We have attempted to reduce the developmental heterogeneity amongst populations of mouse blastocysts by synchronizing embryos to the first visible signs of blastocoel formation. Using embryos timed in this way, we have examined the extent of variation of inside and outside cell number and of inside cell size, nuclear DNA content and developmental potential, between and within embryos of a similar age postcavitation.
The overall impression gained is one of wide heterogeneity in inside: outside cell number ratios and in cell cycling and its relation to cavitation among embryos of similar age postcavitation. However, the simplest explanation of our results suggests that cavitation generally begins at a time when most outside cells are in their sixth developmental cell cycle and that outside cells, as a population, are a little ahead of inside cells in their cell cycling. Additionally we present evidence that, within at least some individual inner cell masses (ICM), there is intraembryo variation in the time at which inside cell developmental potential becomes restricted.
The developmental fate of cells in the early mouse embryo is influenced both by their inherited phenotypic characteristics and by relative cell position (reviewed Johnson, 1985a). Phenotypic and positional differences among cells are first evident in the 16-cell morula which contains distinct subpopulations of outer, polar and inner, apolar cells (Barlow, Owen & Graham, 1972; Handyside, 1981; Reeve & Ziomek, 1981; Johnson & Ziomek, 1981; Ziomek & Johnson, 1981; Kimber, Surani & Barton, 1982; Randle, 1982; MacQueen & Johnson, 1983). Cells occupying an inside position in the morula contribute substantially to the ICM, and outside cells to the trophectoderm (TE), of the early blastocyst (Tarkowski & Wroblewska, 1967; Hillman, Sherman & Graham, 1972; Wilson, Bolton & Cutler, 1972; Kelly, Mulnard & Graham, 1978; Ziomek & Johnson, 1982; Balakier & Pedersen, 1982; Surani & Handyside, 1983). However, these two sets of lineage relationships are not fixed, and cellular traffic between lineages can and does occur, the extent of the traffic being regulated by simple and definable cellular mechanisms (see Surani & Barton, 1984; and Johnson, 1985a,b for discussion). Whilst the mechanisms by which regulation of lineage can operate have become clearer, as a result of experimental manipulation and relocation of individual cells or cell clusters, the extent to which such regulation occurs in situ is less well understood. This deficiency arises in part from the absence of suitable lineage markers for non-invasive labelling of single cells in situ and partly because of the wide variation in the estimates of numbers of inside and outside cells within individual embryos of 16-cells or more. Analysis of 16-cell embryos for the number of inside cells has provided estimates as diverse as rarely more than two (Barlow et al. 1972; Graham & Lehtonen, 1979) and rarely less than four but never more than eight (Handyside, 1981; Johnson & Ziomek, 1981; Reeve, 1981; Balakier & Pedersen, 1982; Surani & Barton, 1984). Analysis of embryos with approximately 32-cells has provided estimates of on average 14 inside cells (Handyside, 1978), 6–7 inside cells (Barlow et al. 1972; Randle, 1982) and 13–16 inside cells (Copp, 1979; Surani & Barton, 1984). One problem with these various studies is that different techniques were used to evaluate cell position, and different strains of mice and relatively crude methods of embryonic staging were employed at a time when there is considerable developmental heterogeneity between embryos (Smith & McLaren, 1977). In this study we (i) examine the extent of mierembryo variability in cell number at the late morula–early blastocyst stage, (ii) apply four techniques for evaluating the variability of inside and outside cell numbers at relatively homogeneous stages of blastocyst formation in individual embryos from two mouse embryo strains, (iii) make estimates of the size heterogeneity within the populations of inside cells at some of the stages analysed, and (iv) examine variability in the DNA content and developmental potential of cells from individual ICMs taken from relatively homogeneous populations of blastocysts.
MATERIALS AND METHODS
Embryo recovery and staging
Embryos were derived from superovulated 3-to 5-week-old HC-CFLP or MF1 females (Hacking & Churchill Ltd or Olac Ltd; 5 i.u. PMS followed after 45–50 h by 5 i.u. hCG, Intervet, and mating with HC–CFLP males). Embryos were recovered as 8– to 16–cell stages at 62–72 h post hCG by flushing oviducts with phosphate-buffered medium 1 or Hepes-buffered medium 2 containing 4 mg-ml−1 bovine serum albumin (PB1+BSA or M2+BSA; Whittingham & Wales, 1969; Fulton & Whittingham, 1978). Embryos were cultured in individual drops of medium 16 plus 4 mg-ml−1 BSA (M16+BSA; whittingham, 1971) under oil in Sterilin culture dishes at 37 °C and 5 % CO2 in air and were scored individually for the earliest signs of blastocoel formation, every 60 or 90 mins between 72 and 100 h post hCG on a Wild inverted phase-contrast microscope at × 60 or × 100. Nascent blastocysts that had formed since the previous inspection were pooled and designated as 0 h old. Pooled groups of blastocysts aged 0 to 12 h were used for the analyses.
