Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Our understanding of the human genome keeps increasing as new technological advances allow us to unravel the genome’s functional elements. Although mapping of the human genome has been completed, it still remains unclear how certain regions are demarcated as coding regions or regulatory elements and what the vast regions without coding account for. The number of genes annotated dwindled from the previously estimated 35,000–100,000 to 25,000 protein-coding regions, which only occupies 1–2 % of the overall genome (Liang et al. 2000). This finding has prompted searches elsewhere for unidentified functional elements. Recently, it has become evident that new classes of molecules, noncoding RNAs, are emerging as crucial functional molecules in development, disease, and physiology. Surprisingly, noncoding transcripts were available in abundance while surveying the human genome but had been conveniently overlooked (Carninci et al. 2005). These noncoding RNA transcripts are classified primarily based on the length of the RNA transcripts. Noncoding RNAs generally fall under two main categories with any length below 200 nucleotide-long classified as small noncoding RNA and anything larger than 200 nucleotide-long classified as long noncoding RNA (lncRNA). Furthermore, evidence supporting the involvement of noncoding RNA in various biological processes, for example, gene expression regulation, is rapidly increasing (Wapinski and Chang 2011; Wang and Chang 2011). Indeed, both small and IncRNAs are involved in bringing about epigenetic changes at a particular locus (Peschansky and Wahlestedt 2014). The significance of noncoding RNAs is increasing with regard to their physiological function as well as their association with diseases (Wapinski and Chang 2011; Maass et al. 2014).

One such widely studied lncRNA is X-inactive specific transcript (Xist), which is indispensable for X chromosome inactivation (X-inactivation). In this chapter, we describe a wide variety of lncRNAs in the X-inactivation center (Xic), which reside on the X chromosome and are required for the induction of X-inactivation, and discuss how these lncRNAs cooperate in inducing monoallelic Xist expression to establish only one inactive X chromosome (Xi) in mammalian females. Finally, the various diseases that arise due to a dysfunctional or skewed X-inactivation, and possible future studies in humans, are discussed. An elucidation of these novel regulators and their interaction networks would provide important insight with respect to the molecular mechanism of diseases such as cancer and help toward designing better therapeutics.

2 X Chromosome Inactivation (X-inactivation)

X-inactivation is a dosage compensation mechanism in female mammals which maintains the balance of X-linked gene expression and is achieved when one of the two available X chromosomes is inactivated at an early stage of embryonic development. Failure of X-inactivation would result in an increased dosage of genes which can alter various pathways pertaining to different vital processes (Lyon 1961; Heard and Disteche 2006; Payer and Lee 2008). In the case of males, only one functional X chromosome exists and is usually referred to as genetic unisomy, whereas in females, two X chromosomes exist and the event of silencing one of them is referred to as functional unisomy. X-inactivation is critical for cellular differentiation, and dysregulation could result in developmental abnormalities (Marahrens et al. 1997). X-inactivation occurs either as imprinted X-inactivation at early embryonic stages and in extraembryonic tissues, wherein the paternal X chromosome is silenced (Takagi and Sasaki 1975; Huynh and Lee 2003; Okamoto et al. 2004), or as an act of random X-inactivation in which both the paternal and the maternal X chromosome have an equal probability of being inactivated in the epiblast (Monk and Harper 1979; Tan et al. 1993). In imprinted X-inactivation, although the paternal X chromosome undergoes complete inactivation around the blastocyst stage, it is reactivated during the peri-implantation stage in the epiblast lineage which then gives rise to a broad range of tissue types in the fetus (Mak et al. 2004). Subsequently, these cells are prone to another wave of random X-inactivation. Random X-inactivation is thought to occur close to the time of implantation in cells of the epiblast at around embryonic days 4.5–5.5 in mice. Once the Xi is established, it is inherited through all subsequent cell divisions.

X-inactivation is characterized by Xist lncRNA coating the Xi (Clemson et al. 1996). As the coating masks the chromosome completely, this observable phenomenon is referred to as the “Xist cloud.” Gene silencing is triggered once Xist is upregulated on the future Xi. This designated future Xi goes through the stages of initiation, progression, and maintenance of repressive chromatin states with the aided participation of multiple proteins and machineries (Table 1) (Wutz 2011). Numerous studies have been carried out with respect to dynamic changes occurring on the Xi during X-inactivation. These studies have shown the remarkable changes happening in the epigenetic landscape as well as the chromatin structure during X-inactivation. The epigenetic histone marks accompanying the chromatin state (euchromatin/heterochromatin) are characteristic of its transcriptional activity (active/silent) and influence strongly the chromatin structure. At the onset of X-inactivation, euchromatin markers such as H3K9Ac, H3K4me2, and H3K4me3 are lost, when Xist RNA starts coating the X chromosome, subsequently leading to global H4 hypoacetylation (Keohane et al. 1996; Chaumeil et al. 2002). Meanwhile, loss of RNA polymerase II and nascent transcripts also occurs post-Xist RNA coating (Chaumeil et al. 2006). Interestingly, a whole new array of repressive epigenetic modifications such as H3K27me3, H3K9me2, H4K20me1, and H2AK119ub1 get enriched on the Xi (Wutz 2011). Strikingly, the histone tri-methylations on H3K27 and H3K9 occur at different regions and are recognized by different cofactors in the Xi indicating two different populations of repressed heterochromatins in humans (Chadwick and Willard 2004; Nozawa et al. 2013). The kinetics of X-inactivation is tightly associated with the dynamics of histone epigenetic marks brought about by protein complexes such as polycomb repressive complex, PRC1 and PRC2, which catalyze the repressive histone modifications, H2A119ub and H3K27me3, respectively, in an Xist RNA-dependent manner (Mak et al. 2002; Silva et al. 2003; Plath et al. 2003; de Napoles et al. 2004; Fang et al. 2004).

