Dock mediates Scar- and WASp-dependent actin polymerization through interaction with cell adhesion molecules in founder cells and fusion-competent myoblasts

Cell–cell adhesion molecules are crucial for the formation of organs and tissue structure during embryonic development and for their homeostasis. Cells express various cell adhesion molecules at their surface, which can either mediate homophilic or heterophilic cell–cell adhesions. Their intracellular tails interact with cytoplasmatic proteins involved in intracellular signaling. During syncytial muscle formation, mononucleated myoblasts must recognize and adhere before they fuse to form multinucleated muscles. In Drosophila, members of the immunoglobulin superfamily (IgSF) mediate the recognition and adhesion of founder cells (FCs) and fusion-competent myoblasts (FCMs) (Abmayr and Pavlath, 2012). FCs determine the position, size and epidermal attachment site of the muscle by the combinatorial expression of muscle identity genes and recruit FCMs. The IgSF molecule Dumbfounded (Duf; also known as Kirre) is exclusively expressed in FCs and functions redundantly with the IgSF molecule Roughest (Rst; also known as IrreC) (Ruiz-Gómez et al., 2000; Strünkelnberg et al., 2001). The extracellular domain of Duf binds to the Ig domain of the FCM-specific IgSF molecule Sticks and Stones (Sns) (Bour et al., 2000). Sns seems to be the main player in FCMs that mediates cell adhesion. The paralog of Sns named Hibris (Hbs) is also exclusively expressed in FCMs (Artero et al., 2001; Dworak et al., 2001). However, hbs mutants do not display any fusion defects and Hbs can rescue only a small amount of fusion in sns mutants (Shelton et al., 2009).

The IgSF molecules Duf, Rst and Sns are expressed in a ring-like structure at cell–cell contact points in FCs and FCMs (Kesper et al., 2007; Sens et al., 2010; Önel et al., 2011; Haralalka et al., 2011). In the center of this structure a dense F-actin focus forms (Kesper et al., 2007), predominantly in FCMs contacting an FC/growing myotube (Sens et al., 2010; Haralalka et al., 2011). In contrast, a thin sheath of F-actin is visible at cell–cell contact points in FCs/growing myotubes (Sens et al., 2010). In the absence of the cell adhesion molecules, F-actin foci fail to form (Richardson et al., 2007), indicating that they trigger the formation of F-actin foci. On a molecular level, recent studies have demonstrated that foci formation depends on the evolutionary conserved Arp2/3 complex (Massarwa et al., 2007; Richardson et al., 2007; Berger et al., 2008), which nucleates branched F-actin. The Arp2/3 complex becomes activated by two nucleation-promoting factors during myoblast fusion: Scar (Richardson et al., 2007; Berger et al., 2008; Gildor et al., 2009; Sens et al., 2010) and WASp (Massarwa et al., 2007; Schäfer et al., 2007). One open and intriguing question is how signaling from the cell adhesion molecules is linked to F-actin formation. Recent co-immunoprecipitation studies on non-muscle Drosophila S2 cells have shown that the SH2-SH3 adaptor protein Crk is able to bind to the intracellular domain of Sns and to the Drosophila WASp-interaction partner Vrp1 (Flybase; Berger et al., 2008) also known as Sltr (Kim et al., 2007) and Wip (Massarwa et al., 2007). Since Arp2/3-based actin polymerization is required in both myoblast types, forming a large actin focus in the FCM and a thin actin sheath in the FC, we have investigated signaling molecules that may be present in both cell types. Based on findings from mammalian Nephrins, we have investigated whether the SH2-SH3 adaptor protein Dock is involved in Drosophila myoblast fusion, and connects both Duf/Rst in the FCs and Sns/Hbs in the FCMs to downstream actin regulators.

Human Nephrins, Neph1 and Nephrin show 33% identity to Duf and Rst and 28% identity to Sns and Hbs (Gerke et al., 2003). They are involved in the formation of the slit diaphragm, a specialized podocyte cell–cell junction in the kidney essential for filtration of the blood (reviewed by Welsh and Saleem, 2010). Recent findings have demonstrated that the intracellular domain of Nephrin can bind to the Src-Homology 2 (SH2)/SH3 domain-containing adaptor protein Nck (Jones et al., 2006). In this study multiple YDxV sites were found in the intracellular domain of Nephrin that can interact with the SH2 domain of Nck.

