EED and KDM6B Coordinate the First Mammalian Cell Lineage Commitment To Ensure Embryo Implantation

Cell culture and reagents.

E14tg2a (E14) mESCs were cultured without a feeder layer in mESC culture medium with 100 IU/ml mouse leukemia-inhibiting factor (mLIF) (ESGRO; Invitrogen, Carlsbad, CA) following procedures described earlier (20). Undifferentiated ESCs from Eed^−/− mice (a kind gift from Terry Magnuson) were maintained on irradiated mouse embryonic fibroblasts (MEFs) with the mESC medium with twice the concentration of mLIF (i.e., 200 IU/ml).

mTSCs were derived following previously described procedures (21). Blastocysts from 6-week-old female 129S2/SvPasCrl (129/Sv) strain mice (Charles River Laboratory, Wilmington, MA) were isolated at 3.5 days postcoitum and transferred on irradiated MEFs to mTSC culture medium consisting of RPMI 1640 (Sigma) supplemented with 20% fetal bovine serum (FBS; Sigma, St. Louis, MO), 100 μM β-mercaptoethanol, 1 mM sodium pyruvate (Cellgro; Mediatech Inc., Herndon, VA), 2 mM l-glutamine, penicillin-streptomycin (100 IU/ml and 100 mg/ml, respectively) in the presence of 25 ng/ml fibroblast growth factor 4 (FGF4; Invitrogen), and heparin (1 μg/ml; Sigma). Blastocyst outgrowths were disaggregated after 5 to 7 days, and a female TSC line, confirmed by a lack of sex-determining region Y (Sry) expression (data not shown), was selected for subsequent experiments. For experimental purposes, mTSCs were expanded in the proliferative state without MEF feeders by culturing in the presence of MEF-conditioned medium and normal mTSC medium (as described above) in a ratio of 7:3 (22).

Human embryonic kidney 293T (HEK293T) cells were cultured in Dulbecco modified Eagle medium supplemented with 10% FBS and 1% penicillin-streptomycin. Transfection of HEK293T cells was done in a similar medium devoid of antibiotics.

Isolation and in vitro culture of preimplantation embryos from mouse, rat, and monkey.

All procedures to isolate preimplantation embryos from mouse, rat, and monkey were performed after obtaining IACUC approvals at the University of Kansas Medical Center and the University of Wisconsin—Madison.

Human preimplantation embryos were collected at Stanford University, obeying all the institutional rules: a two-stage consent process (23) along with the approval from institutional committees at both Stanford University and the University of Kansas Medical Center. Deidentified and discarded human blastocysts were fixed in 4% paraformaldehyde solutions and analyzed for the H3K27Me3 mark by immunostaining. Embryos isolated from other species are described below.

Mouse embryos.

Adult (age, 3 to 4 weeks) CD-1 female mice were superovulated by intraperitoneal injection of 5 IU of P.G. 600 (Intervet, Millsboro, DE), followed by 5 IU of human chorionic gonadotropin (hCG; Sigma) 48 h later. Females were then mated with C57BL/6 males and euthanized either on the following morning to collect one-cell embryos (day 0.5), 2 days later (day 2.5) to collect morulae, or 3 days later (day 3.5) to collect matured blastocysts. Oviducts were removed by surgery; the embryos were harvested in M2 medium (Millipore, Billerica, MA) and cultured in KSOM (Millipore) at 37°C in a 5% CO₂ incubator with a relative humidity of 95%. Embryos were visualized daily, photographed, and sampled for RNA preparation at the morula or blastocyst stage (9).

Rat embryos.

Female dark agouti rats (age, 6 weeks; DA/OlaHsd strain; Harlan Laboratories) were mated overnight with dark agouti males (age, 7 weeks) and then euthanized by cervical dislocation on day 4.5 postcoitum. Uteri were collected from the females after euthanasia and flushed with M2 medium for the collection of blastocysts.

Monkey embryos.

Colonies of rhesus monkey were maintained at the Wisconsin National Primate Research Center (WNPRC; Madison, WI). Monkey oocytes were obtained from the donor monkey after treatment with 30 or 45 IU of recombinant human follicle-stimulating hormone administered twice daily for 7 days, followed by one injection of 1,000 IU of recombinant hCG hormone. Oocytes were retrieved from visible follicles by laparoscopic surgery. These were matured in vitro, followed by in vitro fertilization (IVF) with sperms collected on the same day. IVF-derived zygotes were cultured for 7 days until they formed matured blastocysts. All the surgical procedures were performed in accordance with the NIH guidelines for the care and use of laboratory animals (24) with the approval of the University of Wisconsin Graduate School Animal Care and Use Committee.

