Sol narae (Sona) is a Drosophila ADAMTS involved in Wg signaling

ADAMTS (a disintegrin and metalloproteases with thrombospondin motif) family consists of secreted proteases, and is shown to cleave extracellular matrix proteins. Their malfunctions result in cancers and disorders in connective tissues. We report here that a Drosophila ADAMTS named Sol narae (Sona) promotes Wnt/Wingless (Wg) signaling. sona loss-of-function mutants are lethal and rare escapers had malformed appendages, indicating that sona is essential for fly development and survival. sona exhibited positive genetic interaction with wntless (wls) that encodes a cargo protein for Wg. Loss of sona decreased the level of extracellular Wg, and also reduced the expression level of Wg effector proteins such as Senseless (Sens), Distalless (Dll) and Vestigial (Vg). Sona and Wg colocalized in Golgi and endosomal vesicles, and were in the same protein complex. Furthermore, co-expression of Wg and Sona generated ectopic wing margin bristles. This study suggests that Sona is involved in Wg signaling by regulating the level of extracellular Wg.

and incomplete excision of the P element. On the contrary, ten other white-eyed lines that were complemented by the deficiency lines did not have deletion in the sona coding region. We further cleaned up sona deletion mutants by crossing them with w 1118 for four generations in order to remove any unwanted mutations that might have been generated during the process of imprecise excision of the P element.
Sona is essential for survival and development of flies. All sona mutants exhibited lethality throughout larval and pupal stages. Three sona mutants, sona 13 , sona 18 , and sona 47 , were chosen for further analysis. Rare escapers from these sona mutants could reach adulthood at a very low frequency (< 1%), and exhibited various developmental defects. The percentage of escapers and their morphological defects were overall similar in the sona mutants. Branches in their arista were partially missing or malformed (Fig. 2a,b), and the interommatidial bristles were disoriented in dorsal eye (n = 31; Fig. 2c,d; Supplementary Fig. S2). 40% of the legs had kinks in the femur (n = 36; Fig. 2e,f), and loss of tarsal claw was occasionally observed (not shown). Their wings were crumpled and ~25% smaller than sona heterozygous wings (n = 26; Fig. 2g,h,r). Their wing margin bristles were missing in 80% of wings with average 2.7 anterior bristles (n = 26) (Fig. 2i,j).
We also generated two UAS-sona RNAi lines by using two different regions of the sona cDNA ( Fig. 1a; Supplementary Fig. S1). Expression of these two lines by various Gal4 lines induced similar phenotypes such as lethality and malformation of appendages (Supplementary Table S1). Thus, the sona RNAi phenotypes were due to the loss of sona function but not off-target effects. For instance, expression of sona RNAi-1  (sona RNAi-1 hereafter) driven by nubbin (nub)-Gal4 at 18 °C and 25 °C resulted in small and crumpled wings, and similar wing phenotype was induced by expression of sona RNAi-2  (sona RNAi-2 hereafter) at 29 °C (Fig. 2k-n). When sona RNAi-1 and sona RNAi-2 were driven by engrailed (en)-Gal4 at 18 °C and 25 °C, respectively, rarely obtained adults had smaller wings (Fig. 2o-q). In summary, sona RNAi-1 induced stronger phenotype than sona RNAi-2, and Gal4 lines that drive expression of sona RNAi during embryonic stages induced lethality while those during later stages of development induced loss or malformation of appendages (Supplementary Table S1).
The lethality of sona mutants can be rescued by overexpression of Sona. Similar to the phenotypes induced by sona RNAi expression, overexpression of Sona resulted in lethality or defects in appendages (Supplementary Table S1). This indicated that protease activity of Sona must be tightly regulated for proper development and survival. To test whether the protease activity of Sona is essential for its role in vivo, we generated UAS-sonaE475A, which encodes the protease-dead form of Sona by substituting glutamic acid to alanine at 475 th residue in the conserved site for protease activity (Supplementary Fig. S1). Expression of SonaE475A resulted in neither lethality nor any developmental defects, demonstrating that the protease activity of Sona is essential for the role of Sona in vivo (Supplementary Fig. S3 and Table S1) unlike five mammalian ADAMTS-like (ADAMTSL) proteins that have no protease activity 3,37 .