In some experiments, the extent of cavitation of each embryo was noted during each hourly scoring. Five categories were used: Stage 1, uncavitated morula; 2, blastocoel occupying less than half volume of embryo; 3, blastocoel occupying about half volume of embryo; 4, blastocoel occupying greater than half volume; 5, well-expanded blastocyst (see Fig. 2).
(i) Removal of the zona pellucida was achieved by brief exposure to prewarmed acid Tyrode’s solution, pH 2·5, +4mg-ml−1 polyvinylpyrrolidone (Nicolson, Yanagimachi & Yanagimachi, 1975).
(ii) Immunosurgery was carried out as described by Handyside (1978). Rabbit antimouse species antiserum diluted 1/10 with M2 was prewarmed at 37 °C prior to incubation of early blastocysts (3 min) or expanded blastocysts (10 min). Embryos were washed through three large volumes of M24-BSA before exposure to guinea-pig complement (Difco) diluted 1:3 with phosphate-buffered saline, absorbed with agarose (Cohen & Schlesinger, 1970), stored at −70 °C and thawed immediately prior to use. After 5–7 min embryos were transferred to fresh M2+BSA and left for a further 20–30 min at 37 °C before removal of outer cells by a flame-polished micropipette with an internal diameter slightly larger than the ICM.
(iii) Decompaction of ICMs was achieved either by exposure to Ca2+-free M16 containing 6 mg-ml−1 BSA for 20 min (Pratt, Ziomek, Reeve & Johnson, 1982) or M16+BSA containing cytochalasin D (CCD; Sigma) at 0·5 μg-ml−1 for 20 min.
In vitro culture of ICMs
ICMs or ICM aggregates were cultured individually in 25 μl drops of RPMI 1640 (Flow) supplemented with 10 % foetal calf serum (under oil, at 37 °C in 5 % CO2 in air). Falcon plastic tissue culture dishes were used. ICM cultures were examined using phase contrast optics on a Wild inverted microscope.
Cell counting techniques
Four techniques were used to evaluate numbers of cells at each stage.
(i) Giemsa staining (Tarkowski, 1966). Whole embryos (with or without the zona pellucida intact), or inner cell masses (ICM) from them, were exposed to 0·9 % sodium citrate for 10 or 1 min respectively before transfer to acetone-cleaned slides. Fixation was achieved with a drop of ethanol:acetic acid (3:1) and air drying. Giemsa stain (5 %) was millipored immediately before use and samples allowed to stain for 10 min prior to counting of nuclei. Mitotic figures were counted as one cell. This approach allows direct measurement of inside cell numbers and total cell numbers. Average outside cell numbers can be obtained by subtraction. Some ICMs were spread and fixed as if for Giemsa staining, but stained by the Feulgen reaction for microdensitometry (see below).
(ii) DAPI staining (4, 6-diamidino-2-phenylindole; Reeve & Kelly, 1983). Zona-intact embryos were subjected to immunosurgical procedures up to and including exposure to complement, and then exposed to M16+BSA+DAPI (100 μg-ml−1) for 30 min at 37 °C to stain nuclei of dead trophectoderm and live ICM cells which were retained within the zona. Embryos were then rinsed and transferred individually to separate 10 μl drops of PB1+BSA on a multiwell slide (Baird & Tatlock). The zona pellucida was then removed and the ICM shelled out with a flame-polished micropipette. The dead trophectoderm cells were allowed to dry down on the slide, while the ICM was transferred to a 10 μl drop of 0·9 % Na citrate for 15 min before being disaggregated into small cell clumps and air dried. Dried samples were fixed with 70 % ethanol prior to counting of the fluorescent nuclei on a Zeiss universal microscope equipped with incident source HBO 200 and filter set 48 77 05.
(iii) Disaggregation of inside cell clusters. After immunosurgical isolation, clusters of inside cells were decompacted as described earlier and disaggregated in a small individual drop of decompacting medium. Cells were then counted. This technique allows direct enumeration of inside cells only.
(iv) Fixation and serial section. Intact embryos, within the zona pellucida, were processed for transmission electron microscopy (see below) and embedded in capsules in groups of approximately six in Spurr’s resin. Serial thick sections were cut (2 μm thickness), mounted on slides and stained with 1 % toluidine blue/borax. Samples were viewed on a Zeiss universal microscope with a drawing tube attachment and outlines of cells and nuclei recorded; the following criteria were useful in identifying cells as belonging to the inside or outside, namely cytosolic density, distribution of secondary lysosomes, and cell shape (Fleming, Warren, Chisholm & Johnson, 1984). Numbers of inside, outside and total cells were recorded from three-dimensional reconstructions.
