The utilization of an inhibitor of spermidine and spermine synthesis as a tool for the study of the determination of cavitation in the preimplantation mouse embryo

The earliest process of cellular determination during mammalian development is the progressive establishment of two distinct populations of cells in the late-morula to early-blastocyst stage: the trophectoderm and the inner cell mass (ICM). These two cell types clearly differ from each other in a number of ways which include: cell shape, embryological fate (Gardner, 1971, 1972), [³H]thymidine-labelling indices (Barlow, Owen & Graham, 1972), mitotic rate (Copp, 1978), junctional complexes (Nadijcka & Hillman, 1974; Ducibella, Albertini, Anderson & Biggers, 1975), phosphatase activities (Mulnard, 1974; Mulnard & Huygens, 1978; Vorbrodt, Konwinski, Solter & Koprowski, 1977; Johnson, Calarco & Siebert, 1977; Izquierdo & Marticorena, 1975), esterases isoenzyme (Sherman, 1972), cell surface properties (Calarco & Epstein, 1973), susceptibility to both infective and oncogenic viruses (Glass et al. 1974; Abramczuk, Vorbrodt, Solter & Koprowski, 1978), and synthetic patterns of polypeptides (Handyside & Johnson, 1978).

Two main theories have been proposed to explain the determination of the trophectoderm versus that of the ICM. The first, which is mainly based on histochemical data postulates the existence of a cytoplasmic heterogeneity which confers ‘trophoblastic properties’ to some blastomeres during early cleavage whereas the other blastomeres are predetermined as ICM precursors (Dalcq & Seaton-Jones, 1949; Mulnard, 1955; Seidel, 1960; Dalcq, 1966). The second theory which is based on the experimental data made available by the use of the in vitro culture of preimplantation stages, supports the idea that determination depends on the position of the blastomeres in the morula (Tarkowski & Wroblewska, 1967) such that inner cells become ICM whereas outer cells become trophectoderm. This is the well known ‘outside-inside model’ (Herbert & Graham, 1974), which is almost generally accepted now, at least for the mouse. The problem of the determination of the trophectoderm and the ICM has been reviewed in detail by Mulnard (1966), Herbert & Graham (1974) and Denker (1976).

The formation of the blastocoele in the mouse has been described extensively by Calarco & Brown (1969). It results from the confluence of intercellular spaces progressively filled with fluid released from cytoplasmic vesicles whose number and size have increased subsequent to fertilization. However, little is known about the actual trigger for the appearance of the blastocoele. The work of Tarkowski & Wroblewska (1967), on the development of isolated blastomeres from 4-cell and 8-cell stages, has led to the conclusion that, irrespective of the cell number, cavitation occurs at a precise time after fertilization. More recently, the use of cytochalasin B enabled Smith & McLaren (1977) to conclude that an important factor seems to be either the number of chromosomal divisions or DNA replications after fertilization, or the nucleo-cytoplasmic ratio reached at that time. The use of suitable reversible inhibitors may therefore provide useful models for these studies.

It is now clearly established that polyamines (putrescine, spermidine and spermine) play an essential role in cellular metabolism and proliferation (see Jänne, Pösö & Raina, 1978). Recently, selective inhibitors of polyamines biosynthesis have become available, among which are (i) a-methylornithine (α-MeOrn), a specific inhibitor of ornithine decarboxylase and (ii) methyl-glyoxal-Bis(guanylhydrazone) (MeGAG), an inhibitor of S-adenosyl methionine de-carboxylase (Fig. 1 indicates the precise sites of action of the two inhibitors). Both α-MeOrn (Mamont et al. 1976) and MeGAG (Heby, Marton, Wilson & Gray, 1977) selectively inhibit DNA replication and cell proliferation. In contrast to sea urchin (Brachet et al. 1978) and Polychaete eggs (Emanuelsson & Heby, 1978), preimplantation mouse embryos are much more affected by MeGAG than by a-MeOrn (Alexandre, 1978a) . Moreover, inhibition of spermidine and spermine accumulation apparently induces quiescence in the embryos at the 8- to 16-cell stage and the resumption of their development after transfer to fresh medium is followed by a delay in cavitation. The present work has been performed in order to analyse in more detail the effects of MeGAG on cleavage and blastocyst formation in relation to the determination of cavitation.

