Influence of paternally imprinted genes on development

Evidence suggests that the parental genomes are functionally non-equivalent during development (Surani and Barton, 1983; Barton et al. 1984, 1985; McGrath and Solter, 1984; Mann and Lovell-Badge, 1984; Surani et al. 1984). This is because of a group of ‘imprinted’ genes whose expression is determined by their parental origin (Cattanach, 1986; Solter, 1988; Reik et al. 1990; Surani et al. 1990a). Imprinting confers preferential expression (or repression) of one of the homologous parental alleles, most probably through a heritable epigenetic modification of the chromatin (Solter 1988; Monk, 1988; Holliday 1990; Surani et al. 1990a). It follows therefore that the duplication of maternal or paternal chromosomes (with corresponding deficiencies) harbouring imprinted genes will produce a gain or loss of function and associated phenotypic effects.

Genetic complementation tests reveal both the presence and the effects of subsets of imprinted genes, such as those on chromosome 2, 6, 7, 11 and 17 (Searle and Beechey, 1985; Cattanach and Kirk, 1985; Cattanach, 1986). The identity of three imprinted genes, insulin-like growth factor-2 (Igf2) (DeChiara et al. 1991; Ferguson-Smith et al. 1991) its receptor (Igf2r) (Barlow et al. 1991) and the H19 gene (Bartolomei et al. 1991) has now been established; the paternal allele of the lgf2 and the maternal alleles of Igf2r and H19 are preferentially expressed. Although the phenotypic effects of some of the imprinted genes individually may be too subtle and therefore escape detection, the combined action of some or all of the imprinted genes may produce more dramatic phenotypic effects. It is vital to establish how these genes interact during development.

A potential way to test the cumulative influence of imprinted genes during development is by analysing chimeras in which the experimental and normal cells are allowed to interact. Cell –cell interactions as well as autocrine/paracrine effects, both short and long range, have a key role in cell proliferation, differentiation and pattern formation. The presence of normal cells may effect a ‘rescue’ of experimental cells, allowing them to function to a far greater extent than they are able to do in isolation. Conversely, the experimental cells will exert an influence on development through excess or inappropriate production of some factor(s) and/or because of aberrant interactions with normal cells.

A number of studies have been reported previously using aggregation chimeras between androgenetic (AG ↔F) or parthenogenetic (PG ↔F) embryos and normal embryos (Thomson and Solter, 1987, 1988; Fúndele et al. 1989, 1990; Clarke et al. 1988; Nagy et al. 1989; Surani et al. 1988). Many such chimeras are embryonic lethal especially if the contribution of the experimental cells is very large. The PG ↔F chimeras have been studied most extensively; these are growth retarded and display an uneven distribution of PG cells in the embryo; PG cells are detected consistently in the brain, epidermis and in germ cells which results in viable oocytes, but seldom in skeletal muscle (Stevens, 1978; Fúndele et al. 1989, 1990; Nagy et al. 1989). These phenotypic effects are largely independent of the mouse strain used to derive the PG cells (Fúndele et al. 1991).

Previous studies on AG ↔F chimeras showed that the AG cells were primarily confined to the extraembryonic tissues (Thomson and Solter, 1987, 1988; Surani et al. 1988). However, at least three AG ↔F chimeric embryos have been reported but these embryos were either retarded or dead (Thomson and Solter, 1988; Surani et al. 1988). We have therefore carried out further studies in which chimeras were prepared by injecting the inner cell mass directly into normal blastocysts. These AG/ICM →F chimeras developed more efficiently compared with the AG ↔F aggregation chimeras, and this provides an opportunity to define better the role of the paternal genome during embryonic development. These results can also be compared with the results recently reported by Mann et al. (1990) in which chimeras were prepared with androgenetic embryonic stem cells; these AG/ES → F chimeras developed to term and the AG cells contributed to a large number of embryonic tissues and resulted in severe abnormalities of skeletal elements. Comparisons of the ICM and ES results are vital in order to know if the ‘imprints’ are stable during the derivation and prolonged culture of ES cells, or if these modifications are progressively lost from imprinted loci. It has been suggested recently that the developmental potential of ES cells derived even from normal blastocysts may be affected by the removal of parental imprints during their passage in tissue culture (Nagy et al. 1990). Second, it is also possible, although unlikely, that the phenotypic effects observed with the androgenetic ES cells are due to mutations introduced at some stage of the derivation of these cell lines.

