FEBS Letters 583 (2009) 1721–1727

journal homepage: www.FEBSLetters .org

Minireview

When X-inactivation meets pluripotency: An intimate rendezvous

Pablo Navarro *, Philip AvnerInstitut Pasteur, Unité de Génétique Moléculaire Murine, CNRS, URA2578, F-75015 Paris, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 February 2009Revised 17 March 2009Accepted 18 March 2009Available online 25 March 2009

Edited by Miguel De la Rosa

Keywords:X-inactivationPluripotencyX-inactive specific transcriptNanogOct4Sox2

0014-5793/$36.00 � 2009 Federation of European Biodoi:10.1016/j.febslet.2009.03.043

Abbreviations: ICM, inner cell mass; TE, trophecderm; EPI, epiblast; PGCs, primordial germ cells; Eembryonic stem; iPS, induced pluripotent stem; Xic, Xinactive specific transcript; Tsix, Xist antisense; Xi,immunoprecipitation

* Corresponding author. Fax: +33 1 45 68 86 53.E-mail address: pnavarro@pasteur.fr (P. Navarro).

The integration of X-inactivation with development is a crucial aspect of this classical paradigm ofepigenetic regulation. During early female mouse development, X-inactivation reprogrammingoccurs in pluripotent cells of the inner cell mass of the blastocyst and in pluripotent primordialgerm cells. Here we discuss the developmental strategies which ensure the coupling of the regula-tion of X-inactivation to the acquisition of pluripotency through the regulation of the master of X-inactivation, the non-coding Xist gene, by the key factors which support pluripotency Nanog, Oct4and Sox2.� 2009 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. A historical perspective of X-inactivation regulation during cesses characterize X-inactivation: in the extra-embryonic

early mouse development

In mammals, the emergence of a Y chromosome virtually en-tirely devoted to male sex determination has been associated dur-ing evolution with the appearance of a dosage compensationmechanism that equalizes the level of X-linked gene expressionbetween XY males and XX females. Mary Lyon proposed, basedon cytological and genetic evidence, that to achieve dosage com-pensation of the X-chromosomes between males and females,one of the two X-chromosomes is inactivated early in femaleembryogenesis [1,2].

Following Lyon’s seminal work, the idea that X-inactivation oc-curs during cell differentiation became a long-standing concept.The first signs of X-inactivation in the female embryo were thoughtto appear early in the first tissue to differentiate, the trophoblast,and only later in tissues of the embryo proper [3]. This schemawas compatible with data suggesting that in the early blastocystboth X-chromosomes were active in cells of the undifferentiatedinner cell mass (ICM) [4], while in the trophectoderm one X-chro-mosome in each cell was in an inactive state. Two distinct pro-

chemical Societies. Published by E

toderm; PE, primitive endo-C, embryonic carcinoma; ES,

-inactivation center; Xist, X-inactive X; ChIP, chromatin

lineages, X-inactivation is imprinted with the paternal X-chromo-some always being chosen for inactivation [5], whereas in cellsof the embryo proper both X-chromosomes are targeted at randomby the inactivation process, through the so-called random X-inac-tivation. Cellular differentiation therefore appeared to drive X-inactivation either through imprinted or random mechanismsdepending on the cell lineage in question. A developmentalstem-cell model for X-inactivation was proposed by Monk [6], withX-inactivation occurring, albeit in different forms, at differenttimes and in different cell populations as they differentiate froma pluripotent state. This model was supported by ex vivo X-inacti-vation studies, initially exploiting the differentiation of femaleembryonic carcinoma (EC) cells [7], the stem cells of teratocarcino-mas, later the differentiation of female embryonic stem (ES) cells[8], the stem cells derived from the ICM. In both ex vivo systems,activity of the two X-chromosomes in the female cell is maintaineduntil cellular differentiation is initiated and random X-inactivationestablished, highlighting the existence of a close relationship be-tween the regulation of lineage commitment and the establish-ment of X-inactivation. For over 25 years, the view in the fieldthat prevailed based on such results was that during blastocyst for-mation, imprinted X-inactivation occurs first in the trophectodermand, subsequently, in the primitive endoderm, whereas both X-chromosomes remain active in those cells of the ICM that remainpluripotent. A corollary of this view is that at the onset of themulti-lineage differentiation of the epiblast that generates theembryo proper, X-inactivation is established for the first time

lsevier B.V. All rights reserved.

1722 P. Navarro, P. Avner / FEBS Letters 583 (2009) 1721–1727

and that this random X-inactivation is then stably maintainedthroughout development and the entire life of the organism, withreactivation of the inactive X occurring exclusively in the femalepluripotent germ line [9,10].

Reprogramming experiments further reinforced the idea that X-inactivation is intimately linked to differentiation. Reactivation ofthe inactive X-chromosome carried by female somatic nuclei hasbeen observed after cell fusion with male EC [11] and ES cells[12], as well as after nuclear transfer into the enucleated egg[13]. Importantly, the recent generation of induced pluripotentstem (iPS) cells by forced expression of ES cell regulators [14]has also been shown to be accompanied by the reactivation ofthe inactive X-chromosome of female somatic cells [15]. Basedon such experiments it appears that X-inactivation is establishedwhen the loss of pluripotency occurs and, reciprocally, that X-inac-tivation is reversed following the acquisition of pluripotency.

The recent discovery that imprinted X-inactivation takes placemuch earlier than predicted by the stem-cell model demonstratesthat the simple association of X-inactivation with cellular differen-tiation no longer holds. In a clear paradigm shift, three indepen-dent articles convincingly argued against the conventional viewby demonstrating that X-inactivation initiates several cell divisionsprior to the formation of the blastocyst [16,17]. Imprinted X-inac-tivation, associated with the exclusive inactivation of the paternal

Fig. 1. Developmental dynamics of X-inactivation. Imprinted X-inactivation of thepaternal X-chromosome is first established at the 2–4-cell transition of early femaleembryogenesis. This initial form of X-inactivation is maintained during thecleavage-stages of the morula, as well as during the differentiation of theextraembryonic tissues such as the trophectoderm (TE, in pink) and the primitiveendoderm (PE, in purple). The paternal inactive X (Xi) is then reactivated in thepluripotent cells of the inner cell mass (ICM, in light yellow) of the blastocyst whichallows the establishment of random X-inactivation in the differentiating epiblast(EPI, in orange). This is the form of X-inactivation that will be maintained in somatictissues of the post-implantation embryo and in the adult. The randomly chosen Xi isreactivated in migrating pluripotent primordial germ cells (PGC, in yellow).

