Choice of random rather than imprinted X inactivation in female embryonic stem cell-derived extra-embryonic cells

In eutherian mammals, female cells have two X chromosomes, derived from oocyte and sperm. In adult females, one of the two X chromosomes is inactivated for gene dosage compensation in almost all cell types, with a few notable exceptions. This phenomenon is known as X inactivation (Xi) (Lyon, 1961). Xi is initiated by the expression of the Xist non-coding RNA from one X chromosome, which then coats that X chromosome. The Xist-coated X chromosome recruits polycomb-related complex 2 (PRC2) for the accumulation of tri-methylation of histone H3 lysine 27 (H3K27me3), resulting in the formation of a heritable repressive chromatin state (Silva et al., 2003).

Xi first occurs in the morula stage embryo and selectively inactivates the X chromosome of paternal origin, designated as imprinted Xi (Takagi and Sasaki, 1975). Imprinted Xi is maintained in the trophectoderm (TE), which emerges in 16- to 32-cell stage embryos as the outer layer cells. In early blastocyst stage embryos, the inner cell mass (ICM) initially retains imprinted Xi. However, after the segregation of primitive endoderm (PrE) from epiblast, imprinted Xi is transmitted to PrE but erased in the epiblast, resulting in two active X chromosomes in the pluripotent cell population (Mak et al., 2004). Immediately after implantation, all epiblast cells undergo Xi by the formation of primitive ectoderm in egg-cylinder stage embryos, in which either the maternal or paternal X chromosome is randomly chosen for inactivation, known as random Xi. It is clear that this imprinted Xi mechanism is necessary for extra-embryonic development because paternal Xist knockout embryos cannot induce imprinted Xi in pre-implantation embryos and show severe defects in extra-embryonic tissue development from E6.5, with two active X chromosomes (Marahrens et al., 1997).

The molecular mechanism of imprinted Xi is still obscure. One report has suggested that the selective inactivation of the paternal X chromosome is achieved by epigenetic marks on the maternal X chromosome that are imprinted in the maturating oocyte and allow selective escape of the maternal X chromosome from Xi in TE and PrE (Tada et al., 2000). Others have shown that the epigenetic marks on the paternal X chromosome, similar to the silencing of repetitive sequences, contribute to its selective inactivation (Namekawa et al., 2010). Even if both paternal and maternal X chromosomes carry specific marks to direct imprinted Xi, they should be erased in the epiblast of blastocyst stage embryos and undergo random Xi at later developmental stages.

Embryonic stem (ES) cells are pluripotent cell lines derived from epiblast. They retain the same epigenetic state of the X chromosome as the epiblast, as the female ES cells carry two active X chromosomes and it has been reported that random Xi occurs after the induction of differentiation. Interestingly, ES cells retain the ability to differentiate into extra-embryonic endoderm cells when they form embryoid bodies in suspension culture, in which random Xi is chosen rather than the imprinted Xi that normally occurs in extra-embryonic endoderm in vivo, suggesting that the imprinting marks on X chromosomes are erased in ES cells (Sado et al., 1996). However, it remains ambiguous whether all epigenetic marks for imprinted Xi are erased completely in any context and which type of Xi occurs if ES cells differentiate into TE in conventional culture conditions.

The ability of ES cells to give rise to TE had been opposed by evidence that ES cells never contribute to the TE lineage after injection into pre-implantation embryos (Beddington and Robertson, 1989). However, we previously showed that the forced expression of the homeobox transcription factor Cdx2 triggers the differentiation of ES cells toward TE (Niwa et al., 2005). It is also possible to induce trophoblast stem (TS) cells from ES cells that contribute to all placental lineages after injection into blastocysts, as found in the case of embryo-derived TS cells (Tanaka et al., 1998), indicating proper differentiation. We also confirmed that the forced expression of Gata4 or Gata6 induces the differentiation of ES cells to PrE cells that mimic the character of embryo-derived extra-embryonic endoderm cells (XEN cells), which contribute to parietal endoderm after injection into blastocysts (Kunath et al., 2005; Shimosato et al., 2007). Here, we applied these strategies to female ES cells in which the paternal and maternal X chromosomes can be distinguished by genetic manipulation or polymorphic markers to address which types of Xi occur in these functional extra-embryonic cell types derived from ES cells. We also applied somatic cell nuclear transfer to female ES cells to test their choice of Xi in TE of cloned embryos.