Isolation of ICMs and TEs from blastocysts.

ICMs were isolated by embryo immunosurgery according to the method described earlier (1). Zonae pellucidae were removed using 0.5% pronase (Sigma) for 5 min. The zona-free embryos were then treated with rabbit anti-mouse serum (Sigma) at a 1:100 dilution for 30 min, washed with phosphate-buffered saline (PBS), and incubated with guinea pig complement (Innovative Research, Novi, MI) at 37°C in 5% CO₂ for 30 min. After washing, ICM cells were collected for RNA preparation or chromatin immunoprecipitation (ChIP)–whole-genome amplification (WGA) analyses.

TE samples were acquired under an inverted microscope (a Nikon Ti-U Eclipse microscope equipped with Hoffman modulation contrast optics) by using Leica manual manipulators after microdissection and a customized ultrasharp splitting blade (ESE020; Bioniche Animal Health USA, Inc.). Purified TE samples were further processed for generating samples for either ChIP-WGA or extraction of RNA.

qRT-PCR.

Total RNA was extracted from cells with the TRIzol reagent (Invitrogen) and analyzed for quantitative reverse transcription-PCR (qRT-PCR) following procedures described earlier (22, 25). For analysis of expression in preimplantation embryos, total RNA was isolated using a PicoPure RNA isolation kit (MDS Analytical Technology, Sunnyvale, CA) and processed as described earlier (9). The oligonucleotides used for qRT-PCR are provided in Table S1 in the supplemental material.

Quantitative ChIP and ChIP-WGA analyses.

Real-time PCR-based quantitative ChIP analysis was performed following published protocols (9, 22, 26). In brief, cells were trypsinized; protein-DNA cross-linking was then conducted by treating cells with 1% formaldehyde (Sigma) for 10 min at room temperature with gentle agitation. Glycine (125 mM; Sigma) was added to quench the reaction. Sonicated, cross-linked chromatin fragments were immunoprecipitated with different antibodies (listed in Table S3 in the supplemental material). Mouse IgG1 (Sigma) was used as a negative control of antibody binding in all sets of experiments. Immunoprecipitated chromatins were reverse cross-linked, digested with proteinase K (Sigma), and purified. Quantification of the precipitated DNA was performed using quantitative PCR (qPCR) amplification. The amount of DNA precipitated by each antibody was normalized against 1/5, 1/25, 1/125, and 1/625 of the starting input DNA samples. A list of the primers used for ChIP analyses is provided in Table S2 in the supplemental material.

Quantitative ChIP-WGA analyses with purified ICM and TE samples from preimplantation mouse embryo (≥300) were performed in the same manner described above, with a few modifications. Column-purified immunoprecipitated DNA fragments were amplified using a whole-genome amplification kit (WGA4; Sigma) prior to qPCR analysis.

Western blot analysis.

For Western blot analyses, whole-cell lysates were extracted with a lysis buffer described earlier (1) and electrophoresed in an 8 to 12% polyacrylamide gel (Sigma) using Tris-glycine running buffer (pH 8.3). Proteins were then transferred to polyvinylidene difluoride (PVDF) membranes (Millipore). Nonspecific binding was blocked with 5% nonfat dry milk (Difco, Becton, Sparks, MD) and 0.05% Tween 20 (Sigma) in 20 mM Tris-HCl, pH 7.6 (Tris-buffered saline with Tween [TBS-T]). After incubation with appropriate primary antibodies overnight at 4°C, membranes were washed in TBS-T and reincubated with appropriate secondary antibodies conjugated with horseradish peroxidase. Enhanced chemiluminescence (ECL) signals were developed in films using an ECL-plus reagent (GE Healthcare UK Limited, Little Chalfont, Buckinghamshire, United Kingdom). The primary antibodies along with the dilutions used are listed in Table S3 in the supplemental material.

Immunofluorescence study.