X-chromosome, was shown to be implemented in all cells of thecleavage-stage embryo. This initial form of imprinted X-inactiva-tion was shown, however, to be labile and at the blastocyst stage,the paternal X is reactivated in the ICM [16,17]. This results in bothX-chromosomes being active in undifferentiated cells of the ICMduring a short time-window. Subsequently, random X-inactivationis initiated with either the paternal or the maternal X being chosenfor inactivation. In contrast, X-inactivation remains imprinted inextra-embryonic tissues. Therefore, the critical developmental reg-ulation of X-inactivation is based on the reactivation of the pater-nally-inherited inactive X-chromosome in the ICM, rather than in asimple coupling of the inactivation and differentiation processes.Although it is now clear that imprinted X-inactivation is not perse associated to differentiation, it remains true that random X-inactivation is linked to the differentiation of the epiblast, as illus-trated by differentiating female ES cells.

In summary (Fig. 1), during early female mice development, X-inactivation reprogramming occurs in pluripotent cells of the innercell mass of the blastocyst, when imprinted X-inactivation is re-placed by random inactivation, via a transient stage characterizedby the presence of two active X-chromosomes. Reactivation of theinactive X also occurs in pluripotent primordial germ cells (PGCs)and is also observed in vitro, during the reprogramming of femalesomatic cells mediated by nuclear cloning, by fusion with EC andES cells, and during the generation of iPS cells. Reprogrammingof X-inactivation is therefore associated with the acquisition ofpluripotency both in vivo and in vitro [18].

2. Developmental regulation of Xist, the trigger ofX-inactivation

The initiation of X-inactivation is controlled by the X-inactiva-tion center (Xic), a complex X-linked locus responsible for the inac-tivation of a single X in female cells and an absence of inactivationin male cells [19]. The Xist gene lies within the Xic and produces anessential non-coding RNA with the unique property of coating andsilencing the X-chromosome in cis [20]. Given that only high levelsof Xist RNA can induce X-inactivation, Xist expression has to betightly regulated in order to ensure the dynamics of X-inactivationduring development. In pre-implantation embryos, Xist expressionis imprinted and high Xist RNA levels are exclusively producedfrom the paternal X-chromosome. Drastic changes in Xist expres-sion pattern take place in the ICM, where paternal Xist expressionis efficiently repressed and this correlates with the reactivation ofthe paternal X-chromosome [16,17]. At the onset of random X-inactivation, Xist is upregulated specifically on the future inactiveX, irrespectively of its parental origin. Accordingly, in undifferenti-ated female ES cells, both X-chromosomes produce low levels ofXist RNA. As the cell differentiates, Xist is mono-allelically upregu-lated at random to induce X-inactivation in cis, whilst the secondXist allele of females and the single Xist allele of males are turnedoff.

The randomly chosen inactive X-chromosome is also reacti-vated in the female germ line. Similarly to what has been describedin the ICM, the initial step of the reversion of X-inactivation ap-pears to be the repression of Xist which occurs in migrating PGCs[21,22]. Thus, in the germ line, the repression of Xist expressionwhich leads to the reactivation of the inactive X is again correlatedwith the acquisition of pluripotency.

Two aspects of Xist regulation, the levels of Xist RNA and itschromosomal-origin, appear to be crucial for the developmentalregulation of X-inactivation. In particular, the repression of Xistexpression that characterizes the reprogramming events whichtakes place in pluripotent cells appears as a key developmentalevent in X-inactivation regulation. This notion is further supported

P. Navarro, P. Avner / FEBS Letters 583 (2009) 1721–1727 1723

by in vitro reprogramming experiments, when the reactivation ofthe inactive X is similarly associated with the repression of Xistexpression from the inactive X-chromosome [12,15]. We concludethat some pluripotency-specific activities must be responsible forthe silencing of Xist and the reactivation of the inactive X.

3. Transcriptional aspects of Xist regulation

Using chromatin immunoprecipitation (ChIP) analysis of threecell types representative of different types and stages of theX-inactivation process, we reported major variations in the tran-scriptional machinery recruitment activity of the Xist promoterthat correlate with the X-inactivation status of the correspondingchromosome [23]. The levels of transcription basal machineryassociated with the Xist promoter in cell types expressing high lev-els of Xist RNA, in which X-inactivation has already occurred (atrandom in female mouse embryonic fibroblasts, or imprinted introphectoderm stem cells) is considerably greater than the levelsdetected in cells expressing low levels of Xist, in which X-inactiva-tion has yet to occur, such as ES cells. This indicates that Xistexpression is regulated transcriptionally by the selective recruit-ment of the transcriptional apparatus to its promoter. Analysis ofXist transcription by nuclear run-on experiments carried out inES cells confirmed that Xist transcription is essentially absent priorto differentiation but strongly upregulated on differentiation [24].

The observed correlation in ES cells between low steady statelevels of Xist RNA, the presence of limiting amounts of thetranscriptional apparatus at the Xist promoter, and inefficient Xisttranscription, suggested strongly either the presence of a transcrip-tional repressor or the lack of a specific transcriptional activator insuch cells. Importantly, the repression of Xist transcription in EScells might be part of the mechanism allowing the reactivation ofthe paternal X-chromosome in the ICM, and by extrapolation, inother pluripotent stem cells.

