Identification of JAK/STAT signalling componentsby genome-wide RNA interferencePatrick Mu¨ller1, David Kuttenkeuler2, Viola Gesellchen2, Martin P. Zeidler1& Michael Boutros2Signalling pathways mediating the transduction of informationbetween cells are essential for development, cellular differen-tiation and homeostasis1. Their dysregulation is also frequentlyassociated with human malignancies. The Janus tyrosine kinase/signal transducer and activator of transcription (JAK/STAT) path-way represents one such signalling cascade whose evolutionarilyconserved roles include cell proliferation and haematopoiesis2.Here we describe a systematic genome-wide sur vey for genesrequired for JAK/STAT pathway activity. Analysis of 20,026 RNAinterference (RNAi)-induced phenotypes in cultured Drosophilamela nogaster haemocyte- like cells identified interacting genesencoding 4 known and 86 previously uncharacterized proteins.Subsequently, cell-based epistasis experiments were used to clas-sify these proteins on the basis of their interaction with knowncomponents of the signalling cascade. In addition to multiplehuman disease gene homologues, we have found the tyrosinephosphatase Ptp61F and the Drosophila homologue of BRWD3,a bromo-domain-containing protein disrupted in leukaemia3.Moreover, in vivo analysis demonstrates that disrupted dBRWD3and overexpressed Ptp61F function as suppressors of leukaemia-like blood cell tumours. This screen represents a comprehensiveidentification of novel loci required for JAK/STAT signalling andprovides molecular insights into an important pathway relevantfor human cancer. Human homologues of identified pathwaymodifiers may constitute targets for therapeutic interventions.Developmental genetic screens in Drosophila have identified mul-tiple JAK/STAT pathway components on the basis of their segmenta-tion phenotype4–6, and subseq ue nt analys i s of the pathway hascharacterized evolutionarily conserved roles during immuneresponses, haematopoiesis and cellular proliferation7–10. The JAK/STAT signalling cascade in Drosophila comprises the extracellularligand Unpaired (Upd)5, a transmembrane receptor with homolog yto the interleukin 6 (IL-6) receptor family termed Domeless(Dome)11, a single Janus tyrosine kinase (JAK) called Hopscotch(Hop)4and the Stat92E transcription factor6,12(Fig. 1a). Knownregulators of JAK/STAT signalling, including a family of SOCS-likegenes13, dPIAS/Su(var)2-10 (ref. 14) and STAM (ref. 15), are func-tionally conserved and were identified based on their homology tocomponents originally characterized in mammalian cell culturestudies2. Although successful in identifying the pathway membersUpd, Dome, Hop and Stat92E, it is probable that forward geneticapproaches have missed components, possibly due to non-saturatingmutagenesis, genetic redundancy or phenotypic pleiotropy.In order to identify novel pathway components and circumventlimitations of classical genetic screens, we have undertaken a gen-ome-wide RNAi screen, a powerful technique for the identification ofnew components of diverse cellular pathways16–19. To this end, wedevised a quantitative assay for JAK/STAT signalling activity incultured Drosophila cells by multimerizing a STAT92E-binding sitefrom the Draf promotor20to generate the 6 £ 2 £ DrafLuc fireflyluciferase reporter. Given the role for JAK/STAT signalling in hae-matopoiesis9, we used Drosophila haemocyte-like Kc167cells becauseof their endogenous ability to respond to pathway activation (Fig. 1b).On transfection of the 6 £ 2 £ DrafLuc repor ter and a plasmid toLETTERSFigure 1 | Genome-wide RNAi screen for JAK/STAT signalling factors.a, Schematic representation of the Drosophila JAK/STATsignalling pathway.TF indicates transcription factor. b, Knock down of known JAK/STATcomponents leads to loss of pathway induction by Upd, whereas knock downof lacZ, toll and relish show no effect. The red line indicates an approximately70-fold reporter induction relative to negative control dsRNA. Error barsrepresent standard deviations of six exper iments. c, Schematic diagram ofscreening approach. A total of 20,026 dsRNAs were screened in duplicate in384-well plates before computational analysis and re-testing. FL indicatesfirefly luciferase; RL indicates Renilla luciferase. d, Q–Q plot of normallydistributed quantiles against actual pathway screening result quantiles. A fitto a normal distribution is represented by the red line. Tails of positively andnegatively interacting dsRNAs at each extreme with a z-score threshold of.2 and ,22 represent RNAi experiments with significant phenotypes.1Department of Molecular Developmental Biology, Max Planck Institute for Biophysical Chemistry, Am Fassberg 11, 37077 Go¨ttingen, Germany.2Boveri-Group Signaling andFunctional Genomics, German Cancer Research Center, Im Neuenheimer Feld 580, 69120 Heidelberg, Germany.Vol 436|11 August 2005|doi:10.1038/nature03869871© 2005 Nature Publishing Groupconstitutively express the ligand Upd, a robust induction of reportergene activity was observed (Fig. 1b). We examined whether depletionof known pathway components by RNAi21modifies JAK/STATsignalling activity in Kc167cells. We assessed the effect of double-stranded (ds)RNAs targeting the mRNA of the genes dome, stat92Eand hop, as well as dsRNAs directed against the negative regulatorssocs36E and dPIAS. As shown in Fig. 1b, knock down of JAK/STATcomponents results in significant changes in reporter activity,whereas reporter activity in uninduced cells remains at low levels(Fig. 1b).We then set out to systematically identify genes required for JAK/STAT signalling by generating a library of 20,026 dsRNAs targeting,91% of the predicted transcripts in the Drosophila genome (seeSupplementary Information). Using this librar y we performedduplicate genome-wide screens as outl ined in Fig. 1c and Sup-plementary Fig. S1. After computational analysis to identify candi-date pathway modifiers (Fig. 1d; see also SupplementaryInformation), dsRNAs were resynthesized an d assayed with anindependent reporter, derived from the promoter of the pathwaytarget gene socs36E13, to exclude reporter-specific artefacts. Theseapproaches confirmed the identification of 67 dsRNAs that decreasepathway activity (putative positive regulators) and 24 dsRNAs thatFigure 2 | Analysis of JAK/STAT activity modulators. a, Schematicrepresentation of positive (red) and negative (green) regulator locidistributed within the Drosophila genome. An