Preimplantation embryos were washed with ice-cold PBS and fixed with freshly prepared 4% paraformaldehyde (Sigma) in PBS for 30 min at room temperature. Cells were permeabilized in 0.25% Triton X-100 in PBS and blocked with 1% bovine serum albumin (BSA) for 1 h at room temperature. Appropriate primary antibody, diluted in wash buffer (containing 0.1% BSA and 0.05% Tween 20 in PBS), was incubated overnight at 4°C. Primary antibodies along with the dilutions used are listed in Table S3 in the supplemental material. Washed embryos were labeled with appropriate secondary antibody and conjugated with either Alexa Fluor 488 (green) or Alexa Fluor 568 (red) in a dilution of 1:400 for 1 h in the dark at room temperature. After the labeled embryos were counterstained with DAPI (4′,6-diamidino-2-phenylindole; Invitrogen), they were observed under either a fluorescence microscope (Leica) or a laser scanning confocal microscope (LSM 5; Carl Zeiss Microimaging, Thornwood, NY). Images were acquired using a ×40 water-immersion objective on a confocal microscope (LSM 5 laser scanning microscope; Carl Zeiss Microimaging, Thornwood, NY) operated by LSM 5 Pa software. Alexa Fluor 488, Alexa Fluor 568 fluorophores, and DAPI were excited individually with 488-nm, 543-nm, and 405-nm lasers, respectively, with appropriate filter sets. Uncompressed images were processed with the Zeiss LSM image browser. Where indicated, images were taken as z stacks and rendered as three-dimensional projections for detailed visualization. All immunofluorescence images for a specific set of experiment were captured with an identical set up, including the exposure time.

RNA interference (RNAi) and overexpression via lentiviral transduction.

Short hairpin RNAs (shRNAs) _targeting mouse Kdm6b were cloned in lentiviral vector pLKO.1 (10878; Addgene). Lentiviral supernatants were prepared in HEK293T cells as described earlier (9, 22, 26). Three different shRNA constructs with Kdm6B _target sequences, 5′-CCTGTATATGTCTCTTGTTTA-3′ (shRNA1), 5′-CCTCTGTTCTTGAGGGACAAA-3′ (shRNA2), and 5′-GCTGGATGAATCCATTCGGAA-3′ (shRNA3), were used. Among them, two shRNAs, shRNA1 and shRNA3, were found to be the most effective for successful knockdown. For ectopic EED expression, cDNA for mouse Eed along with a ribosome-binding site was cloned under the control of the human U6 promoter in the pLKO.3G (14748; Addgene) lentiviral vector. The vector also expressed an enhanced green fluorescent protein (EGFP) reporter.

Lentiviral particles were transduced in undifferentiated mTSCs, and cells were selected with puromycin under the undifferentiated TSC culture condition in the presence of FGF4. After 3 to 4 days of lentiviral transduction, experimental analyses were performed. To transduce one-cell preimplantation embryos, lentiviral soups were concentrated using an ultracentrifuge (35,000 rpm for 3 h), and the concentrates were dissolved in EmbryoMax injection buffer (Millipore). About 150 one-cell fertilized embryos in each set were then subjected to subzonal injection. Transducted embryos were grown in KSOM at 37°C and 95% relative humidity in a 5% CO₂ chamber for 3 to 4 days. Embryos were visualized and photographed daily under fluorescence and phase-contrast microscopes. Unmanipulated embryos and embryos infected with lentiviral particles containing pLKO.3G were used as untreated and negative-control experimental sets, respectively.

To transduce TE, early-blastocyst-stage (about embryonic day 2.5 [E2.5] to E3.0) mouse embryos were subjected to subzonal microinjection with lentiviral particles containing either pLKO.3G (control vector) or vectors expressing EED with an EGFP reporter and shRNAs for KDM6B. Transduced embryos were grown in KSOM at 37°C with 95% relative humidity in a 5% CO₂ chamber for 36 h before further analyses.

Analyses of blastocyst implantation efficiency.

Blastocysts were surgically transferred into the uterine horns of E2.5 foster pseudopregnant, 6- to 8-week-old FVB female mice (Charles River) that were previously mated with a vasectomized male. Blastocysts were transferred into a single uterine horn. After 5 days (at E7.5), Chicago sky blue (27) was injected into the tail vein and mice were sacrificed to monitor the sites of embryo implantation.

During preimplantation development, locus-specific mechanisms regulate differential deposition of the repressive H3K27Me3 mark at the chromatin domains of TE regulators.

