Standard Article

You have free access to this content

Mechanisms of X-inactivation

  1. Sarah M Duthie

Published Online: 19 APR 2001

DOI: 10.1038/npg.els.0001194

eLS

eLS

How to Cite

Duthie, S. M. 2001. Mechanisms of X-inactivation. eLS. .

Author Information

  1. Medical Research Council Clinical Sciences Centre, London, UK

Publication History

  1. Published Online: 19 APR 2001

Description

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading

The evolution of mammalian sex chromosomes has resulted in XX females and XY males. The X-chromosome is large and encodes approximately 5% of the haploid genome, while the Y-chromosome is small and mainly heterochromatic. To achieve dosage compensation with males, one of the two X-chromosomes in female mammals is transcriptionally silenced in the developing embryo, by a process known as X-inactivation. The choice of which X to inactivate is normally random and the inactive state is then stably inherited. Female mammals are composed, therefore, of clonal cell populations in which either the maternally derived X (Xm) or the paternally derived X (Xp) is inactive. See also Mammalian Sex Determination, and Reproduction in Mammals: General Overview

Classical genetics

The first steps in the X-inactivation field were taken in the 1950s, during which time several key observations were made. Classical genetic studies revealed that variegated coat-colour phenotypes in female mice were linked to the expression of genes on the X-chromosome. Secondly, female XO mice were shown to be viable and fertile, suggesting that only one X-chromosome is necessary for normal development in mice. Thirdly, the sex chromatin body (Barr body), visible in human female interphase nuclei, was found to consist of a single, condensed X-chromosome. These discoveries prompted the Lyon hypothesis (1961) which stated that one X-chromosome was inactivated at random in diploid female somatic cells. This idea was later modified to state that, regardless of sex chromosome constitution, i.e. XX, XXX, XXY, etc., one X-chromosome should remain active with all others being inactivated. See also History of Classical Genetics, and Human Chromosomes

Observations made in the 1960s on X:autosome (X:A) translocations in female mice produced further discoveries. It was noted that spreading of inactivation from the inactive X (Xi) into autosomal material occurred on only one of the two products of a reciprocal X:A translocation. In addition, the strength of the inactivating signal diminished as it moved further from the translocation break-point with the X-chromosome. These studies gave rise to the concept of a controlling locus or X-inactivation centre (XIC/Xic) from which the silencing signal was proposed to originate and spread bidirectionally, in cis. Subsequent studies demonstrated that the Xic is required for the correct initiation of X-inactivation as well as propagation of the silencing signal. See also X-Chromosome Inactivation

Cytogenetic analyses and classical embryology

Steady progress was made in the inactivation field in the 1970s and 1980s. The development of various cytogenetic techniques provided researchers with the tools needed to assess the physical properties of Xi chromatin. Late replication had long been known as a marker for genetic inactivity and it was eventually shown that the initiation of X-inactivation involved a switch in the replication timing of one of the cell's two X-chromosomes. Studies with X:A translocations revealed that the transcriptionally inactive X was late-replicating compared to its active homologue. This late-replicating X had distinctive staining properties and was visibly different from the active X (Xa) during interphase and at metaphase. See also Cytogenetic Techniques, and Chromosome Preparation and Banding

Cytogenetic analyses of mouse embryos revealed that both X-chromosomes were initially active following fertilization and that initiation of inactivation occurred at a specific stage during early development. Another revelation came with the discovery that genes on Xi and Xa were differentially methylated. The promoters of several genes on Xi were shown to be heavily methylated compared to their homologues on Xa and methylation was known to be stably heritable. These features led to the proposal that methylation might be involved in both the initiation and maintenance mechanisms. Interestingly, marsupial X-linked genes did not show differential methylation and the inactive state was found to be ‘leaky’, with some X-linked genes capable of reactivation. This observation led to the idea that initiation and the subsequent maintenance of the inactive state should be examined separately. See also DNA Methylation, Cleavage and Gastrulation in the Mouse Embryo, and Reproduction in Monotremes and Marsupials