To prove that the deletion in the CG9850 was responsible for phenotype of sona mutants, we checked if the lethality of sona 47 can be rescued by expressing UAS-sona with various Gal4 drivers (Supplementary Table S2). When UAS-sona sona 47 /CyO was crossed with nub-Gal4 sona 47 /CyO-GFP and cultured at 18 °C, about 36% of UAS-sona sona 47 /nub-Gal4 sona 47 were obtained as normal adults. When UAS-sona sona 47 /CyO-GFP were crossed with ptc-Gal4 sona 47 /CyO-GFP and cultured at 18 °C, 23% of UAS-sona sona 47 /ptc-Gal4 sona 47 adults were obtained as normal adults. Two important conclusions were obtained from this rescue experiment. First, CG9850 is the sona gene. Second, the Sona-PA form is sufficient to rescue the lethal phenotype of sona mutants.
Transcription and translation of Sona are dynamically regulated during development. As a first step toward understanding the role of Sona during development, we carried out both Northern and Western analyses to examine the expression pattern of sona mRNA and protein. Multiple sona transcripts were expressed during development, and a 3.4 kb transcript in the late 3 rd instar larval stage was most abundant (Fig. 3a). The amino acid sequence of Sona-PA reported in Flybase was identical to those deduced from a 2.97 kb sona cDNA that had been isolated from a late 3 rd instar larval cDNA library, so the 3.4 kb transcript may encode the Sona-PA form. We focused on the Sona-PA form in this study.
We generated two antisera, 'Sona-Pro' with the prodomain of Sona as an antigen in rabbit and 'Sona-C' with the carboxyl region of Sona as an antigen in mouse as marked in Fig. 1a, and purified them by affinity purification. All analyses were carried out with the purified antibodies. Western analysis with Sona-Pro antibody showed that multiple forms of Sona are dynamically expressed during development (Fig. 3b). For instance, the level of Sona was highly increased in the late third instar stage, which was corresponding to the increased level of sona transcripts in the same stage. 70 kDa Sona protein was most abundant in the late 3 rd instar larval stage (arrow in Fig. 3b), imlying that 70 kDa band is most likely the Sona-PA form (Fig. 3a).
To test whether multiple bands in the Western blot were authentic Sona proteins, UAS-sona RNAi-1 and UAS-sona RNAi-2 were crossed with actin-Gal4, cultured at 18 °C to the third instar stage, shifted to 29 °C for 24 hours, and the larval extracts were prepared. No protein bands were identified with Sona-Pro antibody in the larval extracts. Thus, the bands in the control extract are either different isoforms or processed forms that contain all or some portion of the prodomain (Fig. 3c). Furthermore, full-length Sona was absent in homozygous sona 13 , sona 18 and sona 47 larvae ( Supplementary Fig. S4). The sona 47 extract had a smaller fragment that may be the trun- Sona is expressed in discrete regions and can diffuse far from the expressed site in imaginal discs. We checked expression pattern of sona transcripts in imaginal discs by in situ hybridization and the pattern of GFP driven by sona-Gal4. In the eye-antenna disc of sona > GFP, GFP was expressed at a high level in dorsal peripodial epithelium (arrow in Fig. 4b) and in photoreceptor clusters ( Supplementary Fig. S5a-d). In the leg disc, GFP was expressed in the presumptive region of claw, tibia, and femur 38 (arrow and arrowheads in Fig. 4e). In the pouch of wing discs, sona was expressed in a complicated mosaic pattern (Fig. 4h). Although the expression pattern of sona transcripts was not at high resolution, it was in accordance with the pattern of sona > GFP (arrows and arrowheads in Fig. 4a,d,g).
