Title: "Role of the Xist Gene in X Chromosome Choosing"

Authors: York, Marahrens, Jan Loring, and Rudolf Jaenisch

Cited from: Cell, Vol. 92, March 6, 1998. p. 657-664.

X chromosome inactivation in somatic cells of female mice is thought to be controlled by the Xic(X inactivation center) region on the X chromosome(s). This region contains the Xce(X controlling element) gene, which in wild type cells determines which X chromosome will be inactivated according to allelic differences. It also contains the Xist gene, which in this paper is mutated to analyze the cell's subsequent random or nonrandom inactivation mechanism.

The authors deleted part of the Xist gene, from exon 1 to exon 5, and then derived a line of mice that are heterozygous(Xist+/-)for this deletion(in females, of course). According to the random activation mechanism, half the cells would fail to inactivate an X chromosome and die, so the embryo should develop into a healthy but much smaller organism. Therefore, if it is indeed random, Xist+/- mice should have smaller relative weights compared to wild type organisms.

Figure 1 shows the results of two different experiments. The left panel, A, is a previously published figure showing the difference between relative weights in wild type pups(WT) and pups expressing Searle's translocation(XX16-16x). The newborns heterozygous for Searle's translocation have a lower weight than the wild type newborns, supporting the random inactivation hypothesis. The right panel, B, shows the current experiment, comparing the Xist+/- newborns' weights with wild types'. The relative weights are, surprisingly, equal in both the experimental and wild-type conditions. This suggests to the authors that primary nonrandom X inactivation was occurring, rather than random inactivation. They suggested that another explanation for the equal weights could be compensatory growth, but they discard this theory because it would require gradual growth and loss of cells, which although possible they found to be unlikely.

Figure 2 is a phenotypic comparison of Xist+/- and Xist +/+ female littermates at 7.5, 8.5, and 9.5 days of gestation. Because they had such a small number of mutant female embryos to include as data in figure 1, they felt that they needed to show that the sizes were actually quite similar, and that the relative weight comparison did not skew the figure. This figure shows that the sizes were in fact equal, and they try to contrast these results with the results found in the previous experiments with Searle's translocation, because in that case most of the cells were lost between 7 and 8 days of gestation. At 8.5 days the Xist+/- embryo obviously did not have a significantly lower number of cells than the wild type, so the authors conclude that this is further proof that nonrandom inactivation is occurring.

Figure 3A shows the map of the wild-type(Xist+) and mutant(Xist-) alleles of the Xist gene as well as the full probe(above) and deletion probe(below) used. Figure 3B is a cartoon of the RNA FISH expected results of a wild-type female beginning at about 6 days into gestation. The low level biallelic result shows two pinpoint dots that represent each Xist gene. During X inactivation, Xist RNA from one allele coats the X chromosome, giving a differential biallelic pattern. Finally, the Xist gene on the active X chromosome is silenced, reflecting the high level monoallelic pattern. Figure 3C shows another cartoon FISH expected result using the deletion probe in Xist+/- embryos if random or nonrandom choosing of the wild-type X chromosome occurs. The dashed circles represent the RNA signals that are absent because one X chromosome is not wild-type, which would not show up because the deletion probe only labels the wild-type allele. The actual results for Figure 3A and 3B are found in Figure 4.

Figure 4 is the actual FISH labelled Xist expression in cells from female Xist+/- embryos at 7.5 days of gestation, immediately following X inactivation. Square A shows cells from a Xist- male probed with the full probe. There is low level monoallelic in one of the cells and no labelling in the other. B's cells are also Xist- male, but with a Pgk-1 probe(green labelling) as well as a deletion probe. The Pgk-1 transcript was used as an internal control to show that the absence of an Xist signal was not due to human error. One cell contains green low level monoallelic, and one isn't labelled. C's cells are Xist+/+ female with a full probe. The cells display the high-level monoallelic or differential biallelic expression, which indicates that X inactivation was complete. D's cells were Xist+/- female with a full probe. Because partial Xist RNA cannot coat the inactivate an X chromosome, if the mutant X chromosome had been chosen there would be no high signal expression, which is seen in D as high-level monoallelic and differential biallelic expression. Finally, E's cells were Xist+/- female and labelled with the deletion probe. Because 466/470 cells showed the high-level monoallelic expression, and none showed the low-level monoallelic signal, the authors concluded that the Xist+/- embryos undergo primary nonrandom inactivation of the wind-type X chromosome.

Figure 5 shows two models that could explain X chromosome inactivation in wild-type and mutant female cells. The nonrandom inactivation choice in Xist+/- cells indicates that the Xist gene is necessary for chromosome choosing. This also indicates that a positive element is required for inactivation, and that the deletion of part of Xist means that the binding area for this element had been deleted, so by default the wild-type X chromosome was selected. 5A shows the single initiation factor theory, whereby the initiation factor would recognize a certain sequence within the Xist gene and bind randomly. 5B shows the mutually exclusive initiation factor and blocking factor working together to block all but one X chromosome. In this scenario there would be more than one iniation factor, and the X chromosome that had no blocking factor but did have the iniation factor would be chosen.

The conclusions that the authors reached seemed to be well-founded. A mutation in the Xist gene does seem to be necessary and sufficient to cause the inactivation of the wild-type X chromosome only, as seen in the striking number (466/470) in Figure 4E. The authors don't have the evidence that it is the Xist gene alone that causes this result, for they said that "Xist has so far only been distinguised from the Xce by a single crossover, and the two may yet prove to be synonymous." Their mutation could have affected more than just the sequence of Xist, so I'm not fully convinced, but this paper is definitely a starting point.

As for the figures, I have three complaints. First, they only used 10 mutants vs 31 wild-types in Figure 1, and even though they said that they took this into account by using percentages, comparing 2 vs 7 and 6 vs 19 cannot be numerically significant. Second, they used a lot of "data not shown" references. As we saw in Vaux et al, we have to trust them when they quote numbers and results, and this isn't easy if we don't have the data in front of us. Third and finally, they relied on a lack of evidence that would prove them wrong instead of positive evidence that would prove them right in Figure 4. The only difference in 4C and 4D is that C is wild-type and D is Xist+/-, and they both share the same signal types. Usually we want to see a divergence in data from wild-type to experimental, so this type of result isn't the strongest.

Future experiments should focus on discovering the mechanism behind the inactivation choice, which would help us choose between the models that they suggested. For example, we would want to know what this initiation factor is that is sequence-specific. Because it binds to Xist, we have several options. We could attach Xist to a bead and then see what it binds to when we run everything from gDNA over it. Along the same lines, we could make an anti-idiotype antibody using the sequence as a probe. This could show us the initiation factor as well as the hypothesized blocking factor, which would also bind to Xist. I think it would also be interesting to try to create a homozygous Xist-/- individual for a very small deleted area of Xist, for if it's possible to have a surviving individual, then we know that Xist is not the whole reason for chromosome inactivation.


To return to my homepage CLICK HERE.

If you'd like to(please don't hesitate), send me e-mail!