The developmental potential of parthenogenetically derived cells in chimeric mouse embryos: implications for action of imprinted genes

Parthenogenetic embryos of mammals die during embryogenesis, although this method of reproduction occurs naturally or can be experimentally induced in a variety of nonmammalian species (review: Cuellar, 1987). In the mouse, experimentally produced parthenotes appear to progress normally through preimplantation stages of development and to implant into the uterus (Kaufman, 1983). Shortly after implantation, however, the parthenotes deteriorate, and none survive beyond day 11 of gestation (Kaufman et al. 1977). The embryo itself often appears morphologically normal, whereas the surrounding trophoblastic tissue is extremely underdeveloped (Surani & Barton, 1983; Surani et al. 1984).

Nuclear transplantation studies have revealed that the inviability of parthenotes is due to the absence of the paternally derived genome (Surani & Barton, 1983; McGrath & Solter, 1984; Surani et al. 1984; Mann & Lovell-Badge, 1984). Diploid embryos lacking a maternally derived genome also are inviable. In the case of these androgenetic embryos, the trophoblast is relatively well developed, whereas the embryo is retarded (Barton et al. 1984). The simplest explanation of these results is that there are a number of genes that must be expressed specifically from either the maternal or paternal genome for normal development to occur (Surani et al. 1986). The number of such imprinted genes is not known, but studies using Robertsonian translocations have shown that there are regions of several chromosomes for which the presence of two maternal copies and no paternal copies, or vice versa, causes a variety of phenotypic effects ranging from embryonic mortality to behavioural abnormalities (Cattanach & Kirk, 1985; Searle & Beechey, 1985). This suggests that imprinted genes are diverse in their actions, playing roles in different tissues at different times during development.

One step towards identifying imprinted genes would be to pinpoint the likely time and site of action of such genes. Detailed analysis of chimeras between parthenogenetic or gynogenetic embryos and fertilized embryos may be useful in this regard. Such chimeras have been made and the parthenogenetic or gynogenetic embryos have contributed to somatic and germ-line tissues of the newborn animals (Stevens et al. 1977; Surani et al. 1977; Stevens, 1978; Anderegg & Markert, 1986; Otani et al. 1987). In this report, we describe an analysis at the cellular level of the parthenogenetic contribution to different tissues of chimeras during preimplantation and early postimplantation development, using markers that are detectable in situ. The results show that parthenogenetically derived cells contribute to the trophectoderm (TE) of chimeric blastocysts, but not to the trophoblast of 6·5-day chimeras. In contrast, they contribute at 6·5–7·5 days to all tissues descended from the inner cell mass (ICM). This indicates that the imprinted genes responsible for trophectoderm development must be active before 6·5 days of development.

CD-I mice were obtained from Charles River Canada (Montreal). Mice designated Tg-MβG-1 were obtained from a laboratory stock that is homozygous for a transgenic insertion of about 1000 tandemly repeated copies of a plasmid carrying a mouse β-globin gene (Lo, 1986).

4- to 8-week-old female mice were superovulated by an injection of 5i.u. pregnant mares’ serum (Sigma or Organon) followed 44–48 h later by 5i.u. of human chorionic gonadotropin (hCG; Sigma or Ayerst). 16–17h post-hCG, the mice were killed and the cumulus masses were removed from their oviducts. The masses were incubated for 6–7 min in a 7 % (v/v) solution of ethanol (Cuthbertson, 1983) in M16 medium (Whittingham, 1971) at 37 °C in an incubator containing a humidified atmosphere of 5 % CO₂ in air. They were then washed three times in M2 medium (Quinn et al. 1982) and incubated in M16 medium containing 5 μ g ml ⁻¹ cytochalasin B (Sigma) or 1 μ g ml ⁻¹ cytochalasin D (Sigma) at 37 °C in the CO₂ incubator. After an incubation period of 4 or 8 h, the cumulus cells were removed using hyaluronidase. The cumulus-free eggs were transferred to the oviducts of day 0·5 (= day of plug) pseudopregnant females. On day 2·5, the recipients were killed and the embryos were flushed from the oviducts.

4- to 8-week-old females that received hormone injections as described above were caged overnight with males, and were checked the next morning for the presence of a vaginal plug. On the afternoon of day 1·5, the pregnant females were killed and the 2-cell embryos were flushed from the oviducts. The embryos were incubated overnight in M16 medium at 37 °C in the CO₂ incubator.