We then examined the expression pattern of Sona protein in imaginal discs with both Sona-Pro and Sona-C antibodies that had been tested for their specificity (Supplementary Figs S5 and S6). Sona was more or less evenly distributed except some regions with a higher level of Sona (Fig. 4c,f,i). At higher magnification, Sona was highly enriched in the apical region of photoreceptor clusters (Supplementary Fig. S5a-d) and in the disc proper of wing discs, and was present at a negligible level in sona mutants (Supplementary Fig. S5e-h).
Because Sona protein was ubiquitously present in entire discs in contrast to the localized pattern of sona transcripts, we thought that Sona may be efficiently diffused from the expressed site. To check how far Sona can diffuse, we generated UAS-sona-HA and UAS-sona-HA-mCherry (hereafter 'UAS-sona-mCherry') flies using sona cDNA tagged with HA or HA-mCherry in front of the stop codon. These UAS lines driven by various Gal4 lines could induce lethality and structural defects, indicating that the tags did not compromise the protease activity of Sona ( Supplementary Fig. S7). We generated clones expressing Sona-mCherry by the flp-out method 39 and detected both intra-and extra-Sona-mCherry proteins far from the GFP + Sona-mCherry + clone (Fig. 4k). The cross-section view of the same clone also showed the diffusion of Sona-mCherry proteins away from the clone (Fig. 4l). mCherry signals were genuine because no signal was detected in and around a control GFP + clone at the same level of laser intensity (Fig. 4j). We proved that the mCherry signal is not from the mCherry tag cleaved off from Sona by Western analysis, in which Anti-HA antibody recognized full-length and active form of Sona-HA and Sona-HA-mCherry in  Carboxyl region and prodomain of Sona are not colocalized in ECM. To confirm that the active form but not the full-length Sona are secreted in vivo, Canton S (CS) wing discs were stained to visualize extracellular Sona with Sona-Pro and -C antibodies, as described previously (see Materials and Methods) 40 . Both antibodies recognized extracellular Sona in the presumptive region of wing blade in the same region as lip-shaped but not in the DV midline region on the basal side of the disc proper (Fig. 5a). To prove that the pattern of extracellular Sona is authentic but not due to non-specific binding of Sona antibodies to ECM, Viking-GFP, Collagen IV protein fused to GFP commonly used as an ECM marker 41 , was also examined at the same confocal level. To distinguish intracellular Viking-GFP from extracellular Viking-GFP, the extracellular Viking-GFP was differentially stained with the extracellular staining protocol (Fig. 5b). At a glance, extracellular Viking-GFP was more evenly distributed than extracellular Sona-Pro, which was confirmed by the image at higher magnification ( Supplementary Fig. S9).
A cross-section view of a wing disc also showed that extracellular Sona is present in the basal ECM of both disc proper and peripodial epithelium (Fig. 5c,d). However, there were clear differences between the structures recognized by the two Sona antibodies. The Sona-C antibody recognized some particulate structures that were not recognizable by the Sona-Pro antibody (Fig. 5e). These particulate structures may have active Sona that is devoid of the prodomain. In contrast, Sona-Pro antibody recognized a bunch of string-like structures (Fig. 5e"). Similar pattern was also observed in nub > sona-HA wing discs visualized with both Sona-Pro and HA antibodies (Fig. 5f). A magnified image revealed that a Sona form recognized by only anti-HA antibody, probably an active Sona, was localized in ECM as a separate identity (Fig. 5g).

Loss of sona decreases the level of extracellular Wg.
To understand the in vivo role of sona, we carried out a genetic screen with ethyl methanesulfonate (EMS) as a mutagen to obtain suppressors that could overcome the lethality by Sona overexpression. One of suppressors turned out to have a mutation in the wls gene. This study will be addressed in detail elsewhere (J.-H. W. and K.-O. C., manuscript in preparation). Wls is a transmembrane protein that is required for the secretion of Wnt/Wg 42-44 and interacts with retromer complex for cycling from Golgi to the plasma membrane [45][46][47][48][49] . Consistent with the positive genetic interaction between sona and wls, dpp > wls RNAi flies had notched wing phenotype (n > 50 each), but dpp > wls RNAi sona flies had no notching (n = 20) at 18 °C (Fig. 6a-d). dpp > wls RNAi sonaE475A flies had notched wings, demonstrating that protease activity of Sona is essential to suppress the wls RNAi phenotype. The penetrance was 100% in all cases.