A hybrid F1 CBMS1 female ES cell line was derived from an embryo obtained by mating female Mus musculus musculus (CBA) and male Mus musculus molocimus (MsM) mice. A B142bgeoEG female ES cell line was derived from an embryo obtained by mating genetically modified Mus musculus (C57BL/6 and CBA) mice. Both female ES cell lines were cultured in GMEM (Gibco) containing 14% knockout serum replacement (KSR; Gibco), 1% fetal calf serum (FCS; Thermo Scientific), 1× sodium pyruvate (1 mM; Gibco), 1× non-essential amino acids (NEAA; Gibco), 0.1 mM 2-mercaptoethanol (Nakarai Tesque) and 1000 U/ml leukemia inhibitory factor (LIF) on a 0.1% (w/v) gelatin-coated dish.

Gata6GR or Cdx2ER-IRES-puromycin acetyltransferase vectors (50 μg) (Niwa et al., 2005; Shimosato et al., 2007) were transfected into 1×10⁷ CBMS1 ES cells and 50 μg Gata6GR or Cdx2ER-IRES-hygromycin phosphotransferase vectors were transfected into 1×10⁷ B142bgeoEG ES cells by electroporation. Cdx2ER or Gata6GR stably expressing clones were selected by 1.5 μg/ml puromycin (Nakarai Tesque) or 100 μg/ml hygromycin B (InvivoGen), respectively. Then, the cell lines that could be induced to extra-embryonic lineage-like cells were selected under 1 μg/ml 4-hydroxytamoxifen (Tx) (Sigma) or 100 nM dexamethasone (Dex) (Sigma).

To distinguish random and imprinted Xi events in differentiated cells from ES cells, we used female ES cell lines in which the activities of the paternal and maternal X chromosomes are distinguishable. One line we chose was the hybrid F1 CBMS1 female ES cell derived from CBA and MsM mice. Since there are many genomic polymorphisms available between these different subspecies, the activities of two X chromosomes can be easily confirmed by PCR-restriction fragment length polymorphism (RFLP) analysis (Sugimoto and Abe, 2007). Another line we used was the B142bgeoEG female ES cell line, which carries EGFP and puromycin acetyltransferase on the paternal X chromosome and a fusion of lacZ and neomycin phosphotransferase on the maternal X chromosome. These transgenes on the X chromosomes are inactivated by Xi. ES cells carrying two active X chromosomes can be maintained under puromycin and G418 selection, although it is known that female ES cells tend to lose one X chromosome in long-term culture, and the inactivation of the paternal X chromosome can be monitored by the loss of EGFP fluorescence as detected by microscopy or FACS in living cells.

These two female ES cell lines were transduced with either Cdx2ER or Gata6GR to allow their inducible differentiation toward TE and PrE by Tx and Dex, respectively. To avoid overexpression effects, these extra-embryonic lineage-like cells were induced by transient addition of Tx and Dex, for 3 days and 2 days, respectively. Proper differentiation of these transgenic female ES cell lines was confirmed by following their morphological differentiation (Fig. 1A,B) as well as by monitoring marker gene expression as detected by quantitative (q) PCR or immunostaining (Fig. 2). The loss of pluripotency markers [Oct3/4 (Pou5f1), Nanog, Rex1 (Zfp42)] and upregulation of either TE markers [Dlx3, Elf5, Psx1 (Rhox6) and Hand1] or PrE markers (Gata4, Sox7, Sox17 and Dab2) confirmed their exclusive differentiation toward the extra-embryonic cell lineages (Fig. 2). The maintenance of two X chromosomes in these transgenic female ES cell lines was confirmed by X chromosome painting analysis (Fig. 1C), indicating that they were ready to undergo Xi after induction of differentiation.

