X Inactivation: Pre- or Post-Fertilisation Turn-off?
Anne C. Ferguson-Smith
Dosage compensation of X-linked genes in femalemammals occurs by inactivating one of the two Xchromosomes. Two recent studies have asked whenX inactivation first occurs, with different answers thatleave open the question: does X inactivation occurbefore or after fertilisation?
In mammals, X chromosome expression is balanced infemales and males by shutting off one of the two Xchromosomes in each cell of a female . This processof X inactivation involves epigenetic regulation,characterised by repressive modifications to DNA andchromatin, late replication of the inactive X and theexpression of an untranslated RNA, Xist, which coatsone of the X chromosomes in cis and triggers silencing[2,3]. In 1975, Takagi and Sasaki  first showed that Xinactivation in extra-embryonic tissues of a femalemouse is imprinted the paternally inherited X chro-mosome is exclusively inactivated in the cell lineagesgiving rise to the placenta. In contrast, X-inactivation inembryonic derivatives is random, with either the pater-nal or the maternally inherited X becoming heritablysilenced. Two groups [5,6] have now addressed thequestion of when inactivation of the paternal X actuallyoccurs in the mouse. The results provide insight intothe dynamic process of X chromosome inactivationand suggest that it is governed by imprinting in all cellsduring the earliest embryonic stages. Thereafter repro-gramming occurs in the embryonic derivatives and Xinactivation becomes random.
From their results, Huynh and Lee  claim that oneX chromosome is already silent at the two-cell stagewhen zygotic gene activation occurs, suggesting thatthe X chromosomes are dosage compensated fromconception. Okamoto et al. , on the other hand, claimthat the paternally inherited X is initially active after fer-tilisation, becoming inactivated from early cleavagestages commencing around the four-cell stage ofdevelopment . On the face of it, these studies appearcontradictory. A closer look, however, shows much ofthe data reported in both papers are consistent, noveland important, and this should not be overlooked.
Textbooks state that in female mammals, such asmice or humans, one of the X chromosomes isinactivated after the blastocyst has implanted in theuterine wall . This is consistent with the firstcytological evidence of an inactive X being visiblearound the time of implantation . Classic studiesusing X-linked markers have identified products fromboth the paternally and maternally inherited X
chromosomes at earlier stages . These findings havecontributed to the prevailing view held for over 30years, that X inactivation does not occur until aroundor after implantation.
Nonetheless, transcriptional silencing of the XY biva-lent during male meiosis is a well-establishedphenomenon, raising the possibility that a still silentpaternal X can be inherited by the zygote uponfertilisation . Interestingly, although the inactivatingtranscript Xist is present in germ cells, in Xist mutants,meiotic sex chromosome inactivation still occurs .This suggests that the paternal X is inactivated by aXist-independent mechanism in male meiotic germcells. After fertilisation, Xist expression commences atthe 24 cell stage and is exclusively expressed fromthe paternally inherited X chromosome; Xist RNA coatsthis chromosome and is required for its inactivation .
The ideal experiment for determining if and when, ina female embryo, genes on the maternally (XM) andpaternally (XP) inherited X chromosomes are transcribedwould be to assess expression in an embryo directlyusing the reverse transcriptase version of the poly-merase chain reaction (RT-PCR) from the time of con-ception through early cleavage stages to implantation.This is technically very challenging; not only are tran-scripts that are inherited from the egg present at theearlier stages, but also each embryo needs to be sexedand the expression of several genes needs to be mea-sured in an embryo perhaps with as few as two cells.Furthermore, hybrid embryos between two differentstrains of mice harbouring identified polymorphismsthat distinguish the transcripts from the two X chromo-somes must be used.
Huynh and Lee  did this to some extent and wereable to measure X-linked gene expression in embryosas early as the 816 cell morula stage but not earlier.They analysed the expression of a dozen genesdistributed along the X chromosome both proximal anddistal to the X inactivation centre (the specific regionof the X shown to control most of the steps of X inacti-vation). The results showed that, in female morulae,expression occurred preferentially from XM. Thisindicates that dosage compensation occurs earlier thanpreviously thought, at least as early as the 816 cellstage. Interestingly, the authors found that the closerthe gene to the X inactivation centre, the greater theXM:XP expression ratio. To me, these data suggest thatX inactivation is a progressive process along the XP the authors have beautifully achieved a snap-shot of aprocess that is not yet finished.