Discovery of the Xist gene

Although the inactivation of one X-chromosome in eutherians was known to be stable, several human genes were discovered that escaped inactivation and were expressed from both Xi and Xa. Mapping studies with human and mouse X:A translocations had located the Xic in the region Xq13 in humans or band XD in mice and defined the minimum Xic as approximately 1–2 Mb. The next major step was to find a candidate gene in this region. In 1991 the search led to the discovery of a gene that mapped within the XIC and was transcribed only from Xi. This gene was named XIST for X-inactive-specific transcript (Brown et al., 1991). Characterization of the XIST gene, and its mouse homologue Xist revealed a 17-kb and 15-kb transcript respectively, with no protein-coding potential. The RNA was localized to the nucleus and remained associated with Xi in female interphase nuclei, as demonstrated by fluorescent in situ hybridization (FISH) experiments. In addition, Xist expression was shown to precede X-inactivation in early mouse embryonic development and in differentiating embryonal stem (ES) cells. Based on these unique characteristics, the Xist gene immediately became a candidate for a leading role in X-inactivation. Direct evidence for the involvement of Xist in the global silencing of one X-chromosome came from a gene-targeting experiment which showed that an intact cis-acting Xist gene was both necessary and sufficient for the initiation of X-inactivation (Penny et al., 1996). See also Fluorescence in situ Hybridization

Today, research in the X-inactivation field is focused upon the molecular mechanisms governing the process of counting, initiation of silencing and maintenance of the inactive state.

Mechanism: Chromosome Counting

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading

Blocking factor model

A single X-chromosome remains active in all diploid somatic cells, regardless of sex chromosome constitution. The cell appears capable of counting the number of Xic elements present and hence, the number of X chromosomes. A popular mechanism to account for this phenomenon was proposed by Rastan 1983 and is known as the blocking factor model (Figure 1). The model states that a diploid cell produces just enough of this factor to block one Xic at random, during initiation of inactivation. The blocked Xic is prevented from transcribing Xist RNA and so is marked as Xa. Any additional Xics will remain unblocked and will transcribe the Xist gene by default when the cell differentiates from the totipotent lineage. Although widely favoured as a possible mechanism, no candidate for the blocking factor has been identified.

thumbnail image

Figure 1. Model for initiation of random X-inactivation. (a) Unstable Xist transcript from P0 promoter on both alleles. Both X-chromosomes active (Xa). (b) Upon differentiation, a limited amount of blocking factor binds to only one X-chromosome at random – designated Xa. (c) Transcription switches from P0 to initiation of stable, accumulated Xist RNA from P1/P2 on Xi. Xi inactivates. Unstable P0 transcript still detectable from Xa. (d) Unstable P0 Xist transcript silenced on Xa.

Functional domain of the Xic

Recent mouse transgenic experiments have supported the theory that elements within the Xic are counted by the cell. A 450-kb yeast artificial chromosome (YAC) carrying the Xist gene was introduced into mouse XY ES cells, with several copies integrating in tandem into an autosome (Lee et al., 1996). Upon differentiation the cells were capable of expressing Xist RNA either from the endogenous copy of Xist on the X, or from the ectopic transgenes present on the autosome. This experiment showed that the male XY ES cells were capable of counting elements present within a 450-kb region of the Xic. Similar results were achieved with a 35-kb cosmid containing the Xist gene plus 9 kb of sequence upstream and 5 kb downstream (Herzing et al., 1997). This result suggested that all the functions of the Xic could be reproduced by a comparatively small region surrounding the Xist gene. See also Transgenic Mice Production, and Transcription Activation in Eukaryotic Cells

X-inactivation and Xce

In diploid XX cells, either of the two X-chromosomes has a 50% chance of becoming the inactive X. This random choice can be influenced by a locus known as the X-controlling element (Xce). Genetic analyses mapped the Xce locus to the same region as the Xic; however genetic mapping using microsatellite markers showed that Xist and Xce are genetically separable elements, with Xce mapping downstream of Xist. It is not clear whether Xce alleles exist in human populations, although a recent study based on a three-generation analysis of 36 families, found significant skewing of X-inactivation in several families. In addition, there has been a recent report of a single base pair mutation in the promoter of the human XIST gene, correlating with nonrandom X-inactivation in heterozygous females from two kindreds. These studies imply that, as in mouse, there are human loci that affect the choice of which X to inactivate. See also Short DNA Sequence Repeats, and Genome Mapping

