Molecular Coupling of Xist Regulation and Pluripotency

See allHide authors and affiliations

Science  19 Sep 2008:
Vol. 321, Issue 5896, pp. 1693-1695
DOI: 10.1126/science.1160952


During mouse embryogenesis, reversion of imprinted X chromosome inactivation in the pluripotent inner cell mass of the female blastocyst is initiated by the repression of Xist from the paternal X chromosome. Here we report that key factors supporting pluripotency—Nanog, Oct3/4, and Sox2—bind within Xist intron 1 in undifferentiated embryonic stem (ES) cells. Whereas Nanog null ES cells display a reversible and moderate up-regulation of Xist in the absence of any apparent modification of Oct3/4 and Sox2 binding, the drastic release of all three factors from Xist intron 1 triggers rapid ectopic accumulation of Xist RNA. We conclude that the three main genetic factors underlying pluripotency cooperate to repress Xist and thus couple X inactivation reprogramming to the control of pluripotency during embryogenesis.

In mammals, X inactivation is controlled by the X-linked gene Xist, which produces a noncoding RNA that coats the X chromosome in cis and triggers X inactivation (1). During the cleavage stages of the female preimplantation mouse embryo, X inactivation is imprinted and Xist is specifically expressed from the paternal inactive X chromosome (24). Although imprinted Xist expression is maintained in extraembryonic tissues, in the pluripotent inner cell mass (ICM) of the blastocyst, Xist expression is extinguished and the paternal X chromosome reactivated (3, 4). Subsequently, in the ICM-derived differentiating epiblast, monoallelic Xist expression is randomly up-regulated to establish random X inactivation. To date, the molecular mechanisms specifying X chromosome reprogramming in the ICM have not been defined.

Derived from the ICM, pluripotent embryonic stem (ES) cells express low levels of Xist. Upon differentiation, random X inactivation is established through monoallelic up-regulation of Xist in females but not in males. ES cells are therefore a suitable ex vivo model system for addressing how Xist is repressed in the pluripotent ICM. A major repressor of Xist RNA accumulation has already been characterized in ES cells: the anti-sense noncoding RNA Tsix (5). However, the paternally inherited invalidation of Tsix does not block the normal reactivation of the paternal X chromosome in the ICM (6), and Xist transcription remains repressed in Tsix-mutant undifferentiated ES cells (79). An as yet unknown ES cell–specific repressive activity must therefore be responsible for ensuring Xist down-regulation until cellular differentiation is initiated.

Given the tight correlation between Xist regulation and pluripotency observed in both the embryo and ES cells, the core transcription factors of the genetic circuitry underlying pluripotency—Nanog (10, 11), Oct3/4 (12), and Sox2 (13)—may be involved in the transcriptional repression of Xist. In the ICM of the blastocyst, the reversion of X inactivation follows Nanog expression (3). In three independently generated Nanog-null ES cells [RCNβH-B(t), RCNTK(t), and TβC44cre6] (14), Xist is four- to sixfold up-regulated as compared with the RCN(t) and TβC44 control cell lines (Fig. 1A), whereas restoration of Nanog to RCNβH-B(t) by reparative homologous recombination is accompanied by efficient Xist repression [RCNβH-B(t)-Rescue] (Fig. 1A). Thus, Nanog and Xist expression are inversely correlated in ES cells. Xist up-regulation is rapidly established 48 hours after the induction of Nanog deletion and does not involve Tsix down-regulation (fig. S1), indicating that Xist might be a direct target of Nanog.

Fig. 1.

Nanog binds Xist intron 1 to down-regulate Xist expression. (A) Relative expression levels of Xist in different Nanog mutants: RCN(t) (Nanog+/+), RCNβH-B(t) (Nanog–/–), RCNTK(t) (Nanog–/–), RCNβH-B(t)-Rescue (Nanog+/–), TβC44 (Nanog+/–), and TβC44cre6 (Nanog–/–). Xist RNA levels in RCN(t) were set to one (values and error bars are mean ± SD, N = 5). The difference in Xist RNA levels between Nanog expressing and not expressing cells is statistically significant (P < 0,0001, unpaired t test, N = 15). (B) Map of the Xist/Tsix locus. Xist exons, gray; Tsix exons, white. Each primer pair used in the ChIP assay is indicated by a black dot. The coordinate of the primer pair overlapping the Xist transcription initiation site was set as the origin of the x axis (0 kb). (C) Profile of Nanog binding across the Xist/Tsix region in undifferentiated female ES cells (LF2 cell line). (Inset) Result obtained in parallel experiments at the ES cell–specific Oct3/4 enhancer (25). Both graphs show the percentage of immunoprecipitation (% IP) obtained after normalization to the input (values and error bars are mean ± SD, N = 4). (D) Analysis of Nanog mRNA expression (N = 5) and ChIP signal at Xist intron 1 in Nanog-null ES cells [N = 4 for RCN(t) and RCNβH-B(t), N = 2 for all other cell lines]. The level of Nanog mRNA was calculated and presented as in (A). The relative levels of binding were calculated by dividing the % IP obtained in each cell line with the primer pair providing the maximal value in (C) by that measured in RCN(t) (set to one, all values and error bars are mean ± SD).

