DNA replication and compaction in the cleaving embryo of the mouse

In many developing systems the expression of new developmental information within a cell is associated with a preceding cycle of replication. Therefore it has been proposed that differentiation may be linked causally to the cell cycle, and that developmental cell cycles might provide a measure of time elapsed since fertilization (Holtzer, Weintraub, Mayne & Mochan, 1972). It is probable that the feature of cell replication most important for developmental timing involves DNA, since inhibition of cytokinesis in most syterns, and of karyokinesis in some, does not influence developmental timing (e.g. Smith & McLaren, 1977; Petzoldt, Burki, Illmensee & Illmensee, 1983; Satoh & Ikegami, 1981b). It is not unreasonable to postulate a requirement for DNA replication during devel-opmental change since changing gene expression is associated with local structural changes in the organization and composition of the chromatin, many of which require, or are facilitated by, replication (reviewed Johnson, McConnell & Van Blerkom, 1984).

The earliest phases of development, which involve relatively few cells, are particularly accessible for the study of DNA replication and developmental change, and have begun to reveal evidence suggesting a causal association in both Ascidians (Meedel & Whittaker, 1979; Satoh & Ikegami, 1981a, b) and Xenopus (Newport & Kirschner, 1982a, b). In this paper we report on the effect of inhibiting DNA replication on compaction of the mouse 8-cell embryo. Compaction marks a major morphological transition in which cells flatten on each other to maximize cell contact, change their phenotype from radially symmetrical to polar, and engage in formation of specialized intercellular junctions. There is evidence to suggest that these events represent the earliest progress towards the cell diversification observed in the blastocyst (Johnson, 1985).

Using populations of 2- and 4-cell embryos synchronized to the immediately preceding cleavage division, we have determined the effects of continuous or pulsed administration of aphidicolin, a reversible inhibitor of DNA polymerase alpha, on DNA replication, cytokinesis and the events of compaction when added in Gi of the cell cycle. We conclude that the major features of compaction at the 8-cell stage are not dependent upon proximate DNA replication during the third and fourth cell cycles, but may be dependent upon replication of DNA during the second cell cycle.

4- to 6-week-old MFI strain (Olac) or FjLAC (C57B1×CBA/Ca, bred in the laboratory) female mice were superovulated by an intraperitoneal injection of Pregnant Mare’s Serum gonadotrophin (PMS Intervet 5–10 i.u. depending on size of animal) followed 48 h later by intraperitoneal injection of the same dose of human Chorionic Gonadotrophin (hCG, Intervet). For in vivo fertilization, MFI females were paired with a stud male (HC-CFLP, Hacking and Churchill, Ltd, Alconbury) overnight. The presence of a vaginal plug was taken as a sign of mating. 48 h after the hCG injection, plug-positive mice were killed by cervical dislocation and the embryos recovered by flushing oviducts with prewarmed (37 °C) Medium 2 (Fulton & Whittingham, 1978) with 4 mg ml⁻¹ Bovine Serum Albumen added [M2+BSA]. The numbers of late 2-cell (2c), 3-cell (3c) and early 4-cell (4c) embryos were recorded (dividing cells were recorded as one cell) and all 2-cell embryos were transferred to Medium 16 (Whittingham, 1971) containing 4 mg ml⁻¹ BSA [M16+BSA] under oil and pre-equilibrated at 37°C and 5 % CO₂ in air.

For in vitro fertilization, spermatozoa were collected from the epididymides of male HC-CFLP mice and incubated in 500 μl drops of Whittingham’s medium containing 3 % BSA (w/v) (W+30; Fraser & Drury, 1975) for ⁠. Cumulus masses were recovered from the oviducts of FiLAC female mice at post-hCG and placed in 1ml drops of W+30. Insemination was carried out at 1 post-hCG to give a final sperm suspension of l·4×10⁶ml⁻¹. Fertilized eggs were recovered after 4h incubation and transferred to M16+BSA for further culture. In vitro fertilization using FjLAC female mice was used in all experiments in which early 2-cell embryos were required, since the MFI strain of mice shows a developmental block at the 2-cell stage in vitro (Goddard & Pratt, 1983).

During the experiments, culture dishes were removed from the incubator for the minimum possible period and were kept on a hot plate (at 37°C) or, when viewing, on the heated stage of a Wild binocular dissecting microscope.

