QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Heard, E. Articles by Disteche, C. M. Search for Related Content PubMed PubMed Citation Articles by Heard, E. Articles by Disteche, C. M. Related Content Chromosome Dynamics and Nuclear Architecture Chromatin and Gene Expression Social Bookmarking                   What's this?

REVIEW

Dosage compensation in mammals: fine-tuning the expression of the X chromosome

¹ CNRS UMR218, Curie Institute, 75248 Paris, Cedex 05, France; ² Department of Pathology and Department of Medicine, University of Washington, Seattle, Washington 98195, USA

	Abstract

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

 
Mammalian females have two X chromosomes and males have only one. This has led to the evolution of special mechanisms of dosage compensation. The inactivation of one X chromosome in females equalizes gene expression between the sexes. This process of X-chromosome inactivation (XCI) is a remarkable example of long-range, monoallelic gene silencing and facultative heterochromatin formation, and the questions surrounding it have fascinated biologists for decades. How does the inactivation of more than a thousand genes on one X chromosome take place while the other X chromosome, present in the same nucleus, remains genetically active? What are the underlying mechanisms that trigger the initial differential treatment of the two X chromosomes? How is this differential treatment maintained once it has been established, and how are some genes able to escape the process? Does the mechanism of X inactivation vary between species and even between lineages? In this review, X inactivation is considered in evolutionary terms, and we discuss recent insights into the epigenetic changes and developmental timing of this process. We also review the discovery and possible implications of a second form of dosage compensation in mammals that deals with the unique, potentially haploinsufficient, status of the X chromosome with respect to autosomal gene expression.

[Keywords: X inactivation; dosage compensation; epigenetics; monoallelic regulation; imprinting]

	Genetic content and evolution of the mammalian X chromosome

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

 
The content, regulation, and evolution of the mammalian X chromosome are intimately related to the evolution of the Y chromosome, its partner. The sex chromosomes differ significantly in their gene content: The human X contains 1100 genes, whereas the Y contains 100 genes. This striking divergence results from evolutionary forces that progressively altered an ancestral homologous pair of autosomes or proto-sex chromosomes (Fig. 1A; Ohno 1967). The Y chromosome accumulated male-advantageous genes around the testis-determinant gene SRY and lost many genes by suppression of recombination between the X and the Y in the heterogametic male sex to avoid the production of abnormal sexual phenotypes (Charlesworth 1996; Rice 1996; Charlesworth et al. 2005).

View larger version (22K): [in this window] [in a new window]   Figure 1. Evolutionary pathways of the sex chromosomes. (A) The sex chromosomes derived from a homomorphic pair of chromosomes (proto-sex chromosomes). Once sex was determined by SRY (testis determinant gene) on the Y chromosome, recombination was suppressed between the sex chromosomes. In males, the Y diverged from the X by gene loss and accumulation of male-advantageous genes (blue) around SRY (green), and the X became up-regulated (dark orange). In females, the active X (Xa) became up-regulated (dark orange), and the inactive X (Xi) became subject to X inactivation (black). Some regions of present-day sex chromosomes remain homologous between the X and Y (light orange). (B) Translocation of autosomal material to the sex chromosomes. Translocation of autosomal material (A) to the pseudoautosomal region of the sex chromosomes (PAR) was followed by loss or differentiation (blue dots) of the added region on the Y, and progressive up-regulation (dark-orange dots) and inactivation (black dots) of the added region on the X. (C) Dosage of sex-linked genes. (Top row) Dosage compensation between the autosomes and the sex chromosomes and between males and females is achieved for most genes that have lost their Y paralog by a combination of up-regulation of the active X in males and females (Xa, dark orange) and X inactivation in females (Xi, black). When the X/Y gene pair persists, equal dosage results from expression from the Y-linked gene and escape from X inactivation of the X-linked gene on the Xi (light orange). (Second row) The allele on the Xa may (not shown) or may not be up-regulated already (light orange). (Third row) Unequal expression between the sexes will arise when the Y paralog is lost or differentiated in a male-specific gene but the X-linked gene still escapes X inactivation on the Xi. Note that expression from the up-regulated gene on the Xa may be higher than that on the Xi. For such genes, higher expression in females may play a role in ovarian or other female-specific functions or may not cause any phenotypic differences. (Bottom row) In rare cases of autosome-to-Y translocations, the male would have selective expression of the newly acquired Y-linked gene.