Chromatin immunoprecipitation (ChIP) (15) has demonstrated that Nanog is associated with the first intron of the Xist gene in both female and male undifferentiated ES cells (Fig. 1, C and D), but not in differentiating ES cells, mouse embryonic fibroblasts (fig. S2), or the three Nanog-null ES cell lines (Fig. 1D). As expected, Nanog binding to Xist intron 1 DNA is restored in RCNβH-B (t)-Rescue (Fig. 1D) and is unchanged in the Tsix-truncated ES cell line Ma2L (16) (fig. S3A). We conclude that Nanog displays the appropriate characteristics of the oft-looked-for, ES cell–specific repressor of Xist transcription. However, the four- to sixfold up-regulation of Xist detected in Nanog-null cells is considerably below that generally associated with X inactivation induction in differentiating wild-type (WT) female ES or Tsix-mutant male ES cells (1000-fold, fig. S3B). This suggests the existence of additional factors whose repressive activity would remain unaffected by Nanog invalidation. Prime candidates are the pluripotency-associated transcription factors Oct3/4 and Sox2, whose binding sites overlap substantially with those of Nanog (17, 18).

Similarly to Nanog, Oct3/4 and Sox2 bind within Xist intron 1 DNA in undifferentiated WT and Tsix-truncated ES cells (Fig. 2 and figs. S2 and S3A). In Nanog-null ES cells, both Oct3/4 and Sox2 remain bound to Xist intron 1 DNA (fig. S4), supporting the idea that Oct3/4 and Sox2 prevent complete up-regulation of Xist in Nanog-null cells. To analyze Oct3/4 function, we took advantage of the Oct3/4-null male ES cell line ZHBTc4 (19), in which Oct3/4 expression is sustained from a tetracycline (Tc)–repressible transgene. Tc treatment triggers rapid silencing of Oct3/4 (Fig. 3A) and induces differentiation (fig. S5A). Accordingly, Nanog, Sox2, and Tsix are progressively down-regulated from 48 hours of Tc treatment (Fig. 3A). In contrast, Xist is consistently up-regulated after 24 hours of Tc treatment (Fig. 3A), reaching levels similar to those observed in differentiating female ES cells (fig. S6A) and indicating that Oct3/4 represses Xist. The lack of Xist up-regulation in ZHBTc4 cells differentiated by retinoic acid (RA) in the absence of Tc excludes the occurrence of anomalous up-regulation of Xist, unlinked to Oct3/4 silencing, in differentiating ZHBTc4 cells (fig. S5, D and E). The rapid silencing of Oct3/4 triggered by Tc induces the loss of Oct3/4 from Xist intron 1 DNA within 24 hours, and after 48 hours, this is followed by the loss of Sox2 and Nanog binding (Fig. 3B). In contrast, during the first 48 hours of exposure to RA, Oct3/4, Sox2, and Nanog binding persists (Fig. 3B). This indicates that the major increase of Xist expression observed in Tc-treated ZHBTc4 cells is a consequence of the premature loss of binding, not only of Oct3/4 but also of Sox2 and Nanog. The three factors therefore act synergistically to repress Xist.

Fig. 2.

Direct binding of Oct3/4 and Sox2 to Xist intron 1. (A) Map of the Xist/Tsix locus. Binding of Oct3/4 (B) and Sox2 (C) across Xist/Tsix in undifferentiated female ES cells (LF2 cell line), obtained and presented as in Fig. 1 (values and error bars are mean ± SD, N = 4).

Fig. 3.