Herein, we demonstrate that the Drosophila homolog of Nck, named Dreadlock (Dock), is required for myoblast fusion. Dock is expressed in both myoblast types and interacts genetically and biochemically with the Arp2/3 activators Scar, Vrp1 and WASp. Interaction studies in COS7 and Drosophila S2 cells further reveal that Dock binds to the intracellular domain of all Ig-domain proteins that are known to mediate the recognition and adhesion of FCs and FCMs. Surprisingly, we found that the cell adhesion proteins either bind to the SH2 domain (Hbs), the SH3 domains (Duf) or all domains of Dock (Sns and Rst). However, only duf, sns and hbs show a genetic interaction with dock.

To define the requirement of the Drosophila SH2-SH3 adaptor protein Dock during myoblast fusion, we first studied the subcellular distribution of the Dock protein with an anti-Dock antibody (Clemens et al., 1996). The specificity of the Dock antibody was tested by the expression of a myristylated membrane-bound form of Dock that was driven in the Wingless (Wg) domain (Fig. 1A). Thereafter, we examined whether Dock is expressed in the somatic mesoderm. We observed ubiquitous expression of the Dock protein during embryogenesis (Fig. 1B,B′). The rP298 enhancer trap line expresses β-galactosidase in the nuclei of FCs (Nose et al., 1998). To mark myoblasts and distinguish between FCs and FCMs we labeled wild-type embryos carrying rP298-lacZ with β3-Tubulin, Dock and β-Gal antibodies (Fig. 1B). Dock is expressed in a punctuated manner in both myoblast types, FCs (Fig. 1B,B′ arrow) and FCMs (Fig. 1B,B′, arrowheads). Additionally, Dock seems to be expressed at muscle attachment sites (Fig. 1B,B′, asterisks).

The presence of Dock in myoblasts pointed us to investigate whether dock⁰⁴⁷²³ null mutants display defects in myoblast fusion. We found that the muscle pattern of dock⁰⁴⁷²³ null mutants is reminiscent of that seen in wild-type embryos (Fig. 1C). In podocytes, the SH2–SH3 adaptor proteins Nck and Grb2, which bind to the intracellular domain of phosphorylated Neph1, both cooperate to induce actin polymerization. Since dock null mutants show no mutant phenotype, we generated double mutants with dock and the Drosophila grb2 homolog drk by using the drk^e0A null allele. We expected to see fusion defects in dock drk double mutants if both genes cooperate during myoblast fusion. However, no fusion defects were observed (Fig. 1D).

Previously, we showed that zygotic wasp mutants show no fusion defects due to maternal wasp mRNA (Schäfer et al., 2007). The dock mRNA is also maternally contributed (Desai et al., 1999). For this reason we assessed whether maternal contributed dock might be sufficient to complete myogenesis. By using the dock deficiency Df(2L)ast2 we found that maternal Dock is detectable till stage 15 when myogenesis is almost completed (Fig. 1E,E′).

The Dock protein contains three SH3 domains and a C-terminally located SH2 domain (Fig. 2A). The SH3 domains bind to proline-rich segments whereas SH2 domains bind to phosphotyrosine residues in a target protein (Li et al., 2001). WASp and its interaction partner Wip have been both identified to be downstream effectors of the mammalian Dock homolog Nck (Rivero-Lezcano et al., 1995; Quilliam et al., 1996; Antón et al., 1998). Drosophila possesses only one WASp and Wip protein called Vrp1. Like their mammalian homologs, Drosophila Vrp1 and WASp possess proline-rich sequences (Fig. 2A).

