 · 5 % solids, 0 · 06 μ m particle diameter; Fluoresbrite, Polysciences) was diluted 1:20 in M2+BSA. Intracellular labelling of endocytic organelles was achieved by incubating 2/8 cells in the label for 20 min, then washing in M2+BSA and culturing for 2h. Preliminary experiments confirmed that labelling with latex did not affect development (Fleming & George, 1986).

Surface labelling of exposed cells in 16-cell aggregate embryos was achieved by exposure to concanavalin A conjugated with tetramethylrhodamine isothiocyanate (TRITC-ConA, Polysciences) at Imgml⁻¹ M2+BSA for 1 min (Fleming & George, 1986). To prevent sticking during labelling, embryos were kept clear of the bottom of the culture dish.

Cells were examined only from those embryos where all 16 blastomeres had been recovered successfully. Thus embryos were discarded if they either contained arrested 1/8 cells or suffered some cell loss or lysis during disaggregation. Cells were fixed in 4 % formaldehyde in phosphatebuffered saline (PBS) for 20 min and stored in M2+BSA at 4°C. The fixed cells were transferred to drops of M2+BSA under oil in wells of a tissue-typing slide (Baird and Tatlock) and overlain with a cover slip.

Some aggregate 8-cell embryos were fixed, embedded in JB4 water-soluble resin and sectioned as described in Fleming & George (1986). Sections were viewed by differential interference and fluorescence microscopy.

Material was viewed on a Leitz Ortholux II microscope with filter sets N2 for rhodamine and L2 for FITC. Photomicrographs were taken on Kodak Tri-X film using a Leitz Vario-orthomat photographic system.

Pairs of newly formed 2/8 blastomeres were recovered at hourly intervals from a pool of isolated 1/4 blastomeres. Some of these were labelled with green fluorescent latex, whilst other pairs remained unlabelled. This procedure was continued for up to 7h. Aggregates were made of one labelled pair of cells with three unlabelled pairs of cells. In some aggregates the labelled pair was 4h postdivision and the unlabelled pairs were newly divided (labelled cells designated older), whilst in others the labelled pair was newly divided and the unlabelled pairs were 4h postdivision (labelled cells designated younger). The aggregate embryos were then cultured to the mid- to late-16-cell stage, at which point the embryos were incubated briefly in rhodamine-conjugated concanavalin A to label the exposed outside cells. The embryos were then disaggregated to groups of between one and four cells. The cells were examined for their red and green fluorescent staining patterns, as a result of which each cell was assigned to older or younger groups and to inside or outside positions (Fig. 1). From this information, it was possible to deduce (1) the ratio of inside to outside cells in each aggregate, and (2) the proportion of labelled 1/8 blastomeres that had divided differentiatively.

In both types of aggregate, the number of inside cells varied between 2 and 7 with mean values of 5 · 0 and 4 · 3 inside cells present in aggregates containing labelled older and younger cells, respectively. These figures are comparable with those reported previously for intact MFl-strain embryos (range 2 to 7; mean 5 · 2, Fleming, 1987), suggesting that the aggregate embryos approximate to normal developmental behaviour. The distribution of inside cell numbers in the two groups of aggregates did not differ significantly (Mann-Whitney U-test, P>0 · 05). Since the two populations of embryos were comparable with respect to their patterns of inside: outside cell ratios, they were next compared for the relative allocation of labelled cells to the inside population (Table 1, line 1). It is evident that the older cells contributed relatively more cells to the inside than did younger cells as a result of a significantly greater proportion of differentiative divisions.

Observation of aggregated embryos for the first few hours of their period in culture revealed that the cluster of older cells flattened in advance of the cluster of younger cells (Fig. 2A,B). We examined whether this early flattening might be responsible for the preferential allocation of cells to the inside cell population.