The X chromosome is enriched in genes related to sexual reproduction and brain function, as well as cancer-testis antigen genes (e.g., melanoma-associated antigen genes, MAGE), which encode proteins that are immunogenic in cancer patients, making them potentially useful for immunotherapy (Zechner et al. 2001; Khil et al. 2004; Vallender and Lahn 2004; Ross et al. 2005; Simpson et al. 2005). Genes that enhance male sexual reproduction are thought to have accumulated on the X because recessive mutations expressed in males due to hemizygosity of the X could give rise to novel functions. The X chromosome is enriched for genes expressed in spermatogonia (Wang et al. 2001), but not for genes expressed in later stages of spermatogenesis, likely because of silencing at meiosis (MSCI) (see below) (Khil et al. 2004). Cancer-testis antigen genes are common on the X based on the recent annotation of its entire sequence (Ross et al. 2005). These genes, which are predominantly expressed in normal tests and cancer, will probably fall in the category of genes advantageous to male sexual reproduction. An added peculiarity of the X chromosome is that it is unusually active in retrotransposition (Emerson et al. 2004). Genes that have been retrotransposed from the X to autosomes often retain their function and may serve as a protection from MSCI in males or may have been selected for their meiosis-specific function (Khil et al. 2005). Genes involved in female sexual reproduction are also enriched on the X (Khil et al. 2004); some of these genes could have a dosage-dependent function in females by escaping from X inactivation (see below).

What is the cause of enrichment for brain-expressed genes on the X? This question has led to speculation about mechanisms for selection of genes that confer enhanced cognitive functions (Zechner et al. 2001). Such genes may provide a selective advantage to males in sexual reproduction. Our own studies have shown a higher expression of X-linked genes in brain tissues compared with others (Nguyen and Disteche 2006). This has implication for human diseases, especially X-linked forms of mental retardation, which are common and for which causative mutations have started to be identified (Ropers and Hamel 2005). Imprinting of X-linked genes expressed in brain may explain differences in behavior between XO mice that inherited their single X from their mother or father (Davies et al. 2005; Raefski and O’Neill 2005). In turn, these findings could potentially explain differences in mental function between patients with a single maternal or paternal X and Turner syndrome (Skuse et al. 1997).

In summary, the gene content of the X chromosome reflects its role in sexual reproduction due to divergence between the X and the Y. These evolutionary processes not only influenced which functions would be specifically performed by the X, but also resulted in unique regulatory mechanisms, including X inactivation and X up-regulation, to overcome the presence of a single X in males.

	Initiation of X inactivation: counting, choice, and cis inactivation

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

 
We next discuss mechanistic aspects of X inactivation. The initial differential treatment of the two X chromosomes during early mammalian development is controlled by the Xic, which produces the noncoding Xist transcript responsible for triggering silencing in cis. Classic cytogenetic studies of deleted or rearranged X chromosomes defined this 1-Mb region of the X chromosome to be critical for X inactivation (for review, see Avner and Heard 2001). Only chromosomes carrying the Xic are able to induce XCI, although the Xic sequence requirements for the imprinted and random forms of XCI may differ (Okamoto et al. 2005). Mechanistic insights into the role of the Xic have come from studies of early mouse embryos and embryonic stem (ES) cells, the latter representing a useful tissue culture system, in which differentiation is accompanied by random XCI. Studies in ES cells have shown that, for X inactivation to occur, cells must have at least two Xics (Rastan 1983; Rastan and Robertson 1985). Furthermore, this locus is at the heart of the process that senses, or "counts," the number of X chromosomes, and ensures that only a single X will remain active per diploid autosome set, all extra copies being inactivated. Autosomal ploidy is thought to be important for counting (Jacobs and Migeon 1989), and the most popular model invokes the existence of an autosomal factor produced in limiting quantity, that is sufficient to block one Xic per diploid cell (for review, see Alexander and Panning 2005). The Xic is also involved in the choice of which X chromosome will remain active/be inactivated. The signal produced by the Xic, that triggers cis inactivation of the X chromosome, or even of an autosome, in X–autosome translocations, appears to be the noncoding Xist transcript (Brown et al. 1991). Xist is expressed only from the inactive X chromosome, producing a 19-kb-long, untranslated RNA that coats the X chromosome from which it is produced in cis (Brown et al. 1992; Clemson et al. 1998). Deletions and transgenes demonstrated that Xist is essential for both imprinted and random X inactivation in mice (Penny et al. 1996; Marahrens et al. 1997; Wutz and Jaenisch 2000).