All the embryos used were obtained from 6 to 10-week-old virgin random-bred albino females induced to superovulate by intraperitoneal injections of 5 i.u. PMSG (Gestyl: Organon) at 17.00-18.00 h followed by 5 i.u. hCG (Pregnyl: Organon) 48 h later. They were then caged with males overnight. Fertilization was assumed to occur at 06.00 h on the following day and was therefore considered as time 0 of embryonic development. On Day 2 of pregnancy, the embryos were removed by flushing the oviducts and the 2-cell stages were placed in organ culture dishes (Falcon Plastics) in 50μl drops of Whittingham’s culture medium (Whittingham, 1971), under paraffin oil, and incubated at 37°C in a humidified atmosphere of 5% CO₂ in air.

MeGAG was purchased from Aldrich Chemical Company Inc., Milwaukee, Wisconsin, U.S.A. It was dissolved in the culture medium at concentrations varying from 1·25 to 20 μM.

The effect of MeGAG on RNA synthesis has been tested by autoradiography, after incorporation of 5-[³H]uridine 25 Ci/mM (Radiochemical Centre, Amersham, England). The medium which contained 10 μCi/ml was prepared as described elsewhere (Alexandre, 1977). Some preparations were incubated with either 0·1 mg/ml bovine pancreatic RNAase in Tris-buffer pH 7·4 or 0·2 mg/ml DNAase in Tris-buffer MgCl₂ M/300 pH 7·4, for 2 h at 37°C. All slides were treated with 2% perchloric acid for 30 min at 4°C and rinsed overnight in tap water. They were autoradiographed with Ilford emulsion L4, using the dipping method, and exposed for 5 days.

The embryos were prepared for cell counting and for autoradiography by the air-drying method (Tarkowski, 1966) and were stained with Giemsa.

To determine cell death within the preimplantation embryos, a 0·005% solution of eosin yellow (Merck, AG Darmstadt) in phosphate-buffered saline (Oxoid, London) was used as described by Bellvé (1973). Since eosin Y exclusion by cells is believed to be dependent on the normal functioning of the cell membrane (Goldstein & Okada, 1969), blastomeres which do not exclude the strain may be considered as dead. The embryos were examined for stained blastomeres under a Wild M20 microscope (100 ×) between 2 and 10 min after staining.

Nascent blastocysts were defined using the same criteria as Smith & McLaren (1977), i.e. the presence of one or more intercellular spaces when seen under the dissecting microscope.

Mouse eggs were cultured in vitro from the 2-cell stage for about 65 h either in control medium or in media containing various concentrations of MeGAG. At that time, the majority of the embryos had reached the full grown blastocyst stage and some of them had hatched.

As judged by the percentage of blastocysts (Table 1 and Fig. 2), relatively low doses of MeGAG exert a dramatic effect on cavitation such that at 10 μM or more, MeGAG induces an almost total arrest of development before cavitation while at 5 μM only about 10% of the embryos cavitate.

The few treated embryos which developed until the blastocyst stage cavitated somewhat later than the controls. For instance, in expt 2, about half of the final number of the blastocysts were already formed in the control group after 48 h of culture whereas no cavitation had occurred in the treated groups. Moreover, in contrast to the untreated controls, the ‘small blastocysts’ formed from the 2-cell eggs incubated in the presence of MeGAG never hatched in vitro even when left in culture for an additional period of time. Since it is generally believed that in vitro hatching is a mechanical process that depends on the number of cells in the blastocyst, cell counts were made on embryos treated with MeGAG at concentrations of 5 μM or less, which should allow more than 10% of the embryos to cavitate. The cell counts were performed separately on uncavitated and cavitated embryos and are recorded as ‘morulae’ and ‘blastocysts’ respectively in Table 2.

