Loss of EED in the oocyte causes initial fetal growth restriction followed by placental hyperplasia and offspring overgrowth

Germline epigenetic programming, including genomic imprinting, substantially influences offspring development. Polycomb Repressive Complex 2 (PRC2) plays an important role in Histone 3 Lysine 27 trimethylation (H3K27me3)-dependent imprinting, loss of which leads to placental hyperplasia in mammalian offspring generated by somatic cell nuclear transfer (SCNT). In this study, we show that offspring from mouse oocytes lacking the Polycomb protein Embryonic Ectoderm Development (EED) were initially growth restricted, characterised by low blastocyst cell counts and substantial mid-gestational developmental delay. This initial developmental delay was followed by striking late-gestational placental hyperplasia, fetal catch-up growth and extended gestational length that culminated in offspring overgrowth. This involved remodelling of the placenta, including expansion of fetal and maternal tissues and conspicuous expansion of the glycogen enriched cell population in the junctional zone that was associated with a delay in parturition. Despite this remodelling and offspring catchup growth, fetal/placental weight ratio and fetal blood glucose levels were low indicating low placental efficiency. Genome-wide analyses identified extensive transcriptional dysregulation in affected placentas, including a range of imprinted and non-imprinted genes and increased expression of the H3K27me3-imprinted gene Slc38a4, which regulates transport of essential amino acids in the placenta. Our data provide an explanation for apparently opposing observations of growth restriction and overgrowth of offspring derived from Eed-null oocytes and demonstrate that PRC2-dependent programming in the oocyte regulates fetal and placental growth and developmental outcomes.

In mouse oocytes, PRC2 is required for non-canonical H3K27me3-imprinting, whereby H3K27me3 silences the maternal allele of several genes resulting in paternal allele-specific expression of the affected genes in pre-implantation embryos and extraembryonic tissues (21,22). Oocyte-specific deletion of Eed using Gdf9Cre resulted in loss of H3K27me3imprinting in preimplantation embryos and extraembryonic ectoderm with growth restriction and male-biased lethality in mid-gestation offspring (23). However, in contrast to growth restriction, another study that employed a similar strategy using Zp3Cre to delete Eed in oocytes found that postnatal offspring were overgrown (7). Moreover, somatic cell nuclear transfer (SCNT) embryos lack H3K27me3-imprinting, resulting in bi-allelic expression of several genes that has been linked to placental hyperplasia and fetal overgrowth (24)(25)(26)(27).
In this study, we hypothesised that loss of EED in the oocyte establishes a developmental trajectory that involves initial fetal growth restriction followed by catch-up growth, explaining the overgrowth observed in early post-natal offspring. We demonstrate that while fetuses derived from Eed null oocytes were initially developmentally delayed and growth restricted, they underwent rapid catch-up growth that culminated in postnatal overgrowth. This complex growth curve occurred despite placental hyperplasia and reduced placental efficiency late in gestation. In addition, expression of H3K27me3-imprinted amino acid transporter Slc38a4 was increased, gestational length was extended, contributing to the resolution of fetal growth restriction and ultimately resulting in postnatal overgrowth.
Together, this work reveals an essential role for EED in the oocyte for regulating a complex program of fetal and placental growth in offspring, and a remarkable capacity for fetal growth restriction caused by loss of EED in oocytes to be resolved.

