In somatic cells of mammalian females, one of the two X chromosomes in each cell is inactivated at the blastocyst stage, compensating for the dosage difference between females and males who have only one X chromosome. X chromosome inactivation is a unique developmental regulatory event that results in the cis-limited inactivation of most of the entire chromosome. Lyon first hypothesized X inactivation in 1961, based on the observation that female mice heterozygous for X-linked traits showed mosaic expression.24 Beutler also proposed that females were mosaic on the basis of his studies of the X-linked enzyme glucose-6-phosphate dehydrogenase (G6PD) in males and females.25 The tenets of what is commonly referred to as the Lyon hypothesis are: (a) in normal females, only one of the two X chromosomes present is genetically active, the other being inactivated; (b) X inactivation occurs early in development; (c) the inactive X can be either maternal or paternal in origin and the choice of the X to become inactive is random with respect to parental origin and independent of the choice in other cells in the embryo; and (d) X inactivation is irreversible in somatic cells such that the inactive X in a particular cell remains inactive in all descendants of that cell.24
A number of reviews thoroughly summarize much of the original data on X inactivation, both in humans and in other mammals, and extensively discuss many of the theories proposed to explain various aspects of the inactivation process.26-30 This chapter focuses on the human X chromosome, concentrating on recent developments and advances in the molecular and genetic analysis and understanding of X inactivation, particularly as they relate to medical genetics.
Features of the Inactive X Chromosome
While the inactive X chromosome was first identified cytologically by its heterochromatic nature, there are many features distinguishing the active and inactive X chromosomes. Many of the characteristics appear to be interrelated, and it is as yet unclear which (if any) are causal of inactivation or are, instead, general features of inactive genes and chromatin.30 As well as yielding insight into the process of inactivation, these features are useful for identification of the inactive X. Even before Lyon correlated the heterochromatic sex chromatin body with the inactive X chromosome, Taylor had identified a late-replicating X chromosome in female cells and suggested that it might be heterochromatic.31 Studies of replication timing, originally by tritiated thymidine incorporation32-34 and later by 5-bromodeoxyuridine incorporation followed by fluorescence staining,35,36 showed that the inactive X chromosome starts and finishes replication later in the cell cycle than the active X in the same cell. Molecular studies of replicating timing based on isolation of newly replicated DNA from different stages of the cell cycle have largely supported earlier cytologic analyses, but importantly have provided substantially greater resolution, at the level of individual genes.37 The basis for chromosome-wide differences in replication between active and inactive X chromosomes remains unknown, but suggests an intimate relationship between the regulation of gene expression and the temporal and spatial control of chromosomal DNA replication.38,39
DNA methylation at the 5′ carbon of cytosine residues in cytosine-guanine dinucleotides (CpG) has long been considered an attractive means of maintaining X chromosome inactivation.40-42 DNA from the active X chromosome is more efficient in transforming cultured cells than is DNA from the inactive X,43 indicating that there is some characteristic difference in the DNA between the active and inactive X. Rodent/human somatic cell hybrid lines retaining the human inactive X chromosome can be treated with 5-azacytidine, an inhibitor of DNA methylation, to reactivate genes on the inactive X44 and restore the transforming competence,45 establishing that DNA methylation is responsible for the epigenetic change. Methylation-sensitive restriction enzymes, or, more recently, genomic sequencing, can analyze DNA methylation at specific sites on the X chromosome.46 These data indicated that CpG islands, regions of mammalian DNA unusually high in C + G content with a high frequency of the CpG doublet and found associated with the 5′ end of constitutively expressed genes,47 are methylated on the inactive, but not the active, X chromosome.30 These data are consistent with a role for DNA methylation in at least some aspect of the X chromosome inactivation process (see “Mechanisms of X Chromosome Inactivation” below).
The inactive X chromosome is generally more resistant to the action of DNA nucleases than is the active X chromosome,48,49 and specific sites hypersensitive to nucleases have been identified at the CpG islands of a number of X-linked genes on the active, but not the inactive, X chromosome.50,51 The binding of proteins that specifically bind methylated DNA52,53 and inhibit transcription54 may also contribute to the differential accessibility of the active and inactive X chromosomes to nucleases.55 As an additional correlate of its genetically repressed state, the chromatin of the inactive X chromosome was found to be complexed with histone molecules (histones H3 and H4) that are significantly less acetylated than the histone found on the active X chromosome in the same cell.56-58 Histone hypoacetylation is also a feature of X inactivation in marsupials.59 DNA methylation and histone hypoacetylation are intimately related phenomena and are likely to be partially redundant forms of chromatin modification and gene silencing. The chromatin of the inactive X is also characterized by a novel core histone that is related to histone H2A1, but which has a unique C-terminal sequence.60 This protein, termed macroH2A1, appears to be preferentially associated with the inactive X in female cells, and thus is the first protein component of the inactive X chromosome to be identified.60
The sex chromatin body, or Barr body, is often seen flattened against the inner surface of the nuclear membrane,2 and in situ hybridization with X chromosome “paint” probes shows that sequences from the inactive X tend to be peripheral, with a smoother and rounder shape than those of the active X in the same nucleus.61 The telomeres of the inactive X chromosome are reported to be much more closely associated with each other than are those of the active X chromosome, suggesting that the inactive X chromosome may be a loop formed by telomere association and attachment to the nuclear membrane.62 Recent work, however, has concluded that this formation reflects the free motion of the two arms relative to each other in interphase, rather than a looped structure.63
Although all these features reinforce the long-range and chromosomal nature of X chromosome inactivation, the causal nature of any of these characteristics in X chromosome inactivation remains unproven.
Activity of X-Linked Genes
While it is generally assumed that most X-linked genes are subject to inactivation and that this inactivation occurs at the transcriptional level, evidence for specific genes being subject to X inactivation is limited. Direct evidence for the transcriptional basis of X inactivation has been presented for only a limited number of genes.29,64,65 In addition, there are a growing number of genes that have been demonstrated to “escape” X inactivation and to be expressed from both active and (otherwise) inactive X chromosomes (see “The X Inactivation Profile of the X Chromosome” below). The existence of genes that are not subject to X inactivation was first hypothesized by Lyon for genes in common between the X and Y chromosomes in the pseudoautosomal pairing region to explain the phenotypic effect of X chromosome aneuploidies.66 Cytologic studies have reinforced this concept, as relatively early replicating regions on the inactive X chromosome have been hypothesized to contain genes being expressed from the inactive X.67,68
The original hypothesis of X inactivation was based on the observation of “patchy expression” of X-linked genes,24 and mosaic expression remains one of the most reliable cellular identifiers of genes that are subject to X inactivation. However, there are a number of different criteria (some stronger than others) that have been used as evidence for X-linked genes being subject to X inactivation. The criteria encompass both the direct assay of expression (of either RNA or protein) and the indirect monitoring of phenotype.