To address whether Drosophila Dock is involved in actin polymerization during myoblast fusion, we examined protein–protein interactions using co-immunoprecipitation and yeast-two hybrid assays. We transiently co-transfected COS7 cells with FLAG-tagged Dock, untagged Vrp1 and HA-tagged WASp (Fig. 2B, lanes 1–5 and 6–9). FLAG-tagged Dock, Vrp1 and HA-tagged WASp were immunoprecipitated by anti-FLAG (Fig. 2B, lanes 1–5). These data show that Dock forms a protein complex with Vrp1 and WASp. Moreover, co-immunoprecipitation of FLAG-tagged Dock, Vrp1 and HA-tagged WASp demonstrate that both Vrp1 and WASp are associated with Dock (Fig. 2B, lane 5). We confirmed these results by immunoprecipitating HA-tagged WASp and WASp associated proteins with anti-HA (Fig. 2B, lanes 6–9).

Next, we assessed which domains of Vrp1 and WASp mediate Dock binding by using the yeast two-hybrid system. We generated Vrp1 full length, WASp full length, Vrp1-PPP and WASp-PPP constructs that contain only the middle proline-rich region of Vrp1 and WASp, as marked in red in Fig. 2A, as well as Vrp1 and WASp deletion constructs lacking the proline-rich region. Binding of Dock to Vrp1 and WASp was only achieved when the middle proline-rich region of Vrp1 and WASp is present (Fig. 2C, sectors 1, 3, 7 and 9). In the absence of this proline-rich region, Dock fails to bind to Vrp1 and WASp (Fig. 2C, sectors 5 and 11). To determine which domains in Dock mediate binding to Vrp1 and WASp, we generated deletion constructs that lack only one of the SH3 domains (SH3-1, SH3-2 or SH3-3), all SH3 domains and only the SH2 domain. We found that only the SH3-1 and SH3-3 domains are essential for binding to Vrp1 (Fig. 2D, sectors 1, 3 and 5) and that all SH3 domains are required for binding to WASp (Fig. 2E, sectors 1, 3, 5 and 7).

We next examined whether dock, vrp1 and wasp interact genetically during myoblast fusion. Mutations in vrp1 cause severe fusion defects (Fig. 3A), while some fusion events still take place in vrp1 null mutants (Kim et al., 2007; Massarwa et al., 2007; Berger et al., 2008). The wasp³ mutation carries an intragenic deletion of 13 bp, leading to a truncated protein that lacks the VVCA domain of WASp (Schäfer et al., 2007). Because of a high maternal contribution, homozygous wasp³ mutants show a wild-type muscle pattern (Fig. 3C). We used the protein null allele vrp1^f06715 (Berger et al., 2008) and the wasp³ allele (Ben-Yaacov et al., 2001) to generate double mutants with the dock null allele dock⁰⁴⁷²³. In comparison to the single mutant phenotypes (Fig. 3A,C), we observed enhanced fusion defects in homozygous vrp1^f06715 dock⁰⁴⁷²³ (Fig. 3B) and wasp³ dock⁰⁴⁷²³ double mutants (Fig. 3D). We further quantified the fusion defect of vrp1^f06715 and dock⁰⁴⁷²³ single and vrp1^f06715 dock⁰⁴⁷²³ and wasp³ dock⁰⁴⁷²³ double mutants statistically (Table 1) by examining the number of nuclei of the segmental border muscle (SBM) with anti-Ladybird (Jagla et al., 1998).

A characteristic feature during myoblast fusion is the formation of an F-actin focus at cell–cell contact sites in FCMs (Kesper et al., 2007; Kim et al., 2007; Massarwa et al., 2007; Richardson et al., 2008; Sens et al., 2010; Haralalka et al., 2011). F-actin foci are highly dynamic structures that reach their maximum size within two minutes and dissolve shortly afterwards in less than one minute (Richardson et al., 2007). Interestingly, F-actin foci formation still occurs in mutants affecting Arp2/3-based actin polymerization. However, in some mutants the F-actin foci are enlarged and increased in number (Richardson et al., 2007). Only in scar vrp1 double mutants, in which myoblast fusion is almost completely blocked (Berger et al., 2008), the size of the F-actin focus is severely decreased (Sens et al., 2010). To characterize vrp1^f06715 dock⁰⁴⁷²³ double mutants in more detail, and to determine whether F-actin foci still form in homozygous double mutants, we visualized actin polymerization using phalloidin. In wild-type embryos, an F-actin focus forms at the site of contact between an FCM and a growing myotube (Fig. 3E,E′, arrows). The F-actin focus is still visible in vrp1^f06715 dock⁰⁴⁷²³ double mutants and occurs at the site of the FCM attaching to a growing myotube (Fig. 3F,F′, arrows). Furthermore, F-actin foci seem to be able to dissolve in FCMs, showing an expanded cell–cell contact site (Fig. 3F,F′, gray asterisks).