We report our observations with AG/ICM →F, compare them with the AG/ES →F chimeras and describe some phenotypic effects not reported previously. The results show striking reciprocal phenotypes between AG/ICM →F and GG/PG/ICM →F with regard to embryonic growth and the distribution of cells in the developing embryo.

Mice used in these experiments were (C57BL/6J ×CBA/ Ca)Fj (Gpi-1b/Gpi-1b) (AFRC colony bred from Bantin and Kingman stock), hereafter referred to as F₁, 129/HD (Gpi-la/Gpi-la, dilute pink eyed from OLAC Ltd.) and CFLP outbred albino (Gpi-1a/Gpi-1a), bred from OLAC stock. Animals were housed in standard conditions with a 14 h light and 10 h dark lighting schedule.

To obtain fertilized 1-cell eggs and blastocysts, 3- to 4-week-old females were superovulated by injection of 7.5 i.u. pregnant mare’s serum (PMSG, Intervet Ltd., Cambridge, UK) followed 48 h later by an injection of 7.5 i.u. human chorionic gonadotrophin (hCG, Intervet), caged singly with stud males and checked for vaginal plugs the following morning; the day of finding the plug was designated day 1.

To obtain pseudopregnant recipients, naturally cycling mature F₁ females were mated with vasectomized males of proven sterility.

Fertilized 1-cell eggs were collected from F₁ females mated with 129/HD males 20h after hCG injection. Diploid AG eggs were constructed by nuclear transplantation (Magrath and Solter, 1983) using a Leitz micromanipulator, in phosphate-buffered medium (PB1) with 0.4% bovine serum albumin supplemented with 5 μgml^-1 cytochalasin-B and 1.5 μgml^-1 nocodazole (both from Sigma), using inactivated Sendai virus for membrane fusion (Barton et al. 1987). The 129/HD AG eggs (with F₂ cytoplasm) were then cultured in T6 medium (Quinn et al. 1984; Howlett et al. 1987) with 0.4 % BSA under oil in a humidified atmosphere of 5 % CO₂ at 37.8°C for 5 –6 days to the advanced blastocyst stage. The inner cell masses (ICMs) were isolated by immunosurgery (Solter and Knowles, 1975) and stored in separate drops of PB1 with 10% heat-inactivated fetal calf serum (FCS). A limited number of chimeras were also prepared with androgenetic ICMs derived from (C57BL/6 ×CBA)F₂ blastocysts. These ICMs were introduced into CFLP outbred blastocysts.

A limited number of gynogenones (GG) were prepared mainly for concurrent comparisons with AG. Eggs were collected as above from 129/HD females mated with F₁ males and heterozygous diploid GG eggs constructed (Barton et al. 1987). Because eggs with 129/HD cytoplasm did not develop in culture as well as those with F₂ cytoplasm, the reconstituted eggs after overnight culture were transferred as 2-cell GG 129/HD embryos to the oviducts of day 1 pseudopregnant females; these embryos were retrieved from the uterus 3 days later and cultured for a further day as above (to day 6) before immunosurgery. A small number of gynogenetic eggs were also constructed from (C57BL/6 ×CBA)F₂ eggs. The ICMs from these blastocysts were isolated and introduced into CFLP host blastocysts.

On the morning of day 4 of pregnancy, F₂ blastocysts were flushed from the uteri of superovulated F₁ females previously mated with Fj males. The embryos were kept in T6+0.4% BSA and only well-expanded but zona-intact blastocysts were used for injection.

The two-needle method of blastocyst injection modified from Gardner (1978) and Babinet (1980) was used. The operation was carried out in PB 1+FCS (fetal calf serum), but the ICMs were held briefly in separate drops in the manipulation chamber of PB1+FCS supplemented with 5 μg ml^-1 cytochalasin B and 1.5 μgml^-1 nocodazole. The transfer pipettes were bevelled, flame polished and sharpened and had an outside diameter of 28 microns (inside diameter was approximately 20 microns). Injection was through the side of the blastocyst close to the ICM.

