Dosage compensation in mammals occurs by the inactivation of one of the two X chromosomes in females; the inactive X chromosome is condensed into heterochromatin as a cytologically recognizable Barr body or sex chromatin body. Since trisomic XXX and tetrasomic XXXX cells in culture have two and three Barr bodies, respectively, it might be more accurate to regard the dosage compensation system as keeping one X active, but we will use the more familiar approach and talk about the process of inactivation. Since being proposed by Lyon in 1961, X chromosome inactivation has been repeatedly cited as a prototypical example of an epigenetic effect. While it is an outstanding example of epigenetics and has many features in common with other types of epigenetic effects, as defined in this chapter and in Chapter 3, the process of X inactivation is also quite different from most other epigenetic examples. The process occurs in several distinct steps.
Inactivation initiates at the Xic
In the developing mammalian embryo, one of the two X chromosomes will be inactivated. The earliest differentiation in the mammalian embryo separates the outer layer, known as the trophoblast (and later, the trophoectoderm), from the inner cell mass. The inner cell mass will eventually become the various cell types, tissues, and organs of the organism. The trophoectoderm gives rise to the extra-embryonic tissues such as much of the placenta, the amniotic sac, and the chorion. In these extra-embryonic cells, the paternal X chromosome is always inactivated and the maternal X chromosome is kept active. Thus, there must be some molecular markers on the paternal and maternal X chromosomes that allow them to be distinguished. (In marsupials, the paternal X chromosome is inactivated in all cells, but, in placental mammals, it is only in the extra-embryonic cells.)
The nature of these molecular markers is not clear, but there are many differences between chromosomes that come from the sperm and chromosomes that come from the egg, so there are numerous possibilities. For example, most of the DNA in sperm is condensed with an arginine-rich class of protein, known as protamines, rather than with histones, so sperm chromatin has a different structure from the outset. There is evidence in mice that, at the very earliest stages of inactivation, when the developing embryo consists of only 2–4 cells, the paternal X is highly condensed and inactive; thus, it may be that the paternal X is kept inactive in those cells that will become the extra-embryonic cells but reactivated in the cells that will become the inner cell mass and the embryo proper. By the time the inner cell mass is distinct from the extraembryonic cells, both X chromosomes are decondensed and appear to be active, although transcription at this stage is very low.
In the cells of the inner cell mass, the choice between active and inactive X chromosomes appears to be at random; there is no consistent preference for inactivation of the paternal chromosome over the maternal chromosome, or vice versa. The inactivation begins when the inner cell mass has about 8–12 cells, and each cell inactivates its X separately; at the time of inactivation, neighboring cells are no more likely to have the same X inactivated as non-neighboring cells. The active X is designated Xa, while the inactive X is designated Xi. Inactivation does not occur simultaneously in all cells, but the process is complete by the end of gastrulation. The initiating event for inactivation is not known, and the basis for the choice between the two X chromosomes is also not known at this time.
It is clear, however, that inactivation begins at a particular region near the centromere on the long arm of the X chromosome, termed the X chromosome inactivation center, or Xic. Some DNA sequence variants at the Xic render the chromosome slightly more or less likely to be the inactive X, so there may be a protein that binds at the Xic or a structure that forms there to initiate inactivation. (Conversely, something may occur at the Xic on the active X chromosome that blocks its inactivation.) It does seem that inactivation continues to completion once it begins. Thus, while the initial choice between the two X chromosomes may be random or involve a very subtle difference, the propagation of that choice into inactivation results in one active X and one inactive X. Once an X chromosome is inactivated, it remains inactivated in its mitotic daughter cells, as described in the chapter; this persistence of inactivation is one reason that calico cats have distinct patches of fur color.
The inactivation spreads from the Xic to encompass the entire chromosome. Thus, there is an initiation step, followed by a spreading step. The nature of this spreading step is not known precisely but certainly involves a progressive formation of heterochromatin. Eventually, the entire X chromosome of about 164 million bases is inactivated. As noted in the chapter, some genes, possibly about 5% of X-linked genes, escape inactivation and are expressed from both X chromosomes. The genes that escape inactivation are not precisely the same in all mammals and may not even be precisely the same in different cell types. While it is convenient to imagine that spreading is progressive, like rolling up a rug, a chromosome is a three-dimensional molecule, while a rug is more similar to a two-dimensional object; not all of the genes that escape inactivation are adjacent to each other on the X chromosome, but they may be adjacent to each other on the threedimensional chromosome.
