For a handful of mammalian genes it matters whether it is the maternal or the paternal allele that is expressed. This parental imprinting is achieved through an epigenetic mechanism, which has been extensively studied for a few loci. New work from Shirley Tilghman's group sheds light on the mechanism that ensures maternal expression of a gene cluster on mouse chromosome 7. Importantly, these findings add to the appealing possibility that maternally and paternally imprinted genes are silenced by fundamentally different mechanisms.

Among the imprinted genes on mouse chromosome 7 is Kcnq1 . It encodes a potassium channel and, although its early embryonic expression is exclusively maternal, it later becomes biallelic. Intron 10 of Kcnq1 contains a promoter for a non-coding paternally expressed transcript, Kcnq1ot1 , and it is this promoter that had previously been shown to be required for paternal silencing of several neighbouring genes, including Cdkn1c , which is maternally expressed throughout life. The authors show that a small, 244-bp deletion of the promoter abolishes Kcnq1ot1 transcription. Paternal, but not maternal, inheritance of this deletion abolished silencing of all the imprinted genes in the cluster on the paternal allele, showing that transcription of Kcnq1ot1 is required for their imprinting.

Using homologous recombination, the authors inserted a polyA-based transcriptional stop element to significantly shorten the Kcnq1ot1 transcript. They show that premature termination of the paternal but not the maternal allele abolished imprinting in the cluster, indicating that it is not the act of transcription but the transcript itself, or its elongation, that is necessary for imprinting in this region. The mechanism by which these non-coding RNAs silence their target genes awaits elucidation, although the authors show that the highly conserved repeats in the 5′ end of Kcnq1ot1 are dispensable for silencing.

Two imprinting mechanisms have been described to date: one involves an imprinting control region (ICR) at the well-studied H19 / Igf2 locus. On the maternal chromosome the ICR acts as an insulator to inhibit expression of Igf2, whereas DNA methylation of the ICR on the paternal allele spreads to and silences the H19 gene. The other mechanism involves an ICR that acts as a promoter for a paternally expressed non-coding RNA, first described at the Igf2r locus by Denise Barlow and colleagues. Kcnq1ot1 is the second example of a non-coding RNA with a direct role in the silencing of imprinted genes. The fact that the activity of the ICR in the first mechanism occurs on the maternal chromosome whereas in the second, described by Tilghman and colleagues, the activity applies to the paternal chromosome raises the possibility that the two mechanisms are gender-specific. Understanding how other clusters of imprinted genes are silenced should verify this interesting possibility.