Mosaic Expression of Gene Product.
As originally hypothesized, X inactivation is a stable event that is clonally heritable, such that all progeny of a given cell will have the same X chromosomes active and inactive as in the progenitor cell. Therefore, females are a mosaic of two populations of cells in which different X chromosomes are active (Fig. 61-1). By culturing cells in vitro it is possible to isolate homogeneous clonal lineages in which one X is always active and the other X is always inactive. Definitive proof that a gene is subject to inactivation results from identifying clonal populations with different X chromosomes active and showing that one allele (but not both) of the gene in question is being expressed in each population. To unequivocally demonstrate that a clonal population has a specific X chromosome active or inactive requires the presence of a heterozygous genetic marker whose expression is known to be subject to X inactivation. The original characterizable alleles were null alleles or electrophoretic variants, as in the A and B alleles of G6PD; 69 however, more recent studies have revealed expressed DNA sequence variants as well.70-73 Different clonal populations of cells were isolated from females heterozygous for steroid sulfatase (STS) deficiency (X-linked ichthyosis) and also heterozygous for a G6PD polymorphism. Both G6PD positive and negative clones (that is, clones with one or the other X chromosome active) were found to express STS activity, despite the fact that the mutant STS allele was nonfunctional.74 These data demonstrated that the normal STS gene was expressed even when present on the inactive X chromosome and thus “escapes” X inactivation.
Representation of X chromosome inactivation in female somatic cells. Inactivation early in development is random, with an equal probability a priori that the maternal or paternal X will be active or inactive. However, the distribution of cells expressing alleles from one or the other X chromosome can vary widely, owing either to postinactivation selection against one of the cell types (usually the one expressing a mutant X-linked allele) or to stochastic effects of the choice or subsequent cell proliferation. The phenotype observed in females heterozygous for X-linked traits can also vary widely, with increasing clinical expression or severity correlating with the proportion of cells expressing a mutant allele from the active X chromosome.
It is not always possible to identify doubly heterozygous disease carriers. In such cases, tissues or cultured cells from females heterozygous for an X-linked disease have been shown to be mosaic for the gene product,69,75-94 providing evidence that the gene is subject to X inactivation29 (Table 61-1).
Table 61-1: Selected X-Linked Disorders Demonstrating X Chromosome Inactivation* |Favorite Table|Download (.pdf) Table 61-1: Selected X-Linked Disorders Demonstrating X Chromosome Inactivation*
|Disorder ||Gene ||Map Location |
|X-linked lactic acidosis ||Pyruvate dehydrogenase, E1α subunit ||Xp22.1 |
|Duchenne muscular dystrophy ||Dystrophin ||Xp21.2 |
|Androgen insensitivity syndrome ||Androgen receptor ||Xq12 |
|Fabry disease ||α-Galactosidase ||Xq21.3 |
|Gout, excessive purine production ||PP-ribose-P synthetase 1 ||Xq21-q26 |
|Lesch-Nyhan syndrome ||HPRT ||Xq26 |
|G-6-PD deficiency ||G6PD ||Xq28 |
|Hunter syndrome ||Iduronate sulfatase ||Xq28 |
|Chronic granulomatous disease ||Cytochrome b-245, β polypeptide ||Xp21.1 |
|Ornithine transcarbamylase deficiency ||Ornithine transcarbamylase ||Xp21.1 |
|Hemolytic anemia ||Phosphoglycerate kinase 1 ||Xq13.3 |
|Glycogen storage disease ||Phosphorylase kinase, α2 subunit ||Xp22 |
|Fragile X mental retardation ||FMR1 ||Xq27.3 |
|Adrenoleukodystrophy ||ATP-binding transporter ||Xq28 |
In the absence of X inactivation, females would be expected to have twice as much product from X-linked genes as males. Early studies demonstrating that the products of X-linked genes such as factor IX and G6PD were not twice as abundant in females as males were used to support hypotheses of X chromosome inactivation.25 Recently, dosage at the RNA level was used to argue that the X-linked zinc finger (ZFX) and ribosomal protein S4 (RPS4X) genes escape X inactivation.95,96 Such studies must be rigidly controlled for the quantity and quality of RNA being analyzed and are complicated by the fact that expression of “active” genes from the inactive X may be incomplete.97-99
Gene Expression in Somatic Cell Hybrids.
The isolation of rodent/human somatic cell hybrids that retain the human inactive X without the active X permits the assessment of inactivation for any gene expressed in the hybrid system. By using similar hybrids that retain the active X chromosome, the normal level of expression for the hybrid system can be identified. This method does not require allelic variants of the gene to be studied, but is limited to those genes that are expressed in the hybrids (which are generally of fibroblastoid origin).65,100 Expression of an increasing number of X-linked genes has been analyzed in such systems, providing direct experimental evidence for or against a gene being subject to X inactivation (see “The X Inactivation Profile of the X Chromosome” below).
Mosaic Clinical Expression.
There are many disease genes on the X chromosome that have been identified because of their distinctive X-linked inheritance pattern but that have not been cloned, and whose protein product is not known. There is indirect evidence that many of these genes are subject to X inactivation, although definitive proof requires better knowledge of each gene and its product.
If expression of a gene subject to inactivation is localized, then “patches” of expressing and nonexpressing tissue may be observed (Fig. 61-1). Macroscopically visible patches are identifiable in some heterozygous female carriers for a few X-linked diseases101-106 (Table 61-2). In other diseases, two populations of cells have been identified. These areas of expressing and nonexpressing tissue are assumed to be the result of expression and nonexpression of the disease allele (Fig. 61-1), implying that the gene is subject to X inactivation. However, in none of the cases listed in Table 61-2 (and many others not listed) has it been demonstrated that the patches are, in fact, related to presence or absence of the X-linked gene product.107-112 It is possible that, in some cases, such patterns reflect phenomena unrelated to X chromosome inactivation, and interpretations based on an assumption of X inactivation must be made with caution.