Pairs of newly formed 2/8 blastomeres were harvested and some were labelled, exactly as described above. However, the aggregate embryos were not constructed immediately. Rather, the 2/8 pairs were cultured individually for either 7h (younger) or 11 h (older) by which time intercellular flattening had occurred between the two component cells of each pair, regardless of their relative age. One labelled (older or younger) and three unlabelled (younger or older respectively) 2/8 pairs were then decompacted by brief exposure to medium low in calcium and then aggregated together as above. The aggregates were placed in culture, where all cells were observed to flatten on each other simultaneously. The compacted aggregate embryos were cultured to the mid- to late- 16-cell stage and then analysed as described above. The distributions of inside: outside cell ratios in embryos with labelled older and younger cells did not differ significantly (labelled older aggregates: range 2 to 6, mean 4-0; labelled younger aggregates: range 3 to 6, mean 41; not significantly different by Mann-Whitney U-test, P>0 · 05). The two groups of embryos were therefore compared for the relative allocation of labelled cells to the inside population (Table 1, line 2). The difference between the two types of embryo was not significant.

Aggregate embryos were made in which a labelled 2/8 pair was combined with three unlabelled pairs in reciprocal combinations as described in section Al. Embryos were harvested 7h after aggregation, by which time the 8-cell aggregates were fully compacted. Embryos were fixed and sectioned serially, and the relationship between the labelled and unlabelled cells was determined. Fig. 2C-F shows rep-resentative patterns. In all five aggregates examined in which the labelled cells were older, labelled cells were found deep within the compacted embryo (Fig. 2C,D), whereas this pattern was observed for only two labelled cells in seven aggregates in which the labelled cells were younger, all other labelled cells being located superficially (Fig. 2E,F).

The principle underlying these experiments is very simple. If cells enter a particular developmental cell cycle at different times and if cells acquire a particular property at a certain point in that cell cycle, then the early-formed cells will acquire that property in advance of cells formed later. Temporal heterogeneity will lead to physical heterogeneity. If the cellcycle-linked property affects cell shape or interaction, the physical differences could also generate spatial heterogeneity. Thus, differences in cell division order could generate spatial pattern within the embryo. Such a mechanism may operate to influence the establishment of the embryonic: abembryonic axis of the mouse embryo (Garbutt, Chisholm & Johnson, 1987). Here we examine whether and how the same principle might operate to set up the earlier radial (inside: outside) axis in the mouse embryo.

Previous experiments have suggested that such a mechanism might be operating to influence cell allocation in the mouse blastocyst, early-dividing cells contributing a disproportionate number of progeny to the ICM (Kelly et al. 1978; Surani & Barton, 1984).

Since the allocation of cells to an inside position and thus to the ICM lineage is achieved by the differentiative division of outer polar 1/8 and 1/16 blastomeres, and since it is dividing 1/8 cells that provide the majority of the inside cells that go to form the ICM, we examined whether those 1/8 cells that formed first cleaved differentiatively more frequently. We aggregated four pairs of 8-cell blastomeres differing in age by 4h, a difference within the age range of cells in nonmanipulated embryos (Kelly et al. 1978; Smith & Johnson, 1986). Our results show that 16-cell embryos derived from such aggregates have a similar range and mean number of inside cells to nonmanipulated embryos (Fleming, 1987). The results also support the suggestion that 8-cell blastomeres only contribute cells to the inside by a process of differentiative division, since the labelled pair of 2/8 cells was observed to contribute a maximum of two and a minimum of zero inner cells. Our results also reveal a clear advantage of early-formed cells over cells formed later in their relative contribution to the inside cell lineage.

The inside cells that are generated at the 8- to 16- cell transition give rise to about 75 % of the ICM cells on average (Fleming, 1987). It seems likely that the second smaller allocation of cells to the embryo interior that can occur at the subsequent cell division shows a similar preferential contribution from the early-dividing cells, since a comparison between our results on the proportion of the inside cells at the 16- cell stage generated from early-dividing cells with the results reported by Kelly et al. (1978) for the equivalent proportion of cells in the ICM of the blastocyst reveals that they do not differ significantly (Mann-Whitney U-test, P>0 · 05). We must ask what property of early-dividing cells results in their divisions being differentiative, to yield one inside and one outside cell, rather than conservative, to yield two outside cells.