Glass microscope slides were scored with a diamond pencil to mark out a grid of small squares. The slides were then cleaned in glacial acetic acid and dried using lint-free paper. Embryos were mounted straight from the culture medium onto the slide grid, one per square, and the position of each embryo noted. Excess medium was removed and the embryos allowed to air dry.

A freshly dissected mouse liver was washed in sterile phosphate-buffered saline (PBS) and smeared onto the same slide that held the embryos. Embryos were fixed by placing the slides in a freshly made solution of 3:1 absolute ethanol: acetic acid for 5 min, transfer to a fresh solution of 85:5:10 absolute ethanol:acetic acid: formaldehyde (40 % w/v) for 1 h, air drying and storage at –12 °C.

Samples were stained by the Feulgen reaction as described in Bolton, Oades & Johnson (1984), and stored at –12 °C in the dark for a maximum of one week prior to densitometric quantitation.

Embryonic nuclei and liver nuclei were measured for their integrated optical density at 560 nm wavelength, with the xlOO objective of the Vickers M86 microdensitometer.

At each reading session, between 40 and 60 liver nuclei were read in order to obtain 2C and 4C DNA values and also to ensure that the stain had not faded if the readings were taken over more than one day.

For each embryonic nucleus, the average of two readings was taken. Blurred, damaged or obscured nuclei were excluded (5–10 % of the total). Overlapping nuclei were scanned together, and the average of two readings halved to give the value for each nucleus. This procedure was considered to be legitimate as the values obtained were always in the correct range, and the nuclei were from the same time point. For measurements of the third cell cycle, the percentage of overlapping nuclei was 5–10 % while in the fourth cell cycle it was 20–25 %. The number of nuclei in each embryo and the nuclei undergoing mitosis were noted.

When results from different experiments were pooled, the ratio of the 2C DNA values from each experiment was used as a ‘standardizing factor’ and assigned the value of 100 arbitrary units.

Aphidicolin (supplied courtesy of ICI Limited) was dissolved in dimethylsulphoxide [DMSO] to give a stock solution of 2 mg ml⁻¹. For embryo culture, the stock solution was diluted to 2μgml⁻¹ aphidicolin in M16+BSA.

Embryos were dezonaed in acid Tyrode’s solution (Nicolson, Yanagimachi & Yanagimachi, 1975), washed in M2+BSA and decompacted by incubation in Ca²⁺-free M2 containing 6 mg ml⁻¹ of BSA for 5–10 min. After decompaction embryos were incubated for 5 min at room temperature in tetramethylrhodamine-labelled succinyl concanavalin A (TMRTC-Suc-Con A, 500 μg ml⁻¹; Polysciences). This solution was prepared by dialysing TMRTC-Suc-Con A against two changes of M2 for 8h at 4°C, followed by addition of 6 mg ml⁻¹ BSA. Embryos were then transferred to PHA-coated coverslips forming the base of a small chamber filled with M2+4mgml⁻¹ polyvinylpyrrolidone (M2+PVP) (technique described by Maro, Johnson, Pickering & Flach, 1984). The chamber was then sealed with a second coverslip and spun at 500g for 10 min at 20°C in a Sorvall RC-5 centrifuge using an HB-4 rotor to ensure that the embryos were well attached to the coverslip. The embryos were fixed with 3·7 % formaldehyde in phosphate-buffered saline (PBS) for 30–40 min at 20°C. They were then washed four times in PBS, permeabilized with detergent (0·25 % Triton X-100 in PBS for 10min at 20°C), and finally washed in PBS.

After fixation, the embryos were labelled with affinity-purified antibody against either clathrin or actin as described in Maro, Johnson, Pickering & Louvard (1985) and Johnson & Maro (1984).

The chromatin was stained with Hoechst dye 33258 (gift of Dr B. Hogan) by incubation in M2+BSA containing 50μgml⁻¹ dye for 1 h at room temperature (Hilwig & Gropp, 1972). They were then rinsed and examined under a fluorescence microscope.

Samples were examined using a Leitz Ortholux epifluorescence microscope with the incident source HB200, and equipped with Differential Interference Contrast (DIC) mi-croscopy. Filter set L2 was used for the fluorescein labelling, set N2 for the rhodamine labelling, and set A for the Hoechst staining.

The procedure was essentially that described by Johnson & Ziomek (1982). All solutions were Millipore-filtered immediately prior to use. Circular glass coverslips (diameter = 10 mm) notched to specify orientation and cleaned with absolute ethanol and lint-free paper, were air dried for 1 h before immersion in poly-l-lysine solution [1 mg ml⁻¹, PLL; Sigma type IB] for 2 h. They were then washed 2-3 times in 0-lM-cacodylate buffer (pH 7-6) and placed in individual wells of a Linbro 24-well tissue culture dish (Flow Laboratories) containing 0·lM-cacodylate buffer.