Table 61-2: Selected X-Linked Disorders Demonstrating Mosaicism in Vivo in Heterozygotes, Possibly Reflecting X Chromosome Inactivation |Favorite Table|Download (.pdf) Table 61-2: Selected X-Linked Disorders Demonstrating Mosaicism in Vivo in Heterozygotes, Possibly Reflecting X Chromosome Inactivation
|Disorder ||Gene (If known) ||Mosaicism in Heterozygotes |
|Chondrodysplasia punctata ||Arylsulfatase E* ||Pigmentary changes |
|Amelogenesis imperfecta ||Amelogenin ||Mottled tooth enamel |
|Ocular albinism || OA1 ||Irregular fundus pigmentation |
|Retinitis pigmentosum ||Several X-linked forms ||Patchy retina, tapetal reflex |
|Hypohidrotic ectodermal dysplasia || EDA ||Patchy sweat gland distribution |
|Lowe syndrome ||Inositol polyphosphate-5-phosphatase related ||"Snowflake" lenticular opacities |
|Incontinentia pigmenti ||— ||Swirling melanin, incomplete dentition |
|Hypertichosis ||— ||Patchy, asymmetric hirsutism |
|Choroideremia ||Rab geranylgeranyl transferase* ||Irregular fundus pigmentation |
|Sideroblastic anemia || ALAS2, delta-aminolevulinic synthetase ||Two populations of RBC |
|Alport syndrome ||Collagen type IV, α5 ||Heterogeneous staining with antiglomerular basment membrane antibody |
Because of the long-assumed generality of the initial X chromosome inactivation hypothesis, there is a tendency to interpret all tissue or cell mosaicism for X-linked disease phenotypes in terms of X inactivation. The increasing number of other genes now known to escape X inactivation (see “Genes that Escape X Chromosome Inactivation” below) renders such assumptions unreliable in the absence of direct experimental evidence (as for those disorders listed in Table 61-1).
Clonal Selection in Vivo.
In females heterozygous for some X-linked diseases, in tissues where the gene product is required, cells that have the normal allele inactivated can be at a proliferative disadvantage against cells with the normal allele on the active X.113,114 This results in selection for a population that has only one X chromosome active (Fig. 61-1) and suggests indirectly that the gene is not expressed (or is very poorly expressed) from the inactive X chromosome. For such selection to occur, the defect must be severe and expressed while the affected tissue is still dividing. Such characteristic nonrandom or “skewed” X chromosome inactivation has been observed for a number of disorders, including several X-linked immunodeficiencies,115 Lesch-Nyhan disease,82,83 incontinentia pigmenti,116,117 dyskeratosis congenita,118-120 several disorders of X-linked mental retardation,121-125 and severe cases of G6PD deficiency.126 Importantly, such in vivo selection or differential growth advantage almost invariably results in a clonal population of cells in which the mutant allele is on the inactive X; thus, carrier females for such disorders are characteristically unaffected.114 For some of these disorders, the presence of nonrandom X inactivation has been used in the past for carrier detection.120,124,127,128 However, with the increasing availability of the cloned genes, and with the precise molecular defects in individual families being defined, direct DNA diagnosis is appropriate whenever possible.
Variable Phenotype in Heterozygotes.
With the usual exception of those disorders discussed in the previous section, in which nonrandom inactivation is a characteristic of the carrier state, carriers of any X-linked disease allele may show variable or partial expression resulting from different proportions of the affected tissue having the mutant allele inactivated (Fig. 61-1). Because the initial choice of inactivating the maternal or paternal X chromosome is random early in development, variable mosaicism is expected and would occasionally result in a symptomatic carrier female, a finding often termed “unfortunate lyonization.” For a wide variety of X-linked diseases, this has been observed129-138 and can be used as an indirect indication of X inactivation. However, the demonstration of a variable or intermediate phenotype in heterozygotes is unreliable by itself as a criterion for assessing X inactivation, as expression can be variable due to other genetic and/or nongenetic factors; this is particularly true for X-linked disorders in which the phenotype of affected hemizygous males is variable as well. Many well-established X-linked disorders demonstrate partial expression or occasional complete expression in heterozygous females16 (Table 61-3). Whether these genes are subject to X chromosome inactivation must await confirmation from direct biochemical or molecular studies following firm identification of the gene in question.
Table 61-3: Selected X-Linked Disorders Typically Demonstrating Partial Expression of Clinical Phenotype in Heterozygotes |Favorite Table|Download (.pdf) Table 61-3: Selected X-Linked Disorders Typically Demonstrating Partial Expression of Clinical Phenotype in Heterozygotes
|Disorder ||Phenotype in Heterozygotes |
|Adrenoleukodystrophy ||Approximately 15% of heterozygotes are symptomatic |
|Alport syndrome ||Hematuria less severe than in males |
|Charcot-Marie-Tooth disease ||Less severe than in males |
|Ornithine transcarbamylase deficiency ||Hyperammonemia; wide range of severity |
|Fragile X mental retardation ||Approximately one-third affected; wide range of severity |
|Ichthyosis* ||Skin abnormalities in some carriers |
|Kallmann syndrome* ||Partial anosmia |
|Choroideremia* ||Abnormal visual field |
|Lactic acidosis ||Proggessive neurodegenerative disease; wide range of severity |
|Coffin-Lowry syndrome ||Abnormal digits, mild mental retardation, less severe than in males |
|Duchenne muscular dystrophy ||Mild myopaty in up to 5% of carriers |
For those disorders in which careful analyses were carried out, the data suggest that extreme phenotypic variability in heterozygotes does correlate with clonal variability in X inactivation (e.g., fragile X mental retardation,139 ornithine transcarbamylase deficiency,92,140,141 pyruvate dehydrogenase deficiency,142,143 and Alport syndrome144). For example, pyruvate dehydrogenase (PDH) E1a subunit deficiency presents with a relatively broad spectrum of clinical phenotypes in females, ranging from severe lactic acidosis during the newborn period to a progressive neurodegenerative disease with prolonged survival (see Chap. 100). The severity of clinical presentation and the level of residual PDH activity correlates well, but not perfectly, with the proportion of cells in which the mutant allele is on the active X.142-143 In this and other conditions, however, interpretation is complicated by uncertainty about whether the cells under study are relevant to the disease progression (e.g., leukocytes in fragile X mental retardation139) or representative of the tissue being sampled (e.g., liver biopsies in ornithine transcarbamylase deficiency91). It may be that different tissues in a given individual are marked by different proportions of one or the other X being active.