In Drosophila, myoblast fusion has been characterized at the ultrastructural level by transmission electron microscopy (TEM) analysis. Cell adhesion and alignment of myoblast cell membranes are accompanied by the appearance of electron-dense vesicles and plaques on both sites of the apposing membranes. The fusion of the apposing membranes can be followed by the initiation of cytoplasmic continuity (Doberstein et al., 1997; Massarwa et al., 2007; Berger et al., 2008). Additionally, recent studies using freeze substitution have observed the formation of finger-like protrusions in FCMs that project into the FC/growing myotube (Sens et al., 2010). TEM analyses on conventional fixed vrp1 mutants revealed that homozygous vrp1 mutants stop fusion when multiple discontinuities are visible at apposing membranes (Massarwa et al., 2007). The stronger fusion defect in vrp1^f06715 dock⁰⁴⁷²³ double mutants in comparison to vrp1^f06715 single mutants, let us determine at which ultrastructural level vrp1^f06715 dock⁰⁴⁷²³ double mutants stop fusion. Using conventional TEM fixation methods, we found that FCMs in vrp1^f06715 dock⁰⁴⁷²³ double mutants are still able to form filopodia (Fig. 3G) and that fusion stops when discontinuities are visible at apposing membranes (Fig. 3G′, arrows).

To address whether Dock links cell adhesion with actin polymerization in FCMs we generated dock hbs double mutants and observed severe fusion defects in those mutants (Fig. 4B,B′; Table 1).

To determine whether Hbs and Dock also interact at the protein level, we performed co-immunoprecipitation studies (Fig. 4C) and bimolecular fluorescence complementation (BiFC) assays. FLAG-tagged Dock and HA-tagged Hbs^intra, containing only the cytodomain of Hbs, were transiently transfected into S2 cells. FLAG-tagged Dock was successfully immunoprecipitated with HA-tagged Hbs^intra (Fig. 4C, lanes 2, 3 and 4). Lysates were immunoprecipitated with anti-HA antisera (Fig. 4C).

We then asked which domains are involved in mediating the interaction between Dock and Hbs^intra by using the BiFC vectors N-YFP and C-YFP (Dottermusch-Heidel et al., 2012). The BiFC assay enables direct visualization of protein interactions in living cells (Kerppola et al., 2008). When co-expressed in S2 cells the non-fluorescent N-YFP and C-YFP fragments show no YFP fluorescence. A reconstitution of the split fluorescent protein only occurs when N-YFP and C-YFP are fused to interacting proteins. We cloned full-length Dock, Dock lacking all SH3 domains and Dock lacking the SH2 domain into the N-YFP vector. Hbs^intra was cloned into the C-YFP vector (Fig. 4D). Drosophila S2 cells co-transfected with Hbs^intra and the empty N-YFP vector did not show a reconstitution of the YFP signal (Fig. 4E). Only cells that were co-transfected with Hbs^intra and Dock-Fl showed a YFP signal (Fig. 4F). The YFP signal was still observed in cells co-transfected with Hbs^intra and Dock-ΔSH3, which lacks all SH3 domains (Fig. 4G). In contrast, no YFP reconstitution was observed in S2 cells co-transfected with Hbs^intra and Dock-ΔSH2 (Fig. 4H). Based on these findings we concluded that the SH2 domain of Dock mediates interaction by binding to phosphorylated tyrosine residues within Hbs^intra. The Nck SH2-binding site in Nephrin, the human ortholog of Hbs, has been identified as YDxV motif (Jones et al., 2006). Hbs^intra contains six tyrosine residues that might become phosphorylated (Blom et al., 1999). However, only the tyrosine residue at amino acid position 1088 is a related YDxV motif. To map the Dock-interaction site in Hbs, we replaced this tyrosine residue with phenylalanine (Hbs^{intra T1088F}). As a negative control, we co-transfected cells with mutated Hbs^{intra T1088F} and the empty N-YFP vector (Fig. 4I). When Hbs^{intra T1088F} was co-transfected with Dock-Fl, Dock-ΔSH3 or Dock-ΔSH2, reconstitution of the YFP signal failed (Fig. 4J,K,L), demonstrating that Y1088 is critical for Hbs–Dock interaction.

