An "Xist"ential Review Paper

Review of "Role of Xist gene in X chromosome choosing"
York Marahrens, Jan Loring, Rudolf Jaenisch
Cell Vol. 92, 657-664, March 6, 1998.

Summary

This paper describes a new aspect to the role of the Xist gene in the mechanism of X chromosome inactivation in female mammals. Female mice that are heterozygous for an internal deletion of the Xist gene (from the middle of exon 1 to exon 5) selectively inactivate the X chromosome that carries the wild-type Xist gene, indicating that the Xist gene product performs a positive role in chromosome inactivation.


Previous knowledge

Female mammals have two X chromosomes, one of which is inactivated in each somatic cell. It is believed that a "random choice" mechanism operates to choose one of the X chromosomes for inactivation. The chromosome that is chosen for inactivation is subsequently heterochromatinized.

Previous work had shown that the X inactivation center (Xic) -- a 450 kb region on the X chromosome -- was necessary for X chromosome inactivation. It was known that a gene called the X-controlling element (Xce) -- which maps to the Xic region of the X chromosome -- is the key gene involved in X chromosome choosing, with different alleles of this gene lending the chromosome that bears them a different tendency to be inactivated.

The Xist gene had also been localized to the Xic region of the X chromosome. It was also essential for X chromosome inactivation, but seemed to function downstream of the choosing mechanism since an X chromosome with a large deletion of the Xist gene would be chosen for inactivation but would not become inactive. The Xist gene codes for a 15kb RNA fragment, that is stabilized during X inactivation and coats the inactive X chromosome, leading to the formation of heterochromatin.

Rodents exhibit an interesting variation in this general pattern in their extraembryonic cells. In these cells the maternal X chromosome is never inactivated. This phenomenon is called "maternal imprinting" and is coded for by an unknown marker. Previous investigations had suggested that the "maternal imprinting" mechanism was also upstream of the Xist gene function.


The results of the paper

The authors had previously shown that female embryos heterozygous for a deletion of part of the Xist gene (from the middle of exon 1 to the end of exon 5) have different outcomes depending on which parent they inherit the mutation from. Females who inherit the mutation from their fathers die during embryogenesis. The extraembryonic cells of these females die because both the X chromosomes remain active -- the maternal one due to imprinting, and the paternal one because it has a non-functional Xist gene.

Females heterozygous for the maternally inherited mutation are on the other hand born healthy with the wild-type (paternal) X chromosome inactivated in every cell. The authors had supposed that these pups had lost all cells during development in which the maternal X chromosome had been chosen for inactivation, similar to other observed X chromosome mutations like Searle's X -autosome 16 translocation. However the following figures show that this is incorrect and that the Xist +/- females who inherit the mutation from their mothers actually exhibit primary nonrandom inactivation of the paternal wild-type X chromosome.

Fig. 1A compares the birthweights of wild-type mice (WT) with those of mice heterozygous for Searle's translocation (XX16/16X). The birthweights of the heterozygotes are on average 79% of those of their wild-type littermates. This is consistent with the idea that cells randomly choose either X chromosome for inactivation, and cells that choose the maternal chromosome for inactivation die since they are unable to inactivate the mutant chromosome. The death of cells which choose maternal X chromosomes for inactivation leads to a lower birthweight since the embryos are unable to compensate completely for this considerable cell loss.

Fig. 1B compares the birthweights of wild-type mice (WT) with those of mice heterozygous for the Xist deletion (XXXist-). The two birthweight averages were exactly the same, which seems to indicate that there is no cell loss as observed with the Searle's translocation heterozygotes. The alternative explanation would be that cell loss still occurs, but there is compensatory growth in the embryos, which leads to normal birthweight values. This alternative explanation is addressed in Figure 2.

Fig. 2 compares the size of female Xist +/- and Xist +/+ (wild-type) embryos at 7.5 (A), E8.5 (B) and E9.5 (C) days of gestation [The letter E followed by a number indicates the number of days of gestation]. The two genotypically different embryos do not exhibit any gross differences in size, which makes it unlikely that massive cell loss is taking place at this point. The period of embryogenesis chosen was that in which female embryos heterozygous for the Searle's translocation experience considerable cell loss. Therefore Fig. 2 strengthens the author's claim that unlike in the case of Searle's translocation heterozygotes, Xist +/- females do not undergo cell loss after random X inactivation but rather non-randomly inactivate their wild-type chromosome. Fig. 2 is not conclusive however, because there is once again an alternative explanation. The X inactivation could still be random, but the cell loss (and the compensatory growth) could be much more gradual than for Searle's translocation heterozygotes.