While, in this experiment, an abnormally low percentage (60%) of blastocysts was obtained, the mean cell number of treated blastocysts was significantly lower than that of the controls (43-9, this value is of the same order of magnitude as that found by Smith & McLaren, 1977). An excellent correlation is observed between MeGAG concentration, blastocyst number and the mean cell number of the blastocysts. The presence of vesicular forms with only six and nine cells, recorded as ‘blastocysts’ in Tables 1 and 2, clearly indicates the existence of abnormal forms (mainly in the 5 μM group) which correspond to the false blastocysts, the trophoblastic vesicles and the non-integrated forms exhaustively described by Tarkowski & Wroblewska (1967). On the other hand, it is interesting to note that the MeGAG-blocked embryos recorded in Tables 1 and 2 looked healthy until the end of the culture (3 days) and did not show any signs of degenerative processes (Fig. 2). This point has been confirmed by staining with eosin Y. Samples of 25–40 control embryos or embryos treated with MeGAG at 10 μM were transferred to PBS containing 0-005% eosin Y, at various times after the beginning of treatment (21, 41 and 65 h) and were observed 2 min later for cell death. Their development is shown in Table 3.

Of all the embryos analysed for cell death (see Table 3), only one control blastocyst (an embryo cultured for 65 h) contained any dead blastomeres. Surprisingly none of the treated embryos, even after exposure to the inhibitor for 65 h, contained dead blastomeres. Moreover, the eggs arrested at the 2-cell stage were still able to exclude the stain. Therefore, it can be concluded that MeGAG induces a metabolic quiescence without killing the embryos.

Since it is very difficult, with the dissecting microscope, to determine the precise number of cells in living embryos with more than six to eight cells, air-dried preparations of control and treated embryos (5,10 and 20 μM MeGAG) were examined (Table 4) in the same way as shown in Table 2 for lower concentrations. This enabled us to determine the precise stage at which the embryos were arrested and which were recorded as ‘morulae’ in Tables 1 and 3.2-cell and a few 4-cell-stage embryos were cultured for 65 and 45 h respectively and fixed on slides. It is clear from Table 4 that the treated embryos were blocked at an early stage of cleavage and that at 5 μ M, MeGAG provoked the arrest of development at the early fourth cell cycle (embryos optimally arrested at the 9- to 12-cell stage), whereas at 10 and 20μ M, the arrest occurred during the third cell cycle. No significant difference was observed between the 2- and 4-cell stages when treated with 20 μ M MeGAG. In agreement with the results presented in Table 2, the few treated embryos (mainly with 5 μ M) which cavitated were those that contained the highest cell number.

When a similar analysis is performed at the beginning of the treatment, i.e. during the first cell cycles, it seems that 10 and 20 μ M MeGAG slows down the mitotic rate very early during cleavage (Table 5).

From this experiment and from Table 4, it is evident that 10 μ M MeGAG does not immediately interfere with cleavage since the embryos apparently progress normally through the second cell cycle. However, after the third cell cycle, they are considerably slowed down and they only reach an average cell number of 8·5. The same conclusion can be drawn from experiments with 20μ M MeGAG where the slowing down effect appears earlier and there are very few mitoses, indicating that the cellular arrest is induced during interphase.

Thus, since it induces a total inhibition of cavitation with a minimum of ‘early effects’, 10 μ M MeGAG has been used in all of the experiments performed in order to induce metabolic quiescence.

To examine whether some metabolic processes are progressively switched off by 10 μ M MeGAG during the third cell cycle, the incorporation of [³H]-uridine has been studied in embryos treated in exactly the same way as those of expt 1 in Table 5. Embryos were cultured for either 18 h or 42 h in control or MeGAG containing medium, transferred to the same media containing 5-[³H]uridine for 4 h and then fixed for autoradiography. The embryos were arbitrarily classified as heavily (+ + +), normally ( + + ) or weakly ( + ) labelled; this is illustrated in Fig. 3 and summarized in Table 6.

No differences in either cytoplasmic, nuclear or nucleolar labelling could be detected by autoradiography between control and MeGAG-treated embryos, when they were incubated from 18 to 22 h after the beginning of treatment. In contrast with the normal situation however, no label was seen in the few embryos which were treated with RNAase; this reflects the arrest of DNA synthesis (Alexandre, 1977). RNA synthesis is markedly decreased after long (42–46 h) continuous treatments with MeGAG. Inhibition of RNA synthesis is thus only observed in slowly dividing or arrested embryos where a residual synthesis remains measurable.