Taking a different approach, Okamoto et al.  alsofound that X inactivation occurs earlier than had beenpreviously indicated. Using established chromatin-spe-cific markers for an inactivating X chromosome hypermethylation or hypomethylation and hypoacetyla-tion of specific lysine residues of core histones, and thepresence or absence of polycomb group proteins Eedand Enx they were able to detect an inactive XP in
Current Biology, Vol. 14, R323R325, April 20, 2004, 2004 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cub.2004.03.056
Department of Anatomy, University of Cambridge, DowningStreet, Cambridge CB2 3DY, UK. E-mail: firstname.lastname@example.org
816 cell morulae. By the 32-cell stage, 90% of allfemale cells analysed had an inactive XP. These twostudies show that X inactivation is evident as early asafter just four division cycles from the fertilized egg.
But neither of these approaches can tell us when theXP chromosome is first inactivated, and to address this,less direct approaches were taken. Okamoto et al. used the presence or absence of elongating RNA poly-merase II as a marker of transcriptional activity/inactiv-ity from the Xist-positive nuclear domain. Previously, inES cells, exclusion of polymerase II was found to beone of the earliest X inactivation events. The authorsfound that the 4-cell stage female embryo exhibits theearliest evidence of RNA polymerase II exclusion,though not all blastomeres showed exclusion. Theyconcluded that XP inactivation begins at the 48 cellstage. This conclusion was confirmed using RNAfluorescence in situ hybridisation (FISH) to detectnascent transcripts of the X-linked gene Chic1. At the2-cell stage, when there was no exclusion of RNA
polymerase II, Chic1 expression was detected fromboth X chromosomes.
Taking an alternative approach, Huynh and Lee used Cot-1 DNA, a probe for repetitive sequences, todetect nascent RNA transcripts. Areas of transcriptionalinactivity, including the Xist-marked inactive X chromo-some, were evident as Cot-1 holes. From the 2-cellstage onwards, 83100% of cells had a Cot-1 hole co-incident with Xist positivity. This suggests that one Xchromosome is transcription-poor at this stage.Because this is the time of onset of zygotic genetranscription, the authors conclude that their dataindicate the XX preimplantation mouse embryo is largelydosage compensated from the time of conception.
Okamoto et al.  found that an inactive XP ismaintained in the embryonic cells of isolated inner cellmasses at the early blastocyst stage. By the lateblastocyst stage, however, XP is reactivated in theinner cell masses, allowing chromatin remodelling andrandom X inactivation to proceed in the embryonicderivatives. Similar in vivo results were independentlyobtained by Mak et al. . From their results obtainedwith embryonic and trophoblast stem cells, Huynh andLee  drew the same conclusion. From a develop-mental perspective, the key issue here is that XP isinactive at the time when the trophectoderm and innercell mass lineages are being founded, believed tooccur at the 816 cell stage . This results inextraembryonic tissues having non-random imprintedXP inactivation; later, after the trophectoderm lineageis determined, pluripotent embryonic cells within theinner cell mass are reprogrammed to have random Xinactivation (Figure 1).
The sole inconsistency between the two studies [5,6]thus resides in whether XP enters the egg upon fertili-sation in an inactive state, as Huynh and Lee  claim,or is reactivated and then progressively inactivated,starting around the 4-cell stage, as claimed byOkamoto et al.  (Figure 1). Knowing when XP inacti-vation occurs is important, because it has implicationsfor the mechanism of imprinted X chromosome silenc-ing and the extent of pre-implantation genome repro-gramming issues relevant to our understanding ofthe epigenetic control of genome function.