To test whether any individual cells retain two active X chromosomes soon after X chromosome choosing, the authors used RNA fluorescence in situ hybridization (RNA FISH). The purpose of Fig. 3 is to provide us with some background when looking at Fig. 4. Fig. 3A shows the relation of the probes used in the RNA FISH to the original Xist gene. Two probes were used; one that recognized both the wild-type and the truncated Xist RNA and another that recognized only the wild type RNA.

Fig. 3A shows the expected result for wild-type female cells that have one X chromosome inactivated. Before the X inactivation (about E6), both copies of Xist are transcribed to form an unstable product and this gives a low-level biallelic signal. Later (E7.5 and after) wild-type females have one of their X chromosomes inactivated and therefore a differential biallelic signal or a high signal is expected. In wild-type males and Xist males, the X chromosome is never inactivated and these served as the controls giving us a low-level monoallelic signal. Xist would give us one of the two results shown in Fig. 3C, depending on whether they underwent random or non-random X inactivation. The authors designed this experiment very well, since either possible result would be fairly unambiguous about the nature of X inactivation in Xist +/- females.

Fig. 4 is the cornerstone of the paper and provides the conclusive evidence that X inactivation is indeed non-random in Xist +/- females. The cells in Fig. 4A and 4B are Xist- male embryo cells. As expected the full probe shows a single monoallelic signal in Fig. 4A, indicating that the Xist RNA is unstable for X chromosomes that have not been chosen for inactivation. Therefore Fig. 4A is a positive control for the full probe. Fig. 4B is both a positive control and a negative control. The Pgk-1 probe gives a positive signal and serves as a positive control for the experiment. The deletion probe does not give a signal, showing that it does not bind to the truncated Xist transcript, and it serves as the negative control for the deletion probe. The cells in Fig. 4C are wild-type female embryo cells. The biallelic differential / high level monoallelic signal seen is expected since one of the X chromosomes is inactivated, and its Xist transcript should be present in high levels. Fig. 4D and 4E show the real experimentals, and are Xist +/- female embryo cells. The full probe gives a high level monoallelic / biallelic differential pattern in Fig. 4D and this shows that one of the chromosomes is always inactivated. To confirm that the wild-type chromosome is indeed inactivated, the authors used the deletion probe in the same situation and obtained a high level monoallelic pattern, showing that it is indeed the wild-type Xist transcript that is being overexpressed. Fig. 4 is an example of beautiful experimental design, with all the right controls, and it is able to perform its central role in this paper to perfection.


What the results prove

There are two main points that the authors attempt to make.

Addressing the second point first, I think that it is proven conclusively by Fig. 4. It is proved by a positive result, and the number of cells examined (283 for Fig. 4D and 470 for Fig. 4E) make it unlikely that the results are a statistical anomaly. I also think that Fig. 4 could have been even more convincing if the Pkg-1 probe had been used in all panels, just to be doubly sure of the positive control. On the other hand, there is always some signal from one of the two Xist probes (except in Panel 4B, where the Pkg-1 probe was used to give a positive result), and this serves as a positive control for those panels.

At first glance, the first point about cell death is not conclusively proven. Alternative explanations could be offered for the gross observations. We must however note that Fig. 4 shows that the second point to be true. Since the second point is conclusively proven, the first point becomes irrelevant since it was merely a pointer to the real information that the authors wanted to present. Even if cell death occurs it is most likely not related to the failure of X inactivation.


The authors' hypotheses and their validity

The results of this paper negate the single negative blocking factor hypothesis, because Xist plays a positive role in X chromosome choosing. The authors propose the following hypotheses:


Future directions

The following avenues of research would be interesting to follow to help us understand better the role of Xist in X inactivation.

[Home] [Back to the Davidson Molecular Biology Page]


Comments? Questions? Suggestions? Please email rakarnik@davidson.edu.