Inappropriate Xist up-regulation in male ES cells upon drastic silencing of Oct3/4. (A) Time-course variation of gene expression upon addition of Tc to the male ZHBTc4 ES cell line. The graph shows the ratio of expression levels between Tc-treated and untreated ZHBTc4 cells, after standardization to ArpoP0 mRNA levels. The results were plotted on a log10-scale y axis (values and error bars are mean ± SD, N = 4). (B) Binding of Oct3/4, Sox2, and Nanog in undifferentiated (set here to one) and differentiating ZHBTc4 cells treated with either RA or Tc (values are mean ± SD, N = 3). (C) RNA-FISH analysis of Xist (L510 probe, green) on 4′,6′-diamidino-2-phenylindole–stained (blue) nuclei after 4 days of treatment with Tc. (D) RNA-FISH analysis of Xist (L510 probe, green) and Tsix (E17 probe, red) after 24 and 96 hours of Tc treatment.

Ectopic accumulation of Xist RNA is observed in a fraction of the Tc-treated ZHBTc4 cell population that maximally reaches 10% after 96 hours of treatment (N > 200 counted nuclei) (Fig. 3C). The marked up-regulation of Xist may therefore concern only a limited subpopulation of Tc-treated ZHBTc4 ES cells (fig. S6). Because Tsix blocks Xist RNA accumulation during differentiation (5), the rare Tc-treated ZHBTc4 cells accumulating Xist RNA could have prematurely extinguished Tsix transcription. This would indicate that Tsix, and not Xist, is the primary target of Oct3/4, Sox2, and Nanog. However, Xist up-regulation [like that of Cdx2, a known target of Oct3/4 (20)] is established as early as 24 hours post-Tc treatment (P = 0.0098, paired t test, N = 4 independent experiments) and precedes Tsix down-regulation, whose expression remains unaffected during the first 24 hours of treatment (P = 0.1678, paired t test, N = 4). In addition, RNA–fluorescence in situ hybridization (FISH) revealed that 92 and 36% of the Xist-positive Tc-treated cells still display a Tsix signal after 24 (N = 118) and 96 hours (N = 182) of treatment (Fig. 3D). Thus, Xist repression by Oct3/4, Sox2, and Nanog is probably not a secondary consequence of their direct action on Tsix. In agreement with this, the loss of these three factors in Tsix-truncated cells is accompanied by Xist up-regulation (fig. S3B), demonstrating that Tsix is not required for the repression of Xist mediated by Nanog, Oct3/4, and Sox2. The spatial association of Xist intron 1 DNA with the Xist promoter in ES cells [but not with the Tsix promoter (21)] further indicates that Xist is the direct target of Nanog, Oct3/4, and Sox2.

The heterogeneity of Xist RNA accumulation after Tc addition might alternatively be linked to Nanog mosaicism (14). Indeed, ∼20% of undifferentiated ES cells are Nanog-negative at any given time (14), and Xist is partially derepressed in such cells (fig. S7). Ectopic Xist activation upon acute Oct3/4 extinction in Tc-treated cells could be restricted to this Nanog-negative fraction. This predicts that the Nanog-negative subpopulation of female ES cells would be naturally primed for the initiation of X inactivation on differentiation. In agreement, Xist up-regulation systematically occurs in Nanog-negative WT female ES cells after 48 hours of differentiation (fig. S8).

Ectopic Xist up-regulation in male ES cells in which Nanog, Oct3/4, and Sox2 binding is individually or collectively manipulated (Figs. 1 and 3), together with the correlation at the single cell–level between Nanog-Oct3/4 extinction and Xist RNA accumulation in differentiating WT female ES cells (fig. S8), demonstrates that Nanog, Oct3/4, and Sox2 are critical ES cell–specific repressors of Xist acting synergistically to couple X inactivation regulation to pluripotency. Because in the ICM (from which ES cells are derived) the paternal X is reactivated (3, 4), we propose that these three critical regulators of pluripotency play a direct and pivotal role in X chromosome reprogramming through their mediation of a transient repression of Xist (fig. S9). This proposal argues against the prevalent view that reversion of Xp inactivation in the ICM is a reflection of more global reprogramming events occurring at this stage in support of pluripotency (4, 22). In female primordial germ cells (PGCs), X reactivation is initiated by the extinction of Xist RNA immediately after reexpression of Nanog (23, 24). Because the expression of Oct3/4 and Sox2 also characterizes pluripotent PGCs (22), we speculate that Nanog, Oct3/4, and Sox2 may also be directly involved in X chromosome reprogramming during PGC development. Deciphering whether the molecular triggers of pluripotency play a direct function in other pluripotency-associated epigenetic processes will shed light on how genetic and epigenetic mechanisms interact together to ensure the appropriate progression of embryonic development.

Supporting Online Material

Materials and Methods

Figs. S1 to S10


References and Notes

View Abstract

Navigate This Article