Table 1 Epigenetic modifications associated with the Xi in mice and humans
Full size table

Additionally, the Xi is marked by a series of epigenetic changes such as histone variant macroH2A along with DNA methylation of the CpG islands and promoters (Norris et al. 1991; Costanzi and Pehrson 1998; Sharp et al. 2011). Smchd1 is also involved in delivering the DNA hypermethylation of the CpG islands associated with the Xi, which is required for long-term maintenance of gene silencing (Blewitt et al. 2008; Gendrel et al. 2012). Another group of proteins associated with the Xi is a member of the trithorax group proteins for transcriptional activation, Ash2L, although its functional role on the Xi remains to be elucidated (Pullirsch et al. 2010).

3 X-inactivation Center (Xic)

Early studies of X chromosome truncations and translocations helped identify the X chromosome locus wherein X-inactivation is induced, called an Xic (Rastan and Brown 1990; Brown 1991). The initial mapping of the Xic was first shown by a series of cytological experiments, which used differential staining of X chromosome material at the metaphase stage of mouse embryos. These showed that chromosomal rearrangements between the X chromosome and autosomes led to random inactivation of the segment hosting the Xic. In experiments examining the T16H reciprocal translocation between the X chromosome and chromosome 16, also referred to as the Searle’s translocation, only the translocated 16X chromosome was inactivated, predicting the presence of the Xic (Takagi 1980; Rastan 1983). Subsequently, truncating one of the X chromosomes at the Xic region in embryonic cells resulted in no X-inactivation, suggesting that two Xic’s are required for X-inactivation (Rastan and Robertson 1985). Furthermore, when one of the X chromosomes was truncated leaving behind a significant chunk of the Xic, termed HD3 translocation, random X-inactivation occurred, indicative of the distal boundary of the Xic (Rastan and Robertson 1985). These results thus suggest that the minimum region required for efficiently triggering X-inactivation lies somewhere between the T16H and the HD3 break points, which was followed by further extensive genetic mapping (Keer et al. 1990). While the majority of experiments were established in mice, a similar strategy was used to determine the XIC in humans using mice/human somatic cell hybrids derived from patient samples with human X chromosome translocations and deletions (Brown et al. 1991b). The human equivalent of the mouse Xic seems to be highly conserved and spans approximately a 1 Mb region in Xq13. Studies increasingly propose that a minimal region is required for X-inactivation to take place in humans. Any abnormalities arising due to rearrangements in this demarcated region could lead to improper functioning and haywire regulations.

X-inactivation is a phenomenon controlled exclusively by the events occurring at the Xic. A series of transgene experiments contributed to delineating the Xic region (Heard et al. 1996; Lee et al. 1996; Lee and Jaenisch 1997). In embryonic stem cells, transgenes containing Xist can induce silencing in surrounding regions at its insertion site. Experiments showed that a mouse transgene was sufficient to induce Xist RNA expression and coating in cis and subsequently silence the LacZ reporter within the transgene (Lee et al. 1996). Copy number and expression levels were also found to play an important role for silencing to take place at the integration site for the genome (Heard et al. 1996). The Xic region, including Xist and its flanking regions, harbors a number of lncRNAs and consists of a complex interplay between each lncRNA to regulate the monoallelic expression of Xist (Fig. 1). Apart from the abundance of both positive and negative lncRNA regulators of Xist at the Xic, a number of proteins are also involved in this tight regulation (Wutz 2011).

Fig. 1
figure 1

lncRNAs on the Xic. The figure shows the schematic of Xic and lncRNAs originating from the locus at the onset of X-inactivation. Red and blue arrows show the action of lncRNAs as transcriptional activators and repressors, respectively. Orange box with black arrow indicates the actively transcribed gene and direction of transcription. Blue box with black arrows by dashed line indicates repressed gene. While Jpx and Ftx are known as escape genes (Tian et al. 2010; Chureau et al. 2011; Kobayashi et al. 2013), expression of Tsix and Xite persists longer on the Xi at the onset of X-inactivation (Lee et al. 1999a; Ogawa and Lee 2003). Allelic expression of Tsx, DXPas34, and RepA has not yet been reported. Although the Jpx transgene induces Xist expression, it is not clear whether endogenous Jpx acts both in cis and in trans at the initiation of X-inactivation

Full size image

3.1 Xist

Xist produces a 17 kb transcript in mice and a 19 kb transcript in humans that is processed by polymerase II, polyadenylated, and retained in the nucleus. The Xist lncRNA is exclusively expressed from the Xic on the Xi and a central player of Xic function for both imprinted and random X-inactivation (Brown et al. 1991a; Brockdorff et al. 1991; Penny et al. 1996; Marahrens et al. 1997; Wutz and Jaenisch 2000). Xist lncRNA induces chromosome-wide gene silencing through a cascade of epigenetic modifications on the Xi, which is eventually maintained through multiple rounds of cell division. Xist RNA has multiple functional domains spread across its 8 exons including repeat A-F, which are highly conserved in eutherian mammals, enabling its interaction with transcriptional factors, scaffold proteins, and chromatin-modifying proteins (Sado and Brockdorff 2013).

