Initiation: the inactive X could be either maternally or paternally derived, and the inactivation would be random in any one cell;
Maintenance: X inactivation would be irreversible in a somatic cell and all its descendants.
Unravelling the mechanisms behind these four processes has kept researchers busy for nearly 50 years, and this effort is continuing today. The processes are incredibly complicated and sometimes involve mechanisms that had barely been imagined by any scientists. That’s not really surprising, because Lyonisation is quite extraordinary – X inactivation is a procedure where a cell treats two identical chromosomes in diametrically opposite and mutually exclusive ways.
Experimentally, X inactivation is challenging to investigate. It is a finely balanced system in cells, and slight variations in technique may have a major impact on the outcome of experiments. There’s also considerable debate about the most appropriate species to study. Mouse cells have traditionally been used as the experimental system of choice, but we are now realising that mouse and human cells aren’t identical with respect to X inactivation[99]. However, even allowing for these ambiguities, a fascinating picture is beginning to emerge.
Mammalian cells must have a mechanism to count how many X chromosomes they contain. This prevents the X chromosome from being switched off in male cells. The importance of this was shown in the 1980s by Davor Solter. He created embryos by transferring male pronuclei into fertilised eggs. Males have an XY karyotype, and when they produce gametes each individual sperm will contain either an X or a Y. By taking pronuclei from different sperm and injecting them into ‘empty’ eggs, it was possible to create XX, XY or YY zygotes. None of these resulted in live births, because a zygote requires both maternal and paternal inputs, as we have already seen. But the results still told us something very interesting, and are summarised in Figure 9.3.
Figure 9.3 Donor egg reconstitution experiments were performed in which the donor egg received a male and female pronucleus or two pronuclei from males. Just as in Figure 7.2, the embryos derived from two male pronuclei failed to develop to term. When the nuclei each contained a Y chromosome, and no X chromosome, the embryos failed at a very early stage. Embryos derived from two male pronuclei where at least one contained an X chromosome developed further before they also died.
The earliest loss of embryos occurred in those that had been reconstituted from two male pronuclei which each contained a Y chromosome as the sole sex chromosome[100]. In these embryos there was no X chromosome at all, and this was associated with exceptionally early developmental failure. This shows that the X chromosome is clearly essential for viability. This is why male (XY) cells need to be able to count, so that they can recognise that they only contain one X, and thus avoid inactivating it. Turning off the solitary X would be disastrous for the cell.
Having counted the number of X chromosomes, there must be a mechanism in female cells by which one X is randomly selected for inactivation. Having selected a chromosome, the cell starts the inactivation procedure.
X inactivation happens early in female embryogenesis, as the cells of the ICM begin to differentiate into the different cell types of the body. Experimentally, it is difficult to work on the small number of cells available from each blastocyst so researchers typically use female ES cells. Both X chromosomes are active in these cells, just like in the undifferentiated ICM. It’s easy to roll ES cells down Waddington’s epigenetic landscape, just by subtly altering the conditions in which the cells are cultured in the lab. Once we change the conditions to encourage the female ES cells to differentiate, they begin to inactivate an X chromosome. Because ES cells can be grown in almost limitless numbers in labs, this provides a convenient model system for studying X inactivation.
Initial insights into X inactivation came from studying mice and cell lines with structurally rearranged chromosomes. In some of these studies, various sections of an X chromosome were missing. Depending on which parts were missing, the X chromosome did or did not inactivate normally. In other studies, sections had come off the X chromosome and attached themselves onto an autosome. Depending on which part of the X chromosome had transferred, this could result in switching off the structurally abnormal autosome[101][102].
These experiments showed that there was a region on the X chromosome that was vitally important for X inactivation. This region was dubbed the X Inactivation Centre. In 1991 a group from Hunt Willard’s lab at Stanford University in California showed that the X Inactivation Centre contained a gene that they called Xist, after X-inactive (Xi) specific transcript[103]. This gene was only expressed from the inactive X chromosome, not from the active one. Because the gene was only expressed from one of the two X chromosomes, this made it an attractive candidate as the controller of X inactivation, where two identical chromosomes behave non-identically.
Attempts were made to identify the protein encoded by the Xist gene[104] but by 1992 it was clear that there was something very strange going on. The Xist gene was transcribed to form RNA copies. The RNA was processed just like any other RNA. It was spliced, and various structures were added to each end of the transcript to improve its stability. So far, so normal. But before RNA molecules can code for protein, they have to move out of the nucleus and into the cytoplasm of the cell. This is because the ribosomes – the intracellular factories that join amino acids into long protein chains – are only found in the cytoplasm. But the Xist RNA never moved out of the nucleus, which meant it could never generate a protein[105][106].
This at least cleared up one thing that had puzzled the scientific community when the Xist gene was first identified. Mature Xist RNA is a long molecule, of about 17,000 base-pairs (17kb). One amino acid is coded for by a three base-pair codon, as described in Chapter 3. Therefore, in theory, the 17,000 base-pairs of Xist should be able to code for a protein of about 5,700 amino acids. But when researchers analysed the Xist sequence with protein prediction programs, they simply couldn’t see how it could encode anything this long. There were stop codons (which signal the end of a protein) all through the Xist sequence and the longest predicted run without stop codons was only enough to code for 298 amino acids (894 base-pairs[107]). Why would a gene have evolved which created a 17kb transcript, but only used about 5 per cent of this to encode protein? That would be a very inefficient use of energy and resources in a cell.
But since Xist never actually leaves the nucleus, its lack of potential protein coding is irrelevant. Xist doesn’t act as a messenger RNA (mRNA) that transmits the code for a protein. It is a class of molecule called a non-coding RNA (ncRNA). Xist may not code for protein, but this doesn’t mean it has no activity. Instead, the Xist ncRNA itself acts as a functional molecule, and it is critical for X inactivation.
Back in 1992 ncRNAs were a real novelty, and only one other was known at the time. Even now, there is something very unusual about Xist. It’s not just that it doesn’t leave the nucleus. Xist doesn’t even leave the chromosome that produces it. When ES cells begin to differentiate, only one of the chromosomes produces Xist RNA. This is the chromosome that will be the inactive one. Xist doesn’t move away from the chromosome that produced it. Instead, it binds to the chromosome and starts to spread out along it.