We then asked which domains are involved in mediating the interaction between Dock and Hbs^intra by using the BiFC vectors N-YFP and C-YFP (Dottermusch-Heidel et al., 2012). The BiFC assay enables direct visualization of protein interactions in living cells (Kerppola et al., 2008). When co-expressed in S2 cells the non-fluorescent N-YFP and C-YFP fragments show no YFP fluorescence. A reconstitution of the split fluorescent protein only occurs when N-YFP and C-YFP are fused to interacting proteins. We cloned full-length Dock, Dock lacking all SH3 domains and Dock lacking the SH2 domain into the N-YFP vector. Hbs^intra was cloned into the C-YFP vector (Fig. 4D). Drosophila S2 cells co-transfected with Hbs^intra and the empty N-YFP vector did not show a reconstitution of the YFP signal (Fig. 4E). Only cells that were co-transfected with Hbs^intra and Dock-Fl showed a YFP signal (Fig. 4F). The YFP signal was still observed in cells co-transfected with Hbs^intra and Dock-ΔSH3, which lacks all SH3 domains (Fig. 4G). In contrast, no YFP reconstitution was observed in S2 cells co-transfected with Hbs^intra and Dock-ΔSH2 (Fig. 4H). Based on these findings we concluded that the SH2 domain of Dock mediates interaction by binding to phosphorylated tyrosine residues within Hbs^intra. The Nck SH2-binding site in Nephrin, the human ortholog of Hbs, has been identified as YDxV motif (Jones et al., 2006). Hbs^intra contains six tyrosine residues that might become phosphorylated (Blom et al., 1999). However, only the tyrosine residue at amino acid position 1088 is a related YDxV motif. To map the Dock-interaction site in Hbs, we replaced this tyrosine residue with phenylalanine (Hbs^{intra T1088F}). As a negative control, we co-transfected cells with mutated Hbs^{intra T1088F} and the empty N-YFP vector (Fig. 4I). When Hbs^{intra T1088F} was co-transfected with Dock-Fl, Dock-ΔSH3 or Dock-ΔSH2, reconstitution of the YFP signal failed (Fig. 4J,K,L), demonstrating that Y1088 is critical for Hbs–Dock interaction.

We next examined zygotic mutant combinations of sns^4.3, a hypomorphic allele with a point mutation in the FNIII domain and the dock null allele dock⁰⁴⁷²³ (Fig. 5A–E). We did not observe a dramatic enhancement of the myoblast fusion defects. However, quantification of the unfused myoblasts in confocal Z series of four abdominal hemisegments revealed a statistically significant increase of 30–40% in the number of unfused myoblasts in embryos homozygous for sns^4.3 when either one or both alleles of dock were mutant (Fig. 5F). Furthermore, we generated a sns vrp1 double mutant carrying the hypomorphic sns allele sns^20-15 (Paululat et al., 1995) and the vrp1 null allele wip^D30 (Massarwa et al., 2007). Homozygous double mutants possess a stronger myoblast fusion defect (Fig. 5H,I) than sns^20-15 (Fig. 5G,I) and vrp1 single mutants (Fig. 5I).

We further examined the localization of Dock, by comparison to Sns and Duf in primary myoblasts, which provide better resolution of expression at points of cell contact. Dock colocalizes with Sns and Duf at points of cell–cell contact between primary FCs and FCMs (Fig. 5J–N). In summary, these data support a model in which Sns interacts with Dock to promote Vrp1–WASp-based actin polymerization during myoblast fusion.