However, Xist alone does not account for the multiple functions of the Xic. For example, Xist does not appear to be involved in counting, as deletion of one Xist allele does not prevent the cell from registering the presence of >1 Xic and triggering XCI via the wild-type Xist allele (Penny et al. 1996). Multiple elements 3' to Xist (Fig. 2) appear to be involved in counting and choice functions. Deletions have defined a 37-kb bipartite region, the absence of which results in aberrant inactivation of the single X chromosome in differentiating XO or XY ES cells (Clerc and Avner 1998; Morey et al. 2001, 2004). Furthermore, certain sequences from within this region, when used as transgenes in XX ES cells, can interfere with the normal counting process and block XCI (Lee 2005). Monoallelic regulation of Xist, at least in mice, involves a complex combination of antisense transcription to Xist in the form of Tsix and Xite (Stavropoulos et al. 2001, 2005; Ogawa and Lee 2003) as well as cis-regulatory sequences located in the 3' region of Xist (for review, see Clerc and Avner 2003). In undifferentiated ES cells, Tsix is expressed, along with low-level Xist transcription, from the active X chromosome(s) (Debrand et al. 1999; Lee et al. 1999). Upon differentiation of XX cells, the disappearance of Tsix is accompanied by the accumulation of Xist RNA in cis. The basis for this sudden reciprocal behavior of Xist and Tsix remains unknown. However, two recent studies have revealed that the Xics transiently colocalize, via the Tsix region, during the onset of X inactivation, at the time when counting and choice are thought to occur (Bacher et al. 2006; Xu et al. 2006). This apparent "cross-talk" between the Xics may be required for an exchange of information between Xist/Tsix homologs that ultimately leads to monoallelic down-regulation of Tsix and up-regulation of Xist on one X chromosome and not the other. The Tsix expression pattern, and the fact that targeted deletions/insertions that abolish Tsix transcription result in Xist RNA accumulation demonstrate that Tsix has a repressive effect on Xist during the initiation of X inactivation (Lee et al. 1999; Luikenhuis et al. 2001; Morey et al. 2001; Sado et al. 2001). However, it is still unclear whether it is the Tsix transcript, or the act of transcription, or both, that are involved (Nesterova et al. 2003; Shibata and Lee 2004). Recent studies point to Tsix transcription having a role in modifying the chromatin of the Xist/Tsix locus in ES cells, by participating in the formation of a domain highly enriched in H3K4 methylation, a histone modification associated with open chromatin (Navarro et al. 2005). Tsix has also been shown to participate in generating the silent chromatin status at the Xist promoter on the X chromosome that remains active once XCI has occurred (Sado et al. 2005). In a more recent study, Tsix has been proposed to prevent the recruitment of H3K27 methylation at the Xist locus, as its absence (in a Tsix mutant) results in the appearance of this mark across the Xist locus just prior to Xist up-regulation (Sun et al. 2006). Other loci that affect choice include the X-chromosome controlling element (Xce), which leads to skewed patterns of XCI (for review, see Avner et al. 1998). This has been genetically mapped 3' to Xist, although its exact location and molecular nature remain to be found (Simmler et al. 1993; Chadwick et al. 2006).

View larger version (16K): [in this window] [in a new window]   Figure 2. Features of the Xic. A map of the mouse Xic region is shown, with Xist and its antisense unit, Tsix (shown in color, with other genes not known so far to be involved in X inactivation shown as clear boxes). Above the map the elements involved in counting and choice, including Tsix and Xite, as defined by deletion, are shown. Regions of unusual chromatin enrichment are also indicated (see text for details). Transgenes used to test for Xic function are shown below (in blue). The full extent of the region capable of ensuring autonomous Xic function remains to be defined, as transgenes of up to 460 kb in length are unable to induce counting, choice, and cis inactivation when present as a single copy on an autosome (Heard et al. 1999).

The nature of the protein factors that mediate Xic counting and choice functions have remained elusive. A mutagenesis screen for alleles that affect randomness of XCI has uncovered several possible candidates (Percec et al. 2002). Two studies have pointed to CTCF as being important in the choice function of the Xic. One study reported that CTCF binds the region close to the 5' end of Tsix, and could therefore be important in regulating monoallelic Tsix transcription (Chao et al. 2002). Another study has reported that CTCF binds the promoter region of the human XIST locus (Pugacheva et al. 2005). Thus mutations in CTCF-binding sites could well be implicated in skewing of X inactivation. Furthermore, CTCF may be involved in transient Xic cross-talk during initia- tion of random X inactivation, as it has been shown to be important for nuclear trans-interactions between other loci (Ling et al. 2006). Clearly CTCF mutants will be helpful in deciphering its exact role in the initiation of XCI.

