19. Novick, P. & Guo, W. Ras family therapy: Rab, Rho and Ral talk to the exocyst. Trends Cell Biol. 12,247–249 (2002).20. Cheatham, B. GLUT4 and company: SNAREing roles in insulin-regulated glucose uptake. TrendsEndocrinol. Metab. 11, 356–361 (2000).21. Chamberlain, L. H. & Gould, G. W. The v-and t-SNARE proteins that mediate Glut4 vesicle fusion arelocalised in detergent-insoluble lipid rafts present on distinct intracellular membranes. J. Biol. Chem.277, 49750–49754 (2002).22. Sherman, L. A., Hirshman, M. F., Cormont, M., Le Marchand-Brustel, Y. & Goodyear, L. J. Differentialeffects of insulin and exercise on Rab4 distribution in rat skeletal muscle. Endocrinology 137, 266–273(1996).23. Cormont, M. et al. Potential role of Rab4 in the regulation of subcellular localization of Glut4 inadipocytes. Mol. Cell. Biol. 16, 6879–6886 (1996).24. Shisheva, A. & Czech, M. P. Association of cytosolic Rab4 with GDI isoforms in insulin-sensitive 3T3-L1 adipocytes. Biochemistry 36, 6564–6570 (1997).25. Millar, C. A., Shewan, A., Hickson, G. R., James, D. E. & Gould, G. W. Differential regulation ofsecretory compartments containing the insulin-responsive glucose transporter 4 in 3T3-L1adipocytes. Mol. Biol. Cell 10, 3675–3688 (1999).26. Chang, L., Adams, R. D. & Saltiel, A. R. The TC10-interacting protein CIP4/2 is required for insulin-stimulated Glut4 translocation in 3T3L1 adipocytes. Proc. Natl Acad. Sci. USA 99, 12835–12840(2002).27. Lin, D. et al. A mammalian PAR-3–PAR-6 complex implicated in Cdc42/Rac1 and aPKC signallingand cell polarity. Nature Cell Biol. 2, 540–547 (2000).28. Thurmond, D. C. et al. Regulation of insulin-stimulated GLUT4 translocation by Munc18c in 3T3L1adipocytes. J. Biol. Chem. 273, 33876–33883 (1998).29. Hwang, J. B. & Frost, S. C. Effect of alternative glycosylation on insulin receptor processing. J. Biol.Chem. 274, 22813–22820 (1999).Supplementary Information accompanies the paper on Nature’s website(ç http://www.nature.com/nature).Acknowledgements This work was supported by grants from the National Institutes of Health.We thank T-H. Chun for helpful discussions.Competing interests statement The authors declare that they have no competing financialinterests.Correspondence and requests for materials should be addressed to A.R.S.(e-mail: [email protected])...............................................................NoiseineukaryoticgeneexpressionWilliam J. Blake*, Mads Kærn*, Charles R. Cantor & J. J. CollinsCenter for BioDynamics, Center for Advanced Biotechnology, BioinformaticsProgram, and Department of Biomedical Engineering, Boston University,44 Cummington Street, Boston, Massachusetts 02215, USA* These authors contributed equally to this work.............................................................................................................................................................................Transcription in eukaryotic cells has been described as quantal1,with pulses of messenger RNA produced in a probabilisticmanner2,3. This description reflects the inherently stochasticnature4–9of gene expression, known to be a major factor inthe heterogeneous response of individual cells within a clonalpopulation to an inducing stimulus10–16.HereweshowinSaccharomyces cerevisiae that stochasticity (noise) arisingfrom transcription contributes significantly to the level of hetero-geneity within a eukaryotic clonal population, in contrast toobservations in prokaryotes15, and that such noise can be modu-lated at the translational level. We use a stochastic model oftranscription initiation specific to eukaryotes to show thatpulsatile mRNA production, through reinitiation, is crucial forthe dependence of noise on transcriptional efficiency, highlight-ing a key difference between eukaryotic and prokaryotic sourcesof noise. Furthermore, we explore the propagation of noise in agene cascade network and demonstrate experimentally thatincreased noise in the transcription of a regulatory proteinleads to increased cell–cell variability in the target gene output,resulting in prolonged bistable expression states. This result hasimplications for the role of noise in phenotypic variation andcellular differentiation.To explore the effects of transcriptional variation and control onthe level of noise in eukaryotic gene expression, we used both nativeand artificial modes of transcriptional regulation in the yeast GAL1promoter (Fig. 1a). In its natural context, the GAL1 promoter isactivated in response to galactose (in the absence of preferentiallymetabolized glucose) through an upstream activation sequence(UASG) composed of multiple binding sites for the transcriptionalactivator Gal4p. Like many eukaryotic activators17, Gal4p acts byrecruiting protein complexes involved in chromatin remodellingand the ordered assembly of a transcription preinitiation com-plex18,19. Because Gal4p is a galactose-dependent transcriptionalactivator, activation-based expression from the GAL1 promoter iseffectively modulated with galactose. As a second mode of tran-scriptional control, distinct from the native complexity of theyeast galactose-utilization pathway, we constructed an artificial,Tet-responsive GAL1 promoter (PGAL1*) by inserting tandem tetoperators (2£tetO2) downstream of the GAL1 TATA box. In con-trast to Gal4p-mediated activation, bound Tet repressor (TetR)might act by sterically hindering the assembly of the transcriptionalmachinery, effectively repressing expression from PGAL1*. TetR-mediated repression can be relieved by the addition of the chemicalinducer anhydrotetracycline (ATc), which binds directly to TetR.Constitutive expression of TetR therefore allows rheostat-like con-trol of PGAL1*transcriptional efficiency through the use of varyinglevels of ATc. The gene encoding the yeast-enhanced green fluor-escent protein (yEGFP) was expressed from PGAL1*as a quantifiablereporter, and fluorescence histograms were obtained from flowcytometric measurement of similarly sized cells containing chromo-somally integrated, single genetic copies of each construct.Transcription from PGAL1*is modulated over a broad dynamicrange by both native and artificial modes of regulation (Fig. 1b),allowing a direct comparison between galactose- and ATc-mediatedFigure 1 Transcriptional control of PGAL1*. a, TetR, expressed from PGAL10*, repressesexpression of yEGFP. Anhydrotetracycline (ATc) and galactose (GAL) are required toinduce yEGFP expression. Transcription