Embryos at the 4- to 8-cell stage were exposed to warmed acidified Tyrode’s solution (pH 2·4) for 15 s to dissolve the zonae pellucidae (Nicolson et al. 1975), then washed three times with M2 solution. In some experiments, half of the embryos were fluorescently labelled by incubation for 30 min in a suspension of latex beads (0·2 μ m, yellow-green Fluoresbrite, Polysciences) diluted 1:50 (v/v) in M16 medium at 37·C in the CO₂ incubator (Fleming & George, 1987). Pairs of embryos to be aggregated were transferred to 2·5–5 μ l drops of M16 medium under paraffin oil (Fisher, Saybolt viscosity <158) or silicone oil (Dow Coming 200 fluid, 200cs) in a Petri dish and pushed together using a micropipette. The dish containing the aggregated pairs was left for several minutes on the microscope stage, then returned to the CO₂ incubator. Periodically, the embryos were examined microscopically and pairs that had become separated were pushed back together. They were incubated for 48 h.

After the incubation period, chimeric blastocysts made between fluorescently labelled embryos and unlabelled embryos were examined directly to determine the distribution of fluorescent cells. Blastocysts to be analysed by in situ hybridization were introduced into the ampullae of oviducts removed from day 0·5 pseudopregnant females. Blastocysts that were to be allowed to develop to 6·5 or 7·5 days were transferred to the uterine horns of day 2·5 pseudopregnant foster mothers.

Excised oviducts containing blastocysts were fixed overnight in three parts ethanol: one part acetic acid, and then processed for in situ hybridization as described below. Recipients containing chimeric embryos were killed on day 6·5 or day 7·5 of pregnancy. The embryos were dissected from the decidua and fixed for 30 min in three parts ethanol: one part acetic acid. Then they were washed twice for 30 min each in 100% ethanol and embedded in ester wax (BDH 1960) at 60 °C. Sections were cut at 6/mt, mounted on glass slides previously coated with poly-L-lysine (Sigma) and allowed to dry overnight at 42 °C. The slides were washed in xylene to remove the wax, then in ethanol and air-dried. After processing for in situ hybridization as described below, the slides were counterstained using haematoxylin and eosin.

Probe pMβΔ2 (gift from C. Lo, University of Pennsylvania) was nick-translated in the presence of biotinylated dCTP and dUTP as described (Rossant et al. 1986). In situ hybridization was carried out as described (Rossant et al. 1986; Varmuza et al. 1988), except that in most experiments the slides were incubated before hybridization in 3 % (v/v) H₂O₂ in methanol to block endogenous peroxidase (Burns et al. 1985).

Some slides were subjected to an additional procedure designed to enhance the hybridization signal (Burns et al. 1985). After the in situ hybridization procedure had been completed, the slides were incubated in 0·5 mm-NaAuCl₄ (pH 2·3) for 5 min, washed in water for 5 min, incubated in 20 mm-sodium sulphide neutralized to pH 7·0 for 5 min, and washed in water for 5 min. Finally, the slides were incubated for 3–6 min in a solution containing 2·6mm-ammonium nitrate, 1·2 mm-silver nitrate, 0·3 mm-tungstosilicic acid, 120mm-sodium carbonate, and 0·005% formaldehyde.

To determine quantitatively the contribution of the Tg-MβG-1 cells to the embryonic ectoderm of a chimera, we determined the average number of hybridization signals present in a unit volume of chimeric tissue. An ocular graticule was placed in the microscope and the tissue section was oriented so that the embryonic ectoderm lay under the grid squares of the graticule. At ×300 magnification, each grid square covered an area of 730 μ m², so the volume of tissue sampled in a 6 μ m section was approximately 4400 μ m³. The hybridization sites visible in each grid square that covered embryonic ectoderm were counted and the mean number of hybridization sites per unit volume was calculated. This value was compared to that obtained for a unit volume of tissue from normal Tg-MβG-1 embryos. The ratio of the two values is an estimate of the fraction of the tissue that is contributed by Tg-MβG-1 cells.

Analysis of variance was carried out using standard methods (Snedecor, 1956).

More than 80 % of unfertilized Tg-MβG-1 eggs were parthenogenetically activated by ethanol treatment, as judged by the formation of pronuclei. The subsequent development of the activated eggs was influenced by the duration of the cytochalasin treatment used to diploidize them. When the eggs were exposed to cytochalasin for 4h, they often became fragmented shortly after transfer to cytochalasin-free medium, and only 28% (705/2517) continued development to the 4- to 8-cell stage. When the eggs were exposed to cytochalasin for 8h, they remained intact after transfer to cytochalasin-free medium, and 56 % (135/240) later reached the 4- to 8-cell stage. This difference probably reflects the stage of the cell cycle at which the activated eggs were removed from cytochalasin (Snow, 1973).