	The epigenetics of the inactive X chromosome

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

 
Once the decision to trigger X inactivation has been taken, what is the basis of the differential transcriptional regulation of the two X chromosomes? The inactive X chromosome can be distinguished from its active counterpart in several ways, including Xist RNA coating, chromatin changes such as histone modifications and DNA methylation, as well as asynchronous replication timing. The kinetics of these changes during imprinted and random XCI has been studied in mouse embryos and differentiating ES cells, respectively. Xist RNA accumulation over the X chromosome to be inactivated is the earliest known event in XCI. Gene silencing across the chromosome rapidly ensues, within one or two cell cycles (Kay et al. 1993; Panning et al. 1997; Sheardown et al. 1997; Wutz and Jaenisch 2000; Okamoto et al. 2004). Within Xist, a highly conserved repetitive region (the A-repeats) has been defined as being critical for the silencing function of this RNA (Wutz et al. 2002), but so far its binding partners and its mechanism of action remain a mystery (for review, see Wutz 2002). Using inducible Xist cDNA transgenes, Wutz and Jaenisch (2000) have also shown that Xist RNA-induced silencing can only occur during early ES cell differentiation. This implies either that a developmentally regulated factor must exist to mediate Xist’s action, or else that chromatin is not competent for XCI at later stages of development. This study also showed that during the initial phases of ES cell differentiation, X inactivation can be reversed by switching off the Xist gene, but subsequently the repressed state becomes locked in and is no longer dependent on Xist (Fig. 3). What causes this irreversibility? The earliest chromatin changes observed are the loss of histone modifications associated with active chromatin, such as H3K9 acetylation and H3K4 methylation (Heard et al. 2001; Goto et al. 2002; Okamoto et al. 2004), although whether this is a cause or consequence of the silencing induced by Xist remains to be defined. Subsequently, the X chromosome becomes hypoacetylated for histone H4 (Keohane et al. 1996; Heard et al. 2001; Chaumeil et al. 2002) enriched in H3 Lys-27 trimethylation (Plath et al. 2003; Silva et al. 2003; Okamoto et al. 2004), H3 Lys-9 dimethylation (Heard et al. 2001; Okamoto et al. 2004), and H4 Lys-20 monomethylation (Kohlmaier et al. 2004), as well as H2A K119 monoubiquitylation (de Napoles et al. 2004; Fang et al. 2004; Smith et al. 2004; Hernandez-Munoz et al. 2005). The successive appearance of these histone modifications on the X following Xist RNA coating may underlie the progressive stability and the heritability of the inactive state (Kohlmaier et al. 2004).

View larger version (30K): [in this window] [in a new window]   Figure 3. Epigenetic marks associated with X inactivation. (A) The current status of histone modifications known to be globally associated with the inactive X chromosome (Xi) or its active counterpart (Xa) are shown on amino acid maps of the N (and C) termini of the core histones indicated. (B) A schematic representation of the timing of appearance of epigenetic marks that accompany X inactivation during female ES cell differentiation is shown (see text for details). First, Xist RNA coats the chromosome from which it is transcribed in cis and induces silencing, through unknown mechanisms. Second, histone marks associated with transcriptional activity, such as acetylation or dimethylation of H3K4, are lost either actively or passively. Third, the PRC2 complex is recruited, and H3K27me3 appears in the Xi. A shift to late replication timing appears to follow these early events. MacroH2A becomes enriched on the inactive X from around day 4 onward. Finally, DNA methylation is recruited to the promoters of X-linked genes at later stages.

In the case of the dimethylation of H3 Lys-9 of the Xi, the HMTase responsible has not yet been identified. The knockout of one candidate, the G9a HMTase, does not appear to disrupt X inactivation, although the H3K9me2 mark on the Xi was not actually examined in these mutants (Ohhata et al. 2004). Ezh2 has been reported to have some H3K9 methylation activity; however, the appearance of H3K9me2 on the Xi shows different kinetics from H3K27me3 when examined by immunofluorescence (Okamoto et al. 2004; Rougeulle et al. 2004). Furthermore, the distributions of these two marks on the Xi, as examined by chromatin immunoprecipitation on mouse embryonic fibroblasts, are overlapping but distinct. H3K27me3 is enriched both at the promoter and within the body of genes on the Xi, but not on the Xa, whereas H3K9me2 is enriched only at promoters on the Xi, and is present within the body of X-linked genes both on the active and inactive X chromosomes (Rougeulle et al. 2004). Thus if Ezh2 is the HMTase of H3K9me2, it is likely to involve different partners for this activity compared with H3K27me3. Interestingly, it has been noted that in some human cell lines, trimethylation of H3K9, a modification associated with HP1 binding, particularly at constitutive heterochromatin, characterizes domains of the Xi that are negative for H3K27me3 and XIST RNA (Chadwick and Willard 2004). Thus different states of H3K9 methylation may characterize different chromatin regions on the Xi, illustrating the fact that this large block of facultative heterochromatin is by no means a uniform entity. Furthermore, the exact combination of histone modifications in any one region of the Xi may vary during development, in different lineages (see below) and cell types, and even across the cell cycle. The H4K20 monomethylation mark is one example of cell cycle regulation as it only appears to be enriched in a proportion of cells (Kohlmaier et al. 2004). The enzyme(s) responsible for this mark has not been identified, although PR-Set7 is a strong candidate given its participation in modifying this histone within facultative heterochromatin in other species (Karachentsev et al. 2005).

