Two-color GFP imaging demonstrates cell-autonomy of GAL4-driven RNA interference in Drosophila.

Abstract

RNA interference (RNAi), the specific inhibition of gene expression by homologous double-stranded RNA (dsRNA) (Fire et al., 1998), has proven to be a broadly applicable approach for assessing loss of gene function in a variety of model organisms. An important characteristic of RNAi in plants (called posttransgenic gene silencing, or PTGS), the nematode Caenorhabditis elegans, and invertebrate tissue culture cells is that the inhibitory effect appears to cross cell boundaries. Systemic spreading of RNA interference was first described in plants (Palauqui et al., 1997), where PTGS was shown to transmit from silenced tobacco graft tissue to an unsilenced host. Similar indications of the transmission of dsRNAmediated gene-silencing have been obtained in animals. For example, gene expression is inhibited in the F1 progeny of C. elegans soaked in dsRNA solutions (Tabara et al., 1998) or fed dsRNA-expressing bacteria (Timmons et al., 2001), and gene expression in the body wall of C. elegans can be inhibited by dsRNA transgenically produced in the pharynx (Winston et al., 2002). Gene expression is also inhibited in cultured Drosophila cells exposed to conditioned culture medium from independently dsRNA-treated cells (Caplen et al., 2000). In contrast to these results, our work using the GAL4-UAS system (Brand and Perrimon, 1993) to generate tissuetargeted, transgenically induced RNAi in Drosophila indicates that the effect in vivo of RNAi is cell-autonomous and confined to cells producing dsRNA. We have tested the ability of GAL4-driven RNAi to cross cell boundaries in embryonic Drosophila by expressing dsRNA transgenically in a subset of cells in every segment and assaying its effect on a homologous, ubiquitously expressed marker transgene. The use of GAL4-based transgenic systems to induce tissue-specific RNAi in Drosophila has been described previously (Kennerdell and Carthew, 2000). Although subsequent reports have suggested that GAL4-mediated RNAi is limited to GAL4-expressing body regions (Giordano et al., 2002; Kalidas and Smith, 2002), these have not explicitly investigated the autonomy of RNAi at a single cellular level following localized expression of dsRNA. Characterizing the cellular autonomy of transgenically induced RNAi, however, is critical to understanding both the experimental potential of this technique in Drosophila and the nature of RNA interference mechanisms in higher animals. To induce RNAi in specific tissues, we use transgenic Drosophila carrying an invert-repeat (“hairpin”) sequence encoding both a variant of GFP (YSmGFP6; Haseloff, 1999) and an endogenous Drosophila gene (FKBP12, not relevant to this experiment) under the transcriptional control of UASGAL4 binding sites (UASinvert repeat(YSmGFP6-FKBP12). We drive expression of this sequence in embryonic stripes using an engrailed promoter-GAL4 driver line, en-GAL4 (AHB, K. Yoffe, N. Perrimon, unpublished; Fietz et al., 1995). We assess the efficacy of induced RNAi by imaging the effect of YSmGFP6 dsRNA on expression of an mGFP6-fusion protein driven in all cells from the poly-ubiquitin promoter (PUbq-D-TACC-mGFP6; Gergely et al., 2000). We simultaneously marked all dsRNA-expressing cells by using GAL4 to coexpress a blue-shifted cyan fluorescent protein fusion transgene (UAS-ECFP-actin), which due to codon alteration of ECFP should be unaffected by RNAi against mGFP6. This experimental approach is diagrammed in Figure 1. YSmGFP6 mRNA shares 98% nucleotide identity with mGFP6, differing at only 15 bases, and we observed accordingly that expression of D-TACC-mGFP6 is potently inhibited in YSmGFP6 dsRNA-expressing cells. In contrast, although the human codon-optimized coding sequence of ECFP (ClonTech, Palo Alto, CA) is 77% identical to that of YSmGFP6, ECFP-actin expression is not strongly affected by YSmGFP6 dsRNA (Fig. 2a,b). This confirms previously reported data on the high sequence-specificity of RNA interference in Drosophila embryos (Yang et al., 2000). Furthermore, ECFP is blueshifted in both excitation and emission spectra and can be imaged independently of mGFP6. All cells expressing dsRNA can thus be marked unambiguously. Independent