The 4- to 8-cell parthenogenetic Tg-MβG-1 embryos were then aggregated with 8-cell fertilized CD1 embryos. Aggregation and blastocyst formation occurred as frequently as for aggregated pairs of fertilized embryos (Table 1). To determine the distribution of the cells of the parthenogenetic embryos in the chimeric blastocysts, in some experiments the parthenogenetic embryos were labelled using fluorescent latex beads (Fleming & George, 1987) before they were aggregated with unlabelled fertilized CD-I embryos. As controls, fertilized Tg-MβG-1 embryos were fluorescently labelled and aggregated with unlabelled CD-I embryos. The incubation with fluorescent beads had no effect on the frequency of successful aggregation or blastocyst formation (not shown). Intact chimeric blastocysts were examined using fluorescence microscopy.

Fluorescent cells were present in the mural TE of 48/53 chimeras constructed between parthenogenetic Tg-MβG-1 embryos and fertilized CD-I embryos (P ↔F chimeras), and in 21 of 26 chimeras constructed between fertilized Tg-MβG-1 embryos and fertilized CD-I embryos (F ↔F chimeras). Fluorescent cells were also seen at the embryonic pole containing the ICM and overlying polar TE in 52/53 P ↔F chimeras and 25/26 F ↔F chimeras. These results established that parthenogenetically derived cells contributed to both the ICM and the TE of chimeric blastocysts. It was not always possible by fluorescence examination, however, to distinguish between contributions to the ICM and the polar TE. To address this issue, 10 chimeric P ↔F blastocysts were sectioned and processed for in situ hybridization. Seven of these were found to contain Tg-MβG-1 cells in the polar and mural TE and in the ICM (Fig. 1A). In the remaining three cases, the blastocyst had collapsed, so the location of the labelled cells could not be identified. These results confirmed that parthenogenetic embryos contributed cells to the polar TE of chimeric blastocysts.

Chimeric blastocysts were transferred into the uteri of foster mothers and conceptuses were dissected from the uteri at day 6·5 or 7·5 of development. The P ↔F chimeras implanted less frequently than the F ↔F chimeras (Table 1). Because the transfers of the P ↔F and F ↔F chimeras were not always carried out in the same experiments or by the same investigator, the significance of this difference is unclear, particularly as others have reported that the frequency of implantation is similar between P ↔F and F ↔F chimeras (Nagy et al. 1987). We also found, however, that at day 6·5–7·5 the percentage of implantation sites that contained identifiable embryos was significantly lower for P ↔F chimeras than for F ↔F chimeras (χ²P<0·05). This result is less likely to be influenced by investigator variation, and suggests that chimeras containing parthenogenetic embryos are impaired in their development after implantation as reported previously (Nagy et al. 1987).

The conceptuses were fixed and subjected to in situ hybridization to determine their patterns of chimerism. Preliminary experiments showed that, in tissue sections of normal Tg-MβG-l embroys, >90 % of the nuclei visible in a section were labelled by the hybridization procedure (1500 nuclei counted). The conceptuses derived from the chimeras therefore were judged chimeric if labelled nuclei were present but constituted less than 90% of the total nuclei, as estimated visually. This may underestimate the actual frequency of chimerism, because it would be difficult to detect a small number of unlabelled cells in a large labelled population.

Of the 36 conceptuses analysed that were derived from F ↔F chimeras, 5 contained only unlabelled CD-I cells and 2 contained only labelled Tg-MβG-1 cells. These 7 embryos were classified as nonchimeric. The remaining 29 animals were chimeric in at least one tissue. In 20 of these animals, both the ICM and TE lineages were chimeric (Table 2, Fig. 1B), although the degree of chimerism as estimated visually was not always the same in the two lineages.

When the conceptuses derived from the P ↔F chimeras were analysed, a different pattern emerged. First, 17 of 39 embryos were not chimeric (Table 2). All of these contained only unlabelled CD-I cells, indicating that the parthenogenetically derived cells had become completely eliminated from the embryo. Second, in 20 of the 22 chimeras, the chimerism was restricted to the ICM lineage (Table 2, Fig. 1C). The trophoblast of all of these embryos contained only CD-I cells. One of the remaining embryos contained only a small number of labelled cells (estimated at <5 %) in the trophoblast, while the other contained mainly labelled cells in all tissues and was morphologically abnormal (Fig. 1D). These results indicate that, in almost every case, parthenogenetically derived cells were unable to contribute detectably to the trophoblast of chimeras at day 6·5–7·5 of development.