Recently, several groups reported that the Xi is ubiquitinated, and that this is at least partly due to the monoubiquitination of histone H2A at Lys 119. Recent work by de Napoles et al. (2004) demonstrated that this histone modification is dependent on the Ring1a and Ring1b proteins, particularly the latter, which is also a component of the polycomb PRC1 complex. Several members of the PRC1 complex have now been found associated with the Xi, including Mel18, Bmi1, and HPC2, although there are differences between reports probably due to the antibodies and cell types used (for review, see Heard 2005 and references therein). Whether or not PRC1 members or H2A K119 ubiquitination are critical for the initiation or maintenance of XCI, remains unclear. Indeed, issues of redundancy, not only between epigenetic marks on the inactive X chromosome, but also between mammalian Polycomb group proteins, render functional studies challenging. For example, several potential mouse homologs exist for each of the PRC1 components. Nevertheless, the finding that PRC1 is recruited to the Xi provides an important parallel to the situation in Drosophila, as it has been shown that the H3K27me3 mark deposited by the PRC2 complex could act as a binding site for the chromodomain of the Polycomb protein (Fischle et al. 2003; Min et al. 2003); in this way, PRC2 could result in the recruitment of PRC1 as a further step in maintaining repressed states. In mammals, multiple Pc homologs exist, and the recruitment of several of them to the Xi does appear to be facilitated by the H3 K27me3 mark, although this mark alone may not be sufficient in vivo, and an RNA component may also be required (Bernstein et al. 2006).

Two other changes in chromatin constitution that occur during XCI are the association of the histone H2A variant, macroH2A (Constanzi and Pehrson 1998), and DNA methylation of promoters of X-linked genes (Fig. 3; for review, see Heard et al. 1997). Both of these marks appear to be relatively late events during XCI. The recruitment of macroH2A to the X chromosome appears, despite its late timing, to be Xist RNA dependent (Csankovszki et al. 1999), although no direct association has been demonstrated so far. The role of macroH2A in XCI has remained a puzzle in the absence of mutants. It has been suggested that the apparent recruitment of this histone variant may simply be a reflection of the compaction of the Xi (Perche et al. 2000), but given its role in inhibiting chromatin remodeling associated with activation (Angelov et al. 2003), it seems logical that it could participate in maintenance of the inactive state. MacroH2A is enriched on the Xi in a cell cycle-dependent fashion (S phase and G1), suggesting that its repressive role on the Xi, if any, is only exerted during a certain time window (Chadwick and Willard 2002). This may imply that it participates in ensuring silencing and/or the transmission of epigenetic information, at a time when certain histone modifications are lost.

DNA methylation, on the other hand, has been clearly shown to play an important role in stabilizing the inactive state of the Xi, at least in somatic cells (see Heard et al. 1997 and references therein). Analysis of Dnmt1^–/– mutant embryos has shown that methylation is required for stable maintenance of gene silencing on the Xi in the embryonic lineage (Sado et al. 2000). In extraembryonic lineages, on the other hand, the 5' ends of X-linked genes do not appear to be hypermethylated. Although this correlates with higher rates of sporadic reactivation of X-linked genes (Kratzer et al. 1983), the maintenance of the inactive state, at least in some extraembryonic tissues such as the visceral endoderm, seems to tolerate extensive demethylation in vivo (Sado et al. 2000). As mentioned above, Polycomb group proteins appear to have a more important role in the maintenance of X inactivation in extraembryonic lineages (Wang et al. 2001), which is also consistent with recent findings for some imprinted autosomal loci (Lewis et al. 2004; Umlauf et al. 2004). Nevertheless, it should be noted that even between extraembryonic tissues, differences can be found in the epigenetic marks on the Xi. For example, in cells derived from the primitive endoderm, despite the fact that the inactive state of the X appears to be very stable in this lineage, no, or very low levels of PRC2 and H3K27 methylation are detected on the Xi (albeit at the immunofluorescence level), unlike in the trophectoderm and embryonic ectoderm (Kunath et al. 2005). Furthermore, the role of Eed in maintaining the inactive state in trophoblastic tissues appears to be differentiation dependent (Kalantry et al. 2006).Thus, the Xi may carry different combinations of epigenetic marks in different cell lineages and differentiation stages. The situation in marsupials, where no promoter DNA methylation is observed, will clearly be of interest (Kaslow and Migeon 1987). Examination of other epigenetic marks on the Xi in marsupials, such as the Polycomb group proteins, will provide important insights into the evolution of cellular memory mechanisms in mammals.