Scanning electron microscopy
The procedure of Johnson & Ziomek (1982) was used. All reagents were filtered immediately prior to use. ICMs, or cells therefrom, were fixed in 6 % glutaraldehyde in O-lM-cacodylate buffer (pH7-3; osmolality 730 mOsmol) for 1 h at room temperature, stored in 0·1M-cacodylate for a maximum of 24 h before transfer in 1 % glutaraldehyde to clean glass coverslips that had been coated with poly-L-lysine (Sigma type IB; 1 mg-ml−1 fresh solution) for 15 min followed by two washes in 0·1M-cacodylate buffer and immersion in buffer in wells of a 24-well Linbro (Flow) tissue-culture dish. After dehydration through alcohols (30 min in 30 % and 50 %; overnight in 70 %; 15 min in 80 %, 90 %, 95 % and 100 %) samples were dried from 100 % ethanol via CO2 in a Polaron E3000 critical-point drying apparatus. Coverslips were mounted on stubs with double-sided Sellotape, coated with a 50 μm layer of gold in a Polaron E5000 Diode sputtering system and viewed on a Cambridge Stereoscan 600 electron microscope. The diameter of individual blastomeres was assessed by averaging the greatest and least diameters measured from photographs of non-tilted samples. Grossly distorted blastomeres were not measured. Volumes were calculated from the mean diameters on the assumption that each blastomere was spherical.
Transmission electron microscopy
ICMs or ICM aggregates (after 48 or 72 h of culture in vitro) were fixed in 3 % glutaraldehyde in 0·1 M-cacodylate buffer for 30 min. After washing in cacodylate buffer they were postfixed in similarly buffered 1 % osmium tetroxide (30min), washed in distilled water, dehydrated in a graded ethanol series, and embedded in Spurr’s resin. Ultrathin sections were cut on an LKB U1 tro tome III, mounted on copper grids, stained with uranyl acetate and lead citrate and viewed using an AEI6B electron microscope at 80 kV.
Phase-contrast photographs were taken on 35 mm Ilford Pan F film.
1. Staging of blastocysts
Over 90 % of 8-to 16-cell embryos, that were flushed between 62 and 72 h post hCG and cultured overnight, proceeded to cavitation. Typically about 60 % of HC-CFLP-derived embryos cavitated between 83 h and 93 h post hCG whereas about 50 % of MF1 derived embryos cavitated between 90 and 98 h post hCG (Figs. 1A,B). There is therefore a variation in excess of 10 h in the time at which cavitation is initiated within each strain, a result confirming that reported by Smith & McLaren (1977) for naturally ovulating mice, and there is also variation between strains in the absolute time of these ranges. In addition hourly scoring of the extent of cavitation of blastocysts showed that the rate of blastocoel expansion varied between embryos (see Fig. 2B). Cavitation was detected in 0 h and 1 h blastocysts as a small, oval translucent region located at the abembryonic pole and bound distally by attenuated processes of mural TE and occasionally split into two cavities within a single blastocyst (Fig. 2A). By 11 to 12 h, blastocysts were well expanded, with the ICM appearing as a disc-shaped cell cluster at the embryonic pole (Fig. 2C). At intermediate times the extent of blastocoel expansion varied among blastocysts within each age group (Fig. 2B). In a few embryos, the cavity collapsed after its initial formation and these embryos were excluded from analysis.