Structure of the inactive X chromosome
The inactive X chromosome is an example of heterochromatin, as discussed in Chapter 3. Because it is compacted or condensed, Xi stains brightly with DNA dyes. Genes in vertebrates have a collection of C–G dinucleotides in their upstream regulatory region, known as CpG islands. On the inactive X, the CpG islands are heavily methylated, a characteristic of transcriptionally silenced genes. Particular histone modifications that are characteristic of heterochromatin, such as high levels of H3K9 methylation and low levels of H3K4 methylation and H3 acetylation, are also found on the Xi. (Histone modification are discussed in Chapter 3 and again in Chapter 12.) In addition, a protein known as H2AYF replaces the H2A protein in some of the nucleosomes. Like other regions of heterochromatin, the inactive X chromosome replicates late in the cell cycle. In these ways, the inactive X chromosome is similar to other types of heterochromatin.
None of these modifications or structural changes is likely to be the event that initiates inactivation. Rather, these are the molecular mechanisms that ensure that, once silenced, the chromosome remains heterochromatic and is silenced through subsequent cell divisions. If the initial steps in inactivation are equivalent to closing a door, these steps are analogous to locking it up tightly and keeping it locked.
X chromosome inactivation is generally similar to the formation of heterochromatin in other regions of the genome and in other organisms. However, the most unusual aspect of X inactivation was the discovery of a non-coding RNA called Xist. Xist (for “Xi-specific transcript” and pronounced “exist”) is transcribed from the Xic and exclusively transcribed from the inactive X chromosome. Transcription of Xist may be the triggering event or the master regulator for inactivation—both X chromosomes transcribe Xist at low levels before inactivation, and the chromosome that transcribes Xist at high levels first becomes the inactive one—or it may be an early event indicative of X inactivation. Xist is the only transcript that is specific to the Xi. Mice with deletions of the Xist gene die as very young embryos and fail to inactivate an X chromosome, indicating that the Xist gene has an essential role. Furthermore, if the Xist gene is inserted onto an autosome in a cell line, that autosome is inactivated like Xi, which suggests that Xist (or something that interacts with the Xist gene or its RNA) controls inactivation.
Precisely how Xist is involved in X inactivation is not known, and it may have several roles at different stages of the inactivation process. Xist is a non-coding RNA, so it is not translated into an amino acid sequence and appears to be not capable of being translated. It is spliced and poly-adenylated, like mRNAs and many (but not all) long non-coding RNAs. On the other hand, most noncoding RNAs are relatively short, less than 1 kb in length or shorter than a typical mRNA. By contrast, the Xist RNA is enormous, more than 17 kb long in humans.
Figure A. Xist and the inactive X chromosome. In this image of a fibroblast cell from a female mouse, the DNA is stained blue, while a DNA sequence from the region of the Xic is shown in yellow. This can be used to contrast the structure of Xa and Xi. The Xist RNA is labeled in red; note how it is found along the length of the inactive X chromosome.
The sequence of Xist in humans includes a series of repeated sequences that would be capable of forming complicated intrastrand structures in the RNA, which may play a role in its functions; however, not all of these repeats (and thus, the structures that they can form) are found in the mouse Xist RNA. Certain proteins known to be important in chromatin structure and epigenetic regulation, such as members of the Polycomb family, are recruited to the Xist RNA, so some of its activity may be through its action as a binding site for other factors. It can also form double-stranded RNA hybrids with some other RNA molecules, including an RNA known as Tsix, transcribed using the other strand of the Xic; Tsix (Xist spelled backwards) corresponds to the reverse complement of Xist. Because Xist is a uniquely long non-coding RNA, the mechanisms by which it acts have been somewhat difficult to elucidate; many activities and functions have been found, but the relationship among these and X inactivation are an area of active investigation and speculation. Once Xist is transcribed from the Xic, it spreads along the inactive X, so that eventually the inactive X is coated with Xist. Figure A shows a cell in culture with both an active and inactive X, and the localization of Xist.
X chromosome reactivation
The X inactivation cycle does not end with inactivation, however. While Xi remains inactive in all of the somatic cells of a female, a reactivation process must occur in the germline cells. The mitotically dividing nuclei in the ovary have an inactive X. When these nuclei enter meiosis to form ova, the X must be reactivated or at least decondensed if not actively transcribed. The embryo begins with an active X chromosome from the mother, which implies that her ova had decondensed X chromosomes. This reactivation process has not been studied as intensively as the inactivation process, and not much is known about it.