At an initial phase of X-inactivation, it is proposed that transcriptional factors YY1 (Yin Yang 1) act as an anchoring point to bridge between Xist RNA and Xist gene to serve as a nucleation center for Xist RNA spreading across the Xi (Jeon and Lee 2011). Furthermore, since knockdown of hnRNP U (heterogenous nuclear ribonucleoprotein U, also known as SAF-A and SP120), which has a binding affinity to both DNA and RNA, disrupts Xist RNA localization and X-linked gene silencing, hnRNP U acts as a bridge between matrix/scaffold attached region (MAR/SAR) in the Xi and Xist RNA to facilitate spreading of the silencing machinery such as PRC2 across the Xi (Hasegawa et al. 2010). Recent studies to map Xist RNA and its binding partner, PRC2, on the Xi revealed an orderly fashion of Xist RNA and PRC2 spreading, as well as a strong dependency on Xist RNA for the three-dimensional structure of the Xi (Splinter et al. 2011; Engreitz et al. 2013; Simon et al. 2013). Allele-specific ChIP-seq (chromatin immunoprecipitation with deep sequencing) of a catalytic subunit of PRC2, Ezh2, showed ~50 prominent and ~1500 moderate peaks prior to X-inactivation, suggesting that Ezh2 binds at ~50 strong and ~1500 moderate binding sites (Pinter et al. 2012). The Ezh2 binding sites are frequently associated with canonical H3K4me3/H3K27me3 bivalent domain and CpG islands across the X chromosome. Upon differentiation and induction of X-inactivation, Ezh2 binds to an additional >100 strong and ~4000 moderate binding sites which can then induce spreading of H3K27me3 toward neighboring regions. An Xist RNA binding map produced by combining CHART-seq (capture hybridization analysis of RNA targets with deep sequencing) and RAP (RNA antisense purification) data has also revealed that Xist RNA initially binds to gene-rich regions before spreading to distal, gene-poor regions (Engreitz et al. 2013; Simon et al. 2013). The Xist RNA binding profile overlaps heavily with Ezh2 binding and H3K27me3 density, indicating an Xist RNA-dependent deposition of Ezh2 and H3K27me3. To efficiently induce gene silencing across the entire X chromosome upon differentiation, Xist RNA spreads in a three-dimensional manner away from the Xic toward distal binding sites across the Xi (Lieberman-Aiden et al. 2009; Engreitz et al. 2013; Simon et al. 2013). The repeat A region in exon 1 of Xist RNA is presumed to have an important role in Xist RNA spreading across the gene-rich region of the Xi, since deletion of repeat A has resulted in impairment of this process (Engreitz et al. 2013). It was previously suggested that long interspersed nuclear element 1 (LINE1) repeat elements play a role in assisting Xist RNA to spread along the entire X chromosome, as well as supporting the assembly of heterochromatic nuclear structures and propagation of X-inactivation (Lyon 2000; Chow et al. 2010). However, recent reports show that there is less correlation between the Xist RNA binding site and LINE1 repeats that has been previously speculated (Engreitz et al. 2013; Simon et al. 2013).

4 Long Noncoding RNAs and Elements Controlling Xist Expression

Xist is the central player of X-inactivation: through neighboring lncRNAs, which are involved in the tight regulation of Xist monoallelic expression, Xist induces a repressive chromatin state that leads to X-linked gene silencing along the entire X chromosome. Recent advancements in next-generation sequencing and newer techniques have propelled forward the functional analysis of novel lncRNAs in Xic and increased our understanding of their function and interactions at the molecular level. Here, we describe several lncRNAs in the Xic which regulate Xist expression in both positive and negative ways. Tight regulation in the interplay of these lncRNAs is essential for securing the induction of monoallelic Xist expression and bringing about X-inactivation in only one of the two X chromosomes in females.

4.1 Negative LncRNA Regulators of Xist

4.1.1 Tsix

The noncoding Tsix gene expresses lncRNA antisense to Xist; hence, it is named “Tsix,” which is Xist spelled in reverse order (Lee et al. 1999a). While the Tsix transcript does not coat the X chromosome like its counterpart Xist during X-inactivation, it is detected using RNA fluorescence in situ hybridization as a pinpoint signal at both endogenous loci and is expressed in both male and female undifferentiated cells (Lee et al. 1999a). While monoallelic Tsix expression coincides with Xist silencing on the future active X chromosome (Xa) at the onset of X-inactivation, Tsix extinction and Xist upregulation also occur on the future Xi, thereby suggesting the antagonistic role of Tsix on Xist expression. Tsix expression finally disappears on both X chromosomes at a later stage of differentiation without Xist reactivation on the Xa, suggesting that Tsix represses Xist upregulation at the onset of X-inactivation. Tsix has been reported to play a role for Xist repression both in imprinted and in random X-inactivation (Lee and Lu 1999; Lee 2000; Sado et al. 2001). In Tsix heterozygous mutant female ES cells, a mutation in Tsix always leads to the induction of Xist expression from the Tsix-mutant X chromosome, hence the non-random inactivation of the mutant X chromosome (Lee and Lu 1999). Apart from random X-inactivation, imprinted X-inactivation is controlled by Tsix to prevent the maternal X chromosome from undergoing X-inactivation in the extra embryonic tissues. While female embryos carrying a Tsix mutation on the maternal X chromosome lead to embryonic lethality due to X-inactivation on both X chromosomes, female embryos carrying a mutated Tsix on the paternal X chromosome normally survive to term (Lee 2000; Sado et al. 2001). Since the Tsix-mutant male embryo also carries a Tsix mutant maternal X, this mutation resulted in embryonic lethality due to X-inactivation. These reports demonstrate that Tsix antagonizes Xist expression in cis. Tsix exclusively works to repress Xist by modulating the chromatin structure (Navarro 2005; Sado et al. 2005; Sun et al. 2006). Since the insertion of a polyadenylation signal to produce Tsix truncation at the site close to the 5′ end of Xist effectively abolishes Tsix function, Tsix transcription across the Xist promoter is critical for Xist repression (Ohhata et al. 2008).

