Rapid tissue-specific expression assay in living embryos.

The GAL4 technique (Brand and Perrimon, 1993) has initiated a major change in the way gene function can be studied in Drosophila (reviewed by Phelps and Brand, 1998), vertebrates (e.g., Hartley et al., 2002; Long et al., 2001; Scheer and Camnos-Ortega, 1999) and plants (e.g., Boisnard-Lorig et al., 2001). The major advantage of GAL4-targeted expression is that it allows ectopic expression to be restricted to specific tissues at discrete developmental times. In contrast, the widely used methods of DNA or RNA injection into eggs or embryos result in widespread ectopic expression with little control over the onset of expression. Although targeted expression by GAL4 overcomes these drawbacks, the GAL4 technique lacks the speed of simple DNA or RNA injection. In particular, the generation of transgenic Drosophila lines can take 4–6 weeks. In this report, we show that injection of plasmids carrying UAS-transgenes into GAL4expressing embryos combines the best of both techniques. The timing and site of expression of the transgene is controlled by GAL4 and the subcellular localisation and possible function of the gene can be assayed in the tissue of interest in hours instead of weeks. We first tested if the expression of UAS plasmids depends on the presence of GAL4. To ensure the maximum uptake of DNA into cells, embryos were injected at the syncytial blastoderm stage (2 h after egg lay), when there are no membrane barriers to DNA diffusion or nuclear uptake. Injection of UAS-mGFP5 plasmids (Brand, 1998) into Oregon P flies never results in fluorescent cells (n 15 embryos). In contrast, injection of the same plasmid into the maternal GAL4 line (Hacker et al., 1997) yields fluorescently labelled cells in the ectoderm, CNS, midgut, and muscles in 87.5% of embryos (n 16; Fig. 1b). In crosses of stable transformants, GAL4 drives ubiquitous expression of GFP from gastrulation onwards (data not shown). After injection, fluorescence is first detectable 2 h after gastrulation. The majority of labelled cells are in the ectoderm. Labelled cells are distributed over up to four segments, with the highest number of cells at the site of injection. The distribution of GFP-expressing cells indicates that the injected plasmid diffuses over only a short distance. We also tested if injection of two plasmids carrying different genes would allow co-expression of these genes in the same cells. Co-expression of the gene of interest together with a GFP tagged gene, which encodes a protein that outlines the cell or part of the cytoskeleton, would greatly facilitate phenotypic analysis. 90% (n 10) of GAL4 embryos injected with an equal mixture of UAS-GFP and UAS-lacZ plasmids express both proteins (Fig. 1). Co-expression varies between 50–100%, with two embryos showing a complete overlap. To test if the site and timing of expression is defined by GAL4 expression, we injected UAS-GFP plasmids into the line GAL4 (Hidalgo et al., 1995), which expresses GAL4 in longitudinal glial cells and a subset of central and peripheral neurons from Stage 10 onwards. We also used line GAL4, in which expression in longitudinal glial cells and a subset of peripheral neurons starts at Stage 12. GFP-expressing cells are found in 92% (n 26) of injected GAL4 embryos. Expression is restricted to neurons and longitudinal glial cells. We do not detect any expression before Stage 12, about 2 h after the onset of expression in stable transformants. After injection into GAL4 embryos, GFP expression is restricted to longitudinal glial cells and is first detectable at Stage 15. In contrast to GAL4 and GAL4, in which up to 30 cells over four segments express GFP, we only find one to two GFP-expressing cells per GAL4 embryo (in 14% of embryos; n 14). A reduction in the frequency of expression in GAL4 lines with a late onset of expression may not be unexpected. In Drosophila a small fraction of injected DNA enters the nucleus, where it is replicated only once (Steller and Pirrotta, 1985). The lack of replication coupled with degradation of the cytoplasmic fraction (Steller and Pirrotta, 1985) could reduce the concentration of UAS plasmid present later in embryogenesis. GFP expression generated by plasmid injection is significantly stronger than expression in stable transformants, possibly due to the higher copy number of the injected plasmid. High levels of expression could interfere with the localisation of endogenous proteins or reduce the viability of expressing cells. We investigated this possibility by generating and injecting constructs encoding cytoskeletal proteins, UASp-CFP-actin and