RNA interference from a CAX trinucleotide repeat

The use of RNA interference (RNAi) techniques, combined with a spatially controlled expression of doublestranded RNA (dsRNA), provides a convenient method for producing cell-type-specific perturbations of gene function (Fortier and Belote, 2000; Martinek et al., 2000). While studying the function of the Drosophila miniature-dusky (m-dy) gene complex (DiBartolomeis et al., 2002), we produced transgenic flies expressing a dsRNA that would permit silencing of the dy gene (P{UAS-dy-dsRNA}; Fig. 1a). The m and dy genes share sequence similarity and both encode fly Cuticulin proteins (DiBartolomeis et al., 2002) that are required at the pupal stage of development (Newby and Jackson, 1995) for normal wing differentiation. In the absence of m-dy products, the wing is reduced in size, due to an autonomous effect on the size and/or shape of wing epidermal cells (Newby et al., 1991). Although a reduction in wing size is the only obvious morphological phenotype in m-dy flies, certain mutants exhibit reduced viability and/or fertility (Newby et al., 1991). To examine the phenotypic effects of the UAS-dy-dsRNA transgene, we ectopically expressed it within wing tissues, taking advantage of four different existing Drosophila GAL4 driver strains: P{GawB}30A, P{GawB}34B, MS1096, and A9. The first two strains are described in Flybase, the last two were obtained from M. O’Connor (Univ. of Minnesota). All four strains provide expression in wing and other imaginal discs. Previously characterized dy mutants have small wings of normal proportion ( 2/3 of the wild-type size), whereas m mutants show a more extreme reduction in wing size and the wings are malproportioned compared to the wild type. The relatively less severe phenotype of dy mutants approximates the null phenotype for this gene, as dy/dy and dy/Df flies exhibit similar wing defects (Newby et al., 1991). Expression of UAS-dy-dsRNA with three out of four of the GAL4 drivers (P{GawB}34B, MS1096, and A9) was associated with a wing phenotype, the severity of which was dependent on the particular driver (Table 1). The fourth driver, P{GawB}30A, caused complete lethality at the late pupal stage (Table 1) and pharate adults were not dissected for examination. We were surprised that ectopic expression of UAS-dydsRNA, using either MS1096 or A9 as a driver, produced a wing phenotype more severe than that observed in existing single dy or m mutants (Table 1; Fig. 1; data not shown). Both of these GAL4 transgenes are known to drive expression in the developing wings, although MS1096 has a slightly broader pattern of expression (M. O’Connor, pers. commun.) that might explain the more severe wing phenotype compared to A9. A wing phenotype was observed with these two drivers using three independent UAS-dy-dsRNA insertions (19-4, 28-1, and 38-1), so the phenotype is not a consequence of transgene insertion. In addition, wing abnormalities were not observed in control siblings carrying any one of the GAL4 drivers or the UAS-dy-dsRNA transgene alone (not shown). Interestingly, the size of the wing in MS1096 or A9; UAS-dy-dsRNA flies is similar to that of a m-dy double mutant called Df(1)m (Fig. 1f), which is known to carry a chromosomal deletion removing only m and dy (Roberts and Jackson, 1997). Given the known sequence similarity between m and dy, such a result suggests that the UAS-dy-dsRNA transgene might target both m and dy RNAs for destruction (see below). Other abnormalities, in addition to wing phenotypes, were observed in our GAL4/UAS-dy-dsRNA crosses. Lethality was observed in populations carrying UAS-dydsRNA with any of the four GAL4 transgenes (Table 1) and the degree of lethality was dependent on the GAL4 driver (see Table 1 notes). In many crosses, much of this lethality appeared to occur at the late pupal stage, although we cannot exclude death at earlier developmental stages. As indicated, the combination of P{GawB}30A and UAS-dy-dsRNA was completely lethal, whereas the other drivers were semilethal in combination with the UAS transgene. Some of this observed lethality is attributable to elimination of m-dy function(s), as Df(1)m also exhibits poor viability (Roberts and Jackson, 1997). Although the general morphology of surviving GAL4; UAS-dy-dsRNA flies is relatively normal, and females are fertile (not shown), these flies nonetheless exhibit additional phenotypes that might reflect targeting of genes other than m and dy. A comparison of m and GAL4; UAS-dy-dsRNA flies, for example, indicates that the mor-