After operation the blastocysts were cultured for 1 –2 h in PB 1+FCS and then transferred to one horn of the uterus of day 3 pseudopregnant recipients. Control blastocysts were usually transferred to the contralateral horn.

Potential chimeric embryos were examined from recipient day 12 onwards. Embryos were dissected out, measured, weighed and sometimes photographed, and the majority were dissected out for GPI analysis on Titan III cellulose acetate plates (Helena Labs) (Eicher and Washburn, 1978). The contribution from experimental cells to chimeric embryos was estimated from these GPI analyses as described previously (Barton et al. 1985). A few were fixed and wax embedded for histological processing. A small group of both AG/ICM →F and GG/ICM →F chimeras were left to term. The one very abnormal AG/ICM →F male that survived for 7 days after birth was killed, dissected of its soft parts for GPI analysis, and the remaining carcass processed for whole-mount cartilage and bone staining (McLeod, 1980). GG →F surviving chimeras were test bred for germline chimerism for viable GG-derived oocytes.

The total number of AG and GG eggs and reconstituted blastocysts with the AG or GG inner cell masses are shown in Table 1, together with the number of reconstituted blastocysts that developed after implantation and were found to be chimeric.

The development of chimeras with the androgenetic cells derived from 129/HD strain is summarised in Table 2. Live chimeric conceptuses with a substantial AG component were obtained at a moderate frequency up to about day 14 of gestation; 5 out of 17 five chimeric day 12 and 13 embryos were 50% or more AG. After this stage only chimeras with a minor AG contribution survived and all embryos with 50% or more AG contribution from day 15 onwards were found to be dead.

A number of conceptuses showed chimerism in the yolk sac only with a contribution from androgenetic cells ranging from 5 to 50 %. However, there were no detectable phenotypic effects of this contribution either in the yolk sac itself or the embryo (Table 2 and 3).

The presence of AG cells had a significant effect on embryonic growth (Table 3). This is illustrated in Fig. 1 with respect to embryos obtained on day 13 of gestation. The marked increase in the weights of embryos observed was proportional to the contribution from the AG cells provided these cells were present within the embryo proper. Hence, the 14 conceptuses of all stages in which these cells were present in the yolk sac only did not show these phenotypic effects (Table 2 and 3). With the increase in the contribution of AG cells and embryonic weights, there was also a corresponding increase in the anterior-posterior axial length (Fig. 1). This lengthening produced a characteristic change in shape with the embryo apparently appearing to be thinner and longer with a pronounced elongation of the back. These chimeras were easily distinguishable from non-chimeric siblings and typing by this criteria was subsequently confirmed by GPI analysis. Indeed, the extent of contribution from AG cells could be estimated from the severity of this phenotype.

The changes in the overall size and shape of the chimeric embryos prompted us to carry out a histological examination to determine if there were any major structural differences detectable. This examination was carried out on embryos that were still alive and not on those that were dying due to the severity of the effects. We did not observe any pronounced changes in the size or number of skeletal elements associated with the increase in the anterior –posterior axis in chimeric embryos (data not shown). While some overall differences were evident, no other overt phenotypic effects were seen except in the heart. During dissection of the chimeric embryos, we had observed that the heart was often enlarged especially when the contribution of the AG cells was high; indeed this was a tissue that often had a high contribution from androgenetic cells (see below). The histological sections clearly revealed that there was some enlargement and disorganisation associated with the development of this chimeric organ that could not however, be associated with any other detectable changes in the tissues. We cannot rule out the possibility that examination of additional samples may reveal other effects because these changes are dependent on the extent of contribution as well as the distribution of AG cells.

The distribution of AG cells was analysed using the GPI allozyme marker. Because of the small sample size, the data are presented in a composite format to reveal more easily the overall trends in the distribution of androgenetic cells (Fig. 2). As shown, AG cells contributed significantly to the chimeric conceptuses on day 12 and indeed one embryo was composed of about 60 % AG cells (Fig. 2A). We examined a number of tissues and found that AG cells were present in the majority of these tissues (data not shown). The contribution of AG cells detected in living embryos declined progressively during the course of development. This may not necessarily reflect cell selection since it is probable that embryos that continued to retain a high proportion of AG cells were not viable and were consequently resorbed at later stages. However, the distribution of AG cells was non-uniform as development progressed. The contribution of AG cells to the three major categories, heart, carcass and brain are most informative in this regard (Fig. 2A). The contribution to the heart was very substantial and consistent, a feature shared with the skeletal muscle. By contrast, the contribution of AG cells to the brain was amongst the least that we detected, especially as development progressed beyond day 13 of gestation (see below).