4.1.2 Xite (X-inactivation Intergenic Transcription) and DXPas34

Xite resides between minor and major Tsix promoters and is associated with multiple bidirectional long noncoding transcripts (Ogawa and Lee 2003). Since an Xite heterozygous deletion mutation leads to skewed X-inactivation which favors the mutant X chromosome, Xite is involved in choosing which X chromosome will be inactivated. Further analysis revealed that Tsix is downregulated in the Xite deletion mutant in cis; thus, Xite plays a role in the decision of the Xi by positively promoting Tsix expression, which in turn represses Xist (Ogawa and Lee 2003). Many models have been proposed with regard to how Xite could function in X-inactivation. One model suggests that Xite could act as an enhancer for development-specific Tsix regulation at the onset of X-inactivation. Frequent association of multiple bidirectional transcripts and DNaseI hypersensitive sites with enhancer elements supports the Xite enhancer model (Natoli and Andrau 2012; Lam et al. 2014). Indeed, transient enhancer assays revealed that Xite has development-specific enhancer activity in Tsix expression (Stavropoulos et al. 2005).

DXPas34, another region associated with bidirectional promoter activity and DNaseI hypersensitive sites, is also a positive regulator of Tsix expression (Stavropoulos et al. 2005; Vigneau et al. 2006; Cohen et al. 2007). DXPas34 is a 1.2 kb CpG-rich region containing a 34-mer tandem repeat residing 750 bp downstream of the major Tsix promoter. Transient enhancer assays showed that DXPas34, as well as Xite, enhanced Tsix expression (Stavropoulos et al. 2005). Interestingly, deletion of DXPas34 leads to repression of the major Tsix promoter and activation of Xist expression at the onset of X-inactivation, which is followed by Tsix derepression without reversal of X-inactivation. This suggests that DXPas34 functions as both an enhancer and repressor of Tsix in a differentiation-specific manner.

4.1.3 Tsx (Testes-Specific X-Linked Noncoding RNA)

Another noncoding transcript, Tsx, located nearly 40 kb from the 3′-end of Xist, was found to be specifically expressed in the testes and in a very low concentration in both male and female brains (Simmler et al. 1996; Anguera et al. 2011). Although Tsx was initially reported as a potential protein-coding gene, specifically as an encoder of a 144 amino acid protein (Simmler et al. 1996), further investigation has indicated that Tsx is likely to be a noncoding gene (Anguera et al. 2011). Tsx expresses in ES cells as well as in early embryos and is gradually repressed upon differentiation. The homozygous deletion of Tsx in female mice led to a small decrease in fertility, resulting in skewed sex ratios that slightly favored females. Aberrant upregulation of Xist along with Tsix downregulation was observed in a small population of the Tsx mutant cells; thus, Tsx might promote Tsix expression and indirectly upregulate Xist expression (Anguera et al. 2011).

4.2 Positive LncRNA Regulators of Xist

4.2.1 Jpx/Enox

Jpx/Enox is another important lncRNA, which is located 10 kb upstream of Xist and is expressed in the antisense direction of Xist (Chureau et al. 2002; Johnston et al. 2002; Chow et al. 2003). While transgenes including the X-inactivation hub of lncRNAs such as Xist, Tsix, and Xite could only induce Xist activation inefficiently, the additional upstream region of Xist restored the induction of Xist (Lee et al. 1999b). This suggests that the Xic requires an upstream region flanking Xist. Subsequently, further analysis showed that Jpx is required for the proper expression of Xist (Tian et al. 2010; Sun et al. 2013). Jpx escapes from X-inactivation and is upregulated during X-inactivation (Tian et al. 2010). When Jpx was deleted, no X-inactivation was induced in males; however, Jpx heterozygous mutant females exhibited severe phenotypes with massive cell death, significantly impaired Xist upregulation, and compromised X-inactivation induction. These defects and Xist expression levels were restored to normal with overexpression of Jpx using a transgene. These data thereby suggest that Jpx can act in trans to activate Xist, although the Jpx heterozygous mutation shows mildly reduced Xist expression in cis (Tian et al. 2010). As Jpx overexpression with Tsix disruption efficiently induces aberrant X-inactivation even in male differentiating embryonic bodies, both Jpx and Tsix coregulate Xist as an activator and repressor, respectively. That leaves the question of how Jpx RNA promotes Xist expression. A recent study has indicated the unique function of Jpx RNA in Xist expression at the initiation of X-inactivation (Sun et al. 2013). In undifferentiated cells, transcription factor CTCF is loaded on the CTCF binding sites within the Xist P2 promoter by which Xist is repressed. Whereas overexpression of CTCF blocked induction of Xist upregulation, this repression was rescued by overexpression of Jpx. As CTCF binds to both the Xist promoter and Jpx RNA, it is proposed that Jpx RNA replaces CTCF from the Xist promoter, which is followed by induction of Xist expression on the Xi (Sun et al. 2013). Future work will be needed to elucidate the mechanism by which Jpx RNA selectively replaces CTCF from the Xist promoter in the future Xi to induce monoallelic Xist expression.