To assess whether Sns interacts biochemically with Dock and to define the regions that are critical for binding, we combined co-immunoprecipitation studies with BiFC assays. First, Drosophila S2 cells were transiently transfected with HA-tagged Sns and FLAG-tagged Dock. FLAG-tagged Dock was efficiently immunoprecipitated with HA-tagged full-length Sns (Fig. 6A, lanes 1 and 2). The Dock binding ability of HA-tagged Sns was significantly reduced (Fig. 6A, lane 3) when using a mutated form of Sns that lacks all phosphotyrosine residues and the two consensus PxxP motifs, and is compromised in its ability to rescue sns mutant embryos (Kocherlakota et al., 2008). To map the interaction domains within Dock, S2 cells were co-transfected with full-length HA-tagged Sns and a fragment containing only the FLAG-tagged Dock SH2 domain or the three SH3 domains. Our data demonstrate that both the SH2 and SH3 domains of Dock, in isolation, can bind to Sns (Fig. 6A, lanes 4–7). To evaluate regions of Sns that are critical for this binding, we utilized an HA-tagged intracellular domain of Sns. In contrast to full-length Sns, which bound to the Dock SH2 domain, Sns^intra does not interact with FLAG-tagged Dock SH2 (data not shown). This finding is consistent with the possibility that phosphorylation of tyrosines in the intracellular domain requires full-length Sns at least in S2 cells. In contrast, full-length FLAG-tagged Dock and FLAG-tagged Dock containing all three SH3 domains co-immunoprecipitated efficiently with Sns^intra (Fig. 6B, lanes 8–10). This interaction is reduced to near background levels upon mutation of the two Sns consensus PxxP sites (Fig. 6B, lanes 11–13) and is in line with functional studies showing that Sns lacking all cytodomain tyrosines and the two consensus PxxP motifs severely reduced rescue of sns mutant embryos (Kocherlakota et al., 2008). We interpret these results to suggest that additional non-consensus SH3 binding sites must be present within Sns, and find that binding of Sns^intra and Dock-SH3 is further reduced by mutation of non-PxxP sites in Sns (Fig. 6B, lanes 14–16). These data suggest that Sns and Dock can interact through either the Dock SH2 or Dock SH3 domains, and that the latter binding relies primarily if not exclusively on consensus PxxP sites in Sns.

We further examined the binding domains of both proteins in vivo using the BiFC assay (Fig. 6C). In agreement with our previous co-immunoprecipitation experiments (Fig. 6A, lane 2), S2 cells transfected with the Sns^intra and Dock full-length (Dock-Fl) showed a reconstitution of the YFP signal (Fig. 6D). The reconstitution of YFP was still observed when we used DockΔSH2 instead of the Dock-Fl (Fig. 6E). However, no YFP signal was detectable with the DockΔSH3 deletion construct that lacks all SH3 domains (Fig. 6F). As with the above co-transfection studies, these data suggest that Sns^intra alone is unable to interact with Dock SH2, possibly indicating that this isolated domain cannot be phosphorylated on tyrosine. These data also confirm that Dock interaction with Sns can be mediated through the Dock SH3 via proline-rich motifs in Sns. Consistent with the above studies, the YFP signal was still reconstituted using Sns^intra2xPxxP (Fig. 6G–I) and Sns^{intraF14-2xPxxP} (Fig. 6J–L).

In primary isolated myoblasts Dock colocalizes with the cell adhesion molecule Duf (Fig. 6A,C–E). Mammalian Neph1 is related to Drosophila Duf and Rst and co-operates with Nephrin to induce actin polymerization in kidney podocytes (Garg et al., 2007). Since F-actin is present at cell–cell contact points of both myoblast populations in Drosophila (Kesper et al., 2007; Sens et al., 2010; Haralalka et al., 2011), we also investigated whether Dock interacts with the Drosophila NEPH1 orthologs Duf and Rst. Our yeast two-hybrid data indicate that Dock binds to the cytodomain of Duf (Duf^intra; Fig. 7A) via its SH3 domains (Fig. 7B). The cytodomain of Rst (Rst^intra) did not show an interaction with Dock (data not shown). To verify these results, we performed a BiFC assay by co-transfecting Drosophila S2 cells with the constructs shown in Fig. 7C and with Dock constructs. Co-transfected S2 cells with Duf^intra and Dock full length showed a reconstitution of the YFP signal (Fig. 7D). This reconstitution failed only with a Dock deletion construct that lacks all SH3 domains (Fig. 7F), but not when the SH2 domain was deleted (Fig. 7E). The cytodomain of Duf possesses a small proline-rich region to which the SH3 domains of Dock could bind. We therefore generated a Duf^Δ810-956aa deletion construct that lacks this region and tested it for interaction with Dock full length. No reconstitution of the YFP signal was observed (Fig. 7G), suggesting that the interacting region is located in this fragment.