Loss of maternal EED compromised mid-gestation survival in offspring.
Previous studies demonstrated apparently contradictory findings in that deletion of Eed in oocytes resulted in early embryonic growth restriction (23) and early postnatal offspring overgrowth (7). To understand how these outcomes are realised, we produced offspring from pure C57BL6 females producing Eed-homozygous null (hom), Eed heterozygous (het) or Eed wild type (wt) oocytes and assessed their growth and development throughout the whole gestational period. Initially, Eed fl/fl or Eed fl/wt females were mated to males that were transgenic for Zp3Cre to generate females producing Eed-wt, Eed-het or Eed-hom oocytes (7).
Mating of these females to pure C57BL6 wt males allowed us to generate isogenic wild type or heterozygous offspring as previously described (Fig. 1A) (7). Heterozygous offspring generated from Eed heterozygous oocytes (HET-het offspring) or from Eed homozygous null oocytes (HET-hom offspring) are isogenic. However, the HET-hom offspring were generated from oocytes that completely lacked functional EED and the HET-het offspring were generated from oocytes that had one functional copy of EED and maintained essentially normal gene repression (7 (28). As they are isogenic and heterozygous, comparison of HET-hom with HET-het offspring allowed the identification of differences that resulted specifically from a loss of EED in oocytes in the absence of confounding genetic differences (Fig. 1A). In addition, we generated wild type offspring from Eed-het and Eed-wt oocytes to compare WTwt and WT-het offspring with HET-het and HET-hom offspring, providing controls for differences generated as a result of Eed heterozygosity in offspring (Fig. 1A).
Initially, we used automated time-lapse imaging of individual embryos derived from Eed-wt and Eed-hom oocytes to track their development from zygote to hatched blastocyst stages.
Embryos from Eed-hom oocytes reached the 2-cell stage 1.08h earlier than embryos from Eed-wt oocytes (Fig. 1B), but blastocyst expansion and hatching took 2.67h longer in HET-hom compared to WT-wt embryos (Fig. 1C). All other developmental milestones to blastocyst stage were similar between genotypes (Fig. 1B). Although, viability to blastocyst stage was lower in embryos from females producing Eed-hom compared to Eed-wt oocytes (72.17% vs 88.67%, Fig. 1D), the proportion of embryos surviving to expanded and hatched blastocyst stages did not differ significantly (Supp Fig. 1A).
To determine whether the cell content of blastocysts was affected, we performed cell counts in whole blastocysts and differential staining of inner cell mass (ICM) and trophectoderm (TE) cells (29) (Supp Fig. 1B). This revealed that overall cell content of embryos from Eed-hom females was lower than in Eed-wt controls (Fig. 1E). The number of TE cells was not statistically altered, but the ICM contained fewer cells in HET-hom embryos compared to WTwt controls (Supp Fig. 1C-D). Together, these data indicated that HET-hom embryos contained fewer cells overall than controls and that the ICM was smaller in embryos from oocytes that lacked EED compared to oocytes that maintained EED function (Fig. 1F). As previous reports indicated that maternal deletion of Eed did not compromise development to the blastocyst stage, or cause apoptosis in blastocysts (23), it is likely that lower proliferation explains the reduction we observed in blastocyst cell number, which affected the ICM rather than the TE.
To assess the effect of oocyte-specific Eed deletion on fetal survival and throughout pregnancy, we compared the number of live fetuses and litter size in offspring generated from Eed-hom, Eed-het and Eed-wt oocytes at multiple stages during gestation and after birth.
Using Zp3Cre to delete Eed in oocytes, we found no difference between genotypes in the number of implantations of live embryos in four pregnancies from Eed-null oocytes analysed at E9.5 (Fig. 1G), indicating that HET-hom preimplantation embryos implanted at similar rates and progressed through gastrulation. However, consistent with previous findings (7,23), the number of live fetuses at E12.5, E14.5, E17.5, and the number of live born pups were both significantly lower for Eed-hom mothers than for Eed-het and Eed-wt control females (Fig.   1G). In addition, 71.43% of live fetuses in E17.5 Eed-hom pregnancies were female, indicating male-biased lethality of HET-hom offspring (Fig. 1H), consistent with previous studies (23,30). These observations demonstrated that the major period of fetal loss of HET-hom offspring was between E9.5 and E12.5 in our model or by E6.5 in Inoue, Chen (23), which indicates that either placental function or embryo survival was compromised in both models. To determine whether differences occurred in the fetus and/or placenta, we assessed the growth and development of surviving mid-late gestation offspring. Embryonic day (E)9.5 HET-hom offspring from multiple litters were small compared to WT-wt offspring ( Fig. 2A). Similarly, E9.5 HET-hom offspring, including extraembryonic tissues, were significantly lighter than HEThet, WT-het and WT-wt controls (Supp Fig. 2A). Subsequently, compared to HET-het, WT-het and WT-wt controls, HET-hom embryos contained fewer tail somites at E12.5 and had reduced inter-digital tissue regression in foot plates at E14.5 (Fig. 2B, Supp Fig. 2B), demonstrating that HET-hom offspring were developmentally delayed.

