Effect of egg composition on the developmental capacity of androgenetic mouse embryos

Successful development of mammalian embryos requires the presence of both paternal and maternal genomes. Previous studies have revealed that embryos possessing two paternally derived pronuclei (andro-genones) or two maternally derived pronuclei (gyno-genones and parthenogenones) fail to develop to term (Barton et al. 1984; McGrath and Solter, 1984; Surani et al. 1984). Nuclear transplantation experiments (McGrath and Solter, 1984; Mann and Lovell-Badge, 1984; Surani et al. 1984) and analyses of translocation mutants (review, Cattanach, 1986) have demonstrated that this failure is the result of functional differences between genomes established through ‘genomic imprinting’ rather than cytoplasmic effects. The developmental defects observed in androgenones and gynogenones/parthenogenones are, to some extent, complementary. Androgenetic embryos develop to the 6-somite stage at best and exhibit well-formed extra-embryonic tissues but a poorly developed embryo proper. By contrast, gynogenetic and parthenogenetic embryos can develop to an apparently normal 25-somite stage embryo with poorly developed extra-embryonic tissues (Kaufman et al. 1977; Barton et al. 1984; McGrath and Solter, 1984; Surani et al. 1984; Barton et al. 1985). These differences persist in chimeras formed between androgenetic↔normal, gynogenetic/parthenogenetic↔normal, and andro-genetic↔parthenogenetic chimeras, such that androgenetic cells are either eliminated or excluded from the embryo proper, whereas gynogenetic/parthenogenetic cells contribute extensively to the embryo proper but are progressively eliminated from extraembryonic tissues (Surani et al. 1986; Surani et al. 1988; Thomson and Solter, 1988; Thomson and Solter, 1989). Additionally, parthenogenetic cells are subject to tissue-specific, negative selection as the chimeric embryo develops (Nagy et al. 1987; Fundele et al. 1989; Nagy et al. 1989; Fundele et al. 1990).

The studies to date indicate that androgenones and gynogenones develop normally throughout the preimplantation period. Both types of embryos have been observed to form blastocysts at appreciable frequencies (McGrath and Solter, 1984; Surani et al. 1986). Neither androgenetic nor parthenogenetic embryos exhibit any statistically significant difference in their ability to contribute to the inner cell mass and trophectoderm portions of blastocysts developing from aggregation chimeras (Thomson and Solter, 1989) and embryonic stem cell lines have been successfully established from both parthenogenetic and androgenetic embryos (Evans et al. 1985; Mann et al. 1990). Despite the ability of androgenones to form apparently normal blastocysts, however, some reduction in developmental potential is evident. The frequency with which androgenones form blastocysts is slightly lower than expected (McGrath and Solter, 1984; Surani et al. 1986) as is the percentage of such embryos that implant (Barton et al. 1984; McGrath and Solter, 1984). Furthermore, androgenones exhibit a delay in their formation of outgrowths when placed in cell culture (McGrath and Solter, 1986). These differences are indicative of a possible effect of genomic imprinting on preimplantation development. We now report that the ability of androgenetic embryos to develop successfully to the blastocyst stage is determined, in large part, by the maternal genotype. Our data provide clear evidence of a pronounced, stage-specific defect in the development of androgenetic embryos prior to implantation that is uncovered when the appropriate maternal strain is used.

Adult C57BL/6, DBA/2 and (B6D2)F₁ females 6– 8 weeks of age (Harlan Sprague-Dawley) were superovulated by i.p. administration of 5i.u. of pregnant mare’s serum gonadotrophin (Diosynth) followed 46 h later by 5i.u. of human chorionic gonadotrophin (hCG) (Sigma) and then mated to C57BL/6, DBA/2 or (B6D2)F₁ males as indicated. Embryos were isolated from the ampullae at the 1-cell stage in Hepes-buffered Whittens medium (Whitten, 1971) and cumulus cells removed with hyaluronidase. Embryos were cultured in CZB medium (Chatot et al. 1989) for 2 days and then switched to Whitten’s medium supplemented with 100μM EDTA (Abramczuk et al. 1977). Nuclear transplantations were performed as described (McGrath and Solter, 1983). Male and female pronuclei were identified by their relative sizes and positions in relation to the polar bodies. To ensure the greatest probabilility of correct pronuclear identification, only embryos with clearly asymmetrically located pronuclei of different sizes were used. Embryonic cell numbers were determined as described (Tarkowski, 1966).

