Nature 435, 1275–1279 (2005)

An earlier paper1 by two of us, in which it was concluded that expression of a hairpin RNA homologous to ura4 RNA in Schizosaccharomyces pombe results in the production of siRNAs that bring about methylation of histone H3 on lysine 9 and recruitment of Swi6 to the ura4 gene, has been retracted2. This is because we are unable to reproduce the experiments involving expression of hairpin ura4 RNA because the plasmid sh-ura4SE-280, or fission yeast strains that contain it, do not exist. Attempts to reproduce these observations with other ura4-hairpin constructs have failed.

 As a result, we cannot verify the observations reported in our Letter to Nature and shown in Figures 1, 2 and 4a (lower panel) because these utilize the same plasmid sh-ura4-280 hairpin construct. We therefore retract the conclusions that transcription of an siRNA homologous target is required to bring about silent chromatin assembly, that transcription of an siRNA homologous target by T7 is not sufficient to bring about silent chromatin assembly, and that transcription of an siRNA homologous target is required to allow association of Ago1 with that target.

However, we have confirmed that truncation of the RNApol II CTD (removal of 17/28 heptad repeats) does alleviate silencing at centromeres, that centromeric siRNAs remain detectable (Figure 3), and that RNA pol II and Ago1 co-immunoprecipitate. In addition, the expression-profiling data generated for rpb1-11 are robust.

V.S. has declined to sign this retraction as she maintains that the original conclusions are correct.