4.2.2 Ftx (Five Prime to Xist)

Ftx is another gene which encodes a long noncoding transcript located in the upstream of Xist, a potential activator for Xist (Chureau et al. 2002, 2011). Ftx escapes X-inactivation in both imprinted and random X-inactivation and, like Jpx, is upregulated upon induction of random X-inactivation (Chureau et al. 2011; Kobayashi et al. 2013). Deletion of Ftx in male mouse ES cells showed that the expression pattern of X-linked genes in the vicinity of Ftx was altered through a significant drop in expression levels. Furthermore, increased methylation at the 5´ CpG island of Xist was observed, suggesting a positive role of Ftx in Xist expression. However, Ftx has been reported to be dispensable in imprinted X-inactivation in the mouse embryo (Soma et al. 2014). In spite of targeted deletions of Ftx, neither the survival of female embryos nor the expression of Xist was affected during the preimplantation period in the Ftx mutant mice (Soma et al. 2014). Further investigation would be necessary to conclude whether Ftx is essential for random X-inactivation in female mice.

4.2.3 RepA

Apart from the two major isoforms of Xist RNA, known as long and short forms, there is another lncRNA which is derived from Xist. The 1.6 kb RepA RNA, which is transcribed from the repeat A region of Xist, was identified by a PRC2 immunoprecipitation (Zhao et al. 2008). Multiple roles for RepA RNA have been shown. For instance, RepA embedded within Xist is involved in both the recruitment of the PRC2 complex, Ezh2, and the activation of Xist (Zhao et al. 2008). Furthermore, the RepA region in Xist RNA has been shown to be a binding region for the Ezh2 protein, while other works have suggested that RepA plays a role in PRC2 complex spreading and H3K27me3 modification across the Xi (Plath et al. 2003; Kohlmaier et al. 2004; Zhao et al. 2008; Engreitz et al. 2013).

The Xic harbors a variety of lncRNAs such as Tsix and Jpx that are directly or indirectly involved in the regulation of Xist expression. Since lncRNAs are central players for the proper regulation of X-inactivation, further studies need to be extensively performed in order to fully explore the roles played by each lncRNA as well as the cooperative molecular mechanism involved in their interaction. With advancing technologies and novel approaches, it is likely that more and more novel noncoding RNAs will emerge, allowing for a deeper understanding of Xist’s regulation during X-inactivation and its associated epigenetic modifications.

5 X-inactivation and Disease

Maintenance of proper gene dosage in autosome and sex chromosomes is important for ideal development and survival of organisms (Torres et al. 2008). Aneuploidy is referred to as the condition where cells possess an atypical number of chromosomes and is usually detrimental to the organism (Fig. 2a). Although some patients with autosomal aneuploidies can survive, they are at a high risk of congenital malformations. Some examples are the trisomies in chromosomes 13, 18, and 21, usually referred to as Patau, Edwards, and Down syndromes, respectively. Turner syndrome and Klinefelter syndrome are diseases that arise due to an absence of one of the two X chromosomes (XO) in females and an extra X chromosome (XXY) in males, respectively (Sybert and McCauley 2004; Groth et al. 2013). While only one X chromosome remains active per each cell as a result of X-inactivation, 15 % of X-linked genes on the Xi in humans are known to be actively transcribed even in this chromosome-wide silenced environment; these are called escape genes (Carrel and Willard 2005). Absence and excess of escape gene expression on the Xi is proposed to be a cause of Turner and Klinefelter syndromes, respectively.

Fig. 2
figure 2

Genetic and epigenetic failures induce abnormal gene dosage. Deviation of gene dosage leads to detrimental effects on cell survival and development. a Aneuploidy with excess or loss of certain chromosome. b X-autosomal translocation could induce inactivation of the autosome fused to the X chromosome segment containing the Xic. The segment of X chromosome which lacks Xic in turn fails to be inactivated. c Cells with more than two active X chromosome arise by loss of the Xi followed by duplication of the Xa. Reactivation of the Xi could be a potential cause of a supernumerary Xa. Green, orange, and blue bold lines indicate the autosome, X, and Y chromosomes, respectively. Purple zigzag line indicates Xist RNA-induced gene silencing

Full size image

X-inactivation is a complex enforcement system, which tightly controls the X-linked gene balance between the sexes, as well as secures and maintains a perfect ambience for proper cell differentiation and development. When certain segments from the X chromosome corresponding to the Xic are translocated to an autosomal region, the autosomal genes around the translocation site might be silenced through XIST RNA spreading coupled with accumulation of repressive epigenetic modifications (Fig. 2b) (Brown et al. 1991b), which potentially leads to haploinsufficiency of autosomal genes (White et al. 1998; Giorda et al. 2008; Van Echten-Arends et al. 2013).

While X-linked genes and autosomal genes are present in two copies in females, subsets of genes are expressed only from a single allele in individual cells. X-inactivation and genomic imprinting have allowed us to better understand the mechanism involved in determining whether a particular gene on a single allele will be expressed or repressed, as recently shown by the attribution of lncRNAs as master regulators of monoallelic expression of Xist and imprinted genes (Lee and Bartolomei 2013). The monoallelic expression of female X-linked genes in mammals is crucial for cellular survival and development. In addition, misregulations in proper monoallelic expression of imprinted genes are known to lead to a wide range of diseases such as Beckwith–Weidemann syndrome and Angelman syndrome. Thus, even deviations from normal X-linked gene dosage conditions could potentially give rise to developmental anomalies and disease states (Fig. 2c). In the context of disease conditions induced by dysfunctional X-inactivation, our knowledge remains poor even though the phenomenon is scientifically fascinating.