Loss of EED in oocytes results
This in utero developmental delay contrasted markedly with our previous observation that postnatal HET-hom offspring were overgrown at P2 (7). We therefore examined the temporal developmental trajectory of HET-hom offspring by measuring fetal weight at E12.5, E14.5 and E17.5 and pup weight on the day of birth (P0) and at P3. While weight of HET-hom offspring was significantly lower than HET-het, WT-het and WT-wt controls at E12.5, E14.5 and E17.5, by P0 and P3 HET-hom offspring were heavier than the isogenic HET-het controls (Fig. 2C-H).
However, as we were collecting litters from time mated females, we also observed that gestational length was extended by 1 day in 7 pregnancies, or 2 days in 5 pregnancies of 19 pregnancies examined from Eed-hom oocytes, but gestational length was not extended in litters from Eed-wt and Eed-het oocyte controls (Fig. 2I). To determine the impact of this extended gestational time on offspring weight, we examined offspring weight of pups born on E19.5 days only. This revealed that offspring from Eed-hom oocytes were similar weight to controls (Fig 2J), indicating that HET-hom offspring growth had undergone catch-up growth and weight had been normalised by E19.5. Our previous report and data collected in this study demonstrated that litter size was also decreased from Eed-hom oocytes, raising the possibility that reduced HET-hom litter size may contribute to the normalisation of weight of pups at E19.5 (7). To address this, we compared pup weight from Eed-hom and control litters that contained 5-7 pups and were born on E19.5. This demonstrated that there was no difference in the weight of pups from HET-hom and control litters containing 5-7 pups (Fig 2K), also indicating that fetal growth restriction had been resolved by birth and that this catch-up occurred in utero and in a fashion that was independent of litter size. Together, while HEThom offspring were initially developmentally delayed and under-weight compared to their isogenic heterozygous counterparts and wild type controls, they underwent rapid catch-up growth in utero and were overgrown by P3 (Supp Fig. 2C), consistent with our previous observations of overgrowth at P2 (7).
Despite HET-hom offspring catch-up growth, oocyte-specific loss of EED results in late gestational placental hyperplasia and reduced placental efficiency.
Moreover, comparison of placental to fetal weights revealed that the growth of HET-hom placentas immediately preceded the catch-up growth of late-gestation HET-hom fetal offspring (Fig. 3E, G). While fetal growth rate was initially low between E12.5 and E14.5 and was similar between E14.5 and E17.5 in HET-hom offspring and HET-het controls, HET-hom offspring gained weight at 2.06 times the rate of HET-het controls between E18.5 and birth (E19.5) (Fig 3G).
To further investigate the relationship between placental function and offspring growth, we calculated the offspring body to placental weight ratio, which is indicative of placental efficiency (31). This indicated that placental efficiency was reduced at E12.5, E14.5, E17.5 and E18.5 in HET-hom offspring compared to all other genotypes ( Fig. 3H-K). Histological sections at E17.5 confirmed that HET-hom placentas were obviously larger than HET-het, WT-het and WT-wt controls. This analysis also revealed that the junctional zone was expanded in HEThom placentas, with abnormal projections of periodic acid-Schiff (PAS)-positive spongiotrophoblast cells into the labyrinth (Fig. 3J). . This difference was emphasised in overall cross-sectional area in E17.5 HET-hom placentas compared to all other genotypes, with a 55% increase in cross-sectional area compared to HET-het controls ( Fig. 4A-B). To understand if this increase in placental size was due to changes in a specific layer of the placenta, we measured junctional zone, labyrinth and maternally-derived decidua layer areas. At E14.5, junctional zone area was significantly larger in HET-hom placentas compared to the WT-wt and HET-het controls, however there was no difference the HET-hom decidua or labyrinth areas compared to all other genotypes (Supp Fig. 3B-D). At E17.5, the overall cross-sectional area of all layers of HET-hom placentas was greater than that of HET-het and WT-wt controls (Supp Fig 3E-G). However, as a proportion of the total placental area, only the junctional zone was significantly greater for HET-hom placentas compared to other genotypes ( Fig. 4C-E). Together, these data indicated that the whole HET-hom placenta, including the maternal side, was larger than controls. Moreover, this was associated with a disproportionate expansion of the fetally-derived junctional zone between E14.5 and E17.5.