One of the most exciting characteristics of the Xist gene is itscomplete overlap by a non-coding antisense transcription unit,Tsix, shown to act as a crucial cis-acting repressor of Xist upregula-tion: a tight correlation between antisense downregulation andXist RNA accumulation during random X-inactivation has previ-ously been established [25]. Both deletion of the major Tsix pro-moter and truncation of the antisense transcription unit leads toa complete bias in random X-inactivation towards the mutatedallele [26]. This strongly suggests that Tsix is likely involved inthe chromosomal-origin of Xist expression during random X-inac-tivation. It has similarly been proposed that Tsix is responsible forthe paternal-restricted expression of Xist in extra-embryonictissues exhibiting imprinted X-inactivation [27], although the roleof Tsix in the early, reversible imprinted Xist expression encoun-tered in the cleavage-stages of the morula has yet to be clearlyaddressed.

Given the repressive nature of Tsix transcription across Xist, itwas tempting to propose that Tsix is the crucial repressor of Xisttranscription in undifferentiated ES cells, in which Tsix is highlytranscribed. This would suggest that, in addition to its role in theestablishment of the appropriate chromosomal-origin of Xist tran-scription, Tsix would also be responsible for the suppression of Xisttranscription in pluripotent stem cells. Whilst Tsix is highly tran-scribed in the ICM, and its transcription efficiently restored duringin vitro reprogramming of female somatic cells [12,15], it cannot,however, be the developmentally regulated, pluripotency-specific,repressor of Xist. Indeed, Xist transcription remains repressed inundifferentiated Tsix-mutant ES cells, as evaluated both by ChIPanalysis of the transcriptional machinery at the Xist promoter[23] and by nuclear run-on evaluation of Xist transcription rate[24]. In addition, normal reactivation of the paternal X-chromo-

some in the ICM has been reported in the context of a pater-nally-inherited invalidation of Tsix [28]. The lack of Tsixtranscription observed in female PGCs [22], where Xist transcrip-tion is repressed and the randomly chosen inactive X-chromosomereactivated, further indicates that Tsix cannot be the sought aftergeneral Xist repressor in pluripotent cells.

What then are the other molecular signatures specific to plurip-otent cells that might act as the developmental repressors of Xisttranscription specifying X-inactivation reprogramming in pluripo-tent cells?

4. Direct molecular coupling of Xist repression and pluripotencyin ES cells

Three transcription factors, Nanog [29,30], Oct4 [31] and Sox2[32], are known to be essential for the triggering and maintenanceof the pluripotent phenotype. Given the profound epigenetic repro-gramming that accompanies the acquisition of pluripotency, it waspreviously thought that the transcriptional repression of Xist andthe reactivation of the inactive X-chromosome were likely second-ary reflections of the pluripotent state [17,18]. Thus, X-inactivationreprogramming was view as an indirect epigenetic consequence ofthe action of Nanog, Oct4 and Sox2. The finding that the triumvirate,Nanog, Oct4 and Sox2, bind directly to the chromatin of the Xistgene in undifferentiated ES cells to maintain Xist repression priorto the onset of differentiation [33] argues against this view.

Using ChIP analysis, we demonstrated that Nanog, Oct4 andSox2 bind Xist intron 1 in both male and female undifferentiatedES cells. The binding of the three factors was found to be sharplyreduced in differentiating ES cells, and undetectable in fully differ-entiated mouse embryonic fibroblasts. Importantly, binding of thethree factors was shown to be independent of Tsix transcription, asexpected for Tsix-independent regulators of Xist. Thus, three devel-opmentally regulated transcription factors, whose own function isdedicated to pluripotency, bind the Xist locus with all the charac-teristics required for the pluripotency-specific, Tsix-independent,repressor of Xist transcription [33].

The finding that the Xist gene is a direct target for Oct4, Nanogand Sox2 suggests that these pluripotency-associated transcrip-tional regulators are directly responsible for Xist repression inpluripotent cells. Testing this hypothesis in female ES cells is,however, difficult, as knock-out or knock-down of these factorscan induce or commit such cells to cell differentiation with, as aconsequence, the triggering of Xist upregulation and X-inactiva-tion. In contrast, in differentiating male ES cells Xist expression isnever normally upregulated because additional activities of theXic inhibit the initiation of X-inactivation [19]. Genetic manipula-tion which resulted in the abrogation of the expression of one orother of the pluripotent factors and which resulted in the inappro-priate regulation of Xist in male cells would reveal and confirm aneventual intimate relationship existing between the master genesof pluripotency and Xist. The analysis of male ES cells in which Na-nog was homozygously deleted [34] showed that mutation of Na-nog alone caused a moderate increase in Xist expression, whilstOct4 and Sox2 remained bound, potentially preventing more com-plete activation of Xist transcription [33]. Importantly, this modestupregulation of Xist observed in Nanog�/� cells is an early conse-quence arising after Nanog deletion and is independent of Tsixdownregulation. Restoration of Nanog by homologous recombina-tion mediated rescue of Nanog�/� cells is accompanied by arepression of Xist expression levels to those of wild-type cells.

Genetic invalidation of Oct4 was previously shown to induce thecomplete loss of pluripotency [35]. Accordingly, in an inducibleOct4 invalidation system [35], silencing of Oct4 in male ES cellstriggers the drastic loss from Xist intron 1 not only of Oct4, but also

1724 P. Navarro, P. Avner / FEBS Letters 583 (2009) 1721–1727

of Nanog and Sox2. Strikingly, this induces a large and rapid in-crease in Xist expression to a level similar to that seen in differen-tiating female ES cells, which precedes any measurabledownregulation of Tsix [33]. It appears, therefore, that Nanog,Oct4 and Sox2 synergize to repress Xist transcription in undifferen-tiated ES cells independently of Tsix transcription. The Tsix-inde-pendency of the repressive action of Nanog, Oct4 and Sox2 onXist is further supported by the upregulation of Xist transcriptionobserved in Tsix-mutant ES cells both upon loss of pluripotency in-duced by differentiation [24,33,36] and by siRNA knock-down ofOct4 (PN unpublished observations).