To understand the importance of H3K27Me3 modification during preimplantation mammalian development, we initiated our study by testing the distribution of the global H3K27Me3 mark in preimplantation mouse embryos at different developmental stages beginning from the 2-cell stage. Similar to an earlier report (28), we found that the H3K27Me3 mark exists in the nuclei of all blastomeres at all developmental stages (Fig. 1A). However, at the blastocyst stage, the H3K27Me3 modification is enriched in ICM nuclei compared to TE nuclei (Fig. 1B). To further understand the importance of the H3K27Me3 modification, we tested whether enrichment of the global H3K27Me3 content in the ICM of a blastocyst is a common event for other mammalian species. We analyzed rat, monkey, and human blastocysts. We found that, unlike in a mouse blastocyst, the global levels of the H3K27Me3 mark are very similar among the nuclei of TE and ICM lineage cells in rat, monkey, and human blastocysts (Fig. 1B). Thus, reduced global levels of the H3K27Me3 modification in the TE nuclei compared to those in the ICM nuclei are not a common event for all mammalian embryos.

Although global H3K27Me3 levels vary between the TE and ICM of a mouse embryo and remain similar in rat, monkey, and human embryos, interestingly, expression of TE regulators like CDX2 is suppressed in the ICM of all these species (1). Thus, we next tested whether the H3K27Me3 mark is differentially incorporated between active versus repressed chromatin domains of key TE regulators in the TE and ICM lineages, respectively. To make a comparative analysis in two different mammalian species, we obtained both mouse and rat blastocysts, isolated TEs (Fig. 2A) and ICMs (9), confirmed TE- and ICM-specific gene expression (see Fig. S1A in the supplemental material), and performed ChIP-WGA analyses (as described in references 9 and 29) to determine H3K27Me3 levels at the chromatin domains of key TE regulators. Our analyses showed that the Cdx2, Gata3, and Tead4 promoters as well as the intron 1 region of Cdx2, an enhancer element for Cdx2 transcription (9), harbor low levels of the H3K27Me3 modification in early-stage embryos (8- to 16-cell stage) and the incorporation of the H3K27Me3 mark is highly induced within the ICM at the blastocyst stage, confirming an earlier observation in relation to the mouse Cdx2 promoter (17). However, the incorporation of the H3K27Me3 mark is very low at the Cdx2, Gata3, and Tead4 loci in the TEs of both mouse and rat blastocysts (Fig. 2B). Furthermore, analysis of H3K27 acetylation (H3K27Ac), a chromatin mark that is indicated to have an antagonistic switch with the H3K27Me3 modification and is often associated with transcriptionally active genes (30), showed that the Cdx2, Gata3, and Tead4 loci within the TE harbor increased levels of the H3K27Ac mark (see Fig. S1B in the supplemental material). These results indicated that, in the TE versus the ICM, the differential H3K27Me3 incorporation at the chromatin domains of TE regulators is not dependent upon global levels of H3K27Me3 modification. Interestingly, in contrast to the TE regulators, the chromatin domains of two X-linked genes, cysteine-rich hydrophobic domain 1 (Chic1) and alpha-thalassemia/mental retardation syndrome X linked (Atrx), contain high levels of the H3K27Me3 mark in the TE (Fig. 2C). These results indicate that the machinery essential to incorporate the H3K27Me3 mark at a chromatin domain is functional within the TE lineage. However, that machinery is unable to incorporate the H3K27Me3 mark at the Cdx2, Gata3, and Tead4 loci. Collectively, our results indicate that chromatin domain-specific mechanisms rather than global levels of H3K27Me3 ensure reduced incorporation of the H3K27Me3 mark at the Cdx2, Gata3, and Tead4 loci in the TE lineage of a mammalian embryo.