5.1 X-inactivation and Cancer

While each somatic cell contains a pair of active autosomal chromosomes in mammals, this is not the case in sex chromosomes. When an unfavorable recessive change occurs in an autosomal allele, the secondary chromosome pair can act as a backup to replace the damaged gene; thus, the heterozygous condition might delay or prevent a catastrophic situation. As mentioned previously, mammalian sex chromosomes are characterized as genetic unisomy and functional unisomy in males and females, respectively. Furthermore, the X chromosome is laden with important genes for cellular differentiation and proliferation, as well as those related to cancer. Thus, genetic changes taking place on the delicate sex chromosomes could be immediately detrimental due to the lack of a backup copy and the higher likelihood that the mutations, when carried forward, may be prone to cancer (Spatz et al. 2004).

Aberrant X-inactivation can bring about local as well as chromosome-wide disturbances of X-linked gene silencing to alter the expression of cancer-related and other genes across the Xi which may lead to tumors (Chaligné and Heard 2014). A loss of Xist expression has been reported in many cancer cell lines derived from female breast, cervix, and ovary tumors (Kawakami et al. 2004b; Sirchia et al. 2005; Richardson et al. 2006). In female cancer cells, the frequent disappearance of Xist-expressing cells happens by loss of the Xi followed by a duplication of the Xa instead of reactivation of the Xi. In non-cancerous tissue, reactivation of the Xi by loss of XIST expression could potentially cause a wide range of derepression of X-linked genes, including cancer-related genes, resulting in abnormalities and diseases. X chromosome reactivation in mice is tightly restricted to happen within the inner cell mass at the blastocyst stage followed by random X-inactivation in the epiblast lineage and during the development of the primordial germ cells (Ohhata and Wutz 2013). Thus, reactivation outside of these periods could lead to detrimental effects. Since Xist is constitutively expressed from the Xi in differentiated somatic cells, it is suggestive of Xist’s role in the maintenance of X-inactivation. However, RT-PCR analysis of the mouse/human somatic hybrid cell lines containing human Xi revealed that gene silencing of XIC-lacking human Xi is highly stable, suggesting that no X reactivation takes place on the human Xi without XIC once X-inactivation is established (Brown and Willard 1994) Additionally, when Xist was conditionally deleted in primary mouse embryonic fibroblast cells, the Xi exhibited maintenance of its unique heterochromatic features such as late DNA replication and hypoacetylation on histone H4 even though histone variant macroH2A disappeared due to its Xist RNA-dependent localization (Csankovszki et al. 1999). Based on these observations, it had long been believed that Xist is essential for the initiation of random X-inactivation but dispensable for the maintenance of X-inactivation once the Xi is established. With the recent advantage of high-throughput sequencing and technical improvements, accumulating evidence has indicated otherwise, specifically that depletion of Xist RNA from the Xi can induce partial reactivation of a subset of X-linked genes on the Xi (Csankovszki et al. 2001; Zhang et al. 2007). In conditional Xist-deleted mouse fibroblast, an assessment of individual gene activity revealed that conditional Xist deletion on the Xi led to a slightly increased frequency of reactivation in X-linked GFP and Hprt genes (Csankovszki et al. 2001). This reactivation frequency was largely enhanced by treatment with 5-azadC and trichostatin A, inhibitors for DNA methylation and histone hypoacetylation, respectively, indicating that multiple layers of epigenetic regulation prevent improper reactivation of the Xi. Detailed analysis of conditional Xist knockout in dermal and embryonic fibroblast cells also showed that reactivation of the silenced X-linked genes on the Xi happened in a subset of genes (Zhang et al. 2007). Whereas the repressive histone mark H3K27me3 disappeared on the Xi by conditional Xist knockout, a dearth of active histone marker H3K4me2 remained on the Xi as well, suggesting overall chromosome-wide silencing is somewhat maintained without Xist RNA.

A recent study using human induced pluripotent stem cells (hiPSCs) also supports the role of XIST RNA in the stable repression of X-linked genes. Loss of XIST expression in hiPSCs is significantly correlated with the upregulation of X-linked oncogenes associated with higher growth rate in vitro and poor differentiation in vivo (Anguera et al. 2012). hiPSCs derived from differentiated cells using the Yamanaka factors (OCT-4, SOX2, KLF4, and c-MYC) or its derivatives hold great potential in regenerative medicine (Takahashi and Yamanaka 2006; Takahashi et al. 2007). However, hiPSCs and human embryonic stem cells (hESCs) are known to be genetically and epigenetically unstable (Kim et al. 2010; Bock et al. 2011; Gore et al. 2011); thus, strict validation of hiPSCs would be necessary for therapeutic purposes. Atypical features of female hiPSCs and hESCs are evident in X-inactivation; it is not currently known why these features are not observed in mouse ES cells (Dvash and Fan 2009; Lessing et al. 2013). They are generally classified into three distinct classes based on the status of X-inactivation and XIST expression (Silva et al. 2008; Anguera et al. 2012): Class I lines, which have two Xa’s and can undergo X-inactivation very similarly to mouse ES cells; Class II lines, wherein one X is already inactivated by the XIST cloud and cells may be partially differentiated; and finally Class III lines, wherein X-inactivation is already complete, and however, the expression of XIST is lost. Analysis across different hiPSC cell lines, especially when comparing Class II with Class III lines, showed loss of XIST expression is associated with an upregulation of X-linked genes in the Class III lines. Intriguingly, the upregulated X-linked genes include cancer-related genes such as MAGEA2 and MAGEA6, which are highly expressed in cancers (Rogner et al. 1995). These observations might be a hamper to using hiPSCs as a therapeutic tool. Despite no strong evidence thus far, the changes could be attributed to the loss of XIST expression. Conversely, XIST expression and presence of the Xi could be used as benchmarks to assess hiPSC quality. Furthermore, it should be noted that culture condition to establish and maintain hiPSCs has been improved to create naive Class I hiPSCs with high efficiency (Tomoda et al. 2012; Gafni et al. 2013).