Finally, it has been observed that Xist is upregulated in un-differentiated male ES cells in which a part of Xist exon 1 has beenreplaced with a promoter-less IRES-EGFP cassette [37]. Impor-tantly, in these mutant ES cells, Tsix was not downregulated. Thissuggests that the genetic alteration introduced at the Xist locusimpaired the ES cell-specific, Tsix-independent, silencing mecha-nism of Xist. These unexpected observations lead the authors tohypothese that the deleted region might contain a cis regulatoryelement required for Xist-silencing in undifferentiated cells [37].Interestingly, the 30 boundary of the replaced sequence lies withinXist intron 1. Fine mapping of the deletion boundary in respect tothe pcr amplicon providing maximal binding of Nanog, Oct4 andSox2 in ChIP assays [33] suggests that in such mutant cells bindingof the three factors might be disrupted. Localization of Nanog,Oct4 and Sox2 binding sites within Xist intron 1 using publishedChIP-Seq [38] also indicates that the Nanog and Oct4 binding siteshave been removed in the mutant cells. We conclude that deletionof the binding sequences for Nanog and Oct4 within Xist intron1 induce Xist upregulation in the absence of any measurabledownregulation of Tsix, in agreement with our results [33]. Futureexperiments specifically deleting Xist intron 1 will, however, beneeded to confirm these observations.

We conclude that the basic molecular framework for Xist regu-lation in ES cells depends on the repression of Xist transcriptionthat is ensured by the direct repressive action of Nanog, Oct4 andSox2. Our observations provide novel explanations for both theold observation that random X-inactivation is associated with dif-ferentiation, and for the association of the phenomena of X-inacti-vation reprogramming with the acquisition of pluripotency. In thefollowing two sections we provide two non-mutually exclusivehypotheses that seek to provide a framework for integrating Xistregulation by Nanog, Oct4 and Sox2 with the biology of the ICMand germ cells.

Fig. 2. Hypothetical scenario of the ordered recruitment of Nanog, Oct4 and Sox2 atXist intron 1 during early embryogenesis. (A) During the cleavage-stages of themorula, Xist is highly expressed from the paternal X-chromosome although Oct4 (inblue) and Sox2 (in green) are expressed. We suggest here that Oct4 and Sox2 cannotrepress Xist at this stage because their cognate binding sequences is masked by thenucleosomal array (in gray) spanning Xist intron 1. (B) Nanog is first expressed insome internal cells of the late morula, where it binds to Xist intron 1 and recruitschromatin remodeling complexes such as SWI/SNF (in yellow). (C) During theformation of the blastocyst, Xist intron 1 chromatin is reorganized to unmask thecognate binding sequences for Oct4 and Sox2. (D) In the ICM of early blastocysts,Oct4 and Sox2 can be recruited to Xist intron 1 to suppress Xist transcriptionefficiently, and allow the reactivation of the inactive paternal X-chromosome. Seetext for details.

5. Nanog, Oct4 and Sox2 as the genetic factors specifying Xistrepression in pluripotent cells

Nanog, Oct4 and Sox2 confer on Xist the expression patternappropriate to undifferentiated ES cells. We therefore propose thatthey suppress Xist expression in cells of all the pluripotent com-partments of the early embryo including both the ICM and PGCs.In the ICM, paternal X-inactivation is reverted and random X-inac-tivation prepared [16,17], suggesting that Nanog, Oct4 and Sox2play a direct pivotal role in this process through the transientrepression of Xist transcription. Interestingly, although Oct4 andSox2 are expressed in all cleavage-stages of the morula before theirexpression becomes progressively confined to the ICM during blas-tocyst formation [18], the first cells to reactivate the paternal inac-tive X-chromosome are those internal cells of the late morula thatestablish Nanog expression [16]. Similarly, although Oct4 and Sox2are expressed during the initial stages of PGCs development, therepression of Xist expression and the reactivation of the inactiveX-chromosome is not initiated until Nanog is re-expressed[21,22]. These results, which suggest that Xist repression coincides

temporally and spacially with the acquisition of pluripotency, indi-cate that Nanog probably plays the preponderant role in the repres-sion of Xist.

Surprisingly in this context, in Nanog-null undifferentiated EScells Oct4 and Sox2 remain bound to Xist intron 1, and Xist tran-scription remains strongly repressed. Conversely upon Oct4 silenc-ing, both Nanog and Sox2 are lost from Xist intron 1 beforedowregulation of their respective mRNAs, indicating that Oct4must be required for maintaining the binding of Nanog and Sox2at Xist [33]. It appears, therefore, that the principles governingbinding of Nanog, Oct4 and Sox2 at Xist intron 1 in ES cells may dif-fer from what would be expected from a simple correlative analy-sis of Xist, Nanog, Oct4 and Sox2 expression during earlyembryogenesis (Fig. 2). It is noteworthy that this apparent paradoxis not specific to the regulation of Xist, as it also emerges from aconsideration of the differential in vivo and ex vivo requirementsof the pluripotent phenotype for Nanog. For example whilst Nanogcan be removed from ES cells without severely affecting pluripo-tency [34], the absence of Nanog during early embryogenesis in-duces a failure to establish pluripotent stem cells in the ICM [30]and during PGC development [34]. Similarly, although Nanog is dis-pensable for reprogramming somatic cells back to pluripotency byectopic expression of a minimal cocktail of transcription factorsincluding Oct4 and Sox2, full pluripotency and repression of Xistis not acquired until the endogenous Nanog gene is re-expressed.

P. Navarro, P. Avner / FEBS Letters 583 (2009) 1721–1727 1725

This suggests that Nanog acts as a rheostat mainly required to in-duce and establish pluripotency, rather than to maintain such astate [34]. Accordingly, Nanog has been shown to promote transferof pluripotency after cell fusion [39]. Nanog might therefore play acrucial role in the establishment of Xist repression, but not neces-sarily in its maintenance. It seems therefore reasonable to proposethat Oct4 and Sox2 are able to repress Xist transcription only inNanog-positive cells of the late morula and migrating PGCs, andthat it is this that leads to the complete extinction of Xist RNAand to the reversion of the inactive X-chromosome to the tran-scriptional ground state which characterizes both X-chromosomesin the ICM and in germ cells. This would also explain why Xist isnot repressed during the generation of iPS cells until Nanog isre-expressed [15].