At least four of the conditions listed in Tables 61-2 and 61-3 involve X-linked genes that have been shown to escape X chromosome inactivation. Approximately one-fourth of carriers of X-linked ichthyosis are reported to show mild skin defects145 despite clear evidence that the STS gene at least partially escapes inactivation.74,97 Heterozygotes for X-linked Kallmann syndrome are reported to have mild anosmia and, in some cases,146 abnormal sexual development.16 Yet, studies in somatic cell hybrids have clearly demonstrated that the Kallmann syndrome gene escapes X inactivation.147,148 The third example of a gene escaping X chromosome inactivation is X-linked choroideremia, due to a deficiency of Rab geranylgeranyltransferase, one of a family of GTP-binding proteins regulating vesicular traffic.149 Carrier females often have patchy pigmentation and show uneven degeneration of the retinal pigment epithelium and choroid,110 generally interpreted in terms of random X chromosome inactivation.16 Recent studies, however, have demonstrated that this gene is expressed from both active and inactive X chromosomes and, therefore, escapes X inactivation, at least in a proportion of females.71 The fourth example is X-linked chondrodysplasia punctata (Table 61-2), which can be caused by mutations in one of a family of sulfatase genes localized to Xp22.3, each of which escapes inactivation.23 Thus, in both choroideremia and chondrodysplasia punctata, the clinical results obtained with heterozygotes are misleading, and the patchy pigmentation observed in such X-linked disease carriers must have a different underlying basis representing the degenerative nature of the diseases.
In all these cases, the partial expression of the defect in heterozygotes may reflect incomplete dominance, rather than direct effects of X chromosome inactivation. It may be that clinical expression in each of these disorders has a fairly high threshold, in contrast to typical recessive defects in which the threshold for phenotypic expression is thought to be quite low. On the other hand, it is presently impossible to rule out the possibility that X inactivation contributes, because even partial repression of gene expression from the inactive X (as demonstrated, for example, for STS activity97 and for the choroideremia gene71) may be sufficient to create a functionally significant difference between cells expressing the normal allele from the active X and cells expressing the normal allele from the inactive X.
Nonrandom (Skewed) X Chromosome Inactivation
There are many instances in chromosomally normal females in which X chromosome inactivation appears to be nonrandom rather than random; that is, tissues of an individual consist predominately (if not exclusively) of cells expressing the same X chromosome, rather than being a mosaic of cells expressing the two different X chromosomes. While in some extreme cases this situation is believed to reflect postinactivation selection (see “Clonal Selection in Vivo” above), in many situations this skewed distribution of the two cell types presumably reflects the stochastic variation expected for a random event operating on a relatively small number of progenitor cells.113,114
X chromosome inactivation occurs in the late blastocyst stage of embryogenesis, at about the 32- to 64-cell stage.26 The number of progenitor cells for individual tissues at this time is presumed to be less, as estimated by clonality studies. For example, the number of hematopoietic stem cells has been estimated to be less than 20.150-153 Given these relatively small numbers, one predicts on the basis of the binomial distribution that a significant number of females would show distinctly skewed X inactivation purely by chance. Skewed X inactivation can also arise from a number of cryptic selective events early in embryogenesis. For example, confined placental mosaicism may reflect selection against trisomic cells in the embryo itself; in female embryos, this may result in a reduction in the number of embryonic precursor cells at the time of X inactivation and a concomitant increase in the proportion of embryos that show skewing. Indeed, an increased frequency of skewed X inactivation has been detected in cases of confined placental mosaicism relative to controls, suggesting that skewing secondary to confined placental mosaicism may be a contributor to the few percent of newborns that show extreme skewing.154
A useful assay to measure patterns of X inactivation is a DNA methylation polymorphism assay to distinguish active from inactive X chromosomes.155 In such assays, the two X chromosomes in a sample are distinguished by a DNA polymorphism, while the active and inactive X chromosomes are distinguished by characteristic DNA methylation differences (see “Features of the Inactive X Chromosome” above). Samples in which the two X chromosomes are equally methylated at a particular DNA locus demonstrate random X inactivation, whereas those in which only one of the polymorphic bands is methylated demonstrate skewed inactivation. A number of such assays have been developed at different X-linked loci.155-159 The most informative and widely used assay is based on the androgen receptor (AR) locus,159 and this highly quantitative assay has found widespread usage in both research and clinical settings (Fig. 61-2).
Assay of X inactivation ratio based on androgen receptor ( AR ) DNA methylation differences between the active and inactive X chromosomes, and using the assay of Belmont and colleagues.159 Heights of major peaks on the electropherograms correspond to relative proportion of DNA with one or the other allele at the AR locus. Top panels correspond to total DNA (both active and inactive X chromosomes). Bottom panels correspond to the inactive X chromosome only. DNA has been digested with HpaII, which cleaves unmethylated DNA on the active X, leaving only methylated DNA on the inactive X chromosome. Left: Normal 46,XX female with approximately 60:40 X inactivation ratio. Center: Balanced carrier of X;11 translocation, showing preferential inactivation of the normal X and a 0:100 X inactivation ratio. Right: Female with unbalanced X;14 translocation, showing preferential inactivation of the abnormal der(X) chromosome and a 100:0 X inactivation ratio.
Using such assays, significantly skewed patterns of X chromosome inactivation were demonstrated in fibroblasts or leukocytes in 10 to 20 percent of normal control females.120,122,152,153,159-166 These studies, which in total surveyed nearly 1000 females, are in general agreement with other estimates of skewed X inactivation based on cell cloning151,167 or G6PD polymorphisms.168 These data support nicely the predictions of a binomial distribution, but contradict the often-quoted statement that normal females “should” show 50:50 random X chromosome inactivation. While this may be true on average, there is no such expectation for a given individual or in a given cell lineage. In fact, ratios of 60:40 or 70:30 are common (30 to 40 percent of females in most studies) and more extreme ratios (80:20 or greater) are not rare (at least 10 percent of normal females). Importantly, the proportion of females showing a highly skewed pattern of inactivation (e.g., more extreme than 90:10) increases markedly with age;152,153,163,164 thus, studies of X inactivation patterns need to be carefully controlled for this variable to establish population norms.