The result above prompted us to examine whether secretion of Wg is compromised in the clones expressing sona RNAi-1. Because loss of sona causes cell death (O. T. and K.-O. C., manuscript in preparation), the flp-out clones that coexpressed sona RNAi and caspase inhibitor p35 were generated 50 . The level of intracellular Wg was increased in some clones compared to that of control clones (Fig. 6e,f). Not all clones showed the same phenotype, which suggests that only certain cells express Sona and need the function of Sona. This idea is consistent with the mosaic expression pattern of sona > GFP in wing pouch (Fig. 4h). When sona RNAi-1 was expressed by the apterous (ap)-Gal4 driver in the dorsal wing pouch, the level of intracellular Wg was also increased in the dorsal region regardless of coexpression with p35 ( Supplementary Fig. S10). On the contrary, the level of extracellular Wg was overall decreased in ap > sona wing discs regardless of coexpression with p35 ( Fig. 6g-i). When the sona RNAi-2 was expressed by cubitus interruptus (ci)-Gal4 in the anterior region of the wing disc, the extracellular level of Wg was lower in the anterior region than the posterior region ( Fig. 6j; Supplementary Fig. S11). We also generated sona 13 clones by the FLP-FRT method, and some of them exhibited the decreased level of extracellular Wg (Fig. 6k). Taken together, loss of sona increased the level of intracellular Wg but decreased that of extracellular Wg.
Sona and Wg colocalize in Golgi and endosomal vesicles. Because both Sona and Wg are secreted proteins, and loss of sona decreased the level of extracellular Wg, we examined whether intracellular Sona and Wg are colocalized in S2 cells and wing discs. Both Wg and Sona-mCherry were enriched in the apical region of the wg > sona-mCherry wing disc ( Supplementary Fig. S12), and about 60% (46/75, n = 4) of Sona-mCherry + vesicles contained Wg and vice versa (Fig. 7a). Vesicles containing both Wg and Sona were also observed in S2 cells cotransfected with GFP-wg and sona-mCherry cDNAs (Fig. 7b). To address the nature of the vesicles containing both Wg and Sona, we checked whether Sona + vesicles corresponded to Golgi vesicles in both wing discs and S2 cells. We found that some Sona + vesicles are visualized with a Golgi marker P120 51,52 in S2 cells (Fig. 7c).
Vesicles containing both Rab5-YFP and Sona-mCherry were frequently observed inside of S2 cells (Fig. 7d). About 30% (16/53) of Sona + vesicles in the DV boundary region also contained Rab5-YFP in wg > rab5-YFP wing discs (Fig. 7e). Furthermore, Rab5-YFP + Sona + vesicles also contained Wg (Fig. 7e"'). Rab5, a small GTPase that regulates endocytic vesicle formation and early endosome fusion, is known to significantly co-localize with Wg and is involved in activation of Wg signaling 53 . Co-immunoprecipitation analysis showed that Sona and Wg are present in the same protein complex (Fig. 7f,g). Taken together, Sona and Wg may be secreted together in the same secretory pathway. Sona positively regulates Wg signaling. To further test whether Sona is required for Wg signaling, we checked the effect of sona RNAi-1 on the expression of Sens, Dll and Vg in wing discs. P35 was coexpressed to prevent cell death by sona RNAi expression. Expression level of Sens was reduced in 80% (n = 9), and those of Dll (n = 46) and Vg (n = 25) were reduced in 100% of discs expressing sona RNAi-1 (Fig. 8b,d,f), compared to the control discs (n > 10 for each) (Fig. 8a,c,e). Same results were obtained with sona RNAi-2 driven by ci-Gal4 (Supplementary Fig. S13). The expression level of Wg-LacZ was not decreased upon sona RNAi expression ( Supplementary Fig. S14). Hence, Sona positively regulates Wg signaling by post-transcriptional regulation.