Under this model, it is only when Nanog binds to Xist intron 1that Sox2 and Oct4 either can be recruited or become functionalfor repressing Xist, after which their action becomes Nanog-inde-pendent. A likely mechanism (Fig. 2) could involve a Nanog-depen-dent re-organization of Xist intron 1 chromatin to render accessiblefor binding the cognate Oct4 and Sox2 DNA binding sequences.Interestingly, a developmentally regulated DNAse hypersensitivitysite has been shown to characterize Xist intron 1 in undifferenti-ated ES cells [40], and the BAF155 partner of the nucleosomeremodeling machinery SWI/SNF is found in Nanog-related proteincomplexes [41].

Fig. 3. Hypothetical holistic scenario of Nanog, Oct4 and Sox2 action at Xist duringearly embryogenesis. (A) We suggest here that Oct4 and Sox2 binding sites aredifferentially methylated (black circles) during gametogenesis, with exclusivemethylation at Xist intron 1occurring during spermatogenesis. This blocks bindingof Oct4 (in blue) and Sox2 (in green) at Xist intron 1 in the Xist-expressing malegametes but not in the Xist-repressed female gametes. (B) During the cleavage-stages of the morula, the methylation imprint of Xist intron 1 is protected againstthe active demethylation of the paternal genome, restricting binding of Oct4 andSox2 to the maternal allele. This underlies paternally-restricted Xist transcription.(C) Nanog (in red), which is not expressed in the gametes, is first expressed ininternal cells of the late morula, where it is able to bind to Xist intron 1independently of the methylation imprint. (D) In the ICM, Nanog erases the paternalimprint rendering the paternal Xist allele competent for Oct4 and Sox2 binding. Thisleads to the repression of paternal Xist and to the reactivation of the inactivepaternal X-chromosome. See text for details.

6. A holistic potential role for Nanog, Oct4 and Sox2 in thecontrol of Xist expression during pre-implantationdevelopment

The integration of the dynamics of X-inactivation with earlydevelopmental transitions can also be envisaged using an alterna-tive hypothesis, in which Oct4 and Sox2 would drive imprinted Xisttranscription during the cleavage-stages of the morula (Fig. 3). In-deed, rather than assuming that Oct4 and Sox2 are not able to re-press Xist at these stages because the paternal X-chromosomeexpresses high levels of Xist, it remains possible that, in fact, Oct4and Sox2 repressive action on Xist is restricted to the maternally-inherited Xist allele which remains silent during pre-implantationdevelopment. In this case, an as yet unknown imprinting markresponsible for paternally-restricted Xist transcription would blockbinding of Oct4 and Sox2 to the paternal, but not to the maternalXist locus, therefore providing the basis for the paternally-re-stricted transcription of Xist. Binding of Nanog to Xist intron 1 inthe internal cells of the morula would erase this mark and stimu-late binding of Oct4 and Sox2 to both Xist alleles. This would resultin the bi-allelic repression of Xist by the three pluripotent factors,and the progressive reactivation of the paternal inactive X-chromo-some in the ICM. Once the imprint is erased by Nanog, both Oct4and Sox2 would bind at Xist even in the absence of Nanog, in agree-ment with the observations reported in Nanog-null ES. Under thisscenario, Nanog function would be more related to the reprogram-ming of the epigenetic information specifying imprinted Xist tran-scription, and to rendering the paternal Xist locus competent tobind Oct4 and Sox2, than to the transcriptional repression of Xistper se. Following up on the idea that Nanog acts as a reprogrammerof the imprint responsible for paternal-restricted Xist transcription,it is interesting to note that during PGCs development Nanog-nullcells die around day 11.5 post coitum [34], when DNA demethyla-tion and erasure of imprinting marks is performed in the germ line[18,42]. An interesting working hypothesis based on the predictedfunction of Nanog in the erasure of the putative paternal imprintblocking binding of Oct4 and Sox2 to the paternal Xist intron 1,would involve the generalization of Nanog’s role to the systematicerasure of imprints that occur in the germ line.

Interestingly, Nanog is present in neither male or female maturegametes [43] whereas both Oct4 and Sox2 are present in the oocyteand epigenetically inherited in the zygote as maternal components[32,44]. Moreover, the presence of Oct4 and Sox2 in the oocyte cor-relates with a lack of Xist transcription at this stage [45]. One pos-sibility is that the maternal Xist allele is inherited with both Oct4and Sox2 already bound to Xist intron 1, providing a simple plat-form for establishing maternal Xist-silencing at the time of zygoticgenome activation. Conversely, Xist has been shown to be ex-pressed during spermatogenesis [45] although it has been shownnot to be responsible for the meiotic inactivation of the XY-bodythat characterizes male gametes [46]. In addition, it has beenshown that some Oct4 and Sox2 binding sites are specifically hy-per-methylated in male germ cells, the testes, and in isolatedsperm [47]. Strikingly, this hypermethylation of Oct4 and Sox2binding sites in germ cells was correlated with an absence ofOct4 and Sox2 binding, as evaluated by ChIP [47]. It is unknownwhether Oct4 and Sox2 binding sites at Xist intron 1 are hyper-methylated in the sperm, but if this were the case, then it wouldbe possible that this methylation underlines the expression of Xistthat occurs during male gametogenesis. Further, if the methylationof Xist intron 1 established in male gametes were protected against

1726 P. Navarro, P. Avner / FEBS Letters 583 (2009) 1721–1727

the active demethylation of the paternal genome that takes placeduring the cleavage-stages of the morula [42], as described forother imprinted genes such as H19 and Rasgrf1, then binding ofOct4 and Sox2 would be blocked on the paternal Xist allele effec-tively allowing transcription of the paternal Xist allele duringpre-implantation stages.

In summary (Fig. 2), we propose here as a testable hypothesisthat the Oct4/Sox2 binding sites at Xist intron 1 are differentiallymethylated in female and male gametes, and that this underliesthe lack of binding of Oct4 and Sox2 to the paternally-inherited Xistallele during pre-implantation development. This implies thatbinding of Nanog must somehow induce the loss of the methyla-tion imprint. It cannot, however, be excluded that Xist intron 1methylation is erased in the ICM independently of Nanog but coin-cident with Nanog expression. We expect current testing of thishypothesis to markedly clarify our understanding of Xist regulationduring early mouse embryogenesis.