Embryos were labeled with L-[³⁵S]methionine (1mCiml^-1, >1100 Ci mmol^-1, Amersham) at approximately 120 h after injection of the mothers with hCG. After labeling, embryos were lysed in hot (100°C) SDS lysis buffer (Garrels, 1983). Samples were electrophoresed on high-resolution two-dimensional gels using pH 4– 8 ampholines in the first dimension and 10 % acrylamide in the second dimension as described (Garrels, 1983). Following fluorography, quantitative gel image analysis was performed as described (Garrels, 1989; Garrels and Franza, 1989a). Gel images were entered into a quantitative, two-dimensional gel protein database and matched to a series of gel images representing normal embryos labeled at 3 h intervals from fertilization to blastocyst stage. Using this normal series, the patterns of synthesis of individual proteins during normal preimplantation development can be examined (Latham et al. 1991). Spot intensities, indicative of relative rates of synthesis, were quantified and expressed in units of parts per million (ppm) of total incorporated radioactivity. The degree of similarity between gel images was measured by two methods derived from earlier studies (Garrels and Franza, 1989a). The first relies upon the percentage of detected proteins that differ by at least twofold, while the second method uses the standard deviation (S.D.) of spot intensity ratios plotted as a histogram on a logarithmic scale. The latter method accounts for both the number and magnitude of differences between the two gel images being compared. Variation between duplicate gels is typically 2 % two-fold or greater differences and a S.D. of 0.4, while nearly 30 % two-fold or greater differences and a S.D. of 1.18 are observed between proliferating and quiescent REF52 fibroblasts (Garrels and Franza, 1989b).

Androgenetic and gynogenetic embryos were constructed using fertilized eggs from each combination of matings between C57BL/6 and DBA/2 piale and female mice and their developmental .capacities assessed at approximately 120 h post-hCG injection (Table 1). In all combinations, gynogenetic embryos developed to the blastocyst stage at frequencies approaching those observed for unoperated controls, consistent with previous reports (McGrath and Solter, 1984; Surani et al. 1986). In contrast, androgenones constructed with embryos isolated from DBA/2 mothers formed blastocysts at much lower frequencies, with those constructed with DBA/2×C57BL/6 embryos showing the lowest frequency. Although a few androgenones prepared with embryos from DBA/2 mothers formed small fluid-filled cavities, these typically exhibited an abnormal morphology and the blastocoelic cavities failed to expand fully (Fig. 1). The defect in these androgenones did not result from a simple delay in development since the frequency of blastocoel formation was not improved by prolonged culture to 139h post-hCG injection. More than half of the androgenones constructed with embryos from C57BL/6 or (B6D2)F₁ mothers developed to the blastocyst stage, and most of these formed large blastocoelic cavities (Fig. 1). These data indicate that the observed developmental failure was specific to androgenones constructed with DBA/2 eggs.

To characterize the stage at which the androgenones of DBA/2 origin cease developing, we determined the number of cells contained within androgenetic, gynogenetic and unoperated control C57BL/6 homozygous and DBA/2 homozygous and heterozygous embryos at 120 h post-hCG injection (Table 2). Cell numbers were similar in gynogenones and unoperated controls, whereas androgenones of all types generally had fewer cells. However, androgenones constructed with C57BL/6 homozygous embryos and those constructed with embryos from DBA/2 mothers differed in their ability to progress beyond the 16-cell stage; all of the androgenones constructed with C57BL/6 homozygous embryos contained more than 16 cells at this time, but only 6 of 19 androgenones constructed with eggs from DBA/2 mothers contained more than 16 cells and another 6 remained at the 8-cell stage (Fig. 2). Thus, the majority of androgenones constructed with DBA/2 eggs failed to progress beyond the 16-cell stage and many ceased development at the 8-cell stage.

To determine whether the inability of androgenones constructed with eggs obtained from DBA/2 females to develop to the blastocyst stage results from a cytoplasmic insufficiency or from a difference in nuclear function, we transplanted male pronuclei from pairs of fertilized DBA/2×C57BL/6 embryos into enucleated C57BL/6 eggs. Androgenones produced in this manner consisted of C57BL/6 cytoplasms containing two C57BL/6 male pronulclei that had been exposed to DBA/2 cytoplasm (Table 3). Only one of the 28 such androgenones formed a blastocyst. Seventeen of these androgenones failed to develop beyond the 2-cell stage, two developed to only 3 or 4 cells, and eight reached the 8-cell/morula stage. Thus, development of these androgenones appeared more severely impaired than that of androgenones prepared with fertilized DBA/2 eggs alone. Blastocysts formation was also reduced (4/18) in androgenones constructed by transferring male pronuclei from C57BL/6 homozygous eggs to enucleated eggs from DBA/2 mothers. Manipulated control embryos, in which both maternal and paternal pronuclei were transferred from eggs of one strain to enucleated eggs of the opposite strain, developed to the blastocyst stage at a frequency similar to that observed for unoperated embryos (Table 3).