We then examined adult wings whether Sona can enhance Wg signaling. Since prolonged overexpression of either Sona or Wg caused lethality and developmental defects, we used Gal80 ts system to transiently express GFP-Wg and Sona 54 . Transient expression of either GFP-wg or sona with nub-Gal4 during the late third larval and early pupal stage for 18 hours had no effect on the number of anterior bristles in adult wings or on lethality (n > 40 each; Fig. 8g,h). However, coexpression of GFP-wg and sona in the same culture condition caused lethality in about 85% of animals, and the survivors had multiple ectopic bristles near the anterior wing margin (n > 50; Fig. 8i,j). These phenotypes demonstrated that Sona promotes Wg signaling.

Discussion
ADAMTSs are secreted metalloproteases that are known to be involved in mainly ECM remodeling. Among six ADAMTSs in the fly, Papilin is essential for the formation of basement membrane and fly development 32 , Stall functions in ovarian follicle formation and exhibits positive genetic interaction with Delta 35 , and ADAMTS-A is important for cell migration, especially in detaching cells from the apical ECM in salivary gland 34 . In this report, we have shown that Sona is a fly ADAMTS essential for fly development and survival. Transient coexpression of Sona and Wg increased the number of wing margin bristles, indicating that Sona is positively involved in Wg signaling. Accordingly, loss of sona decreased the level of Wg effector proteins as well as the level of extracellular Wg. Based on these results, we propose that Sona, as an ADAMTS, modifies yet unidentified protein(s) essential for Wg signaling.
During fly development, sona was transcribed at a high level in discrete regions in imaginal discs, which corresponded to the malformed regions in adult appendages of sona escapers (Figs 2 and 4). For instance, dorsal eye disc, the center of antenna disc, and outer ring of leg disc expressed the high level of sona transcripts, and sona escapers accordingly had disoriented ommatidial bristles in the dorsal eye, malformed arista, and kinked femur (Figs 2a-f and 4a-f). Wing disc also exhibited the complicated mosaic pattern of sona transcription, and adult wings of sona escapers were small and abnormally shaped (Figs 2g-j and 4g-i). Involvement of Sona in modulating the level of extracellular Wg may explain why these malformed adult structures are generated in sona escapers because Wg is specifically expressed in eye, wing and leg discs and determines the fate of organs [55][56][57][58] .
The genetic link between Sona and Wg signaling was identified in a genetic screen in which a wls allele could rescue the lethal phenotype caused by the overexpression of Sona. Likewise, wing notching by the loss of wls was rescued by overexpression of sona (Fig. 6c). Furthermore, the loss of sona decreased the level of extracellular Wg (Fig. 6g-i). Taken together, these results raised a possibility that Sona may be involved in Wg signaling by affecting Wg secretion. How may Sona positively regulate Wg secretion? To act on Wg secretion, Sona has to be activated intracellularly, and function in secretory pathways. It has been shown that the prodomains of most ADAMTSs are cleaved in trans-Golgi network to become active. Thus, activated intracellular Sona may cleave unidentified proteins involved in Wg secretion and thereby promote the secretion of Wg. Indeed, intracellular Sona was enriched in the apical region while extracellular Sona are more enriched in the basolateral region. Similarly, intracellular and extracellular Wg are enriched in the apical and basolateral regions, respectively 40 ( Fig. 5d; Supplementary Fig. S12). It has been recently shown that Wg is secreted to the apical side and then reentered cells by endocytosis, and then moves to the basal side and secreted by transcytosis 59 . It will be interesting to figure out whether Sona and Wg may be secreted together by transcytosis.