We next investigated whether Rst is a potential Dock interaction partner. Rst^intra is able to bind to Dock (Fig. 7H). Interestingly, our data indicate that Dock binds Rst via its SH2 and SH3 domains (Fig. 7I,J,K, control). To correlate the observed protein interactions with myoblast fusion, we also performed double mutant experiments. To generate duf; dock double mutants we crossed a duf-RNAi line in the background of dock⁰⁴⁷²³ null mutants. Mesoderm-specific knockdown of duf was achieved using the Mef-GAL4 driver line (Fig. 8A,B). The knockdown of duf with Mef–GAL4 in a wild-type background did not induce any fusion defects (Fig. 8A). In contrast, the knockdown of duf with Mef–GAL4 in a dock⁰⁴⁷²³ mutant background severely disturbed muscle formation (Fig. 8B). A quantitative analysis of the number of nuclei of the SBM showed no fusion defects. Only when we quantitatively analyzed the ventral VA2 muscle that contains up to 9 nuclei (Beckett and Baylies, 2007) the number of nuclei within this muscle was clearly reduced (Table 1). rst; dock double mutants used the loss of function allele rst⁶, which carries a 98 bp deletion in the cytodomain (Ramos et al., 1993). If rst and dock cooperate during myoblast fusion, we expected a severe defect in myoblast fusion. However, the muscle pattern of rst⁶; dock double mutants is reminiscent of rst⁶ single mutants (Fig. 8C,D). From these data, we cannot clearly deduce that the binding of Dock to Rst is essential for myoblast fusion.

Since Scar has been proposed to be responsible for the formation of a thin sheath of actin in FCs (Sens et al., 2010), we also investigated whether Dock may interact with Scar during myoblast fusion. Zygotic mutant combinations of the scar null allele scar^Δ37 (Zallen et al., 2002) and the null allele dock⁰⁴⁷²³ were examined for myoblast fusion defects. scar^Δ37 single mutants show a weak fusion phenotype (Fig. 8E; Table 1); unfused myoblasts were clearly visible in scar^Δ37 dock⁰⁴⁷²³ double mutants (Fig. 8F).

Drosophila Scar possesses like Vrp1 and WASp a proline-rich segment (Fig. 8G, indicated in red). To assess whether Scar serves as a direct interaction partner for Dock we performed a yeast two-hybrid assay. Interestingly, we observed that Scar full-length binds to Dock (Fig. 8G, sector 2). However, this interaction does not seem to depend on the proline-rich region of Scar (Fig. 8G, sector 3).

In response to extracellular signals, cell surface receptors must interact with intracellular signaling pathways to induce a cellular response. In this study we have identified the SH2–SH3 adaptor protein Dock as a new component in myoblast fusion. Based on protein interaction studies, immunostaining results and genetic interaction data, we suggest that Dock transfers the signal from the cell surface receptors to the actin cytoskeleton in FCs and FCMs.

Nephs, Nephrins and their Drosophila orthologs are transmembrane receptors belonging to the immunoglobulin superfamily. Their extracellular domains can interact in trans, forming homodimers and heterodimers, and thus can serve as ligand-receptor pairs (Barletta et al., 2003; Galletta et al., 2004). The function of Nephs and Nephrins has been associated with the maintenance of the kidney filtration barrier (Tryggvason, 2001) and, despite their structural similarities with Drosophila orthologs, it initially appeared that they fulfill different functions during organ formation in vertebrates and insects. However, recent studies have demonstrated a conserved function for Nephs and Nephrins during vertebrate myoblast fusion (Srinivas et al., 2007; Sohn et al., 2009), as well as for their orthologs in Drosophila nephrocytes (Weavers et al., 2009; Zhuang et al., 2009).