2. Interembryo variation in relation to time post hCG and time post cavitation
In an effort to reduce interembryo variation we examined the value of synchronizing embryos obtained from superovulated mice to the earliest signs of cavitation as a means of reducing developmental heterogeneity between embryos (cf. results from Smith & McLaren who used naturally mated mice for the same purpose, 1977). MF1 embryos were recovered at 68h post hCG as 8-to 16-cell morulae, randomized into four groups and cultured until about 20–25% of the total embryo population had undergone cavitation. Each group of embryos was then sorted into subgroups containing those with and those without a blastocoel (blastocysts versus morulae, Table 1). After 90 min the morula subgroups were examined for newly cavitated embryos. Newly cavitated embryos, morulae and blastocysts in group 1 were fixed for counting cells by Giemsa staining. In the remaining groups, newly cavitated embryos were transferred to their blastocyst subgroups. At consecutive 90 min intervals thereafter the same procedure was repeated, so that groups 2–4 were fixed serially as morula, newly cavitated embryo and blastocyst subgroups. Mean cell numbers for each embryo subgroup are given in Table 1. Although overall mean cell number increased with time post hCG, this increase was not distributed evenly among different subgroups of embryos. Mean cell numbers in the morula and newly cavitated subgroups were remarkably constant over the period 91·5 to 96 h post hCG, whereas the substantial increase in blastocyst cell numbers was sufficient to account for the overall rise in mean cell number in the total embryo population (Table 1). The range in cell number within each subgroup was large, but frequency distributions demonstrated that 50–60 % of embryos in the morula subgroups and 80 % of all newly cavitated embryos had between 28 and 35 cells inclusive (data not shown). By contrast, embryos in the blastocyst subgroups showed a fairly even distribution over a broad range of cell number, never having fewer than 24 cells and rarely less than 30. We have therefore used the earliest signs of cavitation, rather than hours post hCG, as a synchronizing device in all subsequent experiments. Embryos (or ICMs from them) are hence described as being aged 0 to 12 h post cavitation.
3. Number and distribution of cells from different stage embryos
The mean values for total cell numbers and for inside and outside cell numbers were obtained from a total of 18 experiments using HC-CFLP mice and are recorded in Table 2. Values for individual Oh blastocysts analysed by the two techniques that permit evaluation of both inside and outside numbers in the same embryo (DAPI staining and serial sections) are recorded in Table 3.
4. Size distribution of inside cells from different stage embryos
The diameters of individual inside cells were estimated in two types of preparation. In one experiment, intact CCD-decompacted clusters of inside cells were examined on the SEM, and the diameters of any exposed cells were measured. The distribution of diameters is plotted in Fig. 3A,B. In a second experiment, single cells derived by decompaction and disaggregation of inside cell clusters were examined on the SEM and the diameters of all non-distorted cells were measured. The distribution of diameters that were recorded is plotted in Fig. 3C,D,E. In neither technique were all cells measured, and it cannot be assumed that losses were random. The distribution of values along each horizontal axis is therefore more useful than the relative heights of peaks as measured on the vertical axis. Also, there is considerable shrinkage of cells during preparation that varies with each set of experimental runs, and so values are expressed in arbitrary linear units (lu).
From Fig. 3, it can be seen that there is a reduction in cell size (assuming cell shape to be spherical) in older ICMs. At 0 and 2 h, diameters are distributed almost unimodally over a range equivalent to volumes of 300–800 lu3 (median value = 500), with evidence in Figs. 3A and C of a second cluster of larger volumes ranging from 1000-1900lu3 (median value = 1200lu3). At 12h, volumes are distributed over the range 100–300 lu3 (median value = 200), with evidence of a second cluster over the range 350–500 lu3 (median value = 450).
5. DNA content of cells from individual ICMs
ICMs were isolated from MF1 blastocysts at hourly intervals from 0 to 12 h after the earliest signs of cavitation (60 min inspection regime) for more detailed analysis of cell numbers and DNA content. Number estimates were made by disaggregation and by both Giemsa and Feulgen staining and results are shown in Table 4. Considerable agreement was found between counts made by all techniques at each time point.
All three techniques gave a mean ICM cell number of 11–12 at 1 h postcavitation. Between 1 h and 3 h postcavitation mean cell number increased by 2–4 whereas it was roughly constant at around 14-16 cells over the period 3–5 h postcavitation (by disaggregation) and 3–7 h postcavitation (by Giemsa and Feulgen staining). By 12 h postcavitation ICMs contained about 20 cells on average. However, the range in cell number among ICMs within each age group was extremely large (Table 4).
Mitotic figures were recorded in stained preparations and mitotic indices (percentage of cells in mitosis) were calculated for each ICM age group (Table 4). In general, mitotic indices were low and individual ICMs rarely showed more than one or two mitotic figures.
In ICMs spread by the Tarkowski method, the majority of cells could be assayed for nuclear DNA content. However, every nucleus could be assayed in very few ICMs due to insufficient nuclear separation. Standard 2C and 4C DNA values were obtained from the liver smear in each experiment. Results showed that in most ICMs, the DNA content of individual nuclei ranged from 2C to 4C, including intermediate values. Since all phases of the cell cycle are present from the 2-cell stage onwards in the mouse embryo (Barlow et al. 1972), each ICM probably contains cells in Gi, S and G2 phases of the cell cycle. Thus cells within each ICM were cycling asynchronously.