In addition to transgenic studies, targeting experiments designed to ‘knock out’ the Xist gene function have contributed towards understanding the counting and choosing mechanisms. Removal of a 65-kb region extending 3′ from exon 6 of the Xist gene has provided some interesting data (Clerc and Avner, 1998). In differentiated XX ES cells, the chromosome bearing the mutant Xist allele was always inactivated. The same deletion in XO cells produced an identical result, showing that X-inactivation occurred even in the absence of a second Xic. These experiments indicated that the region 3′ of Xist plays some part in the counting/choice process. It is possible that this knockout has removed the Xce locus, creating a null mutant. Such a mutant would always inactivate the X carrying the null Xce allele, a phenotype consistent with these results. See also Knockout and Knock-in Animals

Primary or secondary nonrandom inactivation?

Two other knockout experiments involved removing part (Penny et al., 1996), or almost all (Maharens et al., 1997), of the Xist gene coding region in ES cells and transgenic mice. In contrast to Clerc and Avner's knockout, the X-chromosome carrying the mutant Xist allele failed to inactivate and the X carrying the wild-type allele always inactivated, in both experiments. This nonrandom X-inactivation could be due to either primary or secondary mechanisms. Primary nonrandom inactivation would occur with a disruption only in the choice function. In this case, the cell would be able to count the number of X-chromosomes but would be unable to choose the chromosome carrying the mutant allele. Secondary nonrandom inactivation would occur if both counting and choice mechanisms remained unaffected. The cell could choose to inactivate either allele, but would only be able to inactivate the chromosome with the wild-type allele. In this example, those cells that had chosen to inactivate the chromosome carrying the mutant allele would end up with two active X-chromosomes. This situation is lethal in vivo and these cells would be selected against. It is not clear whether Penny's knockout showed primary or secondary nonrandom inactivation. Maharen's experiment, on the other hand, appeared to exhibit primary nonrandom inactivation, suggesting that unknown elements within the Xist gene itself might be involved in the choice mechanism. The counting function was not disrupted in either knockout, implying that elements involved in counting must be found outside the Xist gene coding region.

Imprinted X-inactivation

In most cells, the choice of which X to inactivate is more or less random, depending upon which Xce allele each X is carrying. However, in some cases, cells undergo nonrandom inactivation as a result of a parental imprint. Imprinting is the process by which a chromosome ‘remembers’ its parental origin. In marsupials, the chromosome inherited from the father (Xp) is always the inactive X. A second example is found in the trophectoderm and primitive endoderm of the preimplantation mouse embryo. These cell lineages form extraembryonic membranes around the developing embryo. Random Xist expression in the embryo itself is not seen until a few days later, following erasure of the paternal imprint. In Maharen's targeting experiment, female mice that inherited a mutant Xist allele from their father died early in embryogenesis. In these mice, the trophectoderm cells were unable to undergo nonrandom paternal X-inactivation, thus preventing normal development of the embryo. See also Imprinting (Mammals), and Genomic Imprinting

Mechanism: Initiating Silencing

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading

Stabilization of Xist RNA

Xist RNA is transcribed from both alleles in undifferentiated female ES cells and can be visualized by FISH as a punctate signal localizing to the X-chromosome. In differentiated ES cells and in somatic cells, the transcript appears as a large domain associated with the Barr body at the nuclear periphery. This developmental upregulation of Xist expression from Xi is the result of increased RNA stability rather than an increase in the rate of transcription. Recently, it has been shown that the stable and unstable transcripts are, in fact, different isoforms of Xist RNA (Johnston et al., 1998). Switching between different promoters of the Xist gene generates the different transcripts involved in upregulation (Figure 1). Unstable Xist RNA is transcribed on both X chromosomes from a newly discovered promoter (P0), 5–8 kb upstream of the previously identified start site (P1). Transcription from P0 may be required for activation of the counting mechanism. Upon differentiation, transcription switches to P1 and to a second, recently described, promoter (P2) 1.5 kb further downstream in exon 1. Quantitation of Xist RNA levels has revealed that the internal promoter P2 has a higher transcriptional activity than P1, although the functional significance for two such promoters is unknown. Transcription from the P1/P2 promoters is required for the production of stable, accumulated Xist RNA and subsequent X-inactivation. An antisense RNA, Tsix, transcribed from the opposite strand to the Xist gene, has recently been identified. Tsix is expressed in ES cells and may be involved in the regulation of Xist expression during diffentiation. See also Cell Nucleus, mRNA Stability, Transcriptional Gene Regulation in Eukaryotes, and Small Regulatory RNAs in Bacteria