Loss of EED in
Closer examination of H&E stained E17.5 placentas also showed that there were significantly more PAS-stained glycogen enriched cells counted in the entire HET-hom placenta midline sections compared to all other genotypes ( Fig. 4F-G). To test whether these increased glycogen enriched cells provide an enhanced source of glucose to HET-hom offspring, we measured fetal blood glucose levels. However, consistent with decreased efficiency observed for HET-hom placentas ( Fig 3H-K), HET-hom offspring had reduced blood glucose compared to HET-het controls, indicating that glucose levels are unlikely to contribute to HET-hom fetal catch-up growth ( Fig 4H).
Placental glycogen cells are thought to play a role in inhibiting the production of hormones that promote parturition at term and glycogen cell number usually peaks in number at E16.5 in the mouse, and declines by E18.5, prior to delivery (32)(33)(34). It has been proposed that a higher number of placental glycogen cells towards the end of the pregnancy could delay the production of hormones that promote parturition (33)(34)(35)(36). Consistent with this and the higher number of glycogen enriched cells in HET-hom placentas, gestational length was extended by 1 day or 2 days in Eed-hom pregnancies compared to the Eed-wt and Eed-het controls (Fig.   2I). However, in litters where gestational length was extended by one day from 19.5 days to 20.5 days, 36.36% of the HET-hom offspring were found dead at birth, and in litters where gestational length was extended by two days to 21.5 days, all HET-hom offspring were found dead at birth. Together, junctional zone area and glycogen cell content was increased in all HET-hom placentas and gestational length was extended in almost 2/3 of all litters from oocytes lacking EED. Moreover, despite initial embryonic growth delay, reduced placental efficiency and reduced fetal glucose levels, HET-hom offspring underwent catch-up growth so that their weight was normalised by birth and was further enhanced in early postnatal life.

Widespread gene dysregulation occurs in placentas from Eed-hom oocytes.
As the developmental changes observed in the placenta indicated that transcriptional regulation may be substantially altered in HET-hom placentas we analysed male and female placental tissue from HET-hom, HET-het and WT-wt offspring using RNA-sequencing (RNA- suggesting that the gene dysregulation observed in the HET-hom placenta influences a diverse range of systems and processes. As sequencing was performed on the entire placenta, we compared the Eed placental DEG list with published single cell RNA-seq data from E14.5 C57BL/6 placentas as an indication of the placental cell types in which gene dysregulation occurred in our model (37). Although the E14.5 placental stage assessed by Han, Wang (37) was earlier than the E17.5 placental data collected in our analysis, the majority of mature placental cell types are established by E14.5 (38). We therefore considered the Han, Wang (37) data useful for predicting the spatial expression for our Eed placental DEGs. Of the 2083 Eed placental DEGs identified in the Eed HET-hom placentas, 543 genes were found in the E14.5 mouse placental data set (37). Cellspecific comparisons revealed that of these 543 genes, 166 Eed placental DEGs included genes preferentially expressed in the junctional zone and 57 Eed placental DEGs were preferentially transcribed in the labyrinth, which we defined as 'Eed junctional zone DEGs' and 'Eed labyrinth DEGs', respectively (Fig. 5D-E). In addition, 104 Eed placental DEGs included genes preferentially transcribed in the maternally-derived decidua, which we refer to as 'Eed decidua DEGs' (Fig. 5F). Together these data revealed that gene dysregulation in HET-hom placentas was not isolated to a single region, and that both maternal and fetal-derived cells were altered. Of the Eed junctional zone DEGs the majority (86.1%) were upregulated, whereas a large proportion of Eed labyrinth DEGs and Eed decidua DEGs were downregulated (86% and 97.1%, respectively; Fig. 5G-I). Together these data demonstrate a strong bias for up-regulation of genes expressed in the junctional zone layer of the Eed placenta. This reflects the significant expansion of the junctional zone as a proportion of the total placental size. Eed HET-hom placentas, but only Sfmbt2 dysregulated in placentas of SCNT-derived offspring.
As maternal EED is also required for regulating X-inactivation in pre-implantation embryos, and loss of EED in oocytes caused a male bias in fetal lethality in Eed HET-hom offspring ( Fig.   1H) (23,30), dysregulated X-inactivation may provide another explanation for the gene dysregulation detected in Eed HET-hom placenta. To determine if gene expression from the X-chromosome was preferentially impacted in HET-hom placentas, we examined the representation of the placental DEGs on all autosomes and the X-chromosome. Placental DEGs with increased expression occurred at a similar rate on the X-chromosome and autosomes, indicating that there was no preferential chromosome-wide de-repression of Xlinked genes (Fig. 6C). However, we did note upregulation of X-linked genes Plac1 and Ldoc1, and the imprinted genes Ascl2 and Peg3, which are all known to regulate placental glycogen stores (33) (Supp Fig. 5A). All were increased in the junctional zone DEG list and have been associated with similar placental phenotypes to that observed in here Eed Het-hom placentas (Supp Fig. 5A) (42)(43)(44)(45)(46)(47). In line with its increased transcription, PLAC1 protein was also increased in cells within the junctional zone (Fig. 6D).
Finally, we compared HET-hom growth patterns with available weight data reported for SCNT-derived offspring, in which H3K27me3 imprinting is disrupted (27). As for Eed HET-hom offspring, SCNT-derived offspring were underweight at E14.5, but had caught up by birth (Supp Fig. 6A-B), demonstrating that the growth profiles and placental phenotypes in these models are similar.