Given that a significant proportion of females in the general population show skewed X inactivation, it can be predicted that an equivalent proportion of females heterozygous for X-linked defects should also demonstrate skewed X inactivation. In some cases, therefore, heterozygous females with the mutant allele present on the active X chromosome in a high proportion of cells in a relevant cell lineage are predicted to show symptoms equivalent to those of affected hemizygous males. Indeed, as described earlier, such affected females are occasionally detected for many X-linked conditions.129-138,144,169-172
Recognition of the frequency of skewed X inactivation has significant implications for carrier-detection assays that rely on measurements of X-linked enzyme activities, because samples from females with skewed X inactivation may appear to have normal or near normal activity, despite carrying a mutant allele on the X that is predominantly inactive. Such diagnostic assays are also prone to selective effects in vitro that can result in significant clonality in tissue culture. Studies with both transformed lymphoblasts173 and chorionic villus samples142 have demonstrated a high proportion of samples with extreme skewing of X inactivation, suggestive of clonal outgrowths in culture; thus these samples are likely to be unreliable for diagnostic studies that require an accurate reflection of an individual's X inactivation pattern.
Familial Skewed X Chromosome Inactivation.
In most of the examples cited above, skewed X inactivation is likely to be attributable to stochastic factors. However, in a few instances, for disorders not typically characterized by nonrandom inactivation, such skewing appears to be familial, suggesting the possibility of inherited factors determining the randomness of X chromosome inactivation.113,114 Multiple affected females have been described in families with Duchenne muscular dystrophy,130,174 hemophilia B,175 Lesch-Nyhan disease,176 hemophilia A,177 Fabry disease,178 and Wiskott-Aldrich syndrome.179 While such families are rare and while the basis for familial skewed X inactivation has not been determined definitively in any of the reported families to date, they are important for considering the genetic control of X chromosome inactivation.
Preferential X chromosome inactivation is under control of an X-linked locus in the mouse,30,113,114,180 and thus it might be expected that a similar, perhaps homologous, locus would control X inactivation in humans. Evidence of statistically significant correlations of X inactivation patterns in first-degree female relatives has been presented,181,182 consistent with such a genetically determined trait. In sufficiently large families, it should be possible to genetically map such a locus.123,160,161,166 The X inactivation center itself (see “The X Inactivation Center and the XIST Gene” below) is an attractive candidate for a gene altered in families demonstrating skewed X chromosome inactivation. In the largest study reported to date, Sapienza and colleagues examined X inactivation in 264 females from 38 families and reported a highly significant sister-sister correlation in lymphocytes, consistent with a cis-acting or an imprinting effect.182 Five families had at least two sisters with skewed X inactivation patterns more extreme than 80:20. The effect could be mapped to genetic markers near the X inactivation center and in Xq25-26.182
Skewed X Chromosome Inactivation in Twins.
An oddity of twinning is the apparently common occurrence of discordant phenotypes in monozygotic twin females heterozygous for X-linked disorders. At least six sets of twins discordant for Duchenne muscular dystrophy183,184 and sets of twins discordant for other X-linked conditions, such as Hunter disease185 and red-green color blindness,186 have been described. Nonrandom inactivation of the X chromosome carrying the normal allele in one but not both twins has been suggested as the likely basis, and evidence favoring this mechanism has been obtained in several discordant twin pairs using the DNA methylation assays described earlier. In most instances studied, X inactivation in the phenotypically normal co-twin was skewed in the opposite direction; that is, the X chromosome carrying the mutant allele was inactive in most cells.183,185,186 A different result was reported for monochorionic monozygotic twins, however; the monozygotic twins showed highly correlated X inactivation patterns, indicating that extreme asymmetric splitting of the embryo is not a common mechanism of twin formation, at least not when ascertained because of an affected twin.187 The available data indicate that commitment to X inactivation precedes the twinning event by at least several cell divisions.187-189
Structurally Abnormal X Chromosomes.
In all of the above instances, the X chromosomes were structurally normal, and X inactivation was considered in the context of 46,XX karyotypes. Skewed X chromosome inactivation is nearly always observed in individuals with abnormal karyotypes involving structural abnormalities of an X chromosome, such as interstitial or terminal deletions, isochromosomes, or ring chromosomes. In almost all such patients, the structurally abnormal chromosome is inactive in the vast majority of, if not all, cells, presumably reflecting postinactivation selection against those cells in which the normal X was inactive and the abnormal X active113,114 (Figs. 61-1 and 61-2). As such cells are missing any expressed copy of genes located in the deleted portion of the X, they are presumably at a selective disadvantage. Skewed X inactivation thus has the effect of minimizing the potential genetic imbalance of such abnormalities.190,191 Because of this preferential inactivation of the abnormal X, X chromosome anomalies are tolerated better than similar abnormalities of autosomes and consequently are more frequently observed. In many instances, the phenotype of females with abnormal X chromosomes is that of Turner syndrome, although more severe phenotypes involving mental retardation and multiple congenital anomalies can be observed (see “Cytogenetic Disorders of the X Chromosome” below).
While completely skewed inactivation of the abnormal X is typical of such karyotypes, it is not invariant. Random X inactivation has been documented in several cases of deleted, duplicated, or insertion X chromosomes.192-196 As the clinical consequences can be severe when the unbalanced X chromosome is the active X in a significant proportion of cells, this potential outcome should be considered in counseling and molecular assays of X inactivation performed as appropriate.
Nonrandom X inactivation is also observed in the case of X;autosome translocations.113,114,190,191 If such a translocation is balanced, the normal X chromosome is usually preferentially inactivated (Fig. 61-2), and the two parts of the translocation chromosome remain active, probably as a result of selection against cells in which autosomal genes located adjacent to inactivated X chromosomal material are similarly inactivated.190,191,197 As expected for a genetically balanced condition, carriers of most balanced translocations are usually clinically unaffected and are ascertained either prenatally or because of clinical defects in their unbalanced offspring. Balanced rearrangements (translocations or inversions) involving a region in the middle of the long arm, however, are often accompanied by ovarian failure or amenorrhea,198,199 suggesting that normal ovarian development and maintenance requires one or more genes in this region. In rare instances, usually (but not always) involving X chromosome breakpoints near the ends of the two arms in bands Xp22 or Xq28, the normal X can be active in a proportion of cells, with one of the translocation products (but not both) being inactive. These cells appear to be functionally disomic for the portion of the X chromosome that is active in two copies (one on the normal X and one on the active, translocated X), and such findings are associated with mental retardation and other clinical anomalies.197,200-204 Again, this potential outcome must be recognized in counseling and molecular studies of X inactivation considered for any karyotype with an abnormal X.