Similar to the expression patterns in the TE and ICM lineages, CDX2 and GATA3 are expressed only in the mouse TE-derived mTSCs but are repressed in the mouse ICM-derived mESCs (see Fig. S2A in the supplemental material) (9, 10). Similarly, expression of mRNA of TEAD4, another master regulator of TE fate specification, is severalfold higher in mTSCs than mESCs (1, 13). Thus, mESCs and mTSCs can be exploited for mechanistic studies to understand the transcriptional mechanisms that activate and repress the Cdx2, Gata3, and Tead4 loci in the TE and the ICM, respectively, so we compared the global levels of the H3K27Me3 mark and its incorporation at the chromatin domains of TE regulators in mESCs versus mTSCs. We found that, similar to mouse TE and mouse ICM, the global H3K27Me3 level is low in mTSCs compared to that in mESCs (see Fig. S2A in the supplemental material). Our ChIP analyses also confirmed that activated Cdx2, Gata3, and Tead4 loci in mTSCs contain reduced levels of the H3K27Me3 modification (see Fig. S2B in the supplemental material). However, similar to mouse TE, the chromatin domains of X-linked genes contain high levels of the H3K27Me3 modification in mTSCs compared to those in mESCs (see Fig. S2B in the supplemental material). These results indicated that similar mechanisms might be involved in both mouse TE and mTSCs to reduce the H3K27Me3 mark at the Cdx2, Gata3, and Tead4 loci. Thus, in the next series of experiments, using mESCs and mTSCs as model systems, we asked what molecular mechanism regulates H3K27Me3 deposition at the chromatin domains of TE regulators.

Mammalian blastocysts have reduced EED expression in the TE lineage and high levels of KDM6B expression in both the TE and ICM lineages.

Methylation at the H3K27 residue is mediated by the PRC2 complex (31), and demethylation is mediated by histone demethylases KDM6A (also known as UTX) and KDM6B (also known as JMJD3) (32). A recent study showed that KDM6B is involved in regulating global H3K27Me3 levels in bovine preimplantation embryos (33). In addition, studies in mESCs showed that JARID2 (also known as Jumonji) orchestrates PRC2 function and fine-tunes deposition of the H3K27Me3 mark at chromatin domains (34–36). Therefore, we tested which of these mechanisms reduces the H3K27Me3 mark at the chromatin domains of TE regulators.

We compared expression of PRC2 core components EZH2, SUZ12, and EED in mTSCs versus mESCs. We found that SUZ12 and EZH2 are expressed at very similar levels in mTSCs and mESCs (Fig. 3A). However, EED expression is significantly lower in the mTSCs (Fig. 3A), an observation also reported by an earlier study (19). Next, we compared the expression patterns of histone demethylase KDM6A and KDM6B along with JARID2 in mTSCs versus mESCs. We found that, compared to mESCs, mTSCs express very high levels of KDM6B (Fig. 3A). On the contrary, using two different antibodies, we could not detect KDM6A by Western blot analysis in either mESCs or mTSCs (data not shown). Thus, we concluded that KDM6B is the major H3K27 demethylase present in the mTSCs. Furthermore, compared to JARID2 protein expression in mESCs, JARID2 protein expression was almost undetectable in mTSCs (Fig. 3A). Thus, we also concluded that JARID2 does not modulate PRC2 function in mTSCs.

Our expression analyses in mTSCs versus mESCs revealed two major differences, reduced EED expression and elevated KDM6B expression, that could contribute to the reduced H3K27Me3 mark at the chromatin domains of key TE regulators. Therefore, we next tested expression of PRC2 components along with KDM6B in the TE and the ICM of mouse blastocysts. Our analyses showed that, although Suz12 and Ezh2 mRNAs and proteins are expressed at similar levels in both TE and ICM (Fig. 3B and C), expression of EED is much lower in the mouse TE (Fig. 3B and C). These results indicated that the patterns of expression of core PRC2 components in the TE and ICM are maintained in the stem cell population derived from those lineages. Next, we tested expression of KDM6B in mouse blastocysts and found that, unlike KDM6B expression in the stem cells, KDM6B is abundantly expressed in both the TE and ICM of a mouse embryo (Fig. 3B and C), confirming an earlier observation that showed that Kdm6B expression is repressed during derivation of ESCs from the ICM (37). Interestingly, we also noticed that KDM6B is present in both the cytoplasm and nuclei of TE and ICM cells, with a relatively greater abundance within the cytoplasm.