Discussion
Independent studies have demonstrated that deletion of Eed in mouse oocytes using Zp3Cre or Gdf9Cre results in early growth restriction and offspring overgrowth outcomes in C57BL6 mice (7,23). This study provides a cohesive explanation for these differing reports. We observed time-dependent and inter-related fetal and placental growth trajectories of offspring from Eed-null oocytes. Deletion of Eed in the oocyte initially led to embryonic and fetal developmental delay, followed by pronounced placental expansion and subsequent rapid catch-up and over-growth of the offspring realised in the early postnatal period.
Fetal growth restriction is commonly caused by placental insufficiency and is often investigated using surgical or nutritional interventions in animal models (48)(49)(50)(51). In this study we provide a model that involves only the deletion of Eed in the oocyte and in which HEThom offspring were compared to genetically identical HET-het controls. We detail fetal growth restriction immediately prior to placental hyperplasia. Despite lower placental efficiency, as measured by fetal/placental weight ratio, and low fetal glucose levels in HEThom offspring, placental hyperplasia immediately preceded a period of catch-up growth.
Similar placental hyperplasia and increased fetal weight outcomes were common between at least two models -the Eed oocyte deletion model described here and an SCNT model in which we found published data indicating similar growth restriction occurred prior to placental hyperplasia and subsequent fetal catchup growth was evident. Therefore, in both models, and potentially in others, the data support a model in which initial growth restriction is followed by placental hyperplasia and offspring catch-up growth, despite apparently lower placental efficiencies. In Eed HET-hom offspring, this was combined with extended gestational length that also contributed to increased birth weight. An imperative for offspring survival is for the delivery of sufficiently developed offspring. Considering this, an interesting conjecture may be that an unknown physiological mechanism that results in placental hyperplasia occurs alongside rapid intrauterine catch-up growth and prolonged gestation to rescue the pregnancy. While speculative, such a mechanism could involve epigenetic adaptation, perhaps H3K27me3-dependent or classical imprinting.
While this work has been in review, separate studies demonstrated that loss of Eed in oocytes results in placental hyperplasia due to loss of H3K27me3 imprinting at Slc38a4 (52) and that late gestational placental hyperplasia, enhanced placental essential amino acid transport and increased fetal blood amino acid levels in SCNT-derived offspring is caused by loss of imprinting at Slc38a4 (27). Consistent with these observations, we observed higher Slc38a4 placental expression in HET-hom offspring, indicating that similar enhanced essential amino acid supply is likely to contribute to placental hyperplasia in Eed HET-hom mice. Therefore, while the fetal/placental ratio and fetal glucose levels were indicative of lower placental efficiency, increased amino acid transport across the placenta may explain the fetal catch-up growth in HET-hom mice.
The fetal growth restriction, placental hyperplasia and subsequent offspring overgrowth we observed in Eed HET-hom offspring was similar to that evident in SCNT-derived offspring.
While both models involve loss of H3K27me3-dependent imprinting (23,26,27), the SCNT embryos are different to offspring from Eed-null oocytes because they were derived from oocytes containing functional maternal PRC2. This might indicate that altered H3K27me3-dependent epigenetic patterning of the oocyte genome is responsible for the early developmental delay in these offspring. However, the underlying cause could differ. SCNT embryos are derived by oocyte-driven reprogramming of a differentiated somatic nucleus, a process through which major epigenetic barriers must be overcome and which may compromise preimplantation development, particularly given that the embryos are transferred to recipient female mice (26,(53)(54)(55). Similarly, either loss of maternal PRC2 or H3K27me3 programming caused by Eed deletion in the oocyte could explain the reduced blastocyst cell content, and this may underlie early embryo and fetal growth restriction. One way to separate these issues may be to perform pronuclear transfer from Eed-null oocytes to wild type oocytes to rescue maternal PRC2 in offspring, and vice versa.