In a first group of experiments, embryos were cultured in the presence of MeGAG from the 2-cell stage, transferred at several selected times to fresh medium and then analysed for the appearance of the blastocoele.

As can be seen from Table 7, a good reversibility of the MeGAG treatment was observed, and the arrested embryos have not lost their capacity to cavitate provided that the treatment did not exceed 21 h. The in vitro hatchability was however decreased after 21 h treatment by MeGAG at the concentrations tested. This can again be considered as a result of a lower cell number per embryo.

It is more interesting that the restoration of cavitation by transfer into fresh medium is delayed compared with the controls. For instance, in the experiment with 20 μM MeGAG (Table 7), in spite of the fact that at the end of culture, the same numbers of blastocysts were obtained in both groups, only 7 blastocysts were seen in the group treated for 21 h, while 21 blastocysts had already formed in the control group 54 h after the beginning of culture. Thus, although autoradiography suggests that RNA synthesis is not affected, MeGAG seems to induce a metabolic quiescence with regard to DNA replication and cellular proliferation.

To confirm the existence of a delay induced by transitory quiescence, a systematic examination of embryos treated from the 2-cell stage with 10μM MeGAG for different lengths of time was undertaken, using a dissecting microscope to score the formation of blastocyst.

Figure 4 shows that MeGAG given at the beginning of the culture induces a delay in cavitation which is proportional to the duration of treatment. When the blastocysts which are formed during one additional day are added to the values obtained at the end of the systematic examinations (104 h-old embryos on Fig. 4), it can be seen that an excellent reversion is obtainable for treatments up to 31 h whereas a 40 h-long treatment is almost totally irreversible (Table 8).

In expt 2 of Table 8, nascent blastocysts were recovered every 1 h 30 min and immediately fixed on slides, in order to estimate their mean cell number in relation to the duration of the treatment and the delay recorded on Fig. 4B.

It should be noted, from Table 9, that the mean cell number of ‘delayed nascent blastocysts’ is significantly lower (about 10 cells less) than that of controls (P >0·01); it is of the same order of magnitude for all treated embryos, irrespective of the duration of treatment except for long periods of time (40 h) where we already know that the arrest is irreversible (Table 8). In the latter case, strong cytological aberrations such as chromosomal fragmentation and micronucleation were seen. Such aberrations have also been found in some fixed nascent blastocysts derived from embryos treated from the 2-cell stage for 26 and 31 h but only in the most retarded ones.

In all of the biological systems studied to date (i.e. cultured cells, growing tumours, regenerating liver, kidney or cardiac hypertrophy, etc.), an initial step in proliferation consists of an increase in the activity of the enzymes involved in polyamine biosynthesis (see Jänne et al. 1978, for review). Similarly, increases in the activities of both ornithine decarboxylase (ODC) and S-adeno-sylmethionine decarboxylase (SAMDC) as well as in the level of polyamines have been recorded during the early embryonic development of Amphibians (Russell, 1971), sea urchins (Manen & Russell, 1973; Kusunoki & Yasumasu, 1976) and the nudibranch mollusc Phestilla (Manen, Hadfield & Russell, 1977). However, due mainly to the scarcity of embryonic material, no direct information about changes in either enzyme activity or polyamine content at fertilization and during early development in Mammals is available. In the present work, indirect evidence strongly suggests that polyamines play a key role in genetic activity during the preimplantation development of the mouse.

In Polychaetes (Emanuelsson & Heby, 1978) and Echinoderms (Brachet et al. 1978), the inhibition of putrescine synthesis by α-MeOrn from the onset of development leads to an arrest at the blastula stage whereas treatment with MeGAG has no effect on development. Both the present work and our previous results (Alexandre, 1978 a) indicate that a different situation is encountered in the mouse such that selective inhibition of spermidine and spermine synthesis by MeGAG inhibits cleavage before the occurrence of cavitation, whereas a-MeOrn exerts visible effects on cavitation only at concentrations as high as 20 mM. However, as has been discussed previously (Alexandre, 1978a), the arrest of development cannot be ascribed to a specific inhibition of putrescine synthesis since in this case the osmolarity of the medium had increased to such an extent that the control embryos were also arrested in their development; mouse embryos are known to be very sensitive to this parameter (Brinster, 1965).