In addition to epigenetic modifications, temporal and spatial segregation of the inactive X chromosome may be important means of ensuring the differential activity of the inactive X chromosome. The shift to asynchronous replication timing that accompanies XCI is likely to be important in maintaining the inactive state during each cell cycle, as it provides a temporal segregation of the two X chromosomes (Hansen et al. 1996). This may minimize the exposure of the Xi to transcription factors, and optimize its exposure to the appropriate chromatin-modifying enzymes, thus facilitating the maintenance of transcriptional silence. A striking, but so far unexplained observation is that during random XCI, the Xi becomes late replicating (Takagi et al. 1982), whereas during imprinted XCI it is early replicating (Sugawara et al. 1983). In both cases, the shift in replication timing appears to follow the onset of the histone modifications (Chaumeil et al. 2002; Okamoto et al. 2005), suggesting that it might be a consequence of changes in chromatin structure of the X. Intriguingly, there is also evidence that the silent Xist locus on the active X chromosome may control asynchronous replication timing of the X chromosome, although the mechanistic basis for this remains unclear (Diaz-Perez et al. 2005). The functional importance of asynchronous replication is underlined by the fact that it seems to be one of the best-conserved characteristics of the Xi in mammals (Sharman 1971). However, the role of replication asynchrony in XCI cannot be easily tested without the accompanying perturbation of other epigenetic marks.

The spatial segregation of the inactive X chromosome in the nucleus may also be important in the maintenance, and perhaps even in the initiation, of XCI. The idea that the inactive X chromosome could represent a repressive nuclear compartment has been proposed in the past (Clemson et al. 1996) and is supported by the recent finding that the nuclear scaffold protein SAF-A (Fackelmayer 2005) associates with the inactive X chromosome. Such a compartment may be nucleated through the action of epigenetic marks together with the Xist transcript at every cell cycle. In support of this, a recent study has shown that Xist RNA defines a silent nuclear compartment early on in the X-inactivation process (Chaumeil et al. 2006). Whether distinct domains of heterochromatin define its organization and position in the nucleus (Chadwick and Willard 2004) remains an important question for the future.

The epigenetic marks and segregation mechanisms described above are likely to act in synergy to maintain the inactive state and to provide the cellular memory that enables its heritability through successive cell divisions. Disruption of a single mark, such as DNA methylation, histone hypoacetylation, or Xist RNA coating in somatic cells seems to barely affect the stability of the inactive state, while the combined absence of several marks results in increased rates of sporadic gene reactivation on the Xi (Csankovszki et al. 2001). Nevertheless, even under these conditions, global reactivation of the X chromosome is never seen, suggesting that we may still be missing some critical mark or else that spatial–temporal segregation of the Xi alone can ensure maintenance of its inactive state. Although reactivation of X-linked genes rarely happens in vivo, it can increase with aging and in cancer cells (Wareham et al. 1987; Spatz et al. 2004), suggesting a loosening of epigenetic marks in these situations. Potential new insights into the reasons underlying such sporadic reactivation have come from studies suggesting that perturbation of factors such as BRCA1, ATM, or ATR, which are known to be involved in DNA repair and to act as genome "caretakers," can lead to disruption of XIST RNA coating in the case of BRCA1 (Ganesan et al. 2002) or perturbations in gene silencing in the case of ATM and ATR (Ouyang et al. 2005). Although the molecular basis for this is unclear, the epigenetic stability of the inactive state may be linked to genomic stability in general. The mechanistic nature of this link and its possible implications for cancer represent exciting questions for the future.

	Escape from X inactivation

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

 
Although X inactivation is thought of as a chromosome-wide phenomenon, in fact, some genes can escape X inactivation; that is, they are biallelically expressed in female cells. Escapees with a Y paralog may represent evolutionary remnants from the proto-sex chromosomes (Fig. 1). In human, escape genes are numerous (15% of X-linked genes) and are concentrated in the more recent evolutionary strata, supporting the notion that acquisition of X inactivation is dependent on the loss or differentiation of Y-linked genes (Fig. 4; Jegalian and Page 1988; Lahn and Page 1999; Carrel and Willard 2005). As expected, genes located in the recombining pseudoautosomal region(s) of the sex chromosomes escape X inactivation. One fascinating exception comprises a few genes located in the PAR2 in human that are subject to silencing both on the inactive X in females and on the Y in males. When on the Y, one of these genes, SYBL1, bears epigenetic marks characteristic of genes silenced on the inactive X including DNA methylation, histone modifications, and chromatin condensation, suggesting a Y-specific silencing mechanism independent of Xist (Matarazzo et al. 2002). Since fewer remaining X/Y gene pairs with similar (usually ubiquitous) expression have been found in mouse, the scarcity of mouse escapees is seemingly easy to explain (Disteche et al. 2002). However, many human genes that have lost their Y paralog still escape X inactivation (Carrel and Willard 2005). Thus, the acquisition of X inactivation in the face of Y degeneration/differentiation has proceeded at a different rate in each species. Incomplete silencing of the human X may be due to a barrier effect caused by the centromeric heterochromatin that separates the XIC from the short arm where most escape genes are located (Disteche 1999). In some rodent species with a large block of heterochromatin on the X chromosome, Xist RNA does not spread in this region (Duthie et al. 1999). Nevertheless, the paradoxical persistence of escape genes, especially in human, remains largely unexplained (Disteche et al. 2002; Brown and Greally 2003).