6. Developmental potential of ICMs
In this study, ICMs were cultured singly and were examined at 24 h, 48 h and 72 h after immunosurgical isolation for evidence of trophectoderm or extraembryonic endoderm-like features. Uncommitted ICM cells which differentiate into trophectoderm in vitro are expected to show signs of fluid accumulation and formation of a blastocyst-like vesicle within 24–48 h of the start of culture (Handyside, 1978), whereas endoderm formation should only become apparent by 48–72 h. Thus, both in vitro morphology and the timing of morphological changes may help in assessing the developmental potential of ICMs. Results are shown in Table 5 and Fig. 4.
The majority of ICMs scored as showing trophectodermal development formed blastocyst-like vesicles (Fig. 5A) although some gave ‘non-integrated’ forms (Fig. 5B; Handyside, 1978) or showed only a few vesiculating cells. Trophectoderm formation was usually visible by 24 or 48 h after isolation although it was first observed at 72 h in small number of ICMs. Outgrowth of trophoblast giant cells by vesiculating ICMs was not observed during the 72 h culture period. The majority of ICMs in each ICM age group exhibited trophectodermal differentiation by cells occupying an outside position, although the proportion of ICMs which generated trophectoderm fell with increasing blastocyst age (Fig. 4). Nevertheless 69% of viable ICMs from 12 h blastocysts generated some trophectodermal cells on isolation and only 31 % generated endoderm alone.
Amongst those ICMs classified at the light microscope level as showing endodermal differentiation there was considerable variability of phenotype. These forms fell broadly into four categories: a) structures showing an outer “rind” of cells surrounding an inner solid core, with variable degrees of intra- and extracellular fluid accumulation (endodermal vesicles; Fig. 5C); b) structures lacking a clear outer rind of cells, but nevertheless showing extracellular fluid accumulation which was judged to be endodermal rather than trophectodermal in nature both by morphology and the time (48–72 h) of fluid appearance (endodermal fluid accumulation, Fig. 5D). These ICMs occasionally gave rise to endodermal vesicles if culture was extended to 5–6 days; c) solid embryoid structures, usually showing cell rounding or proliferation, and frequently attached to the culture dish by cells which outgrew within 24 or 48 h of isolation (Fig. 5E); d) a few ICMs spread out to form a monolayer of cells, usually within 24 or 48 h of isolation (Fig. 5F). It is unclear whether this form of development represented trophectodermal or endodermal differentiation. Occasional cells in the outgrowths showed enlarged nuclei (Fig. 5F) although Solter & Knowles (1975) report outgrowth of polygonal endoderm-like cells from isolated ICMs. In this study the very few ICMs which formed monolayer outgrowths were scored as showing endodermal development (Table 5).
A small proportion of ICMs appeared to show evidence of mixed differentiation (at the light microscope level), developing both trophectoderm and endoderm-like cells in outside positions (Fig. 5G). Generally, in these ICMs vesiculating trophec-todermal cells were first noted at 48 h or later postisolation and by 72 h solid endodermal or occasionally endodermal vesicle structures were also present. These mixed character structures were observed to develop from ICMs at any age between 2 and 12 h (with the exception of 11 h ICMs; Table 5). We confirmed this result in a small number of aggregates made by pairing an early (2 h) ICM with that of a late (12 h) ICM in which cells are likely to be predominantly uncommitted and committed respectively to the ICM lineage and might therefore generate a mixed population of trophectoderm and endoderm cells in outside positions in the aggregate.
To check the proposed classification of ICM development made by morphological observations at the light microscope (LM) level, some ICMs and ICM aggregates were scored by LM at 24 h intervals, fixed after 48 h or 72 h in culture and processed for transmission electron microscopy (TEM). Trophectodermal and endodermal cell types in cultured ICMs are known to resemble the trophectoderm and extraembryonic endoderm cell types respectively of intact embryos in their ultrastructural features (Hogan & Tilly, 1978) and are thus clearly distinguishable from one another. Extraembryonic endodermal cells may further be classified as either visceral or parietal by their ultrastructure (Hogan & Tilly, 1981). Electron micrographs of typical trophectoderm and endoderm cells generated by cultured ICMs are shown in Fig 6.
A comparison between LM and TEM classification of ICM development is shown in Table 6. While it is clear that there is reasonably good agreement between LM and TEM classification, there is also some discrepancy in the results. From Table 6 it is apparent that: 1) all ICMs classified by LM as showing only trophectoderm cells without endoderm gave evidence of some trophectoderm at the TEM level, although it was noted that by 72 h many vesiculating structures had collapsed on themselves, with subsequent cell rounding and LM morphology not dissimilar from endoderm; 2) endodermal vesicles showing an outer layer of cells surrounding a solid inner core always possessed an outer layer of endoderm cells (mostly visceral endoderm-like) as expected; 3) evidence was obtained in two ICMs for both trophectoderm and endoderm cells occupying outside positions (Fig. 7), although this did not correlate well with LM classification of mixed-type structures. Both of the ICMs which were judged to be mixed in structure by LM but by TEM showed only endoderm-like cells (Table 6, line 3), contained highly attenuated parietal endoderm-like cells enclosing a cavity. These vesiculated endoderm cells may have been mistakenly identified as late-vesiculating trophectoderm. TEM evidence of mixed development was also obtained for aggregated pairs of ICMs from early and expanded blastocysts (Fig. 8).
Interpretation of these results must take into account the fact that a section of an ICM at any one level allows identification of cell type for only a proportion of cells of that ICM. In this study, ICMs were sectioned at 5 μm and 8 μm for examination, but it remains impossible to examine every cell unless serial sections are made. Nevertheless, from these results we conclude that overt appearance of trophectoderm in LM is a good indicator of the actual presence of trophectoderm cells although they may also be present in situations where there is no vesiculation or the cavity has collapsed and they will therefore not be seen by LM. Thus it is likely that the proportion of ICMs of each age generating trophectoderm is slightly higher and that generating endoderm slightly lower than Table 5 and Fig. 4 suggest. A more rigorous test for the presence of trophectoderm in ICMs would be giant cell outgrowth in vitro (Hogan & Tilly, 1978) or the initiation of a decidual response when replaced in utero (Rossant & Lis, 1979). Additionally, we have shown that both trophectodermal and extraembryonic endodermal cell types can coexist in outside positions in an isolated ICM and in ICM aggregates, although this may occur less frequently than expected from LM classification.
We have confirmed previous observations (McLaren & Bowman, 1973; Smith & McLaren, 1977; Copp, 1979; Rossant & Lis, 1979) that the developmental heterogeneity among expanding blastocysts is considerable, and show that it is in excess of 10 h and probably of the order 15 h. Most of this heterogeneity can be accounted for by the interembryo developmental variability evident at the 8-cell stage, with only a small component of variability being introduced later. Thus, there may be as much as 10 h asynchrony among embryos at first cleavage, of which 2–3 h results from an inherent variability in the length of the first cell cycle and the remainder from differences in the timing of ovulation and/or fertilization of oocytes in vivo (Bolton et al. 1984; Howlett & Bolton, 1985). During the second cell cycle, lasting about 18 h, a further increase in asynchrony of about 2h is introduced among the population of 2-cell blastomeres, with a further increase of about an hour by the 4–8 cell transition (Smith & Johnson, 1985). Indeed, our results suggest that between the generation of 8-cell blastomeres and cavitation (around 32 cells) developmental heterogeneity increases by only 2–3 h further in a total period of around 30h, i.e. by about 1 h per cell cycle. This finding that heterogeneity increases relatively slowly is in accordance with the observation that mean cell cycle time between the 4- and 16-cell stages is independent of the length of the cycle in the progenitor cells at the 2-cell stage, showing that a short second cell cycle is not apparently an heritable feature (Kelly et al. 1978). We attempted to reduce heterogeneity for experimental purposes by examining embryos at regular intervals, and collecting those that had commenced cavitation. This approach should synchronize embryos to the time, within the limits of the inspection interval, at which at least some of their constituent outer cells form an effective zonular tight junctional complex and engage in vectorial transfer of fluid (Ansell & Snow, 1974). Although such an approach might reduce heterogeneity between embryos, it will not eliminate it, since within embryos (a) there is likely to be heterogeneity within the population of outside cells and some outside cells may transport and retain fluid earlier than others (see Surani & Barton, 1984; Fleming et al. 1984), (b) the rate of inside cell development, presumably also itself heterogeneous, may not necessarily be coupled closely to the rate of outside cell development, and (c) cells can cross lineages and may therefore contribute to fluctuating cell ratios if not to total cell numbers. It is probable that the variation in total cell number of the embryos found at any given age postcavitation arises partly for these reasons, although a varying incidence of cell death (Copp, 1979) and loss of cells during manipulation will also account for some variation. Overall, the impression obtained from Tables 2, 3 and 4 is of very considerable variation in the number and distribution of cells within an embryo even when these embryos are synchronized to the same developmental event. Since we do not know whether all embryos are equally viable, regardless of their cellular constitution, it is impossible to assess whether the extremely atypical embryos represent failures of an attempted regulation towards a norm condition that the majority of embryos have achieved successfully (Johnson & Ziomek, 1983; Surani & Barton, 1984).
With these considerations in mind, and despite the considerable inter-embryo variation, the agreement in Table 2 of mean values arrived at independently by the four different techniques used is remarkably good. It is evident that at each point the highest total cell number was always recorded in the serially sectioned zona-enclosed embryos. This result indicates that use of the other three manipulative techniques, however carefully performed, is likely to result in some cell losses. Comparison of the four techniques for inside and outside mean cell numbers hints that the loss might be slightly greater for the former than the latter. The only major discrepancy in Table 2 is within the 12 h blastocyst group, in which the mean number of inside cells is low by DAPI staining and the total (and therefore also outside) cell number is low by Giemsa staining. Undercounting of DAPI-stained inside cells at later stages could be due to the greater difficulty in disaggregation of decompacted 12 h ICMs to single cells. Fluorescent nuclei that overlap are difficult to distinguish. Similarly, the larger total number of nuclei in Giemsa-stained preparations of whole blastocysts could easily lead to underestimates due to an increased chance of superimposition. Additionally, it is possible that the cells undergo mitosis at around 12 h postcavitation, in which case slight interexperiment variations in timing might give large differences in the results observed.
The 0 h and 2 h groups (Table 2) both contain embryos with a mean value of just under 32 cells distributed in an inside:outside ratio of just less than 1:2. These mean values, and the variation about them, are similar to those reported by Smith & McLaren (1977) for nascent blastocysts and agree fairly well with our counts of 0 h total cell numbers in MF1 embryos (Table 1). The distribution of inside cell volumes at these stages is almost unimodal, with some evidence of a minor second peak of cells of approximately double the volume. Taken together, these data suggest that most ICMs at 0 h and 2 h contain cells that are in their sixth developmental cell cycle, with some inside cells in at least some ICMs still in their fifth cell cycle. Presumably those cells in the fifth cell cycle are more likely to be in ICMs derived from blastocysts with less than 32 cells. As is shown in Table 3, most such ICMs contain 7 – 10 cells, and this range shifts to 9 – 14 by the 32-cell stage. As reported in the Introduction, estimates of inside cell numbers at the 16-cell stage range between 1 and 8; 9 – 14 inside cells at the 32-cell stage could be achieved by a simple doubling of 5 – 7 inside cells, but a substantial contribution to the inside cell population by outer cells could be required were the lower estimates of 1 or 2 inside cells at the 16-cell stage to be correct.
Comparative values for outside cell numbers at 0 h and 2 h reveal a shift from around 20 to greater than 20. Since the time of visible evidence of vectorial transport and accumulation of fluid by outer cells seems to occur at about 20 – 24 h after the beginning of the fifth cell cycle (Ziomek, Johnson & Handyside, 1982), and since the fifth cell cycle is 11 – 14 h long (MacQueen & Johnson, 1983), presumably most TE cells in nascent blastocysts are in their sixth cell cycle with some more advanced outer cells entering the seventh cell cycle as blastocyst expansion progresses. This conclusion is supported by the observation that only embryos in which presumptive outer cells have entered their sixth cell cycle will form a blas-tocoel (Smith & McLaren, 1977; Braude, 1979), and by the incidence of mitoses in the outside cell population of more advanced embryos (Table 3). Thus the simplest explanation for the numerology of Oh and 2h blastocysts is that outer cells, as a population, are a little ahead developmentally of inner cells. Direct evidence that the fifth cell cycle is on average 2 h shorter for isolated outer cells than for isolated inner cells has been presented (MacQueen & Johnson, 1983), but indirect evidence for cells in situ indicated the opposite (Barlow et al. 1972). Surani & Barton (1984) have suggested recently that in situ inside cells may act to extend physically the outside trophectoderm cells so that the division of the latter is delayed, or even inhibited. This suggestion might resolve the apparent conflict within published cell cycle data reported above. However, it seems unlikely that delay or inhibition occurs on a large scale as outside cell numbers continue to increase (Table 2), and there is no evidence that this increase is due to an extensive contribution from the ICM (Fleming et al. 1984), indeed mitoses are evident within the trophectoderm throughout expansion of the blastocyst. However, it is clear from Table 2 that when Oh and 12h embryos are compared, no technique demonstrates that a doubling of outside cell number has occurred. There are several possible explanations for this relative deficiency. It is unlikely that consistent losses of outside cells during analysis are involved since all techniques indicate their underrepresentation. It is possible that outside cells have a longer sixth cell cycle than inside cells; however the evidence on this point is contradictory, one study supporting this idea (Barlow et al. 1972) and one other suggesting the reverse, i.e. that the inside cells have a prolonged S phase (Kimura & Kato, 1980). It is also possible that some outside cells either die and are removed, or contribute to the ICM during expansion; however, there is no evidence for this (Copp, 1979). Finally, it is possible that there is heterogeneity among outside cells, some, possibly mural trophectoderm, undergoing mitotic slowing due to their extreme stretching, whilst others, like polar trophectoderm, proliferate (Copp, 1979; Surani & Barton, 1984).
In contrast, the number of inside cells has doubled by 12 h postcavitation (Tables 2 and 4) and their modal volume has halved (Fig. 3), an observation most easily explained if all or most inside cells were “breeding true to lineage” and that most had entered their seventh cell cycle. It is clear from the detailed analysis of ICM cell numbers in Table 4 that ICM cells are not entering the seventh cell cycle synchronously. There is a suggestion of a plateau in ICM cell numbers at 14 – 16 cells over the period 3 – 6 h postcavitation which might represent the completion of the fifth cell cycle by the slowest cells and the period of Gi of the sixth cell cycle for the fastest cells. However, the values observed for total cell number, mitotic indices and DNA C-values for individual embryos combine to present an impression of great heterogeneity spanning almost the length of a full cell cycle.
The finding that some ICMs or ICM aggregates showed both trophectodermal and endodermal differentiation by cells occupying outside positions demonstrates that it is possible for both cell types to be expressed together during in vitro ICM culture and to form an integrated, mixed epithelium. This phenomenon would be expected to result from coexistence in the ICM or ICM aggregate of committed cells (which would then generate endoderm) and uncommitted cells (which can generate trophectoderm) that become exposed to an atypical outside position on ICM isolation. The apparent asynchrony in the timing of cell commitment in some ICMs, coupled with cell cycle asynchrony in most ICMs, is compatible with but not proof of a relationship between the cell cycle and commitment.
Previous reports on ICM morphology during in vitro culture do not record mixed character development, although recently Nichols & Gardner (1984) have demonstrated that both parietal endoderm and trophectoderm cells can arise from the population of outside cells during outgrowth of a single ICM in vitro. However, the usual formation of blastocyst-like vesicles by ICMs does not necessarily preclude frequent coexistence of committed and uncommitted cells within a single ICM, as it is likely that both the ratio of uncommitted to committed cells and the relative positions of such cells within the ICM are crucial factors influencing the type of structure formed by a cultured ICM or ICM aggregate. Trophectoderm cells or outside cells tend to surround and enclose inside cells (Rossant, 1975; Johnson & Ziomek, 1983) and this capacity to envelop is acquired as an extremely early event during the transformation of uncommitted ICM cells into trophectoderm (Fleming et al. 1984). Thus, on ICM isolation, those uncommitted cells which become exposed on the surface of the ICM will almost immediately surround and enclose all committed cells, and also any uncommitted cells occupying internal positions within the ICM which thus become unable to express their potential for trophectoderm formation. A blastocyst-like vesicle will therefore form in which all committed cells become inside cells. Only where the proportion of uncommitted cells is low, or where few are exposed in an outside position, might they fail to enclose all committed cells. Unenclosed committed cells will then express their potential for endoderm formation alongside the trophectoderm cells generated by uncommitted cells, thus giving rise to a structure of mixed character.
The developmental age at which commitment of inside cells takes place is seemingly later than was indicated by other work. Previous estimates of 14 (Spindle, 1978) or 16 – 19 (Rossant & Lis, 1979) inside cells at the time of commitment suggested that it took place when all cells had on average completed their fifth cleavage division and some their sixth. In this study however, the mean ICM cell number at 12 h postcavitation was about 20 and yet 69 % of ICMs apparently generated some trophectoderm on isolation (Fig. 4). This finding suggests that, if commitment is cell cycle related, it takes place late in the sixth or early in the seventh developmental cycle. Since 12 h blastocysts were always expanded, commitment was occurring to a large extent after blastocoel expansion and not during the process of expansion, as suggested by Handyside (1978). Our results are similar to those of Hogan & Tilly (1978) who found that 50 % of ICMs from expanded blastocysts generated trophectoderm in vitro.
Owing to the evident heterogeneity between cavitation time and inside cell development, and between cell cycles within individual ICMs, it will be necessary to find more accurate methods of timing and/or synchronizing inside cell cycling if the relationship between cell cycle and cell commitment is to be investigated effectively.
We wish to thank Ken Thurley, Raith Overhill, Tim Crane, Roger Liles and Gin Tampkins for their technical help, Dr M. Bennett for use of the microdensitometer at the Plant Breeding Institute, Cambridge and our research colleagues for their critical comment on the manuscript. The work was supported by grants from the C.R.C. and M.R.C. to M. H. Johnson, by a grant from Christ’s College to P. D. Warren, and by a grant from the Nuffield Foundation to J. C. Chisholm.