On the assumption that, at least at the 2-cell stage, high amounts of putrescine and low amounts of the two other polyamines could be present in the embryos, these results can be tentatively interpreted as showing that putrescine can be converted into spermidine and spermine in the presence of a-MeOrn, but not in that of MeGAG (see Fig. 1). It seems therefore that, in contrast with the situation described in lymphocytes stimulated by concanavalin A (Morris, Jorstad & Seyfried, 1977), putrescine which is still synthesized in the presence of MeGAG, can not fulfil the role played by spermidine and spermine in early mammalian development.

It seems reasonable to believe that the cause of the arrest of development induced by MeGAG in the present experiments is a specific decrease in spermidine and spermine content. It has been shown that MeGAG exerts some pharmacological effects resulting in a decrease in RNA and protein synthesis when it is used at high doses (millimolar level); however, when MeGAG is used at micromolar doses, these syntheses are not affected (reviewed in Heby et al. 1977). Our autoradiographic analysis of [³H]uridine incorporation during cleavage inhibition by MeGAG is in good agreement with these findings in that a decrease in RNA synthesis is only measurable when the embryos are already arrested. In addition, a residual synthetic activity is still present after the developmental arrest; inhibition of RNA synthesis is thus subsequent to the induction of the arrest of the cellular cycles.

The addition of polyamines together with inhibitors of their synthesis has often been used successfully to demonstrate the specificity of action of these inhibitors. For instance, the effectiveness of MeGAG in the inhibition of initiation of DNA synthesis in 3T3 cells is reduced by spermidine and spermine (Boynton, Whitfield & Isaacs, 1976) while the inhibition of DNA synthesis in activated lymphocytes by a combination of a-MeOrn and MeGAG is suppressed by the addition of putrescine, spermidine and spermine. A few experiments of this type have been carried out on mouse embryos; however, spermine, as well as spermidine, are both toxic for this material at the concentrations normally used (50–500 μM); all the treated embryos were killed and lysed within about 10 h. Nevertheless, it has been possible to show that when the percentage of blastocyst formation is used as the end point, spermidine at concentrations of 10 and 20 μM partially protects (about 50%) against MeGAG treatment, although spermine is far less effective (Alexandre, 1978 c).

It is interesting to find that the arrest of development corresponds to the initiation of metabolic quiescence and not to the death of the embryos. This was first suggested from the healthy appearance of the treated embryos and clearly confirmed from the total absence of dead cells in the ‘arrested morulae’ as shown by the absence of blastomeres stained by eosin Y and the reversibility of the treatment up to 31 h. The transfer to fresh medium might switch on the processes of spermidine and spermine synthesis which are required for traversing the cell cycle in the blastomeres. However, this could be an exceptional situation, since it has been shown that MeGAG blocks rat brain tumour cells irreversibly in G1 and that the treatment of phytohaemagglutinin-stimulated lymphocytes with MeGAG can only be reversed during the time where the lymphocytes have not yet replicated their DNA (Heby et al. 1977).

The search for the signal which initiates the formation of the blastocyst cavity has been undertaken in several ways. Experimental manipulations on the mouse egg have shown that cavitation is not strictly dependent on the number of blastomeres (Tarkowski & Wroblewska, 1967; Smith & McLaren, 1977). The use of cytochalasin B, which when given at the 2-cell stage, prevents the second cleavage division without affecting DNA replication and nuclear division, enabled Smith & McLaren (1977) to suggest a possible role for either the number of DNA replication cycles or the nucleocytoplasmic ratio. In all these experiments, cavitation occurred at the same time in the treated or manipulated eggs as in the control ones, although it has been shown that embryos cultured in vitro cavitate significantly later than embryos developed in vivo (Smith & McLaren, 1977).