View larger version (50K): [in this window] [in a new window]   Figure 4. Escape from X inactivation. (A) Domains of escape from X inactivation are smaller in mouse than in human, as shown for three domains of escape in mouse (one escape gene, Jarid1c, Eif2s3x, Utx) and human (multiple escape genes). (Blue) Genes subject to X inactivation; (yellow) escape genes. (B) Model for the possible role of CTCF at the boundary of escape domains. CTCF (red ovals) binding to the CpG island of escape genes may prevent the establishment of DNA methylation (black circles), resulting in reactivation of the escape genes during development. Histone modifications characteristic of inactivated genes (green box) may be associated with the transient silencing of escape genes during development. At reactivation, such genes would then acquire marks characteristic of active genes (orange). (Blue) Genes subject to X inactivation; (yellow) escape or noninactivated genes. Adapted from Filippova et al. (2005) with permission from Elsevier © 2005.

For a given gene, escape from X inactivation is not necessarily consistent between individuals or between tissues and/or cells within an individual. A comprehensive survey in human confirms the original observation that some genes only escape X inactivation in subsets of cells (Anderson and Brown 1999; Carrel and Willard 2005). Interestingly, many genes (10% of X-linked genes) behave in this manner, resulting in potentially variable expression levels between female tissues and individuals. Whether, in turn, this generates female phenotypic variation is an interesting possibility that remains to be explored. Partial or variable escape from X inactivation is in agreement with progressive incorporation of genes into the X up-regulation/X inactivation systems once the Y paralog degenerated (Fig. 1B).

Molecular mechanisms of escape can be derived from genomic and epigenetic analyses of chromosomal domains containing escape genes. Comparisons of multiple homologous genomic domains in human and mouse indicate that diminishing escape domains typify the mouse X chromosome (Fig. 4A; Tsuchiya and Willard 2000; Tsuchiya et al. 2004; C.M. Disteche, unpubl.). Among factors potentially associated with the shrinking of escape domains on the mouse X are differences in DNA repeat expansion. Escape domains appear to be depleted in long terminal repeats (LTR) (Tsuchiya et al. 2004). Surprisingly, differences in the distribution of L1 elements, which have been proposed as way-stations for the propagation of silencing along the X (Lyon 1998), were not found (Tsuchiya et al. 2004). However, other studies report a correlation between a low density of L1 elements, especially young (recently expanded) L1 repeats in regions that escape inactivation (Bailey et al. 2000; Carrel and Willard 2005).

Genes that escape X inactivation are actively expressed within the context of silenced chromatin. In adult tissues their chromatin structure, including histone modifications and lack of DNA methylation, is characteristic of that of active genes (Goodfellow et al. 1988; Gilbert and Sharp 1999; Boggs et al. 2002; Filippova et al. 2005). Yet, in ES cells, escape genes are marked by specific histone modifications characteristic of biallelically expressed genes (Rougeulle et al. 2003). The existence of actively transcribed domains within inactive chromatin suggests that boundary elements are positioned between domains of escape and inactivation. Accordingly, binding sites for a chromatin insulator element, CTCF, have been discovered at the 5' end of two mouse genes (Jarid1c and Eif2s3x) and one human gene (EIF2S3X); all three genes escape X inactivation and are each adjacent to an inactivated gene (Filippova et al. 2005). CTCF-specific binding between domains of inactivation and escape, but not in a region between two escape genes, suggests that protection from stable silencing may operate at the level of domains that contain one or several escape genes. CTCF binding to a specific DNA sequence both depends on and regulates DNA methylation (Fedoriw et al. 2004; Pant et al. 2004). In the context of X inactivation, which is associated with CpG island methylation, a role for CTCF in escape may involve interference with this methylation process. Insulation of escape genes from adjacent inactive chromatin could be mediated by interactions between CTCF and the establishment and/or cooperative spreading of DNA methylation (Fig. 4B). This is supported by findings of near-complete absence of methylation during early development at the CpG dinucleotides contained within CTCF-binding sites at the 5' end of Jarid1c (Filippova et al. 2005). Interestingly, Jarid1c is thought to be transiently silenced during development (Lingenfelter et al. 1998). A possible explanation for such transient silencing could be that labile chromatin modifications are induced at the locus during the initiation of X inactivation, but that these do not persist and the inactive state cannot be locked in due to lack of CpG island methylation because of CTCF binding. Thus, escape genes may be susceptible to some of the molecular layers that control X inactivation, but they may be protected from certain modifications, resulting in unstable silencing. Incomplete inactivation of some X-linked genes could also be due to low affinity for XIST RNA and its accompanying silencing complex. X;autosome translocations show patchy spreading of silencing in the autosomal regions, apparently dependent on different affinity for XIST RNA (Hall et al. 2002; Sharp et al. 2002). The heterogeneous heterochromatin domains on the human X may also play a role in the distribution of escapees (Chadwick and Willard 2004). Finally, genes that escape may occupy a different nuclear compartment from that of inactivated genes. Chromatin loop structure possibly mediated by chromatin elements such as CTCF could help separate domains on the X chromosome (Fig. 4B). Additional analyses of their chromatin structure as well as their nuclear localization during development will be informative.

