What silences genes — promoter methylation, location within heterochromatin or late cell-cycle replication? The answer seems to lie not with any one of these events but with all of them (and maybe more). A stable gene-silencing event that requires the (mostly unknown) interactions of many gene-silencing processes is X inactivation. To tease apart such interactions, Hansen et al. turned to cells from individuals with defective methylation to investigate the relationship between promoter methylation, replication timing and gene silencing.

The authors used fibroblasts and lymphoblasts from patients with ICF (immunodeficiency, centromeric instability and facial anomalies) syndrome — a human disorder caused by mutations in DNMT3B , which encodes the 'de novo' DNA methyltransferase 3β. These mutations do not cause genome-wide methylation defects but hypomethylation at particular heterochromatic regions, such as at pericentromeric satellite DNA and the inactive X chromosome.

From their studies, Hansen et al. found that in both primary and transformed ICF cells, normally hypermethylated CpG islands at the promoters of several inactive X genes were hypomethylated. However, this hypomethylation did not lead to a wholesale escape from transcriptional silencing — only two of the eleven genes that were studied, G6PD and SYBL1 , showed biallelic expression in ICF cells. The authors then found that an advance in the replication timing of these genes correlated with their escape from inactivation — in ICF fibroblasts, for example, G6PD and SYBL1 replicate earlier in the cell cycle than they do in normal cells.

But the relationships between silencing, replication timing and methylation are far from being this simple — several hypomethylated, inactive X genes from ICF cells also replicate earlier than normal but remain silent, perhaps because their replication is still not as early as that of active X alleles. Also, hypomethylated promoter chromatin is sensitive to nuclease digestion at both reactivated and silent genes on the inactive X from ICF cells. These results indicate that advanced replication can occur without transcription, and that it might therefore be a cause rather than a consequence of escape from inactivation. Late replication on the inactive X might be maintained or established by DNMT3B; in its absence, other factors might assume this role with less fidelity.

Much remains to be learnt about how higher-order genomic organization influences nuclear compartmentalization and replication timing to control gene expression. ICF cells look set to provide a handy tool for such investigations and may provide new insights into how deregulated gene expression contributes to the ICF phenotype.