letters to nature 80 NATURE | VOL 397 | 7 JANUARY 1999 | www.nature.com (Molecular Probes) was dissolved at 1 mg ml -1 in N-N-dimethylformamide and ionophoretically injected into the neuromuscular cleft. Embryos were placed in 1 3 PBS overnight at 4 8C, mounted in 50% glycerol plus 2% peraformaldehyde, and viewed on a Zeiss confocal microscope. Generation of transformants. Restriction fragments from lim3 genomic phage clones 9 were inserted into P[tau-myc] 4 or HZ50PL lacZ 28 plasmids and transformation was done as described 29 . Three independent lim3A-tau-myc lines and one lim3A-lacZ line were tested. All displayed similar expression patterns. Two of the lim3A-tau-myc insertions, denoted 1.8 and 1.13, were crossed into lim3 mutant backgrounds and gave identical results. UAS-lim3 was made by inserting the lim3 cDNA plus 50 base pairs of the Xenopus laevis b-globin 59 leader from plasmid pNB 25 into plasmid pUAS 18 . For the misexpression studies, several independent transgenic lines, each carrying one or two copies of UAS-lim3, were used. The ftz ng -GAL4.20 driver. To improve the expression levels of the original ftz ng -GAL4 line 19 , we generated 25 new insertion lines by transposase-mediated mobilization. Each line was tested for GAL4 expression by crossing it to¯ies carrying a UAS-tau-myc-GFP reporter 30 , where GFP is green¯uorescent protein. One line, ftz ng -GAL4.20, expressed GAL4 at high levels in most, if not all, motor neurons. As assayed by transactivation of the UAS-tau-myc reporter gene, 89% of hemisegments showed labelling in ISNb and 84% showed labelling in ISNd (n 56).