Fig. S1. Characterization of the sns^D1 allele. The hbs⁴⁵⁹ mutant was treated with EMS and screened for mutations in the sns locus, in which the sns^D1 mutant in the hbs⁴⁵⁹ background was identified. We next sought to determine the specific molecular lesion present in the sns^D1 allele. To do this, we sequenced PCR products from gene-specific reverse transcription of sns transcript from RNA lysate (Trizol, Invitrogen) obtained from sns^D1, hbs⁴⁵⁹ and hbs⁴⁵⁹ homozygous embryos aged 6-18 hours AEL. From this analysis, it is apparent that a single base-pair mutation of T1191A of the sns cDNA is the aberration in the sns^D1 allele (A). This mutation creates an amino acid change of W215R in the predicted Ig2 domain of the Sns protein. From sequence alignment of D. melanogaster Sns Ig2 with orthologs from other species or alignment of Sns Ig2 with other non-homologous Ig domains (B), it is clear that the W215R mutation in the Sns^D1 protein alters a highly conserved residue across all Ig domains. To determine whether this newly identified sns mutant is functional, the sns^D1, hbs⁴⁵⁹ mutant was crossed to the sns^XB3, hbs²⁵⁹³ recombinant (C) or to the sns^XB3 mutant alone (D). Embryos were aged 12-18 hours AEL and colorimetrically stained for anti-MHC. It is clear that the sns^D1 mutant is unable to direct proper muscle development, and has a similar phenotype to that of other nonfunctional sns mutants. To determine whether the sns^D1 allele is expressed, wild-type (E) and sns^D1, hbs459 mutant (F) embryos aged 8-14 hours AEL were stained with antibody directed against the Sns cytodomain (Bour et al., 2000). Clearly, there is no difference in Sns protein levels for the mutant compared with the wild type. Scale bars: 20 m.

Fig. S2. Full-length protein sequences for Hbs-HA and SHS-HA. The breakpoint region ± 10 amino acids for the chimeric constructs is shown. The coloring in these sequences indicates the parent protein: Hbs in blue and Sns in orange. For the SETHC chimera, Sns sequence ends at amino acid 1112 and Hbs sequence begins at amino acid 1092 of the respective full-length parent proteins. For the SEHTC-HA chimera, Sns sequence ends at amino acid 1082 and Hbs sequence begins at amino acid 1064 of the respective full-length parent proteins. For the HESTC chimera, Hbs sequence ends at amino acid 1063 and Sns sequence begins at amino acid 1083 of the respective full-length parent proteins. All chimeric constructs have a C-terminal HA tag of sequence YPYDVPDYA.

Fig. S3. Rescue and quantitation of sns/hbs domain swaps with mef2Gal4-driven expression. (A-R) Stage 16 sns^zf1.4/sns^XB3 embryos rescued using mef2Gal4-directed expression of the indicated transgenes (see Fig. S2 for sequences of transgenes). All rescue experiments were performed at 25°C. Embryos were colorimetrically stained with anti-Myosin heavy chain antibody. In all panels, anterior is to the left and dorsal is up. (A,D,G,J,M,P) Dorsolateral view. (B,E,H,K,N,Q) Lateral view. (C,F,I,L,O,R) Ventrolateral view. (S) The average number of unfused myoblasts remaining within three hemisegments of sns mutant embryos rescued with the indicated transgenes. The snslacZ reporter was incorporated into the mef2Gal4 rescue stock for visualization and quantitation of unfused myoblasts. Scale bar: 20 m.

Fig. S4. SNS intracellular and extracellular domains can mediate interaction in cis with full-length Hbs. (A-D) Various deletion constructs of Sns were used to determine whether a specific region of Sns-mediated interaction with Hbs in co-transfected S2 cells. Sns-ΔFN3-HA is lacking the predicted Fibronectin III domain, which corresponds to amino acids 969-1050 of the wild-type protein. Sns-ΔECD-HA is lacking all of the predicted Ig domains and the FNIII domain, corresponding to amino acids 90-1048 of the wild-type sequence, and includes only the signal sequence (amino acids 1-89) followed by the transmembrane and intracellular domains. In Sns-GPI-HA, amino acids 1-1070 are followed in frame by a GPI anchoring sequence. Sns-ICD-HA consists solely of the Sns intracellular domain (amino acids 1106-1482), and is lacking both the Sns extracellular and transmembrane domains. Sns-ΔICD consists of the Sns extracellular and transmembrane domains (amino acids 1-1112), but is lacking the Sns intracellular domain. Where indicated, an HA epitope tag has been inserted in frame after the last amino acid of Sns. Hbs-V5 was created by insertion of the V5 sequence in frame after the last amino acid in Hbs. All constructs were generated in the pUAST vector (Brand and Perrimon, 1993), and expression in S2 cells was achieved by co-transfection with pWAGal4 (Ishimaru et al., 2004). Constructs were transfected in the indicated combinations using Effectene reagent (Qiagen, Valencia, CA) according to the manufacturer’s protocol for adherent cells at a 1:25 ratio of DNA to Effectene reagent. All transfections used 1 g pWAGal4, and 1.5 g each of SNS and Hbs constructs for a total of 4 g DNA per 2.0×10⁷ adherent cells. Cells were harvested 36 hours after addition of DNA complex. Lysate preparation and immunoprecipitation were as described previously (Balagopalan et al., 2006). (A,C) Immunoblots of co-immunoprecipitated samples. (A) Sns was immunoprecipitated via the HA tag, and the corresponding immunoblot probed for Sns using antisera directed against the HA tag and for Hbs using antisera directed against the V5 epitope tag. (C) Hbs was immunoprecipitated via the V5 tag, and the corresponding protein blot probed for Sns with antisera directed against the Sns extracellular domain (Galletta et al., 2004) and for Hbs using antisera directed against the V5 tag. In all immunoprecipitations, the entire eluate was loaded. (B,D) Immunoblots of input lysates for A and C, respectively. Protein amounts loaded for each sample correspond to 3.8% of that used in the accompanying immunoprecipitations. We note that Hbs is not detected upon co-immunoprecipitation with Sns-GPI-HA. This observation is inconclusive, however, since Sns-GPI-HA was not detected in the cleared lysate input (B) and appears to be unstable. Asterisks denote proteins of the expected sizes. Smaller fragments are routinely observed in these studies, and appear to result from cleavage at prominent sites during sample processing. These results demonstrate that all domains within Sns are capable of mediating interaction with Hbs under conditions of protein overexpression. The entire extracellular domain is expendable, and the Sns cytodomain is sufficient to direct interaction with Hbs. This interaction is clearly not specific to the Sns cytodomain, however, since the extracellular domain is capable of directing interaction with Hbs in its absence. These findings are consistent with those for the proteins Cdo and Boc (Kang et al., 2002), two paralogous IgSF members known to associate in cis. We have not confirmed that domains of Hbs act in a similarly promiscuous fashion in directing its interaction with Sns upon overexpression, but anticipate that its overall similarity to Sns and to proteins such as Cdo and Boc supports this hypothesis.