Since a few chimeras with a low AG contribution survived to day 16 of gestation, we allowed some reconstituted blastocysts with small AG ICMs to progress to term. Two chimeras were born. One animal was found dead shortly after birth but when it was recovered no overt abnormalities were observed nor was there any sign of chimerism in the eye. Nevertheless, some of the tissues were dissected from this animal which was partly decomposed; it was found to be chimeric with very considerable contribution of AG cells in some tissues, such as the heart, whilst the contribution to the brain was slight (Fig. 3). The second animal survived, but by day 7 after birth it was obviously retarded and had apparently developed severe skeletal abnormalities (Fig. 4A). We also found deformities affecting the hind limbs which displayed syndactyly. The animal was therefore killed on day 7 after birth and carefully dissected so that all the soft tissues could be analysed for chimerism. Once again, there was widespread distribution of AG cells to many tissues, except for the brain (data included in Fig. 2A). The skeleton was prepared and stained with alcian blue and alizarin red. This preparation clearly revealed severe malformation of the skeletal elements (Fig. 4B). The sternum was considerably compressed which resulted in the ribs being much closer together at their proximal ends. The ribs themselves were enlarged, disorganised and twisted. The alcian blue stain for cartilage revealed an excessive amount of this tissue with reduced ossification shown by the lack of red staining. The vertebrae especially at the posterior end were also somewhat thickened and enlarged. The spine was severely scoliotic. These phenotypic effects on skeletal elements are similar to those in AG/ES →F chimeras reported previously (Mann et al. 1990).

During the course of these studies, we also prepared a number of chimeras in which the androgenetic cells were of the (C57BL/6 ×CBA)F₁ background. Although these chimeras were used for other tissue culture and histological experiments, we did observe enhanced growth in these chimeras by up to approximately 50% just as described above in the case of 129/HD strain. One live chimera was also obtained containing androgenetic (C57BL ×CBA)F₂ cells. At birth, this chimera was 2.3 g compared to the mean weight of 1.8±0.2 for its 6 littermates. However, by day 15 the chimera was smaller than its siblings and thereafter began to lose weight. The chimera appeared to show malformation of the sternum which was very prominent, the rib cage appeared enlarged, the spine scoliotic and the tail thickened. The animal was killed on day 20 and the internal organs were analysed for GPI. Once again androgenetic cells were detected in many tissues and organs especially the heart, liver, spleen and pancreas, and far less in the brain. The carcass is being prepared for histological examination.

The observations described here appear in many respects to be phenotypically reciprocal to those that we and others have described previously for the chimeras containing parthenogenetic or gynogenetic cells. This applies to the size differences as well as to the distribution of PG/GG cells in chimeras. In order to confirm that GG cells of the genotypes used in the present experiments showed the same phenotype, we carried out a limited number of concurrent experiments in which a GG inner cell mass was introduced into normal blastocysts to generate GG/ICM →F chimeras (Table 4). The results are consistent with data that we previously obtained with PG/GG ↔F aggregations (Fundele et al. 1989, 1991). These chimeras were growth retarded and therefore reciprocal in size to the ones that contained AG cells (Table 4, Fig. 5), which confirms the more extensive previous observations (Fundele et al. 1989, 1990; Nagy et al. 1989). We also examined the distribution of GG cells in these few chimeras. As shown previously (Fundele et al. 1989), the results showed that GG cells contributed relatively more to the brain but less to the skeletal muscle and heart (Fig. 3B). Four GG/ICM →F chimeras were allowed to develop to term. Two of these died around birth and two (one male and one female) are healthy and fertile and show germline chimerism in the female (see Table 4), as has been observed previously (Stevens, 1978; Fúndele et al. 1989, 1990; Nagy et al. 1989). In no instance were the gross abnormalities of skeletal elements observed with AG/ICM →F chimeras detected. However, a more detailed examination is required in the case of GG/ICM → F chimeras in order to discover if more subtle abnormalities exist.

