Fig. 3. The LUFS domain of Ssdp proteins mediate interactions with Ldb/Chip that are crucial for nuclear localization. (A) Schematics comparing the amino acid composition and overall structure of mouse (M.m.) Ssdp2 and Drosophila melanogaster (D.m.) Ssdp. There is 90% identity of amino acids over the N-terminal region shown to be sufficient for interactions with Ldb1/Chip (gray box), known as the LUFS domain. The remainders of these proteins are rich in proline, glycine and methionine residues, and share three other small domains that are also highly conserved (small gray boxes, corresponding to Ssdp2 amino acids 232-240, 250-262 and 331-337). (B) Alignment of the primary amino acid sequence of LUFS domains of Ssdp proteins from mouse and Drosophila, and Arabidopsis LEUNIG, showing the LisH domain and the motif P-X-GFL-XX-WW-X-VFWD, which is notable for its conservation in all members and with yeast Flo8. (C) Schematic of Ldb1 and Chip, indicating 94% identity over residues 201-249 of Ldb1 and 387-435 of Chip (hatched boxes). We have called this region the Ldb1/Chip conserved domain (LCCD). Within the LCCD a deletion of 10 residues (Ldb1 amino acids 214-223, corresponding to Chip amino acids 400-409) disrupts the ability of Chip to interact with Ssdp without affecting its ability to homodimerize through the dimerization domain (DD) or bind LIM domains through the LIM interaction domain (LID). The position of the putative nuclear localization sequence (NLS) is indicated. (D-F) Anti-Myc immunofluorescence staining of ventral muscles of stage 16 embryo. In a wild-type background (D), ap^GAL4-driven expression of Myc epitope-tagged SsdpFL reveals discrete localization to the multiple nuclei in each of the muscle cells 21-24. By contrast, Ssdp2-92 fails to localize to the nucleus and instead is found throughout the cytoplasm (E). SsdpFL fails to localize to the nucleus in a Chip^e5.5 null mutant background (F).