In 12 of the 22 P ↔F chimeras, both the primitive ectoderm and primitive endoderm lineages of the ICM were chimeric. In the remaining 10, only one lineage was chimeric, and the other in every case contained only CD-I cells. In contrast, in only 5 of 29 F ↔ F chimeras was one of the lineages chimeric and the other nonchimeric. This difference between P ↔F and F ↔F chimeras may be due to sampling variation. Alternatively, it may be that, in some P ↔ F chimeras, the ICM contained a relatively small number of parthenogenetically derived cells, which contributed only to either the primitive ectoderm or the primitive endoderm. Nevertheless, it is clear that the parthenogenetically derived cells were not specifically excluded at this stage of development from any tissues descended from the ICM.

We attempted to determine whether the abilities of parthenogenetically derived and normal Tg-MβG-1 cells to contribute to the embryonic ectoderm of chimeras were quantitatively similar, using the procedure described in the Methods. The mean number of hybridization sites prer grid square overlying the embryonic ectoderm was calculated for Tg-MβG-1 embryos and for each chimera, and the ratio of the two values was determined for each chimera.

The average value obtained for five Tg-MβG-I embryos was 6·4. The individual embryo means varied by less than 5 % (range: 6·33–6·59; Fig. 2), which suggests that the counting procedure produced consistent results. Three or four sections of each P ↔F and F ↔F embryo were analysed. Analysis of variance showed that the variation between embryos was greater than the variation within embryos (P<0·01), indicating that the differences observed between embryos were not due to sampling variability. Within each embryo, the variation in mean number of hybridization sites was not greater between sections than within sections (P > 0·05) in 22 of 28 F ↔F chimeras and 11 of 16 P ↔F chimeras that contained Tg-MβG-1 cells in the embryonic ectoderm. In the remaining cases, the mean value of chimerism calculated by analysis of three or four sections may represent a reasonable estimate of the true mean value.

As noted above, 7 of the 36 F ↔F conceptuses were nonchimeric. Of the 29 chimeras, three appeared to contain only CD-I cells and three to contain only Tg-MβG-1 cells in the embryonic ectoderm (Fig. 2). Previously, on the basis of visual examination, four of these had been classified as nonchimeric and the other as containing a small CD-I contribution. The remaining 23 chimeras showed varying degrees of chimerism ranging from 7 % to 88 % (Fig. 2). In the P ↔ F conceptuses, 17 of 39 were nonchimeric. Of the 22 chimeras, five did not contain detectable Tg-MβG-1 cells in the embryonic ectoderm. The remaining 17 chimeras showed variable contribution of Tg-MβG-1 cells to the embryonic ectoderm, ranging from 5% to 90%. These results show that, when cells of parthenogenetic origin were present in the embryonic ectoderm, they could contribute as extensively as normal cells at this stage of development.

We also attempted to quantify the contribution of Tg-MβG-1 cells to the visceral endoderm of the chimeras. The variations within and between sections of individual embryos were very high, however, indicating that the distribution of the Tg-MβG-1 cells in this tissue was not uniform, consistent with previous observations (Gardner, 1984; Rossant, 1985). As well, analysis of variance showed that the variation between embryos was not greater than the variation within an embryo (P>0·05). Quantification of the degree of chimerism of the visceral endoderm would require analysis of most or all tissue sections of a chimera and would be better carried out using isolated tissue sheets (Gardner, 1984). Although not quantified, there was no obvious qualitative deficiency in the ability of parthenogentically derived cells to contribute to either the visceral or parietal endoderm at 6·5–7·5 days of development.

We have analysed chimeras made between parthenogenetic and fertilized embryos, using a cell marker that is detectable in situ. The results showed that cells of the parthenogenetic embryo contributed to both the TE and the ICM of chimeric blastocysts. By 6·5 days, however, in almost every chimera, the parthenogenetically derived cells had become completely excluded from the TE lineage. Previously, it has been reported that parthenogenetically derived cells are absent from the trophoblast of 9·5-day (Surani et al. 1988) and 11·5-day (Nagy et al. 1987) P ↔F chimeras. The present results narrow the time of elimination to between the blastocyst stage and 6·5 days of development.