Embryos were dezonaed and decompacted as described earlier and fixed for 30-60 min in 6 % (v/v) glutaraldehyde in 0·lM-cacodylate buffer. They were then transferred to 1% glutar-aldehyde in 0·lM-cacodylate buffer before being attached to the coverslips -the fixative ensured firm attachment of the embryos to the PLL-treated coverslips. Samples were then dehydrated through a graded series of alcohols, transferred to the specimen holder for critical-point drying, and dried from alcohol via CO₂ in a Polaron E3000.

The coverslips were then mounted on stubs using a small drop of DAB (silver glue) and stored in a desiccator until examined. Just prior to viewing, samples were coated with a 50 to 60 nm-thick layer of gold in a Polaron E5000 Diode Sputtering system and then examined in a JEOL 35 CF Scanning Electron Microscope.

Phase photographs were taken on Ilford Pan-F film (exposure time 2–4 s) using a Wild Inverted Phase microscope; for fluorescence photography, the Vario-orthomat photographic system of the Leitz Ortholux epifluorescence microscope was used, the pictures being taken on Kodak Tri-x film.

Curve fitting to graph points was carried out using the Statistical Analysis System -Non-linear Curve Fitting Procedure.

In order to ensure that the round of DNA replication in any given develop-mental cell cycle is inhibited completely, populations of relatively synchronized cells must be used and the inhibitor must be added during the G₁ phase. G₁ is short (0·-5 to 1·5 h) at 2-, 4- and 8-cell stages (Bolton et al. 1984; Smith & Johnson, 1985), and so embryos were examined at hourly intervals for the first signs of division, and any so detected were removed for experimental use.

Compaction occurs in the 4th cell cycle at the 8-cell stage. First we examined the effect on compaction of adding aphidicolin in G₁ of the 3rd cell cycle. Late 2-cells were cultured and scored at Ih intervals. Any embryos in which one or both blastomeres had divided were removed and cultured either in medium containing aphidicolin or in control medium. It was considered legitimate to use both 3- and 4-cell embryos since (a) G₂ of the 2-cell stage is long, being in excess of 8h (Sawicki, Abramczuk & Blaton, 1978), and (b) 97 % of 2-cell embryos will form 4-cells within 4h (Smith & Johnson, 1985). Thus, it is unlikely that the slower blastomeres of 3-cell embryos will be in S-phase of the second cell cycle. Both groups of embryos were cultured for up to 25 h during which time they were scored in three ways:-

• Embryos were sampled for analysis of DNA content. In the control group many blastomeres were at the 16-cell stage, but the aphidicolin group consisted exclusively of 4-cell embryos containing four nuclei. The values for the DNA content of blastomeres in each group are recorded in Table 1. The values for control blastomeres depend upon whether the cells have completed S phase of the 4th cell cycle, have divided to enter the 5th cell cycle or have entered S phase of that cycle. It is clear, however, that aphidicolin-treated embryos contain approximately 2C levels of DNA, and thus two consecutive rounds of DNA synthesis have been suppressed effectively.

• Embryos were scored for the cell flattening that occurs at compaction. Embryos in both groups showed cell flattening (Fig. 1), and the time course of flattening in the aphidicolin-arrested 4-cell embryos was similar to that of controls (Fig. 2).

• Embryos were examined 23–24 h after division to 4-cells for evidence of the development of polarity. The development of cell surface polarity occurred normally in aphidicolin-arrested 4-cell embryos, whether assessed by scanning electron microscopy (Fig. 3) or by binding pattern of fluoresceinated concanavalin A (Table 2; Fig. 4A,C,E,G). Similarly, polar redistribution of actin and clathrin was the same in control and aphidicolin-treated embryos (Fig. 4B,D,F,H).

