X inactivation counting and choice is a stochastic process: evidence for involvement of an X-linked activator

Female mammalian cells achieve dosage compensation of X-encoded genes by X chromosome inactivation (XCI). This process is thought to involve X chromosome counting and choice. To explore how this process is initiated, we analyzed XCI in tetraploid XXXX, XXXY, and XXYY embryonic stem cells and found that every X chromosome within a single nucleus has an independent probability to initiate XCI. This finding suggests a stochastic mechanism directing XCI counting and choice. The probability is directly proportional to the X chromosome:ploidy ratio, indicating the presence of an X-encoded activator of XCI, that itself is inactivated by the XCI process. Deletion of a region including Xist, Tsix, and Xite still results in XCI on the remaining wild-type X chromosome in female cells. This result supports a stochastic model in which each X chromosome in a nucleus initiates XCI independently and positions an X-encoded trans-acting XCI-activator outside the deleted region.     

De novo DNA methylation independent establishment of maternal imprint on X chromosome in mouse oocytes

In female mouse embryos, the paternal X chromosome (Xp) is preferentially inactivated during preimplantation development and trophoblast differentiation. This imprinted X-chromosome inactivation (XCI) is partly due to an activating imprint on the maternal X chromosome (Xm), which is set during oocyte growth. However, the nature of this imprint is unknown. DNA methylation is one candidate, and therefore we examined whether disruptions of the two de novo DNA methyltransferases in growing oocytes affect imprinted XCI. We found that accumulation of histone H3 lysine-27 trimethylation, a hallmark of XCI, occurs normally on the Xp, and not on the Xm, in female blastocysts developed from the mutant oocytes. Furthermore, the allelic expression patterns of X-linked genes including Xist and Tsix were unchanged in preimplantation embryos and also in the trophoblast. These results show that a maternal disruption of the DNA methyltransferases has no effect on imprinted XCI and argue that de novo DNA methylation is dispensable for Xm imprinting. This underscores the difference between imprinted XCI and autosomal imprinting.