Comparison of isogenic HET-hom and HET-het offspring demonstrated that the phenotypic outcomes in this model originate from loss of EED in the oocyte and were largely unaffected by offspring genotype. While HET-het offspring maintained normal fetal growth, placental development, gestational length, and litter size, all these aspects of development were substantially modified in isogenic HET-hom controls. Other studies have reported that preimplantation development proceeds normally in offspring from Eed-null oocytes, in that blastocysts formed at similar rates and were not affected by increased cell death (23).
However, we found that blastocysts derived from Eed-null oocytes contained low cell numbers, and that this primarily affected the inner cell mass. Consistent with this, we observed substantial developmental delay in E9.5-E17.5 HET-hom offspring derived from Eedhom oocytes, but this was resolved by birth and perinatal offspring were ultimately larger than controls.
There are at least two likely explanations for the early developmental delay observed in this model. EED supplied in the mature oocyte is rapidly localised to the maternal and paternal pronuclei within the first few hours following fertilisation (23,28,30,56,57). As the HET-hom offspring derived from Eed-null oocytes lacked maternal EED, it is possible that the lower cell numbers observed at the blastocyst stage resulted from low cell proliferation during preimplantation development due to a lack of maternally-derived PRC2. This is consistent with the established role for PRC2 in driving cell division in stem cells and other cell types (58) and with a previous report that apoptosis was not increased in blastocysts from EED-null oocytes (23). Alternatively, the lower blastocyst cell counts we observed may have resulted from loss of EED-dependent epigenetic programming in growing oocytes. This programming occurs during a transient stage of PRC2 expression in primary-secondary stage follicles prior to the onset of DNA methylation (22,28,59).
We observed widespread morphological changes in HET-hom placental tissue. Placental expansion was observed in both fetally-and maternally-derived placental layers, indicating that while placental hyperplasia may be driven through the spongiotrophoblast cells in the junctional layer, the maternal side of the placenta also expands to accommodate greater placental function. Despite this, there was a disproportionate increase in size of the fetallyderived junctional zone compared to the maternally-derived decidual layer. Furthermore, closer examination of the enlarged junctional zone revealed an increase in glycogen enriched cells. This was unusual as glycogen-enriched trophoblasts peak in number at E16.5, but then decline by approximately 60% by E18.5 (33), an outcome reflected in the decreased placental weights we observed in E17.5 control offspring.
Given that fetal blood glucose levels were low, it seems unlikely that the higher number of glycogen cells in the E17.5 placentas of HET-hom offspring increase glycogen release to the embryo and drive rapid growth catch-up growth. In addition to facilitating increased glucose release and rapid fetal catchup growth, glycogen-producing trophoblasts are also considered to inhibit hormonal release from the placenta to the maternal blood supply to prepare the mother for parturition (33,42). It may be that the increased number of glycogen cells remaining in late-gestation HET-hom placentas delay late gestational hormonal release and prolong pregnancy. Consistent with this, extended gestational length occurred in about two thirds of the pregnancies from Eed-hom oocytes. Moreover, while all pups were naturally born, the extended gestational length observed was also associated with high pup mortality in pregnancies extended by one day and mortality of all offspring in pregnancies extended by two days. Although the mechanisms underlying this extended gestation remain obscure, they appear to provide an extended widow for fetal/developmental catch-up in offspring, perhaps combining with Slc38a4 and amino acid transport to increase the survival of live born pups.
In addition to the morphological alterations observed, we found extensive transcriptional dysregulation in HET-hom placentas, with 2083 genes differentially expressed compared to genetically identical HET-het placentas. This altered gene expression was not isolated to a single region of the HET-hom placenta and was observed in both fetal and maternal layers.