Despite close association between the overexpression of X-linked genes and supernumerary Xa with many human cancers (Liao et al. 2003; Kawakami et al. 2004b; Pageau et al. 2007), it is not clear whether aberrant X-inactivation and reactivation of X-linked genes are a primary cause or just a consequence of cancer transformation and progression. More recently, our understanding of X-inactivation has advanced from its role in dosage compensation to include higher order functions such as tumor suppression. When a conditional knockout of Xist was achieved in the hematopoietic stem cells of mice, highly aggressive forms of hematologic cancer were manifested in a female-specific manner (Yildirim et al. 2013). Histopathological analysis of female mutant mice revealed that Xist deletion in the hematopoietic compartment induced myelodysplastic/myeloproliferative neoplasm (also known as mixed myeloproliferative/myelodysplastic syndrome [MPN/MDS]) (Orazi and Germing 2008). A detailed analysis of Xist deletion mutant mice showed that Xist deletion induced a significant genome-wide gene upregulation, especially in X-linked genes including cancer-related genes, in comparison with autosomal genes. This anomalous gene expression in the Xist mutant mice would potentially promote cancerous effects, suggesting a crucial role of Xist in not only maintenance of X-inactivation but also suppression of cancer. This is the first report that Xist disruption leads to a causal effect on cancer in vivo.

5.2 Skewed X-inactivation and X-linked Diseases

As a consequence of random X-inactivation with an equal probability of either the paternal or maternal originating X chromosome being inactivated, every female’s expression profile is a mosaic with cells having either the paternal or maternal Xi. Such mosaicism is usually advantageous for females since it contributes to physiological diversity and eases the deleterious effects of X-linked mutation. For females who have inherited unfavorable mutations on their X chromosome, there is a chance that approximately half of their somatic cells will express genes from the wild-type allele. This orderly process deviates in a subset of individuals, although the skewing varies by age and cell types (Sharp et al. 2000; Hatakeyama et al. 2004; Minks et al. 2008). It is also known that the clonality of cells varies in different female organs, which could also affect the X-linked phenotype in females (Thomas et al. 1988; Bittel et al. 2008). Since X chromosomes in mammalian female somatic cells are functionally unisomy by random X-inactivation, deviation from random X-inactivation ultimately results in a predominance of either the maternal or paternal X chromosome. This predominance could occur either by chance, by selection after primary choice, or in a predetermined fashion due to the presence of genetic elements or mutations that propel selection bias (Belmont 1996; Minks et al. 2008). The X-linked disorder in female carriers, coupled with skewed X-inactivation in favor of the wild-type X, give rise to greater populations of cells in which the mutated X-linked gene is expressed, therefore manifesting as a severe disease condition (Fig. 3) (Migeon 2006). Less severe variations are more likely to occur in females with smaller populations of mutated X-linked genes.

Fig. 3
figure 3

Skewed X-inactivation and disease phenotype. Female patients carrying a harmful mutation in a single allele show a variety of severity in disease phenotype due to skewed X-inactivation. The skewing of X-inactivation in favor of the wild-type X chromosome specifically leads to a severe phenotype. Orange and blue bold lines indicate the Xa and Xi, respectively. Red X indicates harmful mutation. White and gray circles show healthy and damaged cells, respectively

Full size image

5.2.1 Rett Syndrome

Rett syndrome, an X-linked neurologic disorder which results in severe intellectual disability, primarily affects females and manifests during early childhood, typically occurring between 6 and 18 months of age; prior to onset, development appears to be normal, but is then followed by developmental regression, reduced brain growth, and severe mental retardation (Weaving et al. 2005). In newborn males, the disease is fatal. Rett syndrome symptoms vary from child to child with severity ranging from subtle abnormalities such as loss in muscle tone, difficulty in feeding, and jerkiness of limbs to more complex mental and physical abnormalities. Rett syndrome arises as a result of a heterozygous mutation in the X-linked gene encoding ubiquitous methyl-CpG binding protein 2 (MECP2) (Amir et al. 1999). MECP2 functions as a transcriptional regulator by binding to the genome in a DNA methylation-dependent manner. Female Rett syndrome patients possess mosaic wild-type/mutant MECP2 expression as a result of the random nature of X-inactivation in somatic cells. Random X-inactivation could be affected by chance, selection, or other genetic factors during early embryonic development. Skewed X-inactivation in patients with a MECP2 mutation progresses into a wide range of clinical presentations and manifests as Rett syndrome (Camus et al. 1996; Krepischi et al. 1998; Weaving et al. 2003). Interestingly, since Rett syndrome is not accompanied by neurodegeneration, attempts to express a Mecp2 transgene in postmitotic neurons have partially rescued the neurologic symptoms in both immature and mature Mecp2 mutant mice (Giacometti et al. 2007; Guy et al. 2007).