X-inactivation is developmentally regulated

The events required to produce a single, active X-chromosome occur sequentially over a period of several days in both differentiating ES cells and developing mouse embryos (Figure 2). Diploid male germ cells appear to express Xist RNA at very low levels; however Xist-deficient male mice are fertile, thus Xist does not seem to be involved in silencing during spermatogenesis. Female germ cells do not express Xist RNA once they have entered meiosis. After meiosis, the haploid gametes are either Y or Xp (from the father) and Xm (from the mother) and have no Xist expression. Following fertilization, accumulated Xist expression occurs only from XP (Kay et al., 1993), although at this time, both X-chromosomes are still active and the cells are undifferentiated. See also Meiosis, and Mouse Early Development: Molecular Basis

thumbnail image

Figure 2. The process of X-inactivation. (a) Xist RNA switched off during passage through the germline. (b) Both X-chromosomes active in early cleavage-stage embryo. Stable, accumulated Xist signal associated with paternal X (Xp). (c) By midstage blastocyst, Xp inactive and associated with stable Xist signal in extraembryonic lineages (trophectoderm), maternal X (Xm) active. Unstable Xist RNA visible as punctate dot from both Xist alleles in cells of the inner cell mass (ICM). Both X-chromosomes active. (d) At day 5.5 of development, embryo implants and cells of embryo proper begin to differentiate. Random X-inactivation initiated – most cells now show unstable Xist transcript from Xa and stable, accumulating transcript from Xi. (e) By day 8.5, Xi associated with stable, accumulated Xist transcript, unstable Xist RNA from Xa switched off. Xi inactive and Xa active throughout adult life.

By day 3.5, the embryo (now called the blastocyst) contains two main cell lineages: differentiated cells of the trophectoderm and primitive endoderm and totipotent cells of the inner cell mass (ICM). Trophectoderm cells still express stable Xist signal from Xp, but the chromosome is inactivated during the differentiation process. This observation suggests that other factors involved in transcriptional silencing are required to interact with Xist RNA but are only generated during the process of differentiation. In those cells programmed to become the ICM, the imprint is erased, the transcript from Xp is destabilized and production of unstable transcript from Xm increases until both alleles are exhibiting unstable, punctate signal. It is at this point in development that cells of the ICM can be cultured to produce ES cells.

In the mouse embryo, random X-inactivation is initiated between days 5.5 and 8.5 of development. This transition from unstable Xist expression on both active X-chromosomes to a single, stable, accumulated signal on Xi, occurs via an intermediate state in which the Xa Xist allele continues to produce unstable transcript for a short time. XY embryos never show stable accumulated signal in the trophectoderm since their X-chromosome is inherited from their mothers and carries no imprint. Unstable Xist transcription from Xm persists over the same developmental period as in XX embryos and finally disappears by day 8.5. See also Gene Silencing in Development (Drosophila)

Spreading of the inactivating signal

Following the initiation of transcription from P1/P2, the silencing signal spreads along the X-chromosome from the Xic. FISH techniques have been used to demonstrate that Xist RNA appears as a dense cluster of particles coating Xi in human and mouse female interphase nuclei. The RNA does not seem to associate directly with Xi DNA however, since treatment of cells with DNAase 1 removes chromatin but has no effect on the Xist RNA signal. It has been suggested that the RNA binds either to proteins or other RNA molecules and is involved in higher order chromatin packaging. Analyses of steady-state levels of Xist RNA in different mouse strains have shown that there are not enough RNA molecules to cover the entire X-chromosome, suggesting that Xist may interact with large chromatin domains rather than individual genes. See also Chromatin Remodelling and Histone Modification in Transcription Regulation