Besides the function of intracellular Sona for Wg secretion, presence of active Sona in conditioned medium of S2 cell culture suggests that extracellular active Sona may be involved in Wg signaling by modifying unknown ECM components (Fig. 3e). Immunocytochemical analysis of Sona confirmed that the active form of Sona devoid of the prodomain is present in basal ECM of wing discs ( Fig. 5e-g). Therefore, active Sona may cleave ECM proteins that affect stability or activity of Wg. Well-studied ECM proteins essential for Wg signaling and formation of Wg gradient are Heparan sulfate proteoglycans (HSPG) such as Division abnormally delayed (Dally) and Dally-like (Dlp) [60][61][62][63][64][65][66][67] . These HSPGs can be modified by proteins such as Notum and Matrix metalloprotease 2 (Mmp2). Notum blocks Wg activity as α /β -hydrolase by modifying Dally and Dlp 68 , and Mmp2 cleaves Dlp to inhibit the interaction between Dlp and Wg 69 . Thus, Sona may act on these HSPGs or related ECM proteins to affect the stability or activity of extracellular Wg. Extracellular Sona was highly localized in the presumptive region of wing blade in the basal ECM near the Collagen-IV containing region, but was present at lower level in the DV midline region where Wg is synthesized (Fig. 5). This data suggests that secreted extracellular Sona may not be diffused freely but restricted to a defined region by interacting with some ECM components. Another component of Wg signaling, Frizzled2 (Fz2), is also localized in the presumptive region of wing blade, but Frizzled3 (Fz3) is expressed exclusively in the DV midline 64,[70][71][72] . Interestingly, Fz2 promotes but Fz3 attenuates Wg signaling 71,73 . Thus, these Fz proteins are strategically localized to bind the extracellular Wg in order to regionally regulate strength of Wg signaling. Similarly, Dally promotes but Dlp decreases Wg signaling in the DV midline 63,64 , and extracellular Dlp is present at a lower level in the DV midline region 60 . Taken together, specific localization of these Wg signaling components in ECM may be essential for modulating Wg signaling with regional specificity in wing discs.
Involvement of Sona in Wg signaling raises a possibility that some mammalian ADAMTSs may also be involved in Wnt signaling. Some mammalian ADAMTSs are known to function as positive factors for tumor invasion and progression [74][75][76] . Overexpression of Wnts or downstream components of Wnt signaling also induces various tumors such as colon cancer, breast cancer, and leukemia 77,78 . Wnt signaling is also essential for the growth and remodeling of bones and connective tissues [79][80][81][82] . Overlapping functions of ADAMTSs and Wnt signaling supports our view that some mammalian ADAMTSs may be linked to Wnt signaling. Further work on identifying the intracellular or extracellular substrate(s) of Sona is required to fully understand how Sona is positively involved in Wg signaling. For generation of ectopic clones, hs-Flp, UAS-sona-mCherry/TM6 Tb males and y w ; P[Actin > CD2 > Gal4; w + ]/Cyo-GFP ; UAS-GFP/TM6 Tb females were crossed, their progeny were raised at 18 °C until 48 hr AEL, and then heat shocked at 37 °C for 1 hour as described 3 . They were then kept at 18 °C until dissection. DNA constructs. sona cDNA was originally obtained from a two-hybrid cDNA library screening 84 using the first and second PDZ domains of Discs-Large (Dlg) as a bait 85,86 . Out-of-frame fusion between the Gal4 activating domain and the portion of sona gene that encodes the carboxyl terminus fortuitously generated a perfect but gratuitous PDZ binding motif, which let Sona to be identified in the screen. To generate sona-HA construct, HA tag was attached at Gly638 by removing stop codon with HpaI digestion and ligating amplified HA tag. pUAST-sona-HA-mCherry was generated by tagging mCherry at the downstream of HA tag in the pUAST-sona-HA plasmid. pUAST-sonaE475A was generated by changing GAA (glu) to GCA (ala) by site-directed mutagenesis. pAc-GFP-wg was constructed by recombining the pAc5.1 vector and GFP-wg that was derived from MK33-GFP-wg (a gift from J.P. Vincent, unpublished).

Materials and Methods
Northern blot and in situ hybridization. For Northern analysis, RNA was prepared as described and 20 μ g of RNA each was loaded in 2.2 M formaldehyde containing agarose gel 87 . The RNAs were transferred to nitrocellulose and probed with 32 P-labeled 1.9 kb XhoI fragment that covers the 3′ half of the sona cDNA for Northern analysis. The same 1.9 kb XhoI fragment was linearlized and amplified by polymerase chain reaction using one primer to make digoxigenin-labelled single-stranded DNA probe for in situ mRNA hybridization as described 88 .