The exact manner in which these cells become excluded from the trophoblast is difficult to establish. It seems unlikely that the only viable chimeras we observed were those derived from blastocysts that happened to contain no parthenogenetic TE cells, since nearly all chimeric blastocysts examined showed evidence of parthenogenetic contribution to the trophectoderm. It seems more likely that, on the one hand, chimeras containing a very high parthenogenetic contribution to the trophoblast may have died before 6·5 days. This would be consistent with the observation of a lower survival rate of P ↔F chimeras as compared to F ↔F chimeras. On the other hand, chimeras containing a lower parthenogenetic contribution to the trophoblast may have survived, but the parthenogenetically derived cells were overgrown by normal trophoblast cells.

It is well documented that the trophoblast of parthenogenetic and gynogenetic embryos develops poorly (Surani & Barton, 1983; Surani et al. 1984; McGrath & Solter, 1984; Mann & Lovell-Badge, 1984) . Our results are consistent with these observations, except that we found that parthenogenetic embryos rarely produced any trophoblast at all when part of a chimera. Apparently, parthenogenetic embryos are capable of producing a limited amount of trophoblast, but this ability is not manifested when they are placed in competition with normal embryos in chimeras. The abnormal trophoblast development observed in parthenogenetic and gynogenetic embryos is thought to reflect the absence of paternally inherited copies of imprinted genes (Surani et al. 1985). The fact that, in chimeras, parthenogenetically derived cells are eliminated from the trophoblast by 6·5 days suggests that these genes are expressed before 6·5 days of development.

In contrast to their exclusion from the TE lineage in chimeras, parthenogenetically derived cells could contribute to both the primitive endoderm and primitive ectoderm lineages at 6·5–7·5 days. It should be noted that there was a higher frequency of nonchimerism among the P ↔F conceptuses than among the F ↔F conceptuses, and that contributions restricted to either the primitive endoderm or the primitive ectoderm were more frequent in P ↔F than F ↔F conceptuses. It is unclear whether these differences reflect the absence of paternally inherited genes in the parthenogenetic embryos or deleterious effects of the procedures used to generate them. Despite this uncertainty, it appears that genes expressed from the paternal genome are not required in the ICM lineages of chimeras up to 7·5 days of development. Alternatively, defects in parthenogenetically derived cells in these lineages may be such that they can be rescued by neighbouring normal cells; it should be noted, however, that the embryonic portion of parthenotes is morphologically normal during early postimplantation development (Barton et al. 1985; Surani et al. 1986) .

Our observation that parthenogenetically derived cells contributed as frequently to the visceral endoderm as to the embryonic ectoderm at 7·5 days is in contrast to observations made using older chimeras. At 9·5 and 11·5 days, parthenogenetically derived cells generally are absent from the visceral yolk sac endoderm, which is descended from the primitive endoderm (Nagy et al. 1987; Surani et al. 1988), although they are still present in the embryonic ectoderm lineage. Taken together, these results suggest that parthenogenetically derived cells become eliminated from the visceral endoderm lineage between 7·5 and 9·5–11·5 days of development. It may be speculated that normal development of the visceral yolk sac endoderm depends on the expression of paternally inherited copies of imprinted genes, and that these genes are required between 7·5 and 9·5–11·5 days of development.

Several lines of evidence indicate, in fact, that the paternal genome is required later in development in several lineages derived from the ICM. First, parthenogenetically or gynogenetically derived ICMs cannot form viable mice when surrounded by normal trophoblast (Barton et al. 1985) or in combination with androgenetic embryos (Surani et al. 1987). Second, the parthenogenetic contribution to P ↔F chimeras declines progressively through embryogenesis (Nagy et al. 1987). Third, deleterious effects occur in late stages of fetal development when both copies of specific chromosomal regions are maternally derived (Cattanach & Kirk, 1985; Searle & Beechey, 1985).

All of these results suggest that the development of several tissues in the mouse depends on the expression of imprinted genes. These genes could encode tissue-specific proteins or proteins required in different tissues at different stages of development. Examination of chimeras at later stages of development, using the in situ marker system described here, will reveal whether parthenogenetically derived cells become eliminated from specific tissues at different times during development. Where this is observed, it might indicate that the normal development of the tissue depends on the expression of paternally inherited copies of imprinted genes. In this way, the sites and times of action of imprinted genes can be identified, provided that they act cell-autonomously. This information should aid in the molecular search for imprinted genes.

We thank M. A. H. Surani and D. Solter for useful discussions, and S. M. Darling and A. Gossler for advice on the manuscript. This work was supported by the National Cancer Institute (NCI) and the Natural Sciences and Engineering Research Council of Canada. H.C. is a postdoctoral fellow of the Medical Research Council of Canada. S.V. is a postdoctoral fellow and J.R. is a Research Scientist of the NCI of Canada.