CORRESPONDENCE

67% of the genes recovered by Oh et al. that 78% of 5-kb upstream regions contain Coleen T Murphy contain one or both of these sequences in one or more of these motifs. Thus, a ran- Lewis-Sigler Institute of Integrative Genomics their 5-kb upstream regions, we disagree dom selection of genes should have met this and Department of , Princeton with the authors’ conclusion that their pro- minimal value, regardless of approach. University, Princeton, New Jersey 08544, USA. cedure was substantially more effective than e-mail: [email protected] or Cynthia Kenyon our microarray analyses in recovering these [email protected] sequences. Department of and , Additionally, a search of all C. elegans University of California San Francisco, San 1. Oh, S.W. et al. Nat. Genet. 38, 251–257 (2006). genes (http://rsat.ulb.ac.be/rsat/) shows Francisco, California 94943, USA. 2. Murphy, C.T. et al. 424, 277–283 (2003).

Oh et al. reply: by DAF-16 cannot be determined solely Seung Wook Oh, Arnab Mukhopadhyay, Bharat In their original study, Murphy et al.1 from microarray and bioinformatic L Dixit, Tamal Raha, & Heidi A Tissenbaum listed 514 genes whose expression was analyses. Indeed, numerous studies, Program in Gene Function and Expression, affected by DAF-16 activity. For 58 of including our own, have shown that Program in Molecular Medicine, University these genes, canonical DAF-16 binding not all consensus binding sites for a of Massachusetts Medical School, Worcester, sites were shown. For the remaining transcription factor are occupied in Massachusetts 01605, USA. 456 genes, there was no indication that vivo and that transcription factors e-mail: [email protected] canonical DAF-16 binding sites were can also bind to nonconsensus sites2–4 present, and we assumed erroneously (reviewed in refs. 5,6). For these reasons, Michael R Green Howard Hughes Medical Institute, that these genes lacked a consensus methods based on transcription factor University of Massachusetts Medical School, http://www.nature.com/naturegenetics DAF-16 binding site. Based on the new binding, rather than expression, have Worcester, Massachusetts 01605, USA. information provided by Kenyon and been developed to identify direct targets Murphy, we agree that our estimate of of transcription factors (for example, 1. Murphy, C.T. et al. Nature 424, 277–283 (2003). 11% reflected a misunderstanding of chromatin immunoprecipitation (ChIP), 2. Hickabottom, M., Parker, G.A., Freemont, P., 7 Crook, T. & Allday, M.J. J. Biol. Chem. 277, 47197– their data. We also agree with Kenyon and ChIP-on-chip and ChIP-PET ). In our 47204 (2002). Murphy that the random occurrence of study4, we identified putative DAF-16 3. Ramirez-Carrozzi, V. & Kerppola, T. Mol. Cell. Biol. 23, 1737–1749 (2003). potential DAF-16 binding in promoters is target genes using ChIP, a direct measure 4. Oh, S.W. et al. Nat. Genet. 38, 251–257 (2006). relatively high (78%, according to of DAF-16 binding, and examined their 5. Sikder, D. & Kodadek, T. Curr. Opin. Chem. Biol. 9, their estimate). Thus, the percentage of involvement in DAF-16–dependent 38–45 (2005). 6. Holstege, F.C. & Clevers, H. Cell 124, 21–23 direct DAF-16 targets included among the processes using a variety of functional (2006). list of genes whose expression is affected assays. 7. Wei, C.L. et al. Cell 124, 207–219 (2006). Nature Publishing Group Group Nature Publishing 6 200 ©

398 VOLUME 38 | NUMBER 4 | APRIL 2006 | NATURE GENETICS