In the unbalanced offspring of a balanced carrier, the translocation product is invariably inactivated, while the normal X is active.190,191 These skewed patterns of inactivation have the general effect of minimizing, but not eliminating, the clinical consequences of the particular chromosomal defect. Depending on the extent to which inactivation spreads from the X chromosome to adjacent autosomal material, the clinical phenotype usually depends on the amount and nature of extra autosomal material present. Spreading of inactivation is a well-recognized, but poorly understood phenomenon. Spreading can partially or even completely ameliorate the expected phenotype due to partial autosomal trisomy;190,191,205 failure of spreading, however, can result in severe clinical symptoms.58,206
In rare instances, an unbalanced translocation chromosome is one that lacks the X inactivation center in Xq13 (see “The X Inactivation Center and the XIST Gene” below). In such cases, the translocation product cannot undergo X inactivation and the result is functional disomy. A portion of the X is expressed from two copies, one on the normal active X and one on the translocated portion of the X separated from the X inactivation center.202,207 The existence of such patients and the clinical abnormalities associated with functional disomy reinforce the significance of and requirement for X chromosome inactivation in normal female development.
One consequence sometimes observed in balanced carriers of X;autosome translocations is that the break itself may cause a mutation by disrupting a gene on the X chromosome at the site of the translocation. Such patients present with a severe form of the relevant X-linked disorder, typical of the phenotype seen in hemizygous males, because the only normal copy of the particular gene (on the normal X) is inactivated in most or all cells as a result of skewed X inactivation. Several X-linked genes were mapped to specific regions of the X chromosome when a typical X-linked phenotype was found in a female who then proved to have an X;autosome translocation. Although expression of an X-linked disease gene can occasionally be seen in chromosomally normal females due to skewed X inactivation (see discussion above), high-resolution chromosome analysis is indicated if a female patient manifests an X-linked phenotype normally seen only in males. The finding of a balanced translocation can often explain the phenotypic expression and show the gene's probable map position on the X chromosome.208
Genes that Escape X Chromosome Inactivation
Not all genes on the X chromosome are subject to X chromosome inactivation. Approximately 40 genes on the X chromosome are reported to escape X inactivation, and this number is certain to grow as more genes are identified and studied.64,65,71,74,100,209 Extrapolating the data to date on genes escaping X inactivation suggests that there may be hundreds of genes that are expressed from both active and inactive X chromosomes. The operational definition of a gene escaping X inactivation is that there is significant expression from the inactive X, although there may not be as much expression as from the active X.71,97-99 Expression from the inactive X may result either from the gene being refractory to the X inactivation event early in development, or from the inactivation of the gene being unstable (see “Mechanisms of X Chromosome Inactivation” below).
Many, but not all, of the genes that escape X inactivation have a homologous Y-linked locus,11,17,19,64,65,95,96 consistent with the proposed origin of the sex chromosomes from an original set of autosomes.8,210-212 Comparison of the evolutionary history of different XY-homologous genes indicates that the determination of whether an X-linked gene escapes inactivation or is subject to inactivation is independent for each gene, possibly as an adaptation to the decay of the homologous Y-linked gene.19,210,213 If the Y chromosome homologue is expressed, then dosage equivalence between males and females can be maintained without X inactivation.66 However, for those genes whose Y homologue is not expressed, or when there is not a Y homologue, there must either be an alternative mechanism of dosage compensation or the dosage imbalance between the sexes must be tolerated. A novel variant of dosage compensation has been reported for a synaptobrevin-like gene in the Xq28 pseudoautosomal region. The X-linked homologue is subject to inactivation, while the Y-linked homologue (derived from the inactivated X by meiotic exchange during spermatogenesis) is inactive;214 thus, dosage equivalence is still maintained between XX and XY individuals.
In most cases, the clinical significance of genes that escape X inactivation is unknown. Of the identified noninactivated genes, only four are associated with known X-linked clinical disorders (X-linked ichthyosis, choroideremia, chondrodysplasia punctata, and Kallmann syndrome) (Tables 61-2 and 61-3). The other genes, however, are candidates to explain clinical symptoms in cases of X chromosome aneuploidy, because in such cases their gene products may be either underexpressed or overexpressed relative to karyotypically normal females.
The X Inactivation Profile of the X Chromosome.
As of early 1999, approximately 200 genes had been analyzed for their X inactivation status using the somatic cell hybrid approach described earlier (see “Gene Expression in Somatic Cell Hybrids” above)209 (Fig. 61-3). This number corresponds to about 10 percent of the total expected number of genes on the chromosome.15 In total, nearly 40 genes have been described that escape inactivation. While the genes currently $ known to escape X inactivation are distributed widely along the X chromosome,64,65,95,96,100,209 the distribution appears decidedly nonrandom. First, the pseudoautosomal regions on Xp and Xq contain a number of genes that escape inactivation, and, remarkably, the most distal 10 to 15 Mb of DNA on Xp (consisting of both pseudoautosomal and X-specific regions) contain only genes that escape inactivation (Fig. 61-3). This may reflect the evolutionary origins of this part of the X and suggests that the entire region may have been pseudoautosomal at one point during evolution.8,212 Second, there are clearly many more genes on Xp that escape inactivation than on Xq. Even excluding the pseudoautosomal region, nearly one-third of all Xp genes tested escape inactivation, as compared to only 2 percent of Xq genes tested.209
X inactivation profile of the human X chromosome. Each symbol corresponds to a different X-linked gene, assayed by somatic cell hybrid system.65,100,209 Squares indicate pseudoautosomal loci. A much higher proportion of X-linked genes on Xp escape X inactivation than do genes on Xq.209 This first-generation profile corresponds to an estimated 10 percent of all X-linked genes.
Many of the regions containing genes escaping X inactivation have been described as being early replicating36,67,68 and underacetylated56 on the inactive X chromosome, suggesting that there may be large, cytologically visible domains of genes that escape X inactivation. Consistent with this prediction, a contiguous cluster of at least six genes that all escape inactivation has been described in Xp11.2, one of the regions suspected on cytologic grounds to escape inactivation.215
The nonrandom distribution of genes that escape inactivation suggests that different regions of the chromosome may be organized differently. Lyon has suggested that the density of members of the L1 family of long, interspersed repeated DNA elements on the X chromosome may facilitate the spread of X inactivation along this chromosome,216 a hypothesis that can be examined in more detail as the Human Genome Project progresses. Speculation that the X may have a unique organization is supported by comparative mapping studies that have demonstrated the independent and more recent evolutionary addition of much of the short arm of the X to the modern-day X chromosome, while the long arm (“the ancestral X”) has been X-linked for at least 80 to 130 million years.211,212 This model suggests that sequences on the ancestral X may share properties that render them more likely to be subject to X inactivation than either the more recent Xp sequences or autosomal sequences. Indeed, study of X inactivation in one X;autosome translocation has demonstrated that the frequency of autosomal genes that escape spreading of inactivation is equal to the frequency of Xp genes that escape X inactivation and is, therefore, at least an order of magnitude greater than the frequency of Xq genes that escape inactivation.205 Whether this is a property of all autosomes requires similar investigation of additional translocations.