While one might expect oocyte-specific loss of EED to affect only fetally-derived placental tissue, this ignores the possibility that the maternal tissue may expand or be re-modelled in response to placental hyperplasia in fetally-derived layers -indeed, without maternal remodelling it seems likely that the placenta may not support fetal catch-up growth. On this basis, it is unsurprising that transcription was substantially altered in both fetally-and maternally-derived placental tissues.
Analyses of HET-hom placentas identified cellular and morphological changes in maternallyand fetally-derived tissues and placental DEGs that function in multiple systems. Recent work has demonstrated that loss of H3K27me3-dependent imprinting at Slc38a4 or Sfmbt2, and to a lesser extent, Gab1 or Smoc1, leads to placental hyperplasia in SCNT-derived offspring (25)(26)(27)60). Consistent with this, we observed placental hyperplasia and increased expression of Slc38a4, Sfmbt2, Gab1 and Smoc1 in placentas from Eed-null oocytes, suggesting a loss of H3K27me3-dependent imprinting at these genes. We also observed increased placental expression of the X-linked genes Plac1, Wdr1 and Ldoc1, and immunohistochemistry confirmed that PLAC1 expression was increased in junctional zone cells in HET-hom placentas.
However, deletion of Plac1, rather than increased expression, has been associated with a similar placental phenotype and male biased lethality to that observed in HET-hom offspring (61). Notwithstanding this difference, it seems likely that the fetal growth restriction, placental hyperplasia, and fetal catch-up growth in this model involves one or all of these genes, although additional work is required to understand the mechanisms involved and the interactions between the genes that contribute.
While we observed no difference in the number of surviving hatched or expanded blastocysts from Eed-hom mothers, there was a decrease in the blastocyst cell count, particularly in the ICM, and a marked decrease in HET-hom litter size. Consistent with this and with another study (23), we also observed smaller offspring at E9.5, developmental delay and loss of embryos by E12.5, and reduced litter size at term (7). While the reason for HET-hom offspring loss is yet to be identified, one possibility is that the significant reduction in blastocyst cell numbers and subsequent developmental delay is too profound to support survival during mid-gestation development in the most highly impacted offspring. This may involve impacts within the embryo itself, and/or contributions from a defective placenta. The placenta develops a capacity to exchange nutrients, other supportive metabolites, and waste between the maternal and fetal blood supply at E9.5, and increased embryo loss around this stage could indicate early-mid gestational placental insufficiency (62,63). Embryonic lethality has also been previously linked to placental defects in knockout studies of imprinted and X-linked genes Plac1, Ascl2 and Peg10 (42-44, 46, 47, 64). As Plac1 and Ascl2 transcription was increased in the late-gestation HET-hom placenta, altered expression of imprinted genes such as Plac1 or Ascl2 may contribute to placental defects and fetal death in this model.
In summary, we demonstrate that loss of PRC2 function in the oocyte profoundly affects growth and developmental outcomes in both offspring and placenta via a mechanism that is independent of genetic background. This involves initial embryonic developmental delay and fetal growth restriction, followed by remarkable placental hyperplasia and fetal catch-up growth (Fig 7) associated with low placental efficiency and fetal blood glucose levels. Despite this, the fetal catch-up growth observed in this model may be explained by loss of imprinting for Slc38a4 and increased placental essential amino acid transport. Importantly, rapid fetal growth restriction and fetal catch-up growth have been linked with negative health outcomes later in life, including metabolic conditions (65,66). Moreover, given that mutations in the core PRC2 genes, EED, EZH2 and SUZ12 have all been associated with overgrowth and a range of co-morbidities in Cohen-Gibson, Weaver and Imagawa-Matsumoto syndrome patients (14,15,17,19,20) and we have observed similarities in outcomes of mouse offspring lacking EED in oocytes (7) this work also has implications for understanding these rare conditions.
Together, this research reveals that altered PRC2-dependent programming in the oocyte elicits a complex intrauterine environment for the fetus, which has the potential to mediate impacts on offspring health and disease that may persist into adulthood.

