Human embryonic stem cells (hESCs) are genetically unstable over time in culture, which raises concerns about their clinical safety. Epigenetic changes, heritable controls over gene expression that do not alter DNA sequence, in hESCs may contribute significantly to this instability. Two studies published recently in the Proceedings of the National Academy of Sciences used complementary methods to examine X-chromosome inactivation in hESCs, reporting that this particular epigenetic change shows an incredible amount of variability between hESC lines regardless of the source of derivation.

Cells have evolved to function with only one working copy of each X-linked gene, and X-chromosome inactivation (XCI) is a process by which one of the two X chromosomes is completely silenced in each female cell. XIST is a noncoding RNA that is required to initiate silencing during XCI and is therefore a marker of cells that have undergone this process. Few studies that examine XCI in hESCs have been carried out, and so far the expression patterns seen for XIST have been inconsistent. To use stem cells therapeutically, a strict set of quality control criteria must be met, so epigenetic variability between cell lines could present a significant problem.

Jeannie Lee's group at the Harvard Medical School in Boston examined a panel of hESC lines for the presence of XIST RNA and discovered that hESC lines fall into one of three classes. The first class expressed XIST only upon differentiation, suggesting that both X chromosomes are active in hESCs, and that inactivation is initiated upon differentiation. The second class expressed XIST throughout culture, indicating that these cells may have already completed XCI. The final class was puzzling; these lines did not express XIST at all, even though the cells still contained two X chromosomes. Further experiments showed that in the majority of nuclei, one X chromosome remained in a region with little transcriptional activity, leading to the conclusion that XCI may have occurred, but the cells subsequently lost XIST expression1.

Guoping Fan's group at the David Geffen Medical School, University of California, Los Angeles, came to a similar conclusion after looking at a panel of XCI markers in three hESC lines. They noticed differential expression of the markers, and they also found that the markers were more likely to be lost in particular culture conditions. Surprisingly, most X-linked genes were still expressed monoallelically even when XIST expression was lost, indicating that inactivation had indeed occurred. However, the group also noticed that some X-linked genes were reactivated once XIST expression was lost. This reactivation was accompanied by further epigenetic changes including demethylation of the promoter regions of reactivated genes. The group suggests each individual cell might have its own unique reactivation profile because of inherent genetic and epigenetic differences2.

Both Fan and Lee are concerned that culture conditions may play a vital role in the epigenetic variability they observed, and Lee aims to explore this further. In her experiments, XIST expression was eventually lost in all cell lines. “What does it tell us about how appropriate culture conditions are?” she asks. Indeed, Fan's group even found epigenetic differences between subcultures of the same cell line.

Lee also has concerns that hESCs are inherently flawed and may simply not perform properly ex vivo. “I am cautious about how to move forward with clinical use,” Lee says. Fan adds, “Loss of XCI has been observed in cancer cells, and nonrandom XCI is associated with mental retardation.” In mice, XCI is linked to differentiation, so the conclusions of these studies could have implications for hESC pluripotency. The role of XIST in XCI maintenance and the reason for its irreversible loss in hESCs is unclear; until this is fully understood, the clinical potential of hESCs will remain in question.