In the culture without LIF, all morphologically differentiated cells showed an H3K27me3 focus, a marker of Xi (Heard and Disteche, 2006), in each nucleus (see Fig. S1 in the supplementary material). The homogeneous acquisition of an H3K27me3 focus was also observed in Flk1 (Kdr)-positive mesoderm precursors induced by culture without LIF on a collagen IV-coated surface (Nishikawa et al., 1998). We attempted to confirm Xi in TE and PrE cells derived from these female ES cells carrying Cdx2ER or Gata6GR by Xist RNA fluorescence in situ hybridization (FISH) combined with immunostaining for H3K27me3 (Fig. 1E,F). In undifferentiated ES cells, because both X chromosomes are active, two pinpoint Xist RNA signals that do not merge with H3K27me3 immunostaining were observed. By contrast, in TE and PrE cells, a single, large Xist RNA focus that merged with the H3K27me3 focus was observed, indicating the Xi event. As shown in Fig. 1D, large foci were observed in all nuclei of differentiated extra-embryonic cells by immunostaining for H3K27me3, indicating that all TE and PrE cells carry one inactive X chromosome.

Next, we tested which type of Xi occurred in these cells. For TE and PrE cells derived from CBMS1 ES cells, we performed the PCR-RFLP assay to detect transcripts from paternal and maternal X chromosomes separately. The reverse-transcribed transcripts from genes on the X chromosome were amplified by PCR (for primers, see Table S2 in the supplementary material), followed by digestion with appropriate restriction enzymes that cut either CBA (maternal) or MsM (paternal) derived PCR fragments, resulting in fragments of different sizes as sized on agarose or acrylamide gels. First, we tested Xi in CBMS1 differentiated into embryonic cell lineages. When CBMS1 ES cells differentiated into mesoderm on a collagen IV-coated dish were examined by PCR-RFLP (see Fig. S2 in the supplementary material), since both X chromosomes are active in ES cells, the analysis gave two fragments for each transcript (Fig. 3 and see Fig. S2 in the supplementary material). Both paternal and maternal transcripts were also detected in purified Flk1⁺ mesodermal cells derived from either parental CBMS1 ES cells or the derivatives carrying Cdx2ER or Gata6GR. Since these mesodermal cells exhibited the H3K27me3 focus as a sign of Xi, as shown above (see Fig. S1 in the supplementary material), these data indicated that they had undergone random Xi as in normal development (see Fig. S2 in the supplementary material). Then, we tested TE or PrE cells derived from CBMS1 ES cells induced by activation of Cdx2ER or Gata6GR, respectively. If they underwent imprinted Xi, then only maternal X-linked transcripts would be expressed. However, in both cases, both paternal and maternal transcripts were detected for all X chromosome-linked genes tested, as in the case of the mesodermal cells (Fig. 3). The same results were confirmed in three independently established clones. Since the Xi event was confirmed, as shown above, these data clearly showed that random Xi had occurred in these TE and PrE cells. By contrast, two autosomal imprinted genes, H19 and Igf2r, showed exclusive expression from the maternal chromosome in TE and PrE cells, indicating the proper maintenance of imprinted marks on CBMS1 ES cell chromosomes (see Fig. S3 in the supplementary material).