Mouse strains, animal care and ethics
Mice were housed using a 12h light-dark cycle at Monash Medical Centre Animal Facility, as previously reported (7). Room temperature was maintained at 21-23°C with controlled humidity, and food and water were provided ad libitum.

Genotyping
Genotyping was performed by Transnetyx (Cordova, TN) using real-time PCR assays (details available upon request) designed for each gene as described previously (7).

Collection and culture of pre-implantation embryos
Eight to twelve-week-old female mice were superovulated and mated to C57BL/6 males for one night. Zygotes were collected in handling media (G-MOPS PLUS, Vitrolife) at 37°C (29,69) denuded of cumulus cells with G-MOPS PLUS containing hyaluronidase. All embryos were washed in G-MOPS PLUS and embryo development kinetics was assessed using the EmbryoScope™ (Vitrolife) time-lapse imaging system. Embryos were cultured individually in 25μl of medium, with time-lapse images generated at 15-minute (min) intervals throughout the culture period.

Cell allocation in blastocysts
Following EmbryoScope™ culture, differential staining was performed (70) in hatched blastocysts using propidium iodide to label TE nuclei, while leaving the ICM unlabelled. After fixation, embryos were treated with bisbenzimide to stain ICM and TE, whole-mounted in glycerol and imaged using an inverted fluorescence microscope (Nikon Eclipse TS100). Nuclei were counted using ImageJ.
Placentas were then bisected, and half was fixed in 4% PFA for 72h at 4°C degrees, processed and paraffin embedded with the cut side of the placenta facing the front of the block. The other half of the placenta was rinsed in PBS, snap-frozen on dry ice and stored at -80°C for RNA analysis. P0 and P3 offspring were weighed and euthanized by decapitation.

Placental Histology
Paraffin embedded placentas were sectioned at 5µm using a Leica microtome and sections transferred to Superfrost plus slides (Thermo-Fisher, Braunschweig, Germany). Periodic antigen-shiff (PAS) and hematoxylin and eosin (H&E) staining was performed on placental sections by the Monash Histology Platform (MHTP node). Stained slides were scanned using Aperio slide scanner by the Monash Histology Platform and analysis was conducted using QuPath v0.2.3 (71). Histological analysis was conducted on one section located in the midline of each placenta. Junctional zone, labyrinth and decidual area was calculated in E14.5 and E17.5 H&E stained placentas. Glycogen cell counts were also conducted on the entire midline section of the H&E stained E14.5 and E17.5 placentas using QuPath v0.2.3 (71) . Investigators were blinded for sample genotypes throughout quantitative scoring of placental samples analysed.

Immunohistochemistry
Slides with 5µm thick placental sections were baked at 60°C for 20 mins. Tissue sections were dewaxed in three changes of xylene and rehydrated in three changes of ethanol then rinsed in distilled water. Antigen retrieval was performed in DAKO PT Link in a DAKO Target Retrieval (Low pH) Solution (DAKO, Cat# S1699) at 98°C for 30 min. Slides were then washed in DAKO EnVision Flex Wash Buffer (Cat# K8000) for 5 minutes. IHC was then performed on a DAKO Autostainer Plus in the following steps. Sections were washed once in EnVision Flex Wash Buffer following each subsequent step. Peroxidase Blocking Solution (DAKO, Cat# S2023) was applied for 10 minutes and non-specific binding was prevented with AffiniPure Fab Fragment Goat Anti-Mouse IgG for 1 hour. Mouse anti-PLAC1 (G-1) (Santa Cruz, Cat# sc-365919) primary antibody was applied at an appropriate concentration for 1 hour. EnVision System-HRP Labelled Polymer Anti-Mouse (DAKO, Cat# K4001) was applied for 1 hour. Immunostaining was visualised using DAKO Liquid DAB+ Substrate Chromogen System (Cat# K3468). A counterstain DAKO Automation Haematoxylin Staining Reagent was then applied for 10 minutes. Slides were removed from the Autostainer, transferred to a slide staining rack and rinsed in distilled water. In a fume-hood, slides were then washed in Scott's Tap water and distilled water. Finally, slides were dehydrated in three changes of 100% Ethanol, cleared in three changes of Xylene and mounted in DPX. Slides were either scanned using a VS120 Slidescanner (Olympus).

Placental RNA-sequencing
Placental isolation is described above, and placental RNA was extracted from 4-5 samples/genotype using NucleoSpin RNA Plus columns. RNA quality was assessed on an Agilent Bioanalyser and samples with RIN >7.5 used for library preparation and sequencing on the BGI Genomics platform (BGI Genomics, Hong Kong).

RNA-sequencing data analyses
Adaptor and low-quality sequences in raw sequencing reads were trimmed using   For (b-g) a two-tailed student's t test was used, and error bars represent mean ± SD.               Supp. Figure