Nature 459, 455–459 (2009); doi:10.1038/nature07954

We showed that glycosylation of a histone methyltransferase (MLL5) triggered retinoic-acid-induced granulopoiesis. Although we believe that the key finding and conclusions are still valid, recently detected image manipulation in the published figures undermines our full confidence in the integrity of the study. We therefore wish to retract this Letter. Correspondence should be addressed to S. Kato.