To obtain more information about the determination of blastocoele formation, it should be possible to modify the time of its appearance by the induction of a reversible delay in cleavage. This was partially observed by Alexandre (1974) after acute X-irradiation, but in this case, no specific effect could be ascribed, since X-rays induce mainly irreversible damages resulting in the death of the embryos before they cavitated (Alexandre, 1978b). 5-Biomodeoxyuridine (BUdR), which has often been used for inhibiting differentiation, both reduced the average cell number per embryo and the frequency of blastocysts (Pollard, Baran & Bachvarova, 1976). Such a treatment is also irreversible as are many of the treatments with other metabolic inhibitors such as actinomycin D (Mintz, 1964).

In the present work, we describe a drug-induced metabolic quiescence in preimplantation mouse embryos, which is followed by the resumption of in vitro development until the full grown blastocyst stage. We have found that the delay in cavitation is proportional to the duration of the treatment, and this is precisely what one would expect if cavitation is only induced when the embryos have undergone a sufficient number of cell divisions; from this standpoint, our results are in good agreement with Smith & McLaren’s biological clock theory. However, a significant average reduction of about 10 cells has been found between the cell number of control and delayed nascent blastocysts. This difference, which is identical for all of the treated embryos irrespective of the duration of treatment from 16 to 31 h, indicates that, under those conditions following a period of metabolic quiescence, the required number of cells can be reduced without any apparent consequence on the cavitation process. This reduction is however limited to about 10 cells.

Thus, while there is good evidence in favour of the importance of the nucleocytoplasmic ratio (Witkowska, 1973; Modlinski, 1975; Smith & McLaren, 1977), the present work suggests that another factor might be the actual trigger for blastocoele formation. Kemler et al. (1977) have beautifully shown that since uncompacted 30-cell embryos produced by adding monovalent antibody fragments against F9 antigen (Artzt et al. 1973) are unabled to cavitate, the compaction occurring at the 8-cell stage and involving a given cell surface structure is an essential step in the formation of blastocysts. As MeGAG did not interfere with the compaction of the morulae, we propose that during metabolic quiescence, nuclear division on one hand and cytoplasmic and nuclear maturation on the other have been desynchronized in such a way that some cytoplasmic modifications take place in the absence of nuclear division. This is consistent with the early findings of Mintz (1964) who wrote that: ‘If 2-cell eggs undergo a reversible delay in development from which they can recover within a day, and are still 2-cells the next day, RNA-synthesizing nucleoli make their appearance, as if the normal developmental stage had been reached’. Indeed, we have found that inhibition of the RNA synthesis is subsequent to the arrest of nuclear divisions. In other words, an arrested 8-cell stage is morphologically more advanced than a control 8-cell stage. This is consistent with the fact that the mean cell number of delayed nascent blastocysts is reduced at the same value for a treatment of 16 h and of 31 h.

In conclusion, the cavitation signal could be a cytoplasmic factor which is presumably under nuclear control and which in normal conditions reaches its maximum level at the end of the fifth cell cycle. Although this occurs at a precise nucleocytoplasmic ratio, this ratio could not be the signal by itself. This hypothesis might explain the fact that, occasionally, some embryos with a very small number of cells (sometimes only two) have a vacuolated cytoplasm and can acquire the vesicular forms described by Tarkowski & Wroblewska (1967). This has been seen in control embryos cultured in vitro, as reported by Tarkowski and Snow (discussion of Gardner & Rossant, 1976) and in treated embryos similar to the few ‘blastocysts’ recorded in the continuously or irreversibly (41 h) MeGAG-treated embryos. It is still difficult to distinguish between these two possibilities: either a true cytoplasmic maturation occurring without any DNA replication or a degenerative process. Only an exhaustive ultrastructural analysis could give an answer.

I wish to thank Professor J. Brachet and Dr Th. Vanden Driessche for helpful discussions, Dr J. Osborn for improving the English and Mr D. Franckx for help in the preparation of the figures.

This work was supported by the European Community (Contract Euratom-ULB 099/72/ IBIAB).