View larger version (28K): [in this window] [in a new window]   Figure 5. Imprinted X inactivation in mice. Maternal and paternal germline events that are possibly relevant to the imprint(s) underlying paternal X inactivation in mouse embryos are shown in the upper part of the figure. At 1 h post-fertilization (1 hpf), the genome of the paternal pronucleus (blue) becomes remodeled, and this may facilitate the early activation of the paternal Xist gene. The maternal allele of Xist remains silent through to the morula stage, as a result of an imprint laid down during oocyte growth. The maternal pronucleus is shown in pink. Minor zygotic gene activation occurs at the one-cell stage; major zygotic gene activation occurs at the two-cell stage. X-linked gene primary transcripts are shown as red spots. The onset of Xist RNA accumulation (green) of the Xp occurs from the four-cell stage in every blastomere. The first signs of gene silencing on the Xp can be detected from the eight-cell stage and the recruitment of polycomb group proteins, H3K27 trimethylation, and H3K9 dimethylation occur subsequently. Inactivity of the Xp, albeit leaky, is maintained in the trophectoderm and is found in the primitive endoderm, but the Xp becomes reactivated in the inner cell mass (Mak et al. 2004; Okamoto et al. 2004).

	Sex chromosome inactivation in spermatogenesis

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

	Imprinted X inactivation and reactivation

Top Abstract Genetic content and evolution... Initiation of X inactivation:... The epigenetics of the... Escape from X inactivation Sex chromosome inactivation in... Imprinted X inactivation and... The imprint(s) underlying... The evolutionary origins of... X chromosome hyperactivation:... Conclusions Acknowledgments References

 
Imprinted inactivation of the paternal X chromosome is found in all tissues of marsupials (Sharman 1971; Cooper et al. 1993), and in the extraembryonic tissues of some eutherians, such as mice (Takagi and Sasaki 1975; West et al. 1977). Classical cytogenetic studies in mouse embryos had suggested that the paternal X only became inactivated at the blastocyst stage, accompanying cellular differentiation in the trophectoderm and primitive endoderm (Takagi et al. 1982). However, recent studies have revealed that that the paternal X has already begun to inactivate by the eight-cell stage (Fig. 5; Huynh and Lee 2003; Mak et al. 2004; Okamoto et al. 2004). Following fertilization, the Xp is transcriptionally active at zygotic gene activation (two-cell stage). This was revealed by the chromosome-wide presence of RNA Polymerase II (using immunofluorescence) and by the detection of nascent transcripts of X-linked genes and of Cot-1 repeat-specific transcription, using RNA FISH (Okamoto et al. 2005). Inactivation of the Xp initiates following Xist RNA coating at the four-cell stage (Okamoto et al. 2004, 2005). The chromatin changes induced subsequently on the Xp are similar to those found in differentiating ES cells, with two notable differences: macroH2A is recruited early on, by the morula stage, at a similar time to Ezh2 and H3K27me3 (Constanzi et al. 2000); furthermore, DNA methylation is not found at the promoters of X-linked genes in the trophectoderm (Fig. 2). Indeed, the repressed state of the Xp is much more unstable or "leaky" during imprinted XCI compared with random XCI (Huynh and Lee 2003; Mak et al. 2004; Okamoto et al. 2004).

By the early blastocyst stage, the Xp appears to be globally inactive in all cells of normal female (XmXp) embryos. In the trophectoderm, this inactivity of the Xp is maintained, and presumably further locked in by the shift to asynchronous (early) replication timing (Sugawara et al. 1983). In the ICM of early blastocysts (3.5 days post-coitum [dpc]), the Xp is also inactive. Strikingly, however, during blastocyst growth, the Xp becomes reactivated in the ICM, with cells rapidly losing their Xist RNA coating, Eed/Enx1 enrichment, and the histone modifications characteristic of X inactivation (Fig. 5; Mak et al. 2004; Okamoto et al. 2004). This reactivation precedes subsequent random inactivation of either the maternal or pater