This question was considered by Graham and his colleagues, who showed that early-dividing cells tended to establish more contacts with other cells and to lie deeper within the embryo (Graham & Deussen, 1978; Graham & Lehtonen, 1979). This observation was confirmed in the histological study reported here. Graham and his colleagues therefore suggested that some property associated with cell surface adhesion might be important in determining the preferential internal location of early-dividing cells and their progeny. Indeed, it has been demonstrated that intercellular adhesive properties do change over a defined period during the fourth cell cycle such that after about 7h into the cycle all cells are flattened maximally upon each other (Lehtonen, 1980; Ziomek & Johnson, 1980). The asynchrony of flattening in aggregates of early- and late-dividing cells observed here (Fig. 2A,B) confirms that this cell cycle dependence of the flattening process is indeed cell autonomous and related to the time of entry into the fourth cell cycle. However, a number of other cellular and intercellular features also develop at defined periods during the fourth cell cycle, notably the appearance of functional gap junctions (Goodall & Johnson, 1982, 1984) and the polarization of the cytoskeleton, cytoplasmic organelles and surface of individual blastomeres (reviewed in Johnson & Maro, 1986). Moreover, the changes in cell adhesion that occur during the fourth cell cycle influence the changes in junctional communication and polarization (Goodall, 1986; Johnson, Maro & Takeichi, 1986). The problem is therefore to dissect out which of the various features expressed in a cell-cycle-dependent manner are important for preferential cell allocation.

Exposure to medium low in calcium is a procedure that readily and reversibly inhibits intercellular flattening but does not reverse either gap junctional formation (Goodall, 1986) or the various features characterizing the polarized state (Handyside, 1980; Johnson & Maro, 1984; Maro, Johnson, Pickering & Louvard, 1985). We therefore took 2/8 pairs of cells, all of which were in the latter (post-7 h) half of the fourth cell cycle and all of which had therefore completed intercellular flattening. We reversed the flattening of these cells upon each other by exposure to medium low in calcium, aggregated four pairs together to make asynchronous aggregates and then allowed cells to reflatten synchronously upon each other, thereby removing any flattening advantage of the older cells (as confirmed visually). As a result, the older cells no longer contributed preferentially to the inside cell population at the 16-cell stage. These results therefore support the hypothesis that the preferential contribution by early-dividing cells to the ICM arises from the fact that they are also early-flattening cells (Graham & Lehtonen, 1979). The clear implication is that intercellular flattening influences the orientation of the cleavage plane.

Cell interaction is known to influence the orientation of the cleavage plane in polar 1/16 blastomeres (Johnson & Ziomek, 1983), and recent evidence suggests that this property is used in the embryo to regulate the ratio of inside to outside cells at the 16- to 32-cell transition (Fleming, 1987). In contrast, a comparison of the frequency of differentiative divisions in polar 1/8 cells dividing in isolation with that in polar 1/8 cells dividing in a 2/8 pair does not reveal any very striking difference (unpublished observations by S. J. Pickering, B. Maro, M. H. Johnson & J. Skepper). It seems likely therefore that the clear effect of order of flattening on the incidence of differentiative divisions reported here requires multiple cell packing rather than simple contact per se.

There are two types of mechanism whereby an intercellular interaction might influence cleavage planes. Cell interaction might lead to a ‘marking’ of the internal face of the cytocortex adjacent to the contact point and this marked area could then act as a focus to orient either a pole or the edge of the equatorial plate of the developing spindle (Fig. 3B; Gunning, 1982; Palevitz, 1986). Alternatively, cell flattening could influence cell shape and shape changes could secondarily determine the long axis of the spindle with respect to the polar axis of the cell, thereby effectively determining the orientation of the cleavage plane (Fig. 3C; Meshcheryakov, 1978; Freeman, 1983). In the 8-cell mouse embryo, it seems unlikely that intercellular contact influences directly the orientation in which the nascent spindle is set up (Fig. 3B). Rather, the effect of cell packing on cell shape forces a higher differentiative division rate in early flattening cells (Fig. 3C). It seems likely that such an influence of packing on cell shape is also responsible for regulating cleavage plane orientation at the 16- to 32-cell transition (Johnson & Ziomek, 1983).

We wish to acknowledge the support and stimulation of Tom Fleming, Gin Flach, Brendan Doe and Simon Hanna. This work was supported by grants from Trinity College, Cambridge and the H. E. Durham Fund of Kings’ College, Cambridge to C.L.G., and by grants from the Medical Research Council and the Cancer Research Campaign to M.H.J.