Androgenetic (AG) conceptuses develop to a maximum of 6- to 8-somite stage with relatively normal extraembryonic tissues (Barton et al. 1984; McGrath and Solter, 1984; Surani et al. 1984). However, in the presence of normal cells in chimeras, AG cells undergo more extensive cell differentiation. However, a contribution of AG cells in excess of 50% is usually lethal. More significantly, the phenotypic effects induced by androgenetic cells are reciprocal to those induced by gynogenetic/parthenogenetic (GG/PG) cells in chimeras (Surani el al. 1990b).

Compared to previous studies on aggregation AG →F chimeras (Thomson and Solter, 1988; Surani et al. 1988), there was a substantial improvement in the development of AG →F injection chimeras. We attribute this improvement to several factors. First, there is a much greater selection of AG embryos before chimeras are prepared. Second, deriving the AG ICMs from day 6 –7 blastocysts means that the donor AG cells are considerably more advanced than the host day 4 blastocysts. This aspect may be very significant since we have observed that AG embryos develop more slowly to the blastocyst stage than do normal embryos, which makes it less likely that the AG cells can contribute to the ICM in aggregation chimeras. Finally, producing chimeras by injection is more efficient than by aggregation, so long as contribution to the trophoblast is not an object of the study.

In our studies, the presence of AG cells enhanced embryonic growth by as much as 50%, and altered its shape due to a concomitant increase in the anterior – posterior axial length. These effects were proportional to the contribution from AG cells; the greater the contribution, the greater the effect. Such effects were not reported in the case of AG/ES → F chimeras (Mann et al. 1990). Conversely, in the presence of GG/PG cells in chimeras, embryonic size was reduced by a similar proportion. Growth regulation normally occurs during early postimplantation (Buehr and McLaren, 1974; Lewis and Rossant, 1982; Rands, 1986). Hence, doubling the size of the normal inner cell mass in control blastocysts still produces normal size fetuses (Ferguson-Smith et al. 1991). Our studies show that the AG/ICM →F chimeras were larger than controls on day 12 of gestation and onwards. It is clear that size regulation fails in chimeras, in the presence of AG or GG/PG cells, but the precise mechanism involved is not known.

Classic genetic studies have shown that distal chromosome 7 (Chr. 7) (see below), proximal Chr. 11 and distal Chr. 17 apparently contain imprinted genes that affect embryonic growth (Cattanach and Beechey, 1990). In all these instances (except for the distal Chr. 17 region), the paternal duplication with corresponding maternal deficiencies of these chromosomal regions causes enhanced growth; this is clearly seen in the case of Chr. 11 in which maternal duplication/paternal deficiency leads to growth retardation (Cattanach and Kirk, 1985).

The influence of imprinted genes in chimeras could be cell autonomous or act through interactions between cells. The overall distribution of AG and GG/PG cells in chimeras suggests that the effects of imprinted genes can be lineage specific. AG cells have a tendency to contribute disproportionately to the mesodermal tissues, while the GG/PG cells contribute to the neuroectodermal tissues. For instance, AG cells were detected in skeletal muscle, a tissue to which GG/PG cells make little or no contribution (Fúndele et al. 1989, 1990; Nagy et al. 1989). It is interesting to note therefore that the AG embryonic stem cells when introduced in a subcutaneous ectopic site, produced tumours consisting principally of striated muscle (Mann et al. 1990). Conversely, GG/PG cells in chimeras make a significant contribution to the brain and epidermis to which the AG cells make an apparently reduced contribution. The AG (and probably PG/GG) cells must also produce phenotypic effects as a result of short-range autocrine/paracrine factors since even a very large contribution from AG cells to the yolk sac alone had no detectable phenotypic effects.