The clinical consequences of the X inactivation profile shown in Fig. 61-3 are likely to be substantial. First, the existence of a significant proportion of all X-linked genes that escape inactivation provides a potential explanation for the abnormal phenotypes associated with cases of X aneuploidy and for the well-recognized observation that the phenotype worsens as the number of X chromosomes increases (i.e., 46,XX vs. 47,XXX vs. 48,XXXX vs. 49,XXXXX). This latter finding cannot be explained easily if all X-linked genes undergo inactivation. Second, the profile predicts that imbalance of Xp will be much more poorly tolerated than imbalance of Xq, a prediction that should be accounted for in genetic counseling. Third, it may be that specific phenotypic features of X abnormalities can be associated with overexpression of particular genes. While this concept is not unique to the X chromosome by any means (see Chap. 63), the fact that relatively few X-linked genes escape inactivation reduces the number of potential candidates whose abnormal gene dosage may underlie a particular constellation of clinical findings. In this respect, karyotype-phenotype correlations may be more direct for partial X aneuploidy than for autosomal defects.
The X Inactivation Center and the XIST Gene
Both original and more recent investigations of X chromosome inactivation have emphasized the role of an X-linked locus required in cis for inactivation to occur.191,217-220 The principal evidence for such a locus derives from the observation that when an X chromosome is involved in a balanced translocation to another chromosome, only one of the two products can undergo inactivation, even if inactivation of the other product would have maintained genetic balance.28-30 The simplest interpretation of such data is that the X inactivation signal requires the presence in cis of a critical locus, generally called the X inactivation center (XIC). X inactivation requires the presence of at least two copies of the XIC on different X chromosomes, because males with a duplication of the portion of the X containing the XIC do not demonstrate X inactivation of the duplicated region.221-223 This, together with the observation that altering the relative dosage of autosomes and X chromosomes (as observed in triploid individuals, for example) can influence the number of active or inactive X chromosomes,224 argues strongly that the XIC is involved as part of a mechanism that “counts” or responds to the number of X chromosomes to ensure the appropriate activity state of X-linked genes, normally one active X per two sets of autosomes.225,226
Mapping the X Inactivation Center.
The region on the X chromosome that contains the XIC has been defined cytogenetically and molecularly by analyzing X chromosome rearrangements that are subject to inactivation.191,219,220,227,228 There is no convincing or substantiated evidence for more than a single XIC, and no chromosome without the XIC has been shown definitively to undergo X inactivation. The XIC has been mapped to a small region of <1 Mb within band Xq13.2.220,228 This candidate region is defined by two X chromosome rearrangements, one extending proximally and one distally to the interval. Because genetic imbalance and abnormal phenotypes result from the failure to inactivate structurally abnormal X chromosomes that do not contain this locus, other regions of the X chromosome are apparently not sufficient, in the absence of the XIC, to sustain X inactivation.197,201,202,207 However, such examples of functional X disomy are relatively rare and do not account for the entire remainder of the X chromosome. Nonetheless, the series of abnormal inactive X chromosomes that have been used to map the XIC are sufficient to rule out any unique locus required for X inactivation other than the XIC itself.
Alternatively, this region may be absolutely required for survival, rather than having a direct role in X inactivation per se. Indeed, it is intriguing that deletion of this region has never been reported, either on active or inactive X chromosomes, suggesting a possible function in chromosome integrity or an essential role in early development in both males and females, independent of (or in addition to) any role in X inactivation.
Attempts to map the XIC using patient material have been complemented by ectopic induction of X inactivation by introducing copies of the mouse Xic (including the mouse Xist gene, see below) to murine embryonic stem cells and allowing differentiation and X inactivation to occur in vitro. Detailed deletion analysis of a 450-kb region within the Xic has shown that X inactivation can be recapitulated in this system by a fragment of the Xic as small as 35 to 80 kb.229-231
The XIST Gene and Its Role in X Chromosome Inactivation.
A search for candidate genes mapping to the XIC region led to identification of the XIST gene.232 The XIST gene gets its acronym from the observation that the gene is expressed only from the inactive X chromosome and not from the active X chromosome [thus, inactive X (Xi)-specific transcripts]. In somatic tissues from karyotypically normal individuals, expression is female-specific. However, males with an inactive X chromosome (e.g., 47,XXY individuals) do express XIST, and all normal males express XIST during spermatogenesis when inactivation of the single X chromosome is observed.233-235 XIST expression in females is turned off at about the time of reactivation of the inactive X chromosome during oogenesis.234 The map location and unique expression pattern of XIST strongly suggest that it is either involved in or directly affected by the process of X chromosome inactivation.220,232
Similar data have been reported for the homologous mouse gene, Xist. Xist maps to the corresponding mouse Xic interval and is expressed essentially specifically from the inactive X chromosome.236,237 Both in humans and in mice, the inactive X-specific pattern of expression correlates with methylation of CpG dinucleotides at the 5′ end of the gene only on the active X (i.e., only on the X chromosome that does not express the gene),238-241 consistent with methylation studies of other X-linked genes showing differential methylation between expressed and nonexpressed copies of the gene (see “Features of the Inactive X Chromosome” above). In the mouse (where such studies can be performed), the accumulation of Xist RNA transcripts precedes X chromosome inactivation early in development242 and is likely, therefore, to be a cause, rather than a consequence, of X inactivation. As mentioned earlier, transgenic experiments have demonstrated that the region of the Xic including the Xist gene is sufficient for X inactivation to occur in early development, although the exact extent of the regulatory determinants of Xist function is unknown.229-231,243 The promoters of the human and mouse XIST/Xist genes appear to be unremarkable,238,239,244 suggesting that XIST is under negative, rather than positive, control.244 Thus, the effector sequences responsible for XIST action presumably lie outside the promoter itself, a conclusion that is supported by the finding that ectopic copies of Xist introduced as part of transgenic experiments appear to function only when introduced in multiple copies.229,230,242,243,245 This result may indicate that the fully functional endogenous XIC must have as yet unidentified properties or sequences that are not fully recapitulated by a single copy of the Xist gene itself.245
RNA transcripts from the XIST gene appear to be noncoding, because multiple stop codons were encountered in all potential reading frames, both in the human and mouse genes.231,246,247 XIST transcripts are largely confined to the nucleus, the human XIST RNA being associated with the Barr body, the heterochromatic manifestation of the inactive X chromosome in female interphase nuclei58,247,248 (Fig. 61-4). Similar data have been reported for Xist transcripts on the murine X chromosome, with the additional striking finding that the transcripts remain associated with the inactive X during early metaphase.226,243 Xist RNA is excluded from autosomal material in at least some X;autosome rearrangements249 and shows a banded pattern on the murine inactive X, indicating a nonrandom association of the transcripts with the underlying DNA of the inactive X chromosome.249 While the mechanism of action of the XIST/Xist RNA is open to speculation (see following section), the available sequence data and evolutionary conservation support the hypothesis that at least part of the sequence and/or structure of the RNA itself is important.