To determine whether the developmental block in androgenones constructed with eggs obtained from DBA/2 females correlated with specific differences in gene expression, embryonic proteins were labeled at 120 h post-hCG and analyzed by quantitative, high-resolution two-dimensional gel electrophoresis. The patterns of proteins synthesized by androgenones, gynogenones and unoperated control embryos were entered into a two-dimensional gel protein database and compared to each other and to a normal developmental series (Latham et al. 1991). Examination of the percent of analyzed proteins that differed by twofold or more and the S.D. values indicated that the gel patterns for gynogenetic embryos of both strains were very similar to the control patterns. The gel patterns for androgenetic embryos were slightly less similar to the control patterns (Table 4). The protein synthesis pattern of androgenones constructed with DBA/2×C57BL/6 embryos was as closely related to that of unoperated DBA/2×C57BL/6 controls and gynogenones as was the pattern for androgenones constructed with C57BL/6×DBA/2 embryos to the C57BL/6×DBA/2 control and gynogenetic patterns (Table 4). Furthermore, comparison of the gel pattern of androgenones constructed with DBA/2×C57BL/6 embryos to that of androgenones constructed with C57BL/6×DBA/2 embryos revealed only 11.6% twofold or greater differences and a S.D. of 0.76, values similar to those for comparisons between unoperated DBA/2×C57BL/6 and C57BL/6×DBA/2 controls (11.8% 3⩾ two-fold differences; S.D. 0.82). Thus, the gel image analysis revealed no substantial difference between the overall protein synthesis patterns of androgenones constructed with DBA/2 ×C57BL/6 embryos and unoperated DBA/2×C57BL/6 controls or C57BL/6×DBA/2 androgenones.

Quantitation of the individual rates of synthesis of proteins resolved from androgenones constructed with eggs from DBA/2 females revealed significant differences in the relative intensities of several spots. Eleven spots differed by two-fold or more in duplicate gels of androgenones prepared with DBA/2 ×C57BL/6 embryos as compared with unoperated control DBA/ 2×C57BL/6 embryos (Fig. 3). Eight of these eleven proteins increased in synthesis during normal development to the late blastocyst stage but were synthesized at reduced rates in the androgenetic DBA/2 embryos relative to the control and gynogenetic embryos (Fig. 3). Two of these eleven proteins (no. 229 and 566) were also reduced in androgenones constructed with C57BL/6 homozygous embryos. Three of the eleven proteins (no. 202, 332, 552) decreased in synthesis during normal development from 8-cell to the blastocyst stage but were synthesized at elevated rates in the androgenones constructed with DBA/2×C57BL/6 embryods (Fig. 4). Two proteins (no. 202 and 229) were affected in a reciprocal manner in gynogenetic embryos as compared with androgenenones.

Our results indicate a clear effect of the egg cytoplasm on the developmental capacity of androgenetic embryos. Androgenetic embryos prepared with eggs from C57BL/6 or (B6D2)F₁ females, or other strains (McGrath and Solter, 1984; Surani et al. 1986), develop to the blastocyst stage at a frequency of approximately 45– 55%, while androgenones prepared with DBA/2 eggs are essentially unable to form an expanded blastocyst and tend to block at the 8- to 16-cell stage. This inability of androgenones constructed with eggs from DBA/2 mothers to form blastocysts is independent of the genotype of the male pronucleus. Since C57BL/6 male pronuclei transplanted from DBA/ 2×C57BL/6 embryos to enucleated C57BL/6 eggs remain unable to support androgenetic development to the blastocyst stage, it is unlikely that this defect is attributable simply to some cytoplasmic deficiency peculiar to the DBA/2 strain. Rather, the defect apparently results from a stable, strain-dependent difference in modification of the male pronucleus by the DBA/2 egg cytoplasm that occurs after fertilization. This modification is specific to the male pronucleus, since transplantation of both maternal and paternal pronuclei from a DBA/2×C57BL/6 embryo to an enucleated C57BL/6 egg produces embryos with a normal capacity for blastocyst development. Moreover, the reduced developmental capacity of androgenones constructed by transferring two male pronuclei from C57BL/6 eggs to enucleated DBA/2 eggs indicates that this ability of the DBA/2 cytoplasm to modify negatively a male pronucleus probably remains for some period of time following pronuclear formation.