To test whether reduced expression of EED in the TE is a common event for other mammalian species, we isolated TE and ICM from rat blastocysts and tested mRNA expression. We found that, similar to the mouse TE, Eed mRNA expression in the rat TE is significantly lower than that in the ICM (Fig. 3D) and Kdm6B mRNA expression is similar in both rat TE and rat ICM (Fig. 3D). Unfortunately, we were unable to obtain specific antibodies to study the EED and KDM6B proteins in rat embryos. Nonetheless, as we observed reduced Eed mRNA expression and high levels of Kdm6B mRNA expression in both mouse and rat TE as well as in mTSCs, we next asked whether reduced levels of EED and the function of KDM6B contribute to reduce the H3K27Me3 mark at the chromatin domains of key TE regulators in TE-derived mTSCs, thereby facilitating their transcription.

Reduced levels of EED lead to loss of PRC2 recruitment at the chromatin domains of TE regulators in mTSCs.

EED is implicated in _targeting the PRC2 complex to chromatin through its interaction with the histone H3 tail (38, 39). So, we tested whether reduced levels of EED result in reduced PRC2 recruitment at the Cdx2, Gata3, and Tead4 chromatin domains in mTSCs. Our ChIP analyses revealed that, compared to mESCs (Fig. 4A), PRC2 recruitment is significantly lower at those chromatin domains in mTSCs. Therefore, we hypothesized that reduced expression of EED contributes to reduced PRC2 recruitment and H3K27Me3 incorporation at the chromatin. To test this hypothesis, we ectopically expressed EED in mTSCs (Fig. 4B and C) and asked whether ectopic expression of EED in mTSCs represses TE regulators. We found that, although ectopic expression of EED did not increase the total content of the H3K27Me3 mark in mTSCs (Fig. 4C), the levels of CDX2 and GATA3 mRNA and protein expression were strongly reduced (Fig. 4B and C). Furthermore, ChIP analyses revealed that transcriptional repression of Cdx2 and Gata3 was associated with PRC2 recruitment and enrichment of the H3K27Me3 mark at their chromatin domains (Fig. 4D). Thus, from gain-in-function analyses, we concluded that, in mTSCs, PRC2 recruitment and H3K27Me3 deposition at the Cdx2 and Gata3 loci are dependent upon EED protein levels and reduced levels of EED ensure impairment of PRC2 recruitment and H3K27Me3 deposition at those loci, thereby activating their transcription. Interestingly, despite increased PRC2 recruitment and H3K27Me3 incorporation at the Tead4 chromatin domain (Fig. 4D), TEAD4 expression was not significantly reduced in ectopic EED-expressing mTSCs (Fig. 4B and C). Therefore, we also concluded that, unlike transcription of Cdx2 and Gata3, transcription of Tead4 in mTSCs is not critically dependent upon the EED/PRC2-mediated H3K27Me3 modification.

The EED function in mESCs represses CDX2 and GATA3 expression, thereby contributing to suppression of the trophoblast fate.

EED is expressed in the mouse ICM (40) or ICM-derived mESCs (19). However, EED is dispensable for the development of the ICM lineage (41), and mESCs can be derived from Eed^−/− mouse blastocysts (42). Therefore, the question arises, what is the importance of EED in the ICM or ICM-derived mESCs? We found that, in mTSCs, ectopic induction of EED expression leads to the repression of key trophoblast regulators, CDX2 and GATA3, in mTSCs. Thus, we hypothesized that an EED function might be important to suppress induction of the trophoblast fate in ICM or ICM-derived mESCs by repressing expression of TE regulators. To test this, we took a loss-of-function approach using Eed^−/− mESCs (42). We cultured wild-type E14tg2a (E14) and Eed^−/− mESCs under both mESC and FGF4-containing mTSC culture conditions (22) and determined the expression of TE regulators. We found that the expression of CDX2 and GATA3 was not induced in E14 cells, when cultured under the mTSC culture condition (Fig. 5A and B). We also noticed that expression of TEAD4 was marginally induced in both E14 and Eed^−/− mESCs under the mTSC culture condition. However, expression of both CDX2 and GATA3 was robustly induced in Eed^−/− mESCs, when they were cultured under the mTSC culture condition (Fig. 5A and B). Intriguingly, the induction of CDX2 and GATA3 was also associated with a gradual morphological change of Eed^−/− mESC colonies toward mTSC-like colonies (Fig. 5C). These results indicated that, even though EED is dispensable for development of the mouse ICM or derivation of mESCs, an EED function could be important to repress CDX2 and GATA3 expression and to prevent induction of the trophoblast fate under conditions that facilitate TE/TSC development.

The KDM6B function ensures reduced levels of H3K27Me3 incorporation and maximal transcription at the Cdx2 and Gata3 loci in mTSCs.