Detection of XIST RNA in interphase cells from a normal male, a normal female, and a 49,XXXXX female by RNA fluorescence in situ hybridization. Male cells do not express XIST RNA. Female cells express XIST from the inactive X chromosome, and XIST transcripts remain associated with the Barr body complex. Cells from the 49,XXXXX female have four inactive X chromosomes; each expresses XIST and forms a distinct XIST RNA/Barr body complex.
Mechanisms of X Chromosome Inactivation.
While the precise mechanism(s) of X chromosome inactivation remain to be revealed, the process is generally described in four stages: recognition of the number of X chromosomes (sometimes called “counting”); initiation early in development; promulgation whereby the initial signal is spread to the rest of the chromosome; and maintenance of the inactivating signal through successive cell divisions.28-30 Although this is a theoretically convenient way to visualize events, it may not be accurate in detail, because the hypothesized steps are not generally separable.
The data from studies of X chromosome aneuploidies show that all X chromosomes in excess of one are inactivated (the “n-1 rule”) (Fig. 61-4). This finding suggests that the onset of the inactivation process entails two events, the first to “mark” one X chromosome as the active one, and the second to then inactivate all “unmarked” X chromosomes. The presence of two active X chromosomes in triploid and tetraploid individuals suggests that the marking of the active X involves some effect of the autosomal complement.208 This marking could involve either physical changes in the DNA or epigenetic changes, such as protein binding, membrane attachment, histone deacetylation, or DNA methylation.27 Evidence that Xist itself is involved in the recognition or counting process has come from gene-targeting experiments in mouse that demonstrated that an internally deleted copy of the Xist gene cannot be recognized or chosen.250 Once the single X to remain active has been marked, another factor or inactivation signal could act on any unmarked X chromosomes— none in the case of normal males, one in the case of normal females, and four in the case of an abnormal 49,XXXXX karyotype (Fig. 61-4).
A number of gene-targeting experiments in mouse have demonstrated directly that Xist is necessary for X inactivation to occur. A mutation that eliminates Xist transcription prevents X inactivation on that chromosome,251 while two different deletions of portions of the Xist transcript itself result in primary nonrandom X inactivation.250,252 In combination with the ectopic Xist transgene experiments summarized in the previous section, these mutations clearly demonstrate that Xist, along with adjacent sequences, is both necessary and sufficient for X inactivation to occur and implicate different regions of the Xist gene in both recognition/counting and in initiation of inactivation.226,253 These data are generally consistent with more limited data in humans, including a novel mutation in the promoter of the XIST gene that compromises promoter activity and has been detected in two families with skewed X inactivation.160
Studies of the initiation of X inactivation in murine embryonic stem cells indicate that commitment to inactivation is accompanied by stabilization of the Xist transcripts on the X that will become inactive, followed by transcriptional inactivation (and extensive methylation) of the Xist gene on the X that will remain active.254,255 The factors responsible for converting unstable Xist transcripts from both male and female X chromosomes prior to initiation to the more stable transcripts that associate only with the inactive X have not been identified. However, at least in the mouse, this may be determined in part by a switch from a novel upstream promoter whose transcripts are unstable256 to the constitutive somatic promoter identified earlier in both human and mouse.240,244
One of the most intriguing features of X chromosome inactivation is the ability of the inactivation signal to be promulgated along a single X chromosome in cis, without affecting the active X chromosome present in the same nucleus. The promulgation signal must distinguish those genes that escape X inactivation209,215,257 (see “Genes that Escape X Chromosome Inactivation” above), although formally it isn't known whether these genes are never inactivated or whether they are unable to maintain inactivation. In translocations between the X and autosomes, X chromosome inactivation can spread to the translocated material,205 suggesting that there are not unique X-linked factors associated with each gene subject to inactivation. Promulgation of inactivation could proceed by a locus-by-locus mechanism, although the more commonly proposed mechanisms for a cis-limited spread of inactivation are chromosomal in nature,28,30 presumably involving XIST RNA.249 It is estimated that there are only several hundred to several thousand copies of the RNA per nucleus.258 As this corresponds to less than one copy of the XIST transcript per inactivated gene,249 the RNA presumably acts by modifying higher-order chromatin structure on the inactive X, in the context of a Barr body-XIST RNA complex.248,249
All the properties of the inactive X chromosome described earlier could aid in the maintenance of the inactive state, either singly or cooperatively. DNA methylation is most commonly proposed as a secondary mechanism being used to “lock in” the inactive state.30,39 Extensive methylation of CpG dinucleotides throughout X-linked gene regulatory sequences is well documented46,50 and may interfere with the binding of transcription factors required for gene expression.54,55,259,260
While it is clear that XIST is necessary for the initiation of X inactivation early in development, it is less clear what role XIST or its RNA plays in maintenance of the inactive state. Deletion of the XIST gene from the inactive X either in mouse/human somatic cell hybrids261 or in leukemic cells262 does not result in reactivation of the inactive X, indicating that neither the gene nor its transcripts are required for maintenance. Further, reactivation of the transcriptionally silent XIST gene on an active X in somatic cell hybrids does not induce inactivation of the X;263-265 thus, the association of XIST RNA with the X chromosome, as well as the initiation of inactivation, must require developmentally regulated steps or factors that are absent from somatic cells.