Review of the Role of the Xist Gene in X Chromosome Choosing

from Marahrens, Y., Loring, J., and Jaenisch, R. Cell (1998), 92(5):657-664.


The X inactivation center (Xic) must be present in duplicate for an X chromosome to be inactivated. A random choice mechanism determines which X becomes transcriptionally active and which is silenced. The choice is thought to occur at the X-controlling element (Xce) which has been mapped to the Xic region on the chromosome. Xce is present in different allelic formats which differentially activate one particular X.

Skewed patterns of inactivation noted in some human females have been correlated with point mutations in the XIST gene promoter, suggesting that XIST is also involved in X chromosome inactivation. The Xist gene has also been mapped to Xic, but appears to be separate from Xce. The gene encodes for 15 kilobases of untranslated RNA which is stabilized during X inactivation. During chromosome choosing the Xist RNA transcript coats the inactive X chromosome, suggesting a structural role for Xist in heterochromatinization..

A previously proposed mechanism of inactivation for somatic tissue suggested that a diploid cell produces a negative "blocking" factor which randomly binds one X chromosome and protects it from subsequent inactivation. The unblocked X is therefore inactivated. Xce was identified as the potential target locus for blocking because it has been shown to participate in chromosome choosing. If this were so, the allelic differences noted for Xce could be accounted for by varying affinity for the blocking factor at each Xce locus.

Xist has been shown to be essential for X inactivation in vitro (in cultured cells) and in vivo (in mice). The in vitro study used female ES cell lines which maintain 2 active X chromosomes, but undergo random inactivation of one X when induced to differentiate. A large deletion was made in the Xist gene (Xce was left intact) on one chromosome. Results indicated that the mutant chromosome could be selected for inactivation, but failed to be inactivated. Xist was therefore determined to be necessary for heterochromatinization, acting downstream of the random choice mechanism.

Deletions in the Xist gene in vivo in mice extended 3Õ of the deletion used in the above in vitro study. Female pups with paternally inherited mutated Xist (Xist-) gene died because of failure of X inactivation in imprinted extraembryonic tissue. Imprinting refers to an unidentified mark which always causes extraembryonic tissue in rodent to inactivate the paternal X chromosome, leaving the maternal X resistant to inactivation. Therefore, the female pups with Xist- on the paternal X could not inactivate the mutant chromosome resulting in extraembryonic cell death. In contrast, female pups heterozygous for maternally inherited Xist- were healthy, with the wildtype X inactive in every cell. It was presumed that random inactivation of the X chromosomes in somatic cells produced balanced cells with inactive wildtype X, and unbalanced cells with two active X chromosomes. The unbalanced cells would subsequently be selected against, leaving only the balanaced cells detected in the heterozygous female pups. Previous research has indicated that such selective cell death of unbalanced cells does occur with X-linked mutations when random choice is active (e.g. in mice with SearleÕs X-autosome translocations). The resulting embryos were small due to the massive cell loss, but healthy.

Marahens and his colleagues were interested in explaining the mechanism of inactivation occuring in the female mice heterozygous for the deletion (Xist +/-). If random choice was occuring in the Xist +/- mice, then half of the embryonic cells would be expected to die because they would be unbalanced. The reultant embryos would thus be smaller and have reduced birthweights. Remarkably, their results indicated just the opposite occurance, causing them to reformulate a model by which X inactivation might occur.


Marahrens and company fist tested their hypothesis about the survival of the mutant Xist +/- epibalsts. If random inactivation was occuring in these animals, the resulting cells would be balanced and unbalanced. The unbalanced cells that had chosen, but failed to inactivate the Xist- chromosome would die. Figure 1 is the initial analysis of the occurrence of the cell death by comparison of weights of normal newborn mice to those of Xist +/- mice and mice with maternally inherited SearleÕs X-autosome translocation (previously published). Figure 1A is a histogram of the cell loss following random choice, as evidenced by the lower average relative birthweight of SearleÕs mice as compared to normal mice. Figure 1B is an graphic illustration comparing the relative birthweights of Xist +/- mice to wildtype mice (Xist +/+ females, Xist+ males, and Xist- males). In contrast to the decrease in weight noted for the Searle mice, Xist +/- mice had virtually identical weights to their wildtype counterparts indicating the absence of cell loss following X inactivation. The next experiment aimed to discern if the absence of cell loss in Xist +/- embryos was due to retention of the epiblast because every cell had chosen to inactivate the wildtype X (nonrandom inactivation) or if compensatory growth had occured following random inactivation and subsequent epiblast loss. Random inactivation occurs at about embryonic day 6 (E6). The majority of epiblast loss of unbalanced cells is known to occur between E7 and E8 in SearleÕs translocation. In addition, the reduction in embryo size does not result in the interruption of in utero developement by compensatory growth. Therefore, the Xist +/- mice and their wildtype littermates were examined at E7.5 (conceptus), E8.5 (embryo with amnion and partial yolk sac), and E9.5 for phenotypic differences. Figure 2 demonstrates that the Xist +/- embryos were not smaller than their wildtype littermates and that they bore no other phenotypic differences, indicating that massive unbalanced cell loss was not occuring. However, the possibility of more gradual cell loss with concommitant compensatory growth could not be ruled out.

Flourescence in situ hybridization (RNA FISH) was used to determine the X inactivation status of the individual cells of Xist +/- embryos. Unbalanced epiblast cells are readily detectable at E7 and E8 in SearleÕs translocation embryos. Therefore, if Xist +/- embryos retain cells with two active X chromosomes shortly after X inactivation in unbalanced cells, it would indicate the occurance of random choosing. Two probes were created (Figure 3A)--a full probe, and a deletion probe. The full probe recognizes both the wild type and mutant Xist region, while the deletion probe only recognizes the wild type chromosome.