7. Epigenetic resetting of Xist chromatin in the ICM: renderingall Xist alleles epigenetically indistinguishable

We propose that the binding of Nanog, Oct4 and Sox2 at Xist in-tron 1 forms the molecular basis responsible for the suppression ofXist transcription in pluripotent cells, probably through the repres-sion of the transcriptional machinery recruitment activity of theXist promoter. This could, however, be insufficient to allow theappropriate transition from imprinted to random Xist transcriptionthat occurs in the ICM. In both imprinted and random post-X-inac-tivation cells, the active Xist allele has been shown to be epigenet-ically marked by histone marks associated with euchromatin,whereas the silent Xist promoter on the active X-chromosome isnot [23,36]. These marks are believed to allow the maintenancethrough mitotic cell division of the established chromosomal-ori-gin of Xist transcription. This suggests that in the ICM the chroma-tin of the promoter region of both Xist alleles could carry marksassociated with imprinted Xist transcription, with the paternal Xistpromoter inheriting a euchromatic state that could favor the re-establishment of its activation once Nanog, Oct4 and Sox2 arerepressed, unless these were erased. Both the establishment ofrandom X-inactivation in the epiblast, the random upregulationof Xist in female iPS cells upon differentiation [15], and an apparentabsence of differential chromatin marks in female ES cells [36]indicate that such chromatin marks are likely erased duringreprogramming, prior to their random re-establishment duringdifferentiation. These observations suggest that in pluripotentcells, or at least in those of the ICM and in those obtainedin vitro, the inherited state of Xist promoter chromatin is reset torender both Xist promoters epigenetically indistinguishable andequally competent for activation at the onset of differentiation.

Although Tsix does not directly repress Xist transcription in EScells, antisense transcription across Xist has been shown to inducecomplex chromatin modifications [23,24,36,48]. In particular, Tsixhas been shown to trigger the acquisition of marks associated withgene silencing in undifferentiated ES cells, such as H3K9me3 andCpG methylation, to a CTCF-flanked region of the Xist promoter,blocking in turn any enrichment for euchromatin-associated marks[36]. In the absence of Tsix, the CTCF-flanked Xist promoter chro-matin shifts into euchromatin, but this is not accompanied by in-creased levels of transcriptional machinery recruitment until thecell differentiates [36] and Nanog, Oct4 and Sox2 are lost from Xistintron 1 [33]. Therefore, Tsix transcription, or the produced RNAitself, acts as a Xist promoter chromatin modifier responsible forthe inhibition of euchromatin-associated histone marks.

Tsix, which is highly transcribed from both the paternal andmaternal X-chromosomes in the ICM, in ES cells, and in iPS cells,

could therefore be essential to the resetting of the chromatinmarks required to support the mono-allelic recruitment of thetranscriptional machinery at either the paternal Xist promoter dur-ing pre-implantation, and at the Xist promoter of the randomlychosen inactive X-chromosome of somatic female cells submittedto in vitro genome reprogramming. This chromatin remodelingactivity of Tsix would be involved in providing at both Xist allelesthe epigenetic ground state required to establish, following lossof Nanog, Oct4 and Sox2, new allelic differences supportingmono-allelic Xist transcription.

In conclusion, we propose that the repression of Xist transcrip-tion in pluripotent cells, notably in the ICM, is dependent on therepressive action of Nanog, Oct4 and Sox2 on the one hand, ulti-mately responsible for the control of Xist transcription rate, andon Tsix-mediated chromatin modifications, which reset both Xistalleles to an identical ground state of epigenetic information byerasing inherited marks associated with imprinted (or random inthe case of genome reprogramming in vitro) Xist transcription onthe other. During differentiation, the natural loss of the pluripotentfactors that repress Xist, and the initial mono-allelic silencing ofTsix that designates the future inactive X-chromosome [25], opensa window of opportunity for both the re-establishment of aneuchromatic structure and the recruitment of the transcriptionalmachinery at the Xist promoter of the future inactive X-chromo-some. It will be important, in the future, to understand how Tsixtranscription is specifically established and maintained at high lev-els in the ICM, in ES cells, and in iPS cells.

8. Concluding remarks

We have argued in this review that the connection betweenearly developmental transitions and the regulation of X-inactiva-tion, in particular the systematic repression of Xist and the reacti-vation of the inactive X-chromosome that occurs in pluripotent cellcompartments, is directly specified by the repressive action thatNanog, Oct4 and Sox2, the triumvirate of factors supporting pluri-potency, exert on Xist transcription. This does not, however, implythat current ideas suggesting that X-inactivation reprogramming isa reflection of the pluripotent phenotype on the overall epigeneticstate of the inactive X-chromosome are totally redundant. A keyobservation in this respect is that Xist can be deleted from the inac-tive X-chromosome of female MEFs without triggering anythingother than a minor and very localized reactivation of X-linkedgenes [49]. Although it is suspected that the inactive state of thepaternal X-chromosome characterizing early embryogenesis isnot as stable as that of somatic cells, the suppression of Xist tran-scription by Nanog, Oct4 and Sox2, and the accompanying loss ofXist RNA may not on its own be sufficient to reactivate the inactiveX-chromosome, in particular during both PGC development andiPS generation. Additional particularities of pluripotent cells, possi-bly involving specific chromatin-related activities, may be involvedin facilitating the reactivation of the inactive X-chromosome alongwith the suppression of Xist transcription by Nanog, Oct4 and Sox2.

Similarly, it cannot be excluded that additional factors otherthan Nanog, Oct4 and Sox2 may play a role in Xist upregulationduring the random X-inactivation process that takes place in theearly post-implanted epiblast. The recent demonstration that fe-male pluripotent epiblast derived stem (EpiS) cells are character-ized by random X-inactivation and high Xist expression [50]concomitant with the expression of Nanog, Oct4 and Sox2, sug-gests that additional levels of complexity may be involved in theestablishment of random X-inactivation in the period prior to thelong-term silencing of Nanog, Oct4 and Sox2. This could occurthrough distinct, non-mutually exclusive strategies, such as theblocking of the binding of these factors within Xist intron 1, or

P. Navarro, P. Avner / FEBS Letters 583 (2009) 1721–1727 1727

some other inhibition of their activity. Alternatively, yet unknownactivities could render the Xist promoter insensitive to their action,or restrict the interaction of Nanog, Oct4 and Sox2 to the Xist allelethat has been elected to remain silent. It is noteworthy that humanES cells are arguably more related to mouse EpiS cells than tomouse ES cells. If the plasticity of X-inactivation reported in humanES cells [51] turns out to apply to mouse EpiSC cells, it will betempting to see this as the result of the inconsistent and capriciousrepression of Xist mediated by the pluritency factors. Additionalintersections between the road to pluripotency and the path ofX-inactivation regulation are clearly to be expected.

