Silencing by the imprinted Airn macro lncRNA: Transcription is the answer

The majority of the mammalian genome is transcribed into long non-coding (lnc) RNAs, transcripts longer than 200 nucleotides and devoid of protein-coding potential. Recent genome-wide mapping efforts led to the identification of thousands novel lncRNAs, but so far only few have been functionally analyzed.1 Fifteen years ago, our group discovered Airn (antisense to Igf2r RNA non-coding), the first autosomal lncRNA shown to have a silencing function.2 Airn is an imprinted gene transcribed from the paternal chromosome, in antisense orientation to the imprinted but maternally expressed Igf2r gene. Airn expression controls Igf2r imprinted expression by silencing it in cis on the paternal chromosome,2 but how Airn exerts its silencing function has long been a mystery.3,4 Now, a study published in Science shows that Airn silences Igf2r through transcription alone and not via its RNA product.5 The road leading to this discovery was long and winding. When we first identified Airn and deleted its expressed, paternal promoter, we observed de-repression of the normally silent paternal Igf2r allele.6 After this landmark experiment, three hypotheses were formulated to explain how Airn may silence Igf2r3: (1) the Airn promoter competes with the Igf2r promoter for common transcription factors or enhancers, (2) the Airn lncRNA coats the paternal chromosome, inducing its heterochromatization, (3) Airn transcription through the Igf2r promoter interferes with its expression. The promoter competition model was ruled out some years later by an experiment that prematurely terminated the 118kb-long Airn transcript by inserting a polyadenylation (polyA) cassette 3 kb after its transcription start site (TSS).2 Similarly to the Airn promoter deletion, the Airn truncation relieved paternal Igf2r silencing, indicating that either the Airn transcript or its transcription through the locus, but not the Airn promoter, is needed to silence Igf2r. At that time, no examples of transcriptional interference had been reported in mammals. The lncRNA field was dominated by one exemplary cis-silencing transcript, the Xist lncRNA, which coats one X chromosome in female mammals and recruits chromatin-modifying repressor complexes to inactivate it. We reasoned that if Airn silences Igf2r in an Xist-like fashion, broad domains of allele-specific repressive chromatin should cover the silenced chromosomal region. However, when we analyzed the distribution of repressive histone marks at the Igf2r locus, we found no broad heterochromatic domains, but only focal enrichment at the silent paternal Igf2r promoter.7 If Airn does not behave like Xist, might then mere transcription through the Igf2r promoter, rather than the transcript, be the answer?4 The transcriptional interference hypothesis, by implying no role for the RNA product, would also account for some unusual, “macro” lncRNA-like features of Airn, which, unlike Xist, is a predominantly unspliced, repeat-rich and highly unstable transcript.4 We therefore set out to test our hypothesis, reasoning that if transcription alone is important, then no Airn-specific sequences should be required to silence Igf2r. To perform genetic manipulations of the Airn locus and monitor their effects on Igf2r imprinted expression during embryonic development, we established an embryonic stem cell (ESC) differentiation system. By inserting a polyA cassette at four different positions from the Airn TSS, we then truncated the Airn transcript at its 3′ end, either before or after the Igf2r promoter. Interestingly, only the longer versions of Airn, overlapping the Igf2r promoter, were able to silence it (Fig. 1). Next, to exclude a silencing role for the Airn sequence located at its 5′ end, we moved the Airn promoter immediately before the Igf2r promoter. This modified Airn allele could also silence Igf2r (Fig. 1). Having excluded Airn sequences located both before and after the Igf2r promoter, the only region that remained to test was the one overlapping the Igf2r promoter itself. However, we had previously shown that this sequence can be replaced with an exogenous promoter, with no sequence or structural similarity with the Igf2r promoter, and still undergo Airn-mediated silencing.5 Together, this led us to conclude that Igf2r silencing does not require any part of the Airn lncRNA, but only transcription through the Igf2r promoter. Since we detected less capped Igf2r transcripts, indicative of transcription initiation, we propose that the Airn-transcribing RNA polymerase prevents Igf2r expression by running through its promoter and disrupting transcription initiation complex assembly.5 Figure 1.Airn can silence Igf2r at any time by transcriptional overlap. Left: early embryonic stem cell (ESC) differentiation. Airn silences Igf2r on the paternal allele and the Igf2r promoter is in an open chromatin conformation (top). Truncation ... Transcriptional interference models predict a strong promoter interfering with activity of a weak promoter. In accordance with this model, in another recent study we report that, although Airn transcription can repress Igf2r at any time, it does so less efficiently when paternal Igf2r expression is high, as in late ESC differentiation (Fig. 1).8 Once silenced, the paternal Igf2r promoter normally gains DNA methylation. Interestingly, the same study shows that Igf2r silencing requires continuous Airn expression, but only until DNA methylation is established, at which point Airn becomes dispensable, indicating a role for DNA methylation in safeguarding lncRNA-mediated silencing (Fig. 1).8 In conclusion, 15 years after the first report of the Airn lncRNA, the cumulated evidence indicates that transcription is indeed the answer.5 We hope our work will act as a “wake-up call” for the entire lncRNA field, as many more lncRNAs may act via their transcription alone.