Figure 3B illustrates the expected pattern of Xist expression for random X inactivation in wildtype (wt) cells. When using FISH in wildtype cells, an unstable Xist transcript expressed from both chromosomes prior to X inactivation can be visualized as two dot-like RNA signals (one at each locus). This expression pattern is known as low-level biallelic Xist expression (Figure 3B). Males cells exhibit a similar low-level monoallelic pattern at E6, but with only one RNA dot signal. During X inactivation, Xist RNA from one allele is stabilized and coats an X chromosome (differential biallelic pattern, Figure 3B). Shortly thereafter, the Xist gene on the active X chromosome is silenced and a high-level monoallelic pattern emerges (Figure 3B). Other studies indicate that at E7.5 the expression of the unstable transcript has been silenced in approximately 85% of cells, while the remaining still exhibit a single-dot FISH pattern.

Figure 3C illustrates the expected patterns of random and nonrandom inactivation in female Xist +/- cells using the deletion probe. Prior to inactivation, Xist +/- cells would be expected to demonstrate low-level monoallelic labelling of the single wildtype chromosome. If random choice was the primary mechanism, Xist +/- females would be expected to exhibit cells with both high-level monoallelic patterns, and with no FISH labelling at all because the deletion probe only hybridizes with the wt. If primary nonrandom selection is the primary mechanism, then all the Xist +/- cells would be expected to exhibit high-level monoallelic patterns of labelling.

Figure 4 demonstrates the results of FISH labeling of Xist expression in cells at E7.5. Panels A and B are cells from Xist- males (wt), C is from Xist +/+ females (wt), and D and E are from Xist +/- females. The full probe was used in A, C, and D, while the deletion probe was used in B and E. The full probe, recognizing both wt and mutant Xist RNAs, revealed that the positive control Xist - male embryos displayed low-level monoallelic expression patterns (Figure 4A) that were indistinguishable from wt males (previous results). Also as expected, wt females (Figure 4C) displayed complete X inactivation with high-level monoallelic or differential biallelic expression in every cell. Figure 4D indicates that high-level monoallelic expression was noted in all cells of Xist +/- embryos, indicating that the wt X chromosome was coated with Xist RNA, and that the mutant X was never chosen. When the deletion probe was used on Xist- males (negative control) no Xist signal was seen (Figure B). Application of the deletion probe to the Xist +/- females, hybridization patterns comfirmed that Xist +/- embryos undergo primary nonrandom X inactivation (Figure 4E) as predicted (Figure 3C). A probe recognizing Pgk-1 was used as an internal control to show that any absence of Xist signal was not due to technical failure (Figure 4B). In addition, cells from the female wt embryos produced allelic expression patterns at the expected frequencies--17% differential biallelic, and 83% high-level monoallelic. Xist +/- hybridization with deletion probes produced only high-level monallellic labelling (99%) and no low-level labeling (0%). Therefore, the authors conclude (and I agree) that Xist +/- undergo primary nonrandom inactivation of the wildtype X chromosome.


The observation that Xist +/- embryos undergo nonrandom inactivation is contrary to the prevailing theory of a negative factor blocking inactivation of a chromosome. Instead, the results imply that a positive element (an initiation factor) is required by an X chromosome if it is to be chosen for inactivation. Deletion of this element in the Xist +/- females resulting in the inability of the mutant chromosome to be inactivated. Therefore, the only option remaining is inactivation of the wild type X chromosome.

The authors suggest two roles for Xist in X inactivation. The results here imply that Xist is required for random choosing, although previous studies produced contradictory results with a large deletion of the Xist gene. Marahrens and his coworkers argue that the deletions from contradictory studies damaged the promoter and abolished transcription. The present deletion left all the transcription regulatory elements intact and may thus have produced the contradictory results. In the context of these results, data from the previous study indicate that the Xist RNA transcript may not be essential for chromosome choosing. Rather, heterochromatinization and choice may depend on a DNA element located in the present area of deletion.

The authors postulate two mechanims to explain X chromosome inactivation in wild-type and mutant female cells (Figure 5). The simplest mechanism proposes that a diploid female cell produces a single initiation factor that binds and inactivates one X chromosome (Figure 5A). Here, a unique initiation factor is unable to recognize the mutant Xist allele because the deletion removed the binding locus. Hence, the only available binding site for inactivation is on the wild-type X. The more complex mechanism accounts for the inherent counting mechanims in mammalian cells that prevent activation of more than one X in supernumerary cells (Figure 5B). In this model, a unique blocking factor (the stop sign) reversibly binds an X and alternates between the two chromosomes. During X inactivation, the initiation factor is expressed and binds all chromosomes not bound to the blocking factor. Therefore in the Xist +/- embryos the binding sites for the initiation factors are deleted, forcing the blocking factors onto the mutant allele and producing inactivation of the wildtype X.

Methodical site-directed mutagenisis along the Xist gene will help define the functional elements, especially the DNA binding site for the initiation factor. Deletion of the binding site will produce results similar to what we see in this study. Generation of anti-idiotype antibodies to the binding site may then help isolate the initiation factor that binds here. If such a factor exists, then the antibody will bind to the factor and nothing else, and can then be isolated on an immunoblot or via immunoprecipitation.

Back to homepage

Send your comments, suggestion, fan mail to :