Cell culture and transfection. Drosophila S2 cells were grown in M3 media (Sigma-Aldrich) supplemented with 10% IMS (Sigma-Aldrich) at 25 °C. Transfections were carried out with transfection reagents effectene (Qiagen) or cellfectin (Invitrogen) according to the manufacturers' instructions. For each transfection, a total of 1-2 μ g DNA was used.
To establish S2 sona-HA cell line that stably expresses sona-HA, Hygromycin B selection system was used according to the manufacturers' instructions (Invitrogen life technologies). Briefly, 4 × 10 6 S2 cells were cotransfected with total 1 μ g of two plasmids, pAC sona-HA and pCoHygro (19:1 ratio), for 3 days. Then, the culture medium was changed to selective medium containing 150 μ g/ml of hygromycin B (Invitrogen). The selective medium was replaced every 5 days, and the sona-HA cell lines were established after 3 weeks. The established sona-HA cell line was maintained in selective medium containing Hygromycin B.
Fly larvae were cultured at 25 °C unless stated otherwise. For intracellular staining, imaginal discs were dissected and stained as described 85 . The samples were incubated with primary antibodies in washing buffer (50 mM Tris pH6.8, 150 mM NaCl, 0.5% NP-40, 1 mg/ml BSA) for overnight at 4 °C or room temperature and washed with washing buffer several times. Then, samples were incubated with secondary antibodies and washed several times before mounting. For extracellular staining of Wg and Sona, we followed the method as described 87 . The wing imaginal discs were dissected in M3 media at 4 °C. The samples were incubated with about 3-4 fold more primary antibodies than for intracellular staining in cold M3 media for 2 hrs. Then, samples were washed with cold M3 and PBS twice, fixed for 40 minutes in 4% paraformaldehyde/PBS, and then socked in non-detergent blocking buffer (60 mM Tris pH6.8, 150 mM NaCl, 5 mg/ml BSA) for 3 hrs at 4 °C. Samples were then incubated with secondary antibodies with non-detergent washing buffer (50 mM Tris pH6.8, 150 mM NaCl, 1 mg/ml BSA) and washed with non-detergent blocking buffer four times and detergent blocking buffer twice before mounting. For HA and GFP extracellular staining, the following antibodies, HA (Roche, rat, 1:30) and GFP (Abcam, rabbit, 1:30) were used. Fluorescent images were captured using Zeiss LSM laser scanning confocal microscope and presented using Adobe Photoshop.
Western analysis and co-immunoprecipitation. For Western analysis, samples were mixed with 4 × SDS sample buffer and boiled at 95 °C for 10 min. Samples were then separated by 10-12% SDS-PAGE and transferred to the nitrocellulose membrane (Millipore). Membranes were blocked with 5% nonfat milk in TBST buffer (10 mM Tris pH 7.4, 0.8% NaCl, 0.1% Tween-20), and probed with the antibody. After washing membranes with TBST several times, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibody in TBST with 5% nonfat milk. After washing, protein bands were visualized using the ECL system (AbFrontier).
For co-immunoprecipitation, cells were lysed in HEPES buffer (20 mM HEPES, 70 mM KCl, 2 mM DTT, 0.1% NP40, 8% Glycerol, 1 mM PMSF, 10 mM EDTA, 10 mM EGTA, and protease inhibitor cocktail (Roche)) on ice. The lysates were precleared by incubating with protein G-sepharose beads (Amersham Bioscience) for 30 min at 4 °C. A new set of G-sepharose beads were incubated with anti-GFP (Abcam, rabbit) or Sona-Pro (rabbit) for coupling at room temperature for 2 hrs. The precleared lysates were then incubated with coupled protein G-sepharose beads for overnight at 4 °C. The protein G-sepharose beads were washed with HEPES buffer and Western blots were performed as described.