The increased contribution of androgenetic cells to the mesodermal derivatives also caused abnormalities of the skeletal elements. The most significant of these effects was observed in the sternum, which was compressed, and the ribs, which were enlarged, twisted and fused, with considerable amounts of cartilaginous material. Such abnormalities were first observed in AG/ES/ICM →F chimeras, and may be associated with an increased proliferation of chondrocytes (Mann et al.1990). We propose that the whole population of these cells, including the normal cells, may proliferate abnormally in response to the altered dosage of autocrine/paracrine factors in this and other tissues such as the heart. This inference is based on the relative intensities of the GPI-1 markers representing the control and experimental cells which remained equal. It is also possible that the abnormal dosage of imprinted genes may be involved in disruption of positional information resulting in the skeletal and axial anomalies. However, more detailed analysis with in situ markers is essential to deduce the precise fate of AG and GG/PG cells in particular tissues and organs.

Some of the phenotypic effects on growth could be attributed to the imprinting of the insulin-like growth factor-2(Igf2) (DeChiara, 1991; Ferguson-Smith et al.1991) and one of its receptors (Igf2r) (Barlow et al. 1991). These genes are co-ordinately expressed during development immediately after implantation in the extraembryonic and embryonic tissues, principally in the mesodermal derivatives in the rat and the mouse (Beck et al. 1987; Stylianopoulou et al. 1988; Senior et al. 1990; Lee et al. 1990; Ohlsson et al. 1989). Indeed, there is evidence that the Igf2 is overexpressed in AG conceptuses while its expression is undetectable in PG conceptuses (R. Ohlsson, A. Glaser, A.F-S., S.C.B., M.A.S., unpublished). Igf2 is regarded as an embryonal mitogen (Daughaday and Rotwein, 1989), whose excess expression in AG cells with two active paternal alleles could clearly produce enhanced embryonic growth in AG/ICM →F chimeras. Furthermore, AG cells contribute preferentially to the extraembryonic tissues, as well as to the mesodermally derived tissues such as heart and muscle, the very tissues in which the Igf2 gene is expressed (Stylianopoulou et al. 1988; Lee et al. 1990).

It is important to note that although there is some evidence to show that Igf2r is capable of signal transduction, its primary role may be to regulate the levels of Igf2 by removing any excess ligand (Senior et al. 1990; Haig and Graham, 1991). The fact that the ligand and the receptor are expressed, both temporally and spatially in a coordinated fashion during development, is consistent with the view that Igf2 may be involved in short-range autocrine/paracrine control of growth, and has its biological effect via another receptor such as Igf1r. The androgenetic cells are therefore likely to have a growth-enhancing effect both because of the excess production of the Igf2 and the absence of the Igf2r. Conversely, the PG/GG cells will be deficient in the production of lgf2 ligand but will have an excess of Igf2r. This may partly account for the reduction in the size of the GG/ICM →F chimeras as the deficiency of the Igf2 ligand is coupled with the greater capacity for removing the ligand due to the excess Igf2r present on GG/PG cells. This proposal is testable although there is no direct evidence for this notion at present. We recently demonstrated that the paternal duplication of distal Chr. 7 (PatDi7), which harbours the Igf2 gene, in the PatDi7/ICM →F chimeras resulted in approximately 50 % increase in the size of embryos (Ferguson-Smith et al. 1991). This is analogous to the phenotype in the AG/ICM →F chimeras.

While the effects of AG cells derived from ICMs in our study on skeletal elements are similar to those obtained with AG embryonic stem cells (Mann et al. 1990), the effects on embryonic growth were not reported previously. Hence at least some genes must remain stably imprinted in embryonic stem cells. Therefore, while the androgenetic stem cells may prove a valuable source of material to investigate the role and mechanism of imprinting, we cannot at present entirely rule out the possibility that some ‘imprints’ may not be stable in these cells. It is not possible at present to address fully the question of the influence of strain differences, although from the limited number of our observations, the results with the 129/HD and (C57BL ×CBA)F₁ are on the whole similar. A more extensive series of observations on parthenogenetic cells from a number of strains demonstrated that the strain differences do not affect the phenotype very markedly (Fundele et al. 1991). The task now is to identify the other imprinted genes and to determine their phenotypic effects and their role during embryonic development, together with the molecular mechanism of genetic imprinting.

We thank Walter Mills for help with the analysis of chimeric tissues and Dianne Styles for preparing the manuscript. We are grateful to all our colleagues for their help and advice particularly Peter Jones, Nick Allen and Wolf Reik. A.C.F-S is a Babraham Research Fellow. R.F. was supported by a grant from the Wellcome Trust.