Acknowledgments

We thank Philip Clerc and Ian Chambers for stimulating discus-sions. P.N. and P.A. were supported by recurrent funding from theCNRS and the Institut Pasteur, and by funding from the EU Epige-nome Network of Excellence.

References

[1] Lyon, M.F. (1961) Gene action in the X-chromosome of the mouse (Musmusculus L.). Nature 190, 372–373.

[2] Lyon, M.F. (1962) Sex chromatin and gene action in the mammalian X-chromosome. Am. J. Human Genet. 14, 135–148.

[3] Plotnick, F., Klinger, H.P. and Kosseff, A.L. (1971) Sex-chromatin formation inpre-implantation rabbit embryos. Cytogenetics 10, 244–253.

[4] Gardner, R.L. and Lyon, M.F. (1971) X chromosome inactivation studied byinjection of a single cell into the mouse blastocyst. Nature 231, 385–386.

[5] Takagi, N. and Sasaki, M. (1975) Preferential inactivation of the paternallyderived X chromosome in the extraembryonic membranes of the mouse.Nature 256, 640–642.

[6] Monk, M. (1978) Biochemical studies on X-chromosome activity inpreimplantation mouse embryos. Basic Life Sci. 12, 239–246.

[7] Martin, G.R., Epstein, C.J., Travis, B., Tucker, G., Yatziv, S., Martin Jr., D.W., Clift,S. and Cohen, S. (1978) X-chromosome inactivation during differentiation offemale teratocarcinoma stem cells in vitro. Nature 271, 329–333.

[8] Rastan, S. and Robertson, E.J. (1985) X-chromosome deletions in embryo-derived (EK) cell lines associated with lack of X-chromosome inactivation. J.Embryol. Exp. Morphol. 90, 379–388.

[9] Gartler, S.M., Andina, R. and Gant, N. (1975) Ontogeny of X-chromosomeinactivation in the female germ line. Exp. Cell Res. 91, 454–457.

[10] Monk, M. and McLaren, A. (1981) X-chromosome activity in foetal germ cellsof the mouse. J. Embryol. Exp. Morphol. 63, 75–84.

[11] Takagi, N., Yoshida, M.A., Sugawara, O. and Sasaki, M. (1983) Reversal of X-inactivation in female mouse somatic cells hybridized with murineteratocarcinoma stem cells in vitro. Cell 34, 1053–1062.

[12] Tada, M., Takahama, Y., Abe, K., Nakatsuji, N. and Tada, T. (2001) Nuclearreprogramming of somatic cells by in vitro hybridization with ES cells. Curr.Biol. 11, 1553–1558.

[13] Eggan, K., Akutsu, H., Hochedlinger, K., Rideout 3rd, W., Yanagimachi, R. andJaenisch, R. (2000) X-chromosome inactivation in cloned mouse embryos.Science 290, 1578–1581.

[14] Takahashi, K. and Yamanaka, S. (2006) Induction of pluripotent stem cellsfrom mouse embryonic and adult fibroblast cultures by defined factors. Cell126, 663–676.

[15] Stadtfeld, M., Maherali, N., Breault, D.T. and Hochedlinger, K. (2008) Definingmolecular cornerstones during fibroblast to iPS cell reprogramming in mouse.Cell Stem Cell 2, 230–240.

[16] Mak, W., Nesterova, T.B., de Napoles, M., Appanah, R., Yamanaka, S., Otte, A.P.and Brockdorff, N. (2004) Reactivation of the paternal X chromosome in earlymouse embryos. Science 303, 666–669.

[17] Okamoto, I., Otte, A.P., Allis, C.D., Reinberg, D. and Heard, E. (2004) Epigeneticdynamics of imprinted X inactivation during early mouse development.Science 303, 644–649.

[18] Surani, M.A., Hayashi, K. and Hajkova, P. (2007) Genetic and epigeneticregulators of pluripotency. Cell 128, 747–762.

[19] Clerc, P. and Avner, P. (2003) Multiple elements within the Xic regulaterandom X inactivation in mice. Semin. Cell Dev. Biol. 14, 85–92.

[20] Ng, K., Pullirsch, D., Leeb, M. and Wutz, A. (2007) Xist and the order ofsilencing. EMBO Rep. 8, 34–39.

[21] de Napoles, M., Nesterova, T. and Brockdorff, N. (2007) Early loss of Xist RNAexpression and inactive X chromosome associated chromatin modification indeveloping primordial germ cells. PLoS ONE 2, e860.

[22] Sugimoto, M. and Abe, K. (2007) X chromosome reactivation initiates innascent primordial germ cells in mice. PLoS Genet. 3, e116.

[23] Navarro, P., Pichard, S., Ciaudo, C., Avner, P. and Rougeulle, C. (2005) Tsixtranscription across the Xist gene alters chromatin conformation without

affecting Xist transcription: implications for X-chromosome inactivation.Genes Dev. 19, 1474–1484.

[24] Sun, B.K., Deaton, A.M. and Lee, J.T. (2006) A transient heterochromatic state inXist preempts X inactivation choice without RNA stabilization. Mol. Cell 21,617–628.

[25] Lee, J.T., Davidow, L.S. and Warshawsky, D. (1999) Tsix, a gene antisense toXist at the X-inactivation centre. Nat. Genet. 21, 400–404.

[26] Lee, J.T. and Lu, N. (1999) Targeted mutagenesis of Tsix leads to nonrandom Xinactivation. Cell 99, 47–57.