The lack of effect on compaction of inhibition of the third round of DNA replication led us to investigate the effect of inhibition of DNA replication in the preceding cell cycle. Previous work has shown that addition of aphidicolin to newly formed 2-cell embryos results in inhibition of DNA replication and of cleavage to 4-cells (Bolton et al. 1984). Therefore late 1-cell eggs were examined at 1 h intervals, any 2-cell embryos formed during the preceding hour were removed and allocated to one of four groups. Group-1 embryos were placed directly into aphidicolin. Group-2 embryos were placed into aphidicolin but were removed 10 h later after DNA replication was completed in control embryos (Bolton et al. 1984); the embryos were then washed thoroughly and transferred to M16+BSA. Group-3 embryos were placed in M16+BSA for 10 h and then transferred to aphidicolin. Group-4 embryos were placed in M16+BSA. As expected, 2-cell embryos exposed continuously to aphidicolin did not undergo further cleavage, and each nucleus had a DNA content of only a little above 2C (Table 3, Group 1). Control embryos developed normally (Table 3, Group 4). The majority of embryos in Group 3, which were placed in aphidicolin after the period of 2-cell DNA replication divided, and, as expected, all further DNA replication was blocked, the 2-cell nuclei having a 4C and the 4-cell nuclei a 2C DNA content.

Embryos in Group 2, pulsed with aphidicolin over a 10h period that included the S phase of the second cell cycle, remained as 2-cells but with a DNA content of 4C. Therefore, embryos in this group were examined at 5 h intervals after removal from aphidicolin to determine whether they replicated their DNA immediately after withdrawal from the drug, or whether their DNA replication occurred at the same time as replication was occurring in control embryos at the early 4-cell stage. At the time of withdrawal from aphidicolin, the DNA content of Group-2 embryos was 2C (67 nuclei examined); 5h and 10h later the DNA content had doubled to 4C (74 and 73 nuclei examined respectively). Thus, ‘catch-up’ replication occurred during the late 2-cell stage, but no further rounds of replication occurred, despite the absence of aphidicolin.

When the control embryos had developed to the late 8-cell stage, embryos were taken from each of the four groups and examined at light and electron microscope levels for evidence of flattening and the development of polarity. The results are summarized in Table 4. The control group 4 showed the expected high incidence of flattening and polarization except in three retarded embryos. The embryos exposed to aphidicolin continuously (Group 1) showed no evidence of flattening or polarization (Fig. 5A,C). Group-2 embryos placed in aphidicolin for the first 10 h of the 2-cell stage showed almost complete suppression of flattening (one embryo) and polarization (two blastomeres). Moreover, incubation of these embryos for an extra 10 h or 24 h did not result in any increase in polarization indicating that the drug had exerted an inhibitory rather than a delaying effect. In contrast, Group-3 embryos, exposed to aphidicolin from the mid-2-cell stage onwards, showed a high incidence of polarization and a substantial degree of flattening, whether or not they blocked at 2- or 4-cells (Fig. 5B,D-F). The polarity in this group of embryos differed from controls in that in control embryos, the pole of Con A binding and microvilli is opposite to the point of intercellular contact, whereas in the aphidicolin-treated embryos poles were often eccentrically placed (Fig. 5D,E). Also, in many blastomeres a sharply defined ‘ridge’ separated the microvillous and non-microvillous domains.

In early embryos of most species studied, a close relationship exists between the occurrence of identifiable developmental events and the number of developmental cell cycles elapsed since fertilization. In the mouse embryo, for example, the first expression of the embryonic genome, the changing morphology at compaction and the beginnings of blastocoel expansion each occur at characteristic times, namely in the second, fourth and sixth developmental cell cycles respectively (Flach et al. 1982; Lehtonen, 1981; Smith & McLaren, 1977). A number of studies have investigated the extent to which this association is incidental rather than causal, and, should the latter be true, the nature of the underlying time-keeping mech-anism. In this paper, we report on our examination of the relationship between DNA replication in the second, third and fourth cell cycles and compaction in the fourth cell cycle.

The major developmental event of compaction was found to be remarkably independent of proximate replication. Thus, complete suppression of DNA replication in the third and fourth cell cycles is without detectable effect on compaction. This observation makes it unlikely that ‘developmental’ time is measured solely by replication cycles. Indeed, we noticed that when embryos in the control population were cleaving from 4- to 8-cells, the aphidicolin-treated embryos remained as 4-cells but showed surface deformation as though attempt-ing unsuccessful division. Similar observations have been made on enucleated Xenopus, sea-urchin, newt and mouse eggs (Yoneda, Ikeda & Washitani, 1978; Sawai, 1979; Hara, Tydeman & Kirschner, 1980; Waksmundzka et al. 1984), leading to the suggestion that a ‘cytoplasmic clock’ can run for a few cell cycles in the absence of a nucleus. Whether the run down of this putative clock, that occurs several cell cycles after enucleation, is due simply to the lack of adequate metabolic support in the absence of a nucleus, or whether the nucleus gives a short-term periodicity to the cytoplasm that can then run for a while after enucleation is not clear (Satoh, 1982).