We then examined the Xi events in B142bgeoEG ES cell-derived TE and PrE cells by FACS analysis to monitor EGFP expression from the paternal X chromosome. This analysis confirmed their homogeneous EGFP expression, indicating that the paternal X chromosome is active in all cells. After culture without LIF on collagen IV-coated dishes, ∼50% of the Flk1⁺ mesodermal cells showed EGFP expression, indicating the occurrence of random Xi (Fig. 4E,F and see Fig. S4 in the supplementary material). The B142bgeoEG derivatives carrying either Cdx2ER or Gata6GR also showed homogeneous EGFP expression, confirming the activity of the paternal X chromosome (Fig. 4A,B). If imprinted Xi occurs in TE and PrE cells derived from these ES cells, then they would lose EGFP fluorescence completely, whereas half of them would retain the EGFP signal after random Xi. The data shown in Fig. 4C,E,F unequivocally show the latter to be the case in both TE and PrE cells, indicating that they chose random Xi (Fig. 4C,E,F). The same results were confirmed in three independently established clones. These data indicated that extra-embryonic lineage cells induced from ES cells chose random Xi, as in Flk1⁺ mesodermal cells. Therefore, differentiation toward the extra-embryonic cell lineages is not coupled with the choice of imprinted Xi in this context.

The data obtained using the ES cell in vitro differentiation system indicated that female ES cells carrying two active X chromosomes lost all the imprinted marks on the paternal and maternal X chromosomes that might otherwise direct imprinted Xi in TE and PrE. To confirm the complete erasure of X-linked imprinting, we asked whether random Xi occurs in TE cells in cloned embryos generated by the transfer of B142bgeoEG nuclei into enucleated oocytes. We found that 30-40% of nuclear-transferred oocytes developed to blastocysts, which is comparable to that observed for male ES cell-derived nuclear transfer (Wakayama et al., 1999). Homogeneous EGFP expression was detected in the ICM, indicating activation of the paternal X chromosome (Fig. 4D and see Fig. S5 in the supplementary material). By contrast, a mosaic pattern of EGFP expression was observed in TE, although these cells carried the H3K27me3 focus in each nucleus, suggesting that they chose random rather than imprinted Xi, as in the ES cell in vitro differentiation system. Therefore, the female ES cell nuclei chose random Xi during differentiation to TE, even in the context of embryonic development.

It has been reported that the inactive X chromosome in female somatic cell nuclei is reactivated after nuclear transfer but is then selectively inactivated in extra-embryonic cell lineages, in which imprinted Xi occurs in normal embryos (Eggan et al., 2000), suggesting that the reactivation of the X chromosome is imperfect in cloned oocytes and that the residual marks on randomly inactivated X chromosomes are traced for the initiation of the first Xi in morula as the imprinted Xi based on the parental imprinted marks. By contrast, the female ES cell nuclei undergo random Xi in cloned embryos, indicating that all marks that trigger selective Xi are completely erased from both maternal and paternal X chromosomes.

Recent reports applied the ES cell system to analyze the molecular mechanisms governing the initiation of Xi and revealed that pluripotency-associated transcription factors, such as Oct3/4, Sox2 and Nanog, are involved in the repression of Xist expression in the undifferentiated state (Donohoe et al., 2009; Navarro et al., 2008). Such regulation might confer the initiation of Xi after induction of cell differentiation. However, in pre-implantation embryos, the paternal X chromosome is activated in 2-cell embryos and then starts to be inactivated at the 4-cell stage, although such cells express Oct3/4 and Sox2 from both maternal transcripts and by zygotic gene expression (Avilion et al., 2003; Pesce et al., 1998). By contrast, zygotic expression of Nanog starts in 16- to 32-cell embryos and is retained in the ICM of blastocysts. In late blastocysts, paternal X reactivation occurs only in the Nanog-expressing ICM cells. Since the reactivation of the paternal X chromosome does not occur in the ICM of Nanog-null embryos, Nanog is evidently required for X reactivation, although it remains unclear whether the reactivation is achieved by the repression of Xist by Nanog in cooperation with Oct3/4 and Sox2 (Silva et al., 2009). The molecular mechanisms of the complete erasure of the imprinted marks that accompanies reactivation of the paternal X chromosome as triggered by Nanog will be the next question to address in the future.

We thank Takashi Sado and Hiroyuki Kugoh for providing the Xist exon probes; Michihiko Sugimoto and Kuniya Abe for sharing information on primers and MsM genomic sequences used in PCR-RFLP analysis; and Yoko Futatsugi-Nakai for helpful comments on the manuscript. This research was supported by a RIKEN grant.