[27] Lee, J.T. (2000) Disruption of imprinted X inactivation by parent-of-origineffects at Tsix. Cell 103, 17–27.

[28] Kalantry, S., Mills, K.C., Yee, D., Otte, A.P., Panning, B. and Magnuson, T. (2006)The Polycomb group protein Eed protects the inactive X-chromosome fromdifferentiation-induced reactivation. Nat. Cell Biol. 8, 195–202.

[29] Chambers, I., Colby, D., Robertson, M., Nichols, J., Lee, S., Tweedie, S. and Smith,A. (2003) Functional expression cloning of Nanog, a pluripotency sustainingfactor in embryonic stem cells. Cell 113, 643–655.

[30] Mitsui, K., Tokuzawa, Y., Itoh, H., Segawa, K., Murakami, M., Takahashi, K.,Maruyama, M., Maeda, M. and Yamanaka, S. (2003) The homeoprotein Nanogis required for maintenance of pluripotency in mouse epiblast and ES cells.Cell 113, 631–642.

[31] Nichols, J., Zevnik, B., Anastassiadis, K., Niwa, H., Klewe-Nebenius, D.,Chambers, I., Scholer, H. and Smith, A. (1998) Formation of pluripotent stemcells in the mammalian embryo depends on the POU transcription factor Oct4.Cell 95, 379–391.

[32] Avilion, A.A., Nicolis, S.K., Pevny, L.H., Perez, L., Vivian, N. and Lovell-Badge, R.(2003) Multipotent cell lineages in early mouse development depend on SOX2function. Genes Dev. 17, 126–140.

[33] Navarro, P., Chambers, I., Karwacki-Neisius, V., Chureau, C., Morey, C.,Rougeulle, C. and Avner, P. (2008) Molecular coupling of Xist regulation andpluripotency. Science 321, 1693–1695.

[34] Chambers, I., Silva, J., Colby, D., Nichols, J., Nijmeijer, B., Robertson, M., Vrana,J., Jones, K., Grotewold, L. and Smith, A. (2007) Nanog safeguards pluripotencyand mediates germline development. Nature 450, 1230–1234.

[35] Niwa, H., Miyazaki, J. and Smith, A.G. (2000) Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat.Genet. 24, 372–376.

[36] Navarro, P., Page, D.R., Avner, P. and Rougeulle, C. (2006) Tsix-mediatedepigenetic switch of a CTCF-flanked region of the Xist promoter determinesthe Xist transcription program. Genes Dev. 20, 2787–2792.

[37] Sado, T., Hoki, Y. and Sasaki, H. (2006) Tsix defective in splicing is competentto establish Xist silencing. Development 133, 4925–4931.

[38] Chen, X., Xu, H., Yuan, P., Fang, F., Huss, M., Vega, V.B., Wong, E., Orlov, Y.L.,Zhang, W., Jiang, J., Loh, Y.H., Yeo, H.C., Yeo, Z.X., Narang, V., Govindarajan, K.R.,Leong, B., Shahab, A., Ruan, Y., Bourque, G., Sung, W.K., Clarke, N.D., Wei, C.L.and Ng, H.H. (2008) Integration of external signaling pathways with the coretranscriptional network in embryonic stem cells. Cell 133, 1106–1117.

[39] Silva, J., Chambers, I., Pollard, S. and Smith, A. (2006) Nanog promotes transferof pluripotency after cell fusion. Nature 441, 997–1001.

[40] Tsai, C.L., Rowntree, R.K., Cohen, D.E. and Lee, J.T. (2008) Higher orderchromatin structure at the X-inactivation center via looping DNA. Dev. Biol.319, 416–425.

[41] Wang, J., Rao, S., Chu, J., Shen, X., Levasseur, D.N., Theunissen, T.W. and Orkin,S.H. (2006) A protein interaction network for pluripotency of embryonic stemcells. Nature 444, 364–368.

[42] Reik, W., Dean, W. and Walter, J. (2001) Epigenetic reprogramming inmammalian development. Science 293, 1089–1093.

[43] Yamaguchi, S., Kimura, H., Tada, M., Nakatsuji, N. and Tada, T. (2005) Nanogexpression in mouse germ cell development. Gene Exp. Patterns 5, 639–646.

[44] Pesce, M., Wang, X., Wolgemuth, D.J. and Scholer, H. (1998) Differentialexpression of the Oct-4 transcription factor during mouse germ celldifferentiation. Mech. Dev. 71, 89–98.

[45] McCarrey, J.R. and Dilworth, D.D. (1992) Expression of Xist in mouse germcells correlates with X-chromosome inactivation. Nat. Genet. 2, 200–203.

[46] Marahrens, Y., Panning, B., Dausman, J., Strauss, W. and Jaenisch, R. (1997)Xist-deficient mice are defective in dosage compensation but notspermatogenesis. Genes Dev. 11, 156–166.

[47] Imamura, M., Miura, K., Iwabuchi, K., Ichisaka, T., Nakagawa, M., Lee, J.,Kanatsu-Shinohara, M., Shinohara, T. and Yamanaka, S. (2006) Transcriptionalrepression and DNA hypermethylation of a small set of ES cell marker genes inmale germline stem cells. BMC Dev. Biol. 6, 34.

[48] Sado, T., Hoki, Y. and Sasaki, H. (2005) Tsix silences Xist through modificationof chromatin structure. Dev. Cell 9, 159–165.

[49] Csankovszki, G., Panning, B., Bates, B., Pehrson, J.R. and Jaenisch, R. (1999)Conditional deletion of Xist disrupts histone macroH2A localization but notmaintenance of X inactivation. Nat. Genet. 22, 323–324.

[50] Guo, G., Yang, J., Nichols, J., Hall, J.S., Eyres, I., Mansfield, W. and Smith, A.(2009) Klf4 reverts developmentally programmed restriction of ground statepluripotency. Development 136, 1063–1069.

[51] Silva, S.S., Rowntree, R.K., Mekhoubad, S. and Lee, J.T. (2008) X-chromosomeinactivation and epigenetic fluidity in human embryonic stem cells. Proc. Natl.Acad. Sci. USA 105, 4820–4825.