Arp2/3-based actin polymerization is essential for both myoblast fusion (reviewed by Schejter and Baylies, 2010; Önel et al., 2011; Abmayr and Pavlath, 2012) and podocyte formation (Welsch et al., 2001). Since previous studies report that the intracellular domain of Neph1 and Nephrin become tyrosine phosphorylated, a modification that is required for their interaction with the SH2–SH3 adaptor proteins Nck and Grb2 (Jones et al., 2006; Garg et al., 2007), we assessed the role of the Drosophila homolog of Nck during myoblast fusion. Because Drosophila dock, drk (corresponding to Nck, Grb2) double mutants did not show any fusion defects, we concluded that the Drosophila homologs of Nck and Grb2 do not cooperate during myoblast fusion as they do during actin polymerization in mammalian podocytes (Garg et al., 2007). However, our biochemical and genetic data in this study support the notion that Dock functions in both myoblast types by binding to different motifs in Duf, Hbs and Sns (Fig. 9).

Two actin nucleation-promoting factors are required during myoblast fusion to activate the Arp2/3 complex and promote F-actin sheath and foci formation in FCs and FCMs: WASp and Scar. Here we show that Dock biochemically interacts with Scar, Vrp1 and WASp (Fig. 9). The functional relevance of these biochemical interactions is also supported by genetic interaction studies with dock, scar, vrp1 and wasp. A direct interaction of Dock with mammalian Wip and WASp has been suggested in earlier studies (Quilliam et al., 1996; Antón et al., 1998). However, we found that the binding domains between mammalian and Drosophila Nck/Dock differ. Whereas the SH3-2 domain of Nck is essential for binding mammalian Wip (Antón et al., 1998) and the SH3-3 domain of Nck binds mammalian WASp (Quilliam et al., 1996), our study revealed that the SH3-1 and SH3-3 domain are required for binding to Drosophila Vrp1 and all SH3 domains are necessary for the interaction with Drosophila WASp.

A direct interaction between Dock and Scar has not been reported yet. Mammalian Nck has been identified as an interaction partner for the Nck-associated protein 1 (Nap1) (Kitamura et al., 1996), which is known as kette in Drosophila. Nap1/Kette is a member of the Scar complex that regulates Scar activity during Arp2/3-based actin polymerization (reviewed by Pollard, 2007). The activation of the Scar complex might be achieved by binding to activate Rac and Nck. However, the nature of the Scar complex regulation is still not clear yet and remains a matter of debate. The loss of rac1, 2 and kette disturbs Drosophila myoblast fusion (Hakeda-Suzuki et al., 2002; Schröter et al., 2004; Gildor et al., 2009). Rac (Vasyutina et al., 2009) and Nap1 (Nowak et al., 2009) are also essential for mammalian myoblast fusion, supporting the notion of a conserved function for Nck/Dock in mammalian and Drosophila myoblast fusion.

Our data further show that F-actin foci in FCMs are still present in dock vrp1 double mutants. This might be due to the presence of the SH2–SH3 adaptor protein Crk. Crk has been shown to bind the FCM-specific cell adhesion molecule Sns and Vrp1 by co-immunoprecipitation studies using Drosophila non-muscle S2 cells (Kim et al., 2007). Additionally, Crk is able to bind the FCM-specific protein Blow via its SH2 domain, which has been reported to compete with WASp for Vrp1 binding (Jin et al., 2011). Taking these data into consideration, it is possible that Dock may act independently of Crk in FCs, whereas it cooperates with Crk in FCMs to mediate actin polymerization.

The dock⁰⁴⁷²³ null allele, the deficiency Df(2R)ast2, the drk^e0A null allele, the rst⁶ allele and wingless–GAL4 were obtained from the Bloomington Stock Center. The UAS–duf RNAi line (Dietzl et al., 2007) was ordered from the Vienna Drosophila RNAi Center. The hbs⁴⁵⁹ allele was kindly provided by Mary Baylies (Sloan-Kettering Institute, New York). UAS–dockmyr was obtained from Sven Bogdan (Münster University). Eyal Schejