Mechanism of size regulation in mouse embryo aggregates

Genetic chimaeras produced by aggregating early embryos have proved very useful for studying development and differentiation in the mouse (Mintz, 1974; McLaren, 1976), and have also provided insights into embryonic growth regula-tion. Aggregates of two (Tarkowski, 1961; Mintz, 1962), three (Markert & Petters, 1978) or four embryos (Petters & Markert, 1980) all produce viable offspring of normal size, indicating that there are growth control mechanisms in the embryo which can compensate for increased preimplantation size. Buehr & McLaren (1974) presented evidence that size regulation occurred in double embryo aggregates shortly after implantation around the time of proamniotic cavity formation. However, no attempt was made to determine mechanisms responsible for such regulation. The aim of the present study was to use serial reconstructions of double embryo aggregates to:

• (1) pinpoint clearly the period of time over which size regulation occurred,

• (2) determine whether a single control mechanism could account for size regulation or whether tissues regulated size independently.

• (3) determine the most likely mechanism of size regulation.

Results of the study indicated that size regulation occurred in all tissues at the same time and was accomplished by an alteration in cell cycle length.

All embryos were obtained from natural matings of random-bred Ha(ICR) mice. PB1 medium (Whittingham & Wales, 1969) supplemented with 10% foetal calf serum (FCS) was used for recovery, manipulation and transfer of embryos. Embryos were cultured overnight in α-modified MEM (GIBCO) + 10% FCS at 37 °C in an atmosphere of 5 % CO ₂ in air.

Eight-cell embryos were flushed from the oviducts of pregnant females be-tween 12.00 and 15.00 h on the second day following vaginal plug formation. Zonae pellucidae were removed by a brief incubation in acidified Tyrode’s solution (pH 2·5) (Nicolson, Yanagimachi & Yanagimachi, 1975). Embryos were washed and transferred singly or in pairs to drops of α-MEM under light liquid paraffin in Falcon bacteriological petri dishes. Double embryo aggregates were made by gently pushing two uncompacted 8-cell embryos together. Once aggregation had begun, the culture dish was placed in the incubator and cultured overnight. The following afternoon, well-integrated aggregates and control embryos had formed late compacted morulae or early blastocysts. Five double embryos were transferred to one uterine horn of each recipient female on the second day after vaginal plug formation. Five control single embryos were transferred to the opposite horn. In all further analysis embryos were staged by the time postcoitum of the pseudopregnant recipient rather than by the age of the embryos. Mating was assumed to occur at the midpoint of the dark period (2.00 a.m.).

Recipients carrying double and control embryos were killed every 8 hours from 4 days, 0 h to 6 days, 16 h post coitum (p.c.). Uteri were dissected and fixed overnight in AFA fixative (acetic acid: formalin: alcohol: water; 1:1:3:5). After dehydration and embedding in paraffin wax, sections were cut at 7 μm, mounted and stained with haematoxylin and eosin.

Recipient females carrying both double and control embryos were injected with [³H]thymidine and killed at 5 days, 22 h and 6 days, 2 h p.c. Females were given four intraperitoneal injections of 50 μCi α-methyl[³H]thymidine (Amersham, specific activity, 20 μCi/mmole) at hourly intervals and were killed 1 h after the last injection (Solter, Skreb & Damjanov, 1971). Uteri containing decidual swellings were fixed and processed as before. Wax sections were mounted on glass slides, taken down to water and treated with 10% trichloro-acetic acid for 10 min. The slides were then dipped in Kodak NTB3 nuclear track emulsion, exposed for four weeks at 4 °C in the dark and then developed, fixed, and stained with haematoxylin and eosin.

Several recipient females were given a single intraperitoneal injection of 0·6 ml of a 0·1 mg/ml solution of demecolcine (Sigma) in phosphate-buffered saline two hours prior to killing (Copp, 1978). These recipients were killed every 8 h between 5 days, 16 h and 7 days, 0 h p.c. Uteri containing decidua were fixed and processed as described for untreated embryos.

Camera-lucida drawings were made of all sections which contained embryonic material. The boundaries of tissue types and all mitotic figures were recorded on the drawings. Cell number, mitotic index and labelling index were calculated for the various tissues. Figure 1 illustrates which tissues were included in cell number estimates at different stages. In the preimplantation embryo, total cell numbers of trophectoderm and inner cell mass (ICM) were calculated, but, in the postimplantation embryo, trophoblast giant cells plus adjacent parietal endoderm and ectoplacental cone were excluded because it was difficult to discern the boundaries between these tissues and maternal tissues. Extra-embryonic ectoderm was the only postimplantation trophectoderm derivative included in the calculations.

Direct counting of cell number was clearly impossible in later stage embryos and so cell number throughout was estimated by calculating the number of nuclei in a small area of tissue and multiplying this by the total tissue area per embryo as estimated from planimeter measurements of the camera-lucida drawings. Abercrombie’s formula (Abercrombie, 1949) was used to estimate number of nuclei per unit area.

Abercrombie’s formula is P = A × M/L + M, where

M was known to be 7 μm. A was estimated for each tissue by obtaining the average of several counts of nuclear fragments in different small areas (20–50 planimeter units). No significant difference in A could be found between ICM, embryonic ectoderm, and extraembryonic ectoderm at different times or be-tween single and double embryos (mean number of nuclear fragments/ten planimeter units in all tissues except endoderm = 5·0, range 4·7–5·3). Thus, the same value of A was used for calculating P in these tissues at all stages. How-ever, the value of A for endoderm differed significantly from the other tissues and also varied significantly with time (range of number of nuclear fragments/ ten planimeter units = 2·3–3·7). Thus, separate values of A, and hence separate values of P, were used for calculating total endoderm cell number at all stages of development. Nuclear length measurements varied little from stage to stage or tissue to tissue (range 7·8–8·2 μin) and a mean nuclear length, L, of 8 μm was used for all calculations. Once a value of P per unit area was derived for each tissue, total cell number was readily estimated by multiplying by the total section area.

The mitotic index (MI) was calculated for untreated and colcemid-treated embryos by counting the number of complete metaphases and anaphases and expressing them as a percentage of total cell number. The labelling index (LI) was obtained from [³H]thymidine-treated embryos by scoring the percentage of labelled cells in the population. Background labelling was low (> 2 grains overlying the cytoplasm of a cell) and a cell was considered labelled if more than three silver grains overlaid the nucleus (Copp, 1978).

No difference in the rate of successful development was observed between double and control embryos. For double embryo aggregates, 39 % of all embryos transferred to recipients that became pregnant developed successfully, while 40 % of control embryos developed after transfer. Double and control embryos appeared morphologically normal and there was no obvious increase in the amount of cell death in double embryos at any stage of development. Treatment with colcemid and [³H]thymidine had no obvious deleterious effects on the embryos. Only pregnant mice containing successful implants in both horns were used for further analysis.

Analysis of cell numbers of double and single embryos revealed that the double embryos were at least twice as large as single embryos early in develop-ment (Table 1). Cell numbers of double and single embryos remained signifi-cantly different until 6 days, 16 h. When the ratio of cell numbers of double to control embryos was plotted against time (Fig. 2) it was clear that the ratio remained around 2:1 until 5 days, 16 h. Over the following 24 h, there was a rapid shift towards a 1:1 ratio, so that cell numbers were virtually identical at 6 days, 16 h. The MIs of all embryos were low (Table 1) and no significant differences could be found between double and control embryos.

Analysis of cell number data for individual tissues showed that size regulation occurred in all tissues over roughly the same time period. Both ICM and trophectoderm derivatives showed size regulation by 6 days, 16 h, with the shift in cell numbers occurring between 5 days, 16 h and 6 days, 16 h (data not shown).

When data for embryonic ectoderm, extraembryonic ectoderm and endoderm were considered separately (Table 2), no clear differences in timing of size regulation could be observed. A statistically non-significant difference between cell numbers of double and control embryos occurred slightly earlier in embryo-nic ectoderm and extraembryonic ectoderm than in endoderm. However, visual inspection of the graph of double: single embryo cell number ratio against time (Fig. 2) revealed no obvious ‘leading’ tissue in size regulation.

Labelling of embryos was successfully achieved after injection of [³H]thymidine in all experiments. Both control and double embryos showed high labelling indices in all tissues after [³H]thymidine treatment between 5 days, 18 h and 6 days, 2 h (Table 3) and there was no significant difference between single and double embryos. Therefore, no large non-dividing and/or dying cell population was present at the beginning of the period of size regulation in double embryos.

Embryos treated with colcemid showed cell numbers which were very similar to those of untreated embryos and size regulation occurred at the same time (data not shown). However, use of colcemid greatly amplified the MI and allowed meaningful comparison of the mitotic activity of double and single embryos. Control embryos showed a dramatic increase in mitotic activity between 5 days, 16 h and 6 days, 16 h (Fig. 3), peaking at 6 days, 8 h and then declining. Mitotic activity was highest in the embryonic ectoderm where the MI after 2 h colcemid treatment reached 53·6 at 6 days, 8 h. Double embryos failed to show this increase in mitotic activity; large differences in MI were observed between double and control embryos in all tissues, particularly at 6 days, 0 h and 6 days, 8 h (Fig. 3). The differences between MIs of double and single embryos were significant at 6 days, 0 h and 6 days, 8 h (P > 0·05, using Student’s t test) for all tissues except extraembryonic ectoderm. MIs were not significantly different in any tissue at 5 days, 16 h or 6 days, 16 h.

The mitotic index produced after colcemid treatment was used to estimate cell cycle length in double and control embryos over the period of size regulation (Table 4), using the following formula

(1) that all cells are dividing,

(2) that all cells are dividing at the same rate,

(3) that there is no synchrony of cell division,

(4) that no cells are entering or leaving the population,

(5) that colcemid is not toxic to cells.

Continuous labelling with [³H]thymidine revealed that > 90% of the cells (Table 3) were synthesizing DNA and, thus, were presumably mitotically active. Assumptions (2), (3) and (4) are unlikely to be entirely true (Snow, 1977) but are common to all methods of estimating cell cycle lengths. Although colcemid is known to be toxic to cells, we observed no deleterious effects over the 2 h period used and Copp (1978) has shown that blastocysts can survive exactly the same colcemid treatment as used here. Thus, we consider that the cell cycle lengths derived from colcemid-induced MIs represent reasonable estimates on which to base comparisons between double and single embryos.

In all the previous analysis, embryos were grouped according to gestational age rather than by their stage of development, because it was difficult to re-cognize distinctive stages of egg cylinder development. However, there are two morphological events which can be readily scored: proamniotic cavity forma-tion and primitive streak development. Proamniotic cavity formation occurred around 5 days, 6 h in control embryos, prior to the time of size regulation. Mesoderm was first observed around 6 days, 16 h, after size regulation was complete. When these events were scored in double embryos, it became obvious that proamniotic cavity formation occurred earlier in double than control embryos. The majority of double embryos had formed the proamnion by 5 days, 8 h. The timing of proamnion formation was also studied in half embryos, produced by destruction of one blastomere of a 2-cell embryo. We have pre-liminary data (unpublished) that these embryos also size regulate around the same time as double embryos, although regulation may not be complete until 7 days of development. Proamnion formation was delayed until 6 days, 0 h in the majority of half embryos (Table 5). The mean cell numbers of double, control and half embryos were very similar at the time when the majority of embryos had formed the proamnion. Embryos were also scored for the formation of mesoderm, which occurred at the end of the size regulation period. Double, control and half embryos formed mesoderm at the same time and at the same cell number (Table 5).

In this study, double embryos produced by aggregating eight-cell mouse embryos were shown to remain twice the size of single embryos until at least 5 days, 16 h p.c. Size regulation occurred rapidly over the next 24 h period; there was no difference in cell number between double and control embryos by 6 days, 16 h p.c. (Table 1, 2; Fig. 2). The timing of size regulation found in this study differs from that observed by Buehr & McLaren (1974). They reported that size regulation was complete by proamnion formation, which is approximately 24 h earlier than observed in this study. This discrepancy may result from the different strains of mice used, but may also reflect the small sample sizes and irregularly spaced sampling points used by Buehr & McLaren. Sample size in the present study was never less than four and often much higher, and samples were taken every 8 h over the complete post-implantation period studied.

Size regulation was found to occur in all tissues of the embryos over roughly the same time period (Fig. 2). No marked differences in either onset or comple-tion of size regulation were observed. This suggests that there is a single regula-tory system controlling size in the early embryo. This differs from the size regulation seen in later embryos treated with Mitomycin-C where different tissues regulate for cell loss at different times (Snow & Tam, 1979).

There are several mechanisms that could account for the size regulation observed in double embryos (Rossant, 1977):

1. Double embryos might experience enhanced cell death. Cell death is difficult to assess in the light microscope (Saunders, 1966) but no obvious in-creased level of necrosis was observed in double embryos. Also, continuous labelling with [³H]thymidine revealed that > 90 % of cells in double embryos were synthesizing DNA (Table 3) and were presumably viable.

2. A large non-dividing population of cells might occur in double embryos around the time of size regulation. Although viable, these cells would be blocked in some stage of the cell cycle, while other cells of the embryo divided at the normal rate. Labelling with [³H]thymidine did not reveal any such population during the initial stages of size regulation (Table 3). Nearly all cells in both double and single embryos were active in DNA synthesis.

3. Double embryos might show an increase in cell cycle length compared to that of controls. Cell cycle lengths calculated from colcemid-induced MIs (Table 4) provide strong evidence to support this proposed mechanism. Over the period of size regulation double embryos showed a much lower MI in all tissues and hence longer cell cycle length than single embryos. Control embryos showed a proliferative burst, which peaked around 6 days, 8 h, while double embryos showed a gradual shortening of cell cycle times but no dramatic peak of mitotic activity (Fig. 3). The proliferative burst observed in control embryos has been reported previously by Snow (1977) and the cell number and cell cycle lengths reported here correspond well with his results.

It is difficult to prove conclusively that the difference in MI observed between double and control embryos is sufficient to account for the entire process of size regulation. The MI at one time point cannot be used to predict the cell number at a later point since the cell cycle length changes continuously between the two time points. However, a rough calculation using differences in cell cycle length of the order observed serves to illustrate that the differences are of the right magnitude to account for size regulation. If the cell cycle length of the single embryos were 6 h and of the double embryos were 8 h, then in 24 h, the double embryos would undergo three doublings while the single embryos would experience four cell doublings. If double embryos began with 2x cells and controls with x, both would contain 16x cells after 24 h.

These experiments have clearly established that the length of the cell cycle can be regulated in the early post-implantation mouse embryo in order to com-pensate for increased preimplantation cell number. The mechanism by which cell cycle length is altered remains to be elucidated but it is interesting to note that size regulation occurs when the normal embryo is undergoing a dramatic increase in mitotic activity. This suggests that this may be an important control point in development at which cell division rate can be increased or decreased to compensate for any deviations from normal cell number. It is not possible to determine from the present study whether size regulation is intrinsic to the embryo or whether the maternal environment plays a role. Study of size regula-tion of double embryos in vitro will be necessary to answer this question.

The results of this study also provide some information about the timing of differentiation in the early postimplantation embryo. The first differentiative event in postimplantation development which can be readily scored morpho-logically is proamnion formation. This event occurred before size regulation but the time when a majority of embryos had formed a proamnion was found to differ, with most double embryos forming the cavity before controls and most controls before halves. However, it is interesting to note that the mean cell numbers of double, control and half embryos were almost identical at the time when a majority of each of them had formed the proamnion (Table 5). This suggests that the timing of differentiation in the postimplantation embryo may depend on the absolute cell number. This contrasts with differentiation in the preimplantation embryo where blastocoel formation has been shown to occur independently of total cell number. This event is believed to be timed by the number of nuclear divisions undergone since fertilization or by some other intrinsic cytoplasmic clock set in motion at fertilization (Smith & McLaren, 1977; Johnson, 1981). Experiments investigating differentiative events at a biochemical rather than a morphological level will clearly have to be undertaken in order to verify that differentiation in the postimplantation embryo utilizes a different timing mechanism from the preimplantation embryo.

We should like to thank Dr V. E. Papaioannou for useful discussion. The work was supported by the Canadian Natural Sciences and Engineering Research Council.