As KDM6B is highly expressed in the TE lineage and TE-derived mTSCs, we next asked whether the KDM6B function contributes to reduction of the H3K27Me3 mark and transcription of TE regulators in mTSCs. To that end, we tested whether KDM6B physically occupies the Cdx2 Gata3, and Tead4 chromatin domains in mTSCs. Our ChIP analyses detected KDM6B recruitment at the Cdx2 and Gata3 chromatin domains. However, KDM6B recruitment was not observed at the Tead4 promoter or the promoter region of the X-linked gene Atrx (Fig. 6A). To test whether KDM6B is functionally important to regulation of Cdx2 and Gata3 transcription in mTSCs, we knocked down KDM6B expression via RNAi (Fig. 6B and C). We found that depletion of KDM6B increases the total H3K27Me3 mark in mTSCs (Fig. 6C) and strongly represses expression of Cdx2 and Gata3 (Fig. 6B and C) but does not affect TEAD4 and SUZ12 expression. ChIP analyses in KDM6B-depleted mTSCs showed that loss of KDM6B recruitment at the Cdx2 and Gata3 chromatin domains is associated with increased deposition of the H3K27Me3 mark (Fig. 6D). Moreover, a combination of KDM6B depletion and ectopic expression of EED in mTSCs abrogated expression of CDX2 and GATA3 (Fig. 6E). Analyses of the Cdx2 and Gata3 chromatin domains indicated that combinatorial ectopic EED expression/KDM6B depletion further increased H3K27Me3 deposition, strongly reduced H3K27Ac deposition (see Fig. S3A in the supplemental material), and inhibited RNA polymerase II recruitment (see Fig. S3B). We also found that loss of CDX2 and GATA3 expression upon combinatorial EED induction/KDM6B depletion is associated with induced expression of differentiated trophoblast markers (Fig. 6F) like placental lactogen I (PL-I) and PL-II (markers for trophoblast giant cells), trophoblast-specific protein alpha (Tpbpa; spongiotrophoblast marker), and glial cell missing 1 (Gcm1; syncytiotrophoblast marker). These results indicate that mTSC self-renewal can be negatively affected by altered KDM6B/EED expression.

Ectopic EED expression along with KDM6B depletion in preimplantation embryos inhibits CDX2 and GATA3 expression in the TE lineage.

Our analyses in mTSCs confirmed that two distinct mechanisms, reduced PRC2 recruitment due to repressed expression of EED and the function of KDM6B, prevent H3K27Me3 deposition at the Cdx2 and Gata3 chromatin domains and activate their transcription. Therefore, we next analyzed whether coordinated regulation of EED and KDM6B is important to ensure CDX2 and GATA3 expression and proper development of the TE lineage in a preimplantation mouse embryo. First, we tested whether expressions of EED and KDM6B is dynamically regulated during the course of mouse preimplantation development. Earlier studies (43, 44) indicated that EED mRNA could be detected in preimplantation mouse embryos and the EED protein can be detected within the nuclei of mouse TE and ICM lineages (40, 45), with a greater abundance found within the ICM (46). However, none of these prior studies performed any quantitative analyses of EED expression between the TE and ICM lineages. Therefore, for measuring EED and KDM6B expression, we performed quantitative RT-PCR analyses with embryos isolated at different developmental stages (from 4-cell to blastocyst stages).

We found that Eed and Kdm6B mRNAs are present at all developmental stages. Interestingly, Eed mRNA expression was strongly induced within the ICM lineage during the morula-to-blastocyst transition (Fig. 7A), whereas KDM6B induction was a common event in both TE and ICM (Fig. 7B). A similar pattern for Kdm6B expression was also observed during bovine preimplantation development (33). As expected, we noticed that the induction of EED expression during the morula-to-blastocyst transition occurs solely in the ICM lineage (Fig. 7A). We also compared PRC2 recruitment at the Cdx2 and Gata3 loci and found that, compared to cells in morula-stage embryos, cells in the ICM lineage have high levels of the PRC2 complex at the Cdx2 and Gata3 chromatin domains (Fig. 7C), whereas PRC2 recruitment at those loci within TE cells is reduced. Collectively, these results indicate that the relative concentrations of EED and KDM6B are altered in the nascent TE and ICM lineages and relatively less EED expression correlates with the loss of PRC2 recruitment and H3K27Me3 incorporation at the Cdx2 and Gata3 loci in the nascent TE lineage. Thus, we next tested whether alteration of EED and KDM6B expression alters CDX2 and GATA3 expression within the developing TE lineage and affects proper preimplantation mouse development.