In contrast to the resistance of compaction to aphidicolin applied early in the third cycle, suppression of the second round of DNA replication prevented compaction. This inhibition was not due to the inability of embryos containing only two cells to undergo compaction, as revealed by the few arrested, but nonetheless compacted 2-cell embryos observed when aphidicolin was added early in G₂ of the second cell cycle (see also Goddard & Pratt, 1983). Neither was the inhibitory effect of aphidicolin likely to be due merely to its prolonged presence from early in G₁ of the 2-cell stage, since the inhibitory effect was not observed when the drug was added only 10h later (i.e. was present for about 32h instead of 42 h) but was observed when the drug was present for a total of only 10 h during the first half of the long 2-cell stage. Strikingly, in this latter group of embryos, DNA replication occurred over the 5 h period immediately following withdrawal of the drug. Thus, the inhibition of compaction was only observed under conditions in which the second round of DNA replication did not take place at the appropriate time in the second cell cycle. It is possible that a mismatch between cytoplasm and replicating DNA results in a failure of some modifying action of the cytoplasm on the DNA and thereby prevents the events of compaction at the 8-cell stage from occurring. However, interpretation of an inhibitory action of a drug beyond its immediate target, in this case DNA replication, is difficult. For example, mis-matched replication also resulted in a failure to divide to 4-cells, and whilst aphidicolin does not directly inhibit division (e.g. see Group-3 embryos in Table 3), it does not necessarily follow that the failure of Group-2 embryos to divide resulted from a direct effect of the drug on DNA replication. The only conclusion that can be drawn with confidence from these results is that the third and fourth rounds of DNA replication are irrelevant to compaction.

This independence from proximate replication of a key developmental event in early mouse embryogenesis has also been reported with respect to the major activation of many embryonic genes that occurs at the mid-2-cell stage in the absence of the preceding second round of DNA replication (Bolton et al. 1984). Moreover, when the fourth round of DNA replication was delayed by 8h, blastocoel formation occurred, but did so slightly earlier, and with half the number of cells, and therefore presumably after one less round of DNA replication (although this was not tested directly: Dean & Rossant, 1984). These results on the mouse are reminiscent of those reported for Ascidian embryos, in which nuclear replication cycles appear to be critical for developmental events observed two or more cell cycles later (Meedel & Whittaker, 1979; Satoh & Ikegami, 1981a, b).

There are two conclusions that might be drawn from these results. The conservative conclusion is that when drugs such as aphidicolin only have developmental effects after relatively long exposure periods, the results are artefactual and should be questioned seriously. Despite the controls for long exposure to the drug described above, this view cannot be denied. A less-conservative conclusion is also possible, and its value lies in directing attention to more definitive experimental approaches. There is evidence to support the notion that changes in the organization of chromatin may be required if the potential for subsequent transcriptional activity of genes is to be achieved (e.g. Groudine & Weintraub, 1982; Burch & Weintraub, 1983). The actual expression of the genes in the form of utilizable transcripts, and the use of these transcripts, may occur later in response to temporal or spatial signals. Many of the relevant changes to chromatin organization that yield the potential for expression require, or are facilitated by, replication (e.g. Razin & Riggs, 1980; Leffak, 1984; Poccia et al. 1984). In some types of early embryo, there is evidence that the reduction in nucleocytoplasmic ratio that occurs during cleavage without growth can ‘dilute out’ maternally derived cytoplasmic components and thereby permit nuclear changes (Newport & Kirschner, 1982a, b). Such a mechanism appears to be less likely in the mammalian embryo (reviewed Johnson et al. 1984), and perhaps more likely is the temporally programmed synthesis of a cytoplasmic factor that could interact with replicating chromatin at the 2-cell stage (e.g. see Bolton et al. 1984; Howlett & Bolton, 1985). It should be possible to test directly these alternative mechanisms and this we are now attempting.

We wish to thank Gin Tampkins, Sue Pickering, Ian Edgar, Raith Overhill, Gerald Moore and Susan Few for their technical assistance in this work and in its preparation for publication. We also thank our research colleagues and especially Ginny Bolton for their critical comments, and Dr Mike Bennett at the Plant Breeding Institute, Cambridge, for allowing us to use the microdensitometer. The work was supported by grants from the Cancer Research Campaign and Medical Research Council to M.H.J.