One theory to explain the propagation of inactivation requires the presence of X-specific elements or ‘waystations’ spaced at intervals along the X. Waystations are predicted to facilitate the spread of inactivation and are weaker, less frequent, or absent, on autosomes and near genes that escape inactivation (Figure 3). FISH studies have revealed that Xist RNA hybridizes to discrete foci on the X, rather than coating the entire chromosome indiscriminately and has only very limited spread into cis-linked autosomal material in X:A translocations. These findings suggest that if waystations do exist, they should be concentrated in specific domains on the X such as the gene-rich Giemsa-light (G-light) bands. In contrast, investigations into an Xist YAC transgene integrated into mouse chromosome 12 showed that Xist RNA completely coated the autosome with no evidence of discrete foci and was capable of silencing some autosomal genes. These results suggested that X-specific elements are unnecessary for the propagation of silencing. Several other Xist YAC transgenes have not resulted in silencing of autosomal material. To account for such contradictory data, it is possible that Xist RNA waystations, or similar elements, are present on some autosomes at a high enough frequency to allow spreading of the inactivating signal from a cis-linked Xist gene. See also Karyotype Analysis and Chromosome Banding

thumbnail image

Figure 3. Model for spreading of X-inactivation. (a) Active X-chromosome (Xa) showing waystation elements present in gene-rich Giemsa-light (G-light) bands. Waystations are absent or inaccessible in condensed, gene-poor G-dark bands. (b) Xist RNA transcribed from the Xic region spreads bidirectionally via waystation elements along the X. Binding causes local changes in chromatin conformation, resulting in condensation of G-light bands. (c) Xist RNA stably associated with condensed, inactive X-chromosome (Xi).

Mechanism: Maintaining Silencing

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading

Does Xist RNA play a role in maintenance?

X-inactivation is stable and clonally heritable, therefore the cell needs a means of maintaining the inactive state. Somehow the cell marks Xi so that the same chromosome remains inactive after every cell division. The use of FISH techniques has allowed visualization of the Xist transcript at different stages of the cell cycle. In humans, XIST RNA paints Xi at interphase but dissociates from Xi just prior to mitosis, while in rodents the RNA remains associated with Xi during metaphase and dissociates just before telophase. The RNA is then retranscribed almost immediately in the new daughter cells. Thus Xist RNA cannot be the marker for memory of the inactive state but could still be involved in its maintenance, since it is expressed throughout adult life. However, two studies showed that when somatic cells (which had already undergone X-inactivation) lost the region carrying the Xic locus, Xi did not reactivate, implying that Xist RNA is not required for maintenance of gene silencing. The association of Xist with the X-chromosome may be required solely to initiate a sequence of events during a specific window of time in development. A third study found that Xist RNA was capable of associating with Xa in somatic cells under certain experimental conditions, but could not induce the chromosome to inactivate. Such an observation implies that once the ‘inactivation window’ has been passed, Xist RNA can associate with the X-chromosome but cannot induce inactivation. If Xist RNA is not required for maintenance of inactivation, why is it transcribed continually in somatic cells? There is, as yet, no answer to this question. See also Cell Cycle

Xi shares properties of heterochromatin

The heterochromatin of Xi shares many properties with constitutively heterochromatic elements such as centromeres. These include condensed, transcriptionally silent chromatin, late replication and hypoacetylation of histones H3 and H4. Such features are possible candidates for involvement in the maintenance of inactivation. The transition to late replication is an early event in the X-inactivation process, occurring at about the same time as stabilization of Xist RNA but preceding gene silencing. It is possible that propagation of the Xist transcript may somehow alter the replication timing of Xi by recruiting proteins that regulate heterochromatin formation to the replication origins. It has also been suggested that replicating later in S phase than Xa effectively prevents Xi from competing with Xa for proteins such as transcription factors. See also Heterochromatin and Euchromatin, and Heterochromatin and Euchromatin