5.2.2 X-inactivation in Other Diseases

Similarly, the relation between skewed X-inactivation and phenotype severity in X-linked disease has been shown in a number of X-linked cutaneous genetic diseases such as incontinentia pigmenti, which is associated with characteristic patterns of lines and swirls appearing on the patient’s body called Blaschko’s lines (Happle 2006; Sun and Tsao 2008). Blaschko’s lines are believed to indicate the migration path of ectodermal skin cells during development illuminated by female cells containing a mix of wild-type and mutant Xi (Happle and Frosch 1985). For example, incontinentia pigmenti caused by heterozygous mutation in X-linked NFkB essential modulator (NEMO) (Smahi et al. 2000) commonly develops vesicles that later progress to verrucous and finally to hyperpigmentation in Blaschko’s lines across the trunk. While phenotypic outcome varies in incontinentia pigmenti patients, skewed X-inactivation has been reported in females with heterozygous NEMO mutation (Martinez-Pomar et al. 2005), implicating the correlation between skewed X-inactivation and phenotypic severity of X-linked disease. Skewed X-inactivation is also implicated in X-linked disease conditions such as autoimmune deficiency (Puck et al. 1987), Duchenne muscular dystrophy (Yoshioka et al. 1998; Viggiano et al. 2013), and cancers (Kristiansen et al. 2002, 2005). Clarification of the molecular mechanism that affects probability of skewing toward either X chromosome could help us to develop therapeutic approaches and potential drug targets to improve the condition of these patients.

6 Therapeutic and Diagnostic Applications Using X-inactivation

Characteristic features of Xist RNA-mediated X-inactivation could potentially be harnessed for clinical purposes. Interestingly, the long-range inactivation potential of Xist RNA has already been employed to suppress the extra chromosome 21 in iPS cells established from cells of Down syndrome patients (Jiang et al. 2013). The detrimental condition in Down syndrome arises as a result of three copies of chromosome 21. The manifestations of such an aberration lead to numerous birth defects, stunted growth, reduced intellectual abilities, mental retardation, congenital heart defects, and many physical abnormalities (Mégarbané et al. 2009; Gardiner 2010). To correct gene dosage of chromosome 21, zinc finger nucleases (ZFN) was used to insert an inducible XIST into the DYRK1A locus on one of the three chromosome 21s of iPS cells derived from a Down syndrome patient. Subsequently, the XIST transgene initiated accumulation of the repressive histone markers H3K27me3, H4K20me1, and H2AK119ub1 onto the modified chromosome. This action was associated with hypermethylation of the CpG islands at promoter, and gene repression across the modified chromosome. These results suggest that genes across chromosome 21 undergo chromosome-wide gene silencing by induction of transgenic XIST similar to X-linked gene silencing on the Xi. Most importantly, XIST induction from the transgene on the extra copy of chromosome 21 rescued the defects in proliferation and neural development. A similar application could be used to develop potential therapeutics for other diseases with an extra number of chromosomes such as Edwards syndrome (Trisomy 18) and Patau syndrome (Trisomy 13).

Potential application of XIST as a biomarker has been proposed in testicular germ cell tumors (TGCTs) (Kawakami et al. 2004a). Plasma samples obtained from patients with TGCTs showed hypomethylation of the 5′-end of CpG sites in XIST, while somatic cells showed complete methylation through the CpG sites. The XIST gene is silenced on the Xa in males and usually methylated at its 5′-end. Detection of unmethylated XIST would be due to the extranumerical X with partial inactivation in TGCT patients (Kawakami et al. 2003). Therefore, the methylation profile of XIST might also be used as a potential biomarker for diagnosis of male patients suffering from TGCTs. More recently, advances in the use of Xist expression have led to its potential as a biomarker attributing to its high levels in the urine of patients with membranous nephropathy (MN). MN is an autoimmune-induced glomerular nephritis and the most common cause of nephrotic syndrome in humans (Huang et al. 2014). The primary experiments carried out in MN model mice showed significant upregulation of Xist and long noncoding gene Neat1 in tubular epithelial and glomerular cells. Interestingly, Xist expression levels detected in urine but not serum of MN mice are strongly correlated with the severity of MN. Significantly, this finding could be applied to identify human patients developing different types of glomerular nephritis, particularly as upregulation levels of XIST were detected in the urine, but not in serum samples, of human patients confirmed to have glomerular nephritis. Thus, XIST is a potential noninvasive biomarker to detect this disease.

There is support for examining the validity of underutilized molecular markers for diagnostic purposes. In the most recently released data for the GENCODE project (version October 22, 2014, http://www.gencodegenes.org), the number of long noncoding genes in humans is nearly comparable with the number of conventional protein-coding genes and outnumbers miRNA (Henry and Hayes 2012; Hayes et al. 2014). This offers a new variety of possible diagnostic biomarkers to investigate. Thus, combining the expression profiles of several lncRNAs, including XIST, might allow us to develop better and more reliable diagnostic tools in the future.

7 Conclusions and Future Perspectives

The discovery of novel lncRNAs and their varied functions is emerging at a high rate owing to the advancement of novel techniques used to detect and investigate them. Thus, our understanding of abundant lncRNAs has increased over the decades. In this chapter, we describe lncRNAs residing within the Xic in mice and implicated the role of X-inactivation in the initiation and progression of diverse disease conditions. To further understand the functional mechanism of lncRNAs in a variety of physiological processes, identification of the protein partners in the ribonucleoprotein complex would be inevitable. Recent introduction of techniques such as CHART, RAP, ChIRP (chromatin isolation of RNA purification), RIP-seq (RNA immunoprecipitation-sequencing), and CLIP (crosslinking and immunoprecipitation) help our understanding of the molecular interaction and function between lncRNA and protein partners (Hafner et al. 2010; König et al. 2010; Chu et al. 2011; Simon et al. 2011). Growing evidence linking pathological conditions and developmental anomalies to lncRNAs is emerging, especially in cancer development and progression. Thus, lncRNAs could become valuable therapeutic targets and promote the development of rapid diagnostic tools. Future studies into X-inactivation, a paradigm of lncRNA-mediated gene regulation, will provide additional insight to the molecular mechanisms behind lncRNA function, which will in turn further contribute to lncRNA research and its clinical and diagnostic applications.