From these two new studies [5,6] two likelymechanisms emerge. The first involves meiotic XP
silencing in the paternal germline which is carried overinto the zygote a mechanism likely to be indepen-dent of Xist. The alternative mechanism, proposed byOkamoto et al. , involves Xist-dependent inactivationof XP commencing at the 24 cell stage in response toa germline-derived imprint. It will be interesting toassess the pre-inactivation model in female zygotesfathered by Xist deficient males. These conceptuses dieearly in embryogenesis because they fail to undergo Xinactivation in the extraembryonic lineages ; earlycleavage stage embryos of this genotype would still,however, be informative for assessing whether the pre-inactivation hypothesis involves Xist; perhaps usingRNA-FISH for X-linked genes. The gradient of expres-sion along XP observed by Huynh and Lee  suggestsa role for Xist in the pre-inactivation model. So althoughtechnically challenging, it would be informative to
Figure 1. Two models of imprinted X inactivation.
(A) In the pre-inactivation model of Hyunh and Lee , the XP
retains most of its inactivity throughout preimplantation stages.(B) In the zygotic inactivation model of Okamoto et al. , theXP is active at the 2-cell stage, commencing re-inactivationaround the 4-cell stage and progressively acquiring full inacti-vation status. In both models, the XP is fully inactivated by theblastocyst stage, remaining so in trophectoderm lineages. Theepiblast cells of the inner cell mass (ICM) become repro-grammed to have random X inactivation. Black ovals representan inactivated X and grey ovals, a mostly inactivated X. Theactive XM is represented by a pink oval, and the active XP incells of the ICM is represented by a pale blue oval.
Sperm Egg xSperm Egg
Both Xs active
compare the ratio of XM:XP transcripts at the 48 cellstage with the published 816 cell stage data, and todetermine whether a gradient of silencing from the Xinactivation centre is also seen at the earlier stage. Thenext instalment in this story is eagerly awaited.
References1. Lyon, M. (1961). Gene action in the X chromosome of the mouse
(Mus musculus). Nature 190, 372-373.2. Avner, P., and Heard, E. (2001). X chromosome inactivation; count-
ing, choice and initiation. Nature Rev. Genet. 2, 56-97.3. Penny, G., Kay, G., Sheardown, S., Rastan, S., and Brockdorff, N.
(2002). Requirement for Xist in X chromosome inactivation. Nature379, 131-137.
4. Takagi, N., and Sasaki, M. (1975). Preferential inactivation of thepaternally derived X chromosome in the extraembryonic mem-branes of the mouse. Nature 256, 640-642.
5. Huynh, K., and Lee, J.T. (2003). Inheritance of a pre-inactivatedpaternal X chromosome in early mouse embryos. Nature 426, 857-882.
6. Okamoto, I., Otte, A., Allis, D., Reinberg, D., and Heard, E. (2004).Epigenetic dynamics of imprinted X inactivation during early mousedevelopment. Science 303, 644-649.
7. Wolpert, L. (1998). In Principles of Development pp 379-380 (OxfordUniversity Press).
8. Sugowara, O., Takagi, N., and Sasaki, M. (1985). Correlationbetween X chromosome inactivation and cell differentiation infemale preimplantation mouse embryos. Cytogenet. Cell Genet. 39,210-219.
9. Kratzer, P., and Gartler, S. (1978). HGPRT activity changes in preim-plantation mouse embryos. Nature 274, 503-504.
10. Solari, A. (1974). The behaviour of the XY pair in mammals. Rev.Cytol. 38, 273-317.
11. Turner, J., Mahadevaiah, S., Elliot, D., Garchon, H.-J., Pehrson, J.,Jaenisch, R., and Burgoyne, P. (2002). Meiotic sex chromosomeinactivation in male mice with targeted disruptions of Xist. J. CellSci. 115, 4097-4105.
12. Mak, W., Nesterova, T., Napoles, M., Appanah, R., Yamanaka, S.,Otte, A., and Brockdorff, N. (2004). Reactivation of the paternal Xchromosome in early mouse embryos. Science 203, 666-669.
13. Johnson, M., and Ziomek, C. (1981). The foundation of two distinctlineages within the mouse morula. Cell 24, 71-80.
14. Marahrens, Y., Panning, B., Dausman, J., Strauss, W., and Jaenisch,R. (1997). Xist-deficient mice are defective in dosage compensationbut not spermatogenesis. Genes Dev. 11. 156-166.