Hypoacetylation of core histones

Histones are components of nucleosomes involved in the packaging of DNA and all four histones of the nucleosome core particle have acetylated lysine residues in their N-terminal domains. Hypoacetylation of core histones seems to be a general property of silent heterochromatin where it may act as an epigenetic marker signalling the activity status of the chromatin from one generation to the next. In both human and mouse, centromeres and Xi were found to label weakly with antibodies to acetylated isoforms of the core histones H3 and H4. Recently, a new gene family encoding a core histone called MacroH2A, has been discovered. A subtype of this histone has been shown to associate with Xi, linking X-inactivation with an X-specific nucleosome component. The transition to deacetylation on Xi occurs as a relatively late event during inactivation, following silencing of X-linked genes. This indicates that histone deacetylation may be involved in maintenance rather than initiation of inactivation. However, hypoacetylation of Xi has also been observed in marsupials which show unstable inactivation, suggesting that other factors must also be involved in maintenance. See also Histones: Gene Organisation and Post-translational Modification, Chromatin Remodelling and Histone Modification in Transcription Regulation, and Nucleosomes: Structure and Function

Methylation

Methylation of cytosine residues in CpG island sequences, commonly found clustered in the promoters of many genes, is another marker of gene silencing. It is also stable and clonally heritable due to an enzyme which acts preferentially on hemimethylated DNA following replication. Genes on Xa and those genes that escape inactivation on Xi are not methylated and methylation is not found in the CpG islands of the marsupial Xi, which exhibits unstable inactivation. In adult mice, the Xist gene promoter is methylated and silent on Xa and hypomethylated and expressed from Xi. This methylation imprint is reversed in germ cells, with methylation of the Xist promoter in the female germline and hypomethylation in the male germline. Experiments that inhibited methylation enabled XY ES cells to express stable Xist RNA, implying that DNA methylation is required to repress Xist expression and maintain a transcriptionally active Xa (Panning and Jaenisch, 1996). In male embryos, however, demethylation had no effect on Xist expression in 95% of cells. It seems that methylation is required to play some part in the inactivation process but the precise nature of its role is still uncertain. See also DNA Methylation, and Genes: Definition and Structure

Summary

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading

It is clear that the combination of the early classical genetic data and the current molecular approach has produced important advances in the field of X-inactivation in recent years. It has been established that the global silencing of one X-chromosome in female mammals is initiated by transcription of the Xist gene located within the Xic region. X-inactivation is a developmentally controlled process involving a switch from an unstable to a stable isoform of the Xist gene. The Xist gene encodes a functional RNA molecule that forms part of the nuclear matrix and is capable of associating with Xi only in cis. Finally, the functional domain of the Xic has been narrowed down to a region spanning only 35 kb, facilitating the search for elements involved in counting and choice. In spite of this progress, there are still many puzzles left unsolved. For example, what is the exact role of Xist RNA and how does it associate with cis-linked chromatin? What are the elements involved in counting and choice of the X-chromosome and what other genes are involved in the process? How is silencing maintained throughout adult life? The Xist gene is still the only gene with a definite role in a very complex process.

Glossary
Cis-acting

The ability of a gene to affect genes on the same chromosome.

Epigenetic

The mitotic or meiotic transmission of a pattern of gene expression that does not have its basis in alterations in DNA sequence.

G-banding

Unique banding patterns that appear when chromosomes are stained with Giemsa dye.

Heterochromatin

Densely staining, condensed chromosomal regions, believed to be, for the most part, genetically inert.

Nucleosome

The basic unit of eukaryotic chromosome structure; a ball of eight histone molecules wrapped about by two coils of DNA.

Reciprocal translocation

A translocation in which part of one chromosome is exchanged with part of a separate, nonhomologous chromosome.

References

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading

Further Reading

  1. Top of page
  2. Description
  3. Mechanism: Chromosome Counting
  4. Mechanism: Initiating Silencing
  5. Mechanism: Maintaining Silencing
  6. Summary
  7. References
  8. Further Reading