We infected one-cell mouse embryos with lentiviral vectors expressing ectopic EED or shRNAs against KDM6B individually or in combination. We monitored preimplantation development by culturing infected embryos. We used lentiviral vectors expressing only EGFP as a control. We found that the presence of EGFP-expressing lentiviral vectors does not significantly inhibit development of preimplantation mouse embryos and they form matured blastocysts when cultured for 96 h (Fig. 8A and B). When infected with ectopic EED- and KDM6B shRNA-expressing vectors individually, we saw a partial but significant reduction in blastocyst maturation and found that 65 to 70% of the embryos developed to the blastocyst stage (Fig. 8B; see Fig. S4A in the supplemental material). However, when we infected embryos with lentiviral vectors expressing Kdm6B shRNAs along with ectopic EED-expressing vectors, we noticed stronger inhibition in blastocyst development and that ∼45% of the embryos did not develop to the blastocyst stage (Fig. 8A and B), and we found that expression of both the CDX2 and GATA3 proteins was strongly repressed in those development-arrested embryos (Fig. 8C).

Next, we isolated TEs from the ectopic EED-expressing/KDM6B-depleted embryos which matured to the blastocyst stage. We compared mRNA expression with that of TEs from control embryos and also performed ChIP-WGA to determine H3K27Me3 deposition at the chromatin domains of the Cdx2 and Gata3 loci. We found that Gata3 and Cdx2 mRNA expression was strongly repressed in TEs that were isolated from ectopic EED-expressing/KDM6b-depleted blastocysts (Fig. 8D) and transcriptional repression was associated with increased H3K27Me3 deposition at their chromatin domains (Fig. 8E). Interestingly, similar to our observation in mTSCs, we did not notice a significant reduction in Tead4 or Suz12 mRNA expression in ectopic EED-expressing/KDM6B-depleted TEs.

Blastocysts with altered expression of EED and KDM6B fail to implant.

Both CDX2 and GATA3 are implicated in proper TE function (9, 11), and earlier studies (10, 47–49) showed that the absence of CDX2 in a mouse blastocyst results in ultrastructural abnormalities within the TE, leading to the failure of embryo implantation to the uterus. We found that, although several of the ectopic EED-expressing/KDM6B-depleted embryos formed blastocysts, they showed defective hatching compared to the control embryos expressing only vectors (see Fig. S4B in the supplemental material). Therefore, we next tested whether the ectopic EED-expressing/KDM6B-depleted blastocysts show reduced implantation efficiency. We transferred blastocysts expressing ectopic EED or shRNAs against KDM6B individually or in combination into the uterine horns of E2.5 pseudopregnant female mice to test their implantation efficiency. We noticed a partial defect in the implantation efficiency of blastocysts with ectopic EED expression as well as KDM6B depletion (Fig. 9A and B). However, a complete failure of implantation was observed in blastocysts with both ectopic EED expression and KDM6B depletion (Fig. 9A and B).

To further define the importance of TE-specific regulation of EED and KDM6B, we tested whether TE-specific induction of EED and depletion of KDM6B result in the loss of CDX2 and GATA3 expression and negatively affect implantation efficiency. We took early blastocyst-stage mouse embryos and performed subzonal injection to specifically transduce TE (Fig. 10A) with lentiviral particles to ectopically induce EED expression and deplete KDM6B expression (Fig. 10B). We found that TE-specific alteration of EED and KDM6B expression resulted in a significant loss of Cdx2 and Gata3 mRNA expression (Fig. 10B). Immunofluorescence analyses also indicated strong inhibition of GATA3 and CDX2 protein expression in several blastocysts (Fig. 10C). Furthermore, the loss of CDX2 and GATA3 expression was associated with a reduced efficiency of blastocyst hatching (Fig. 10D and E) and implantation in surrogate females (Fig. 10F). Collectively, from our results, we concluded that coordinated functions of EED and KDM6B regulate the spatial expression patterns of key TE regulators, thereby leading to proper development and function of the TE, ensuring successful implantation.