Strain-dependent differences in expression have been reported for certain transgenes. Expression of the CMZ12 transgene locus, for example, is enhanced when (C57BL/6×CBA)F₁ transgenic males are mated to DBA/2 females but suppressed when the same males are mated to BALB/c females (Surani et al. 1990). Suppression also occurs in the reciprocal cross when transgenic females are mated to DBA/2 males, further indicating that expression is dependent upon maternal genotype (Surani et al. 1990). A similar pattern of expression is observed with the TKZ751 transgene locus; expression is enhanced in DBA/2 eggs but suppressed in BALB/c eggs (Allen et al. 1990; Surani et al. 1990). In addition, suppression occurs only when the transgene is supplied by the paternal genome and not when inherited from transgenic females mated to BALB/c males (Surani et al. 1990). The molecular basis for this strain-dependent variation in .transgene expression may be related to differences in methylation (Allen et al. 1990; Surani et al. 1990). In fact, Engler et al. (1991) recently described a strain-specific modifier locus on mouse chromosome 4 that is responsible for differential methylation of a transgene inherited through C57BL/6 as compared with DBA/2 mothers.

The strain-dependent expression of these transgenes has led to the proposal that germ line-specific modifications (imprinting) of the genome interact with strainspecific ooplasmic factors (modifiers) following fertilization or, alternatively, with the products of the activated maternal genome, such that the level of expression is dependent upon the maternal genotype (Sapienza, 1989; Surani el al. 1990). Such modifiers could, in principle, account for the effect of the maternal genotype on androgenone development reported here. This would be the first demonstration that strain-specific modifiers influence endogenous genes that are expressed in the early embryo.

If such modifiers do operate in the early embryo, it is clear that they do not lead to extensive alterations in the overall pattern of gene expression. The only reproducibly detected differences in protein synthesis patterns of androgenones were a few quantitative alterations observed primarily in androgenones constructed with DBA/2 eggs. These differences indicate that a portion of the normal progression of changes in gene expression is either delayed or interrupted as a result of a straindependent difference in egg constitution. Whether these alterations are a direct effect of the action of maternally derived factors and are themselves responsible for the morphological block at the 8- to 16-cell stage, or an indirect consequence of this developmental block remains to be determined.

Our results superficially resemble those reported for DDK mice in that the embryos from DDK females mated to non-DDK males also block between the 8-cell and blasotcyst stage (Renard and Babinet, 1986) and that this defect results from modification of the male, but not the female, pronucleus (Renard and Babinet, 1986; Babinet et al. 1991). In addition, the DDK egg cytoplasm retains the ability to affect the development of embryos bearing non-DDK male genomes even up to the 8-cell stage (Babinet et al. 1991). The effects of DDK cytoplasm differ significantly from those of DBA/2 cytoplasm, however, in that they are dependent upon the paternal genotype and they are, to some degree, reversible following transplantation of pronuclei from DDK to non-DDK eggs (Renard and Babinet, 1986). Furthermore, non-DBA/2 male pronuclei do not adversely modify the DBA/2 egg cytoplasm, as observed with DDK eggs fertilized by non-DDK sperm, since the transfer of maternal and paternal nuclei together from C57BL/6 eggs to enucleated DBA/2 eggs that had been fertilized by C57BL/6 males produces embryos that develop with high efficiency.

That many of the androgenones constructed by transferring two male pronuclei from fertilized DBA/2 eggs to enucleated homozygous C57BE/6 eggs fail to progress beyond the 2-cell stage indicates that exposure of a male pronucleus to DBA/2 egg cytoplasm may reduce its ability to function within a C57BL/6 cytoplasm, producing a more severe phenotype than that observed with androgenones constructed with fertilized DBA/2 eggs alone. One possible explanation for this is that strain-dependent modifications of male pronuclei are a general feature of early mouse development, with the eggs of each strain imposing slightly different modifications that render the male pronuclei compatible with a particular cytoplasm. These data further indicate that genomic imprinting may affect genes that are involved in early preimplantation as well as postimplantation development so that, for certain strains of eggs, a female pronucleus is essential for successful development even as early as the 2-cell stage.

We thank Frank Rauscher and James McGrath for critical reviews of the manuscript. This work was supported in part by US Public Health grants HD-17720, HD-23291, and HD-21355 from NICHD and CA-10815 from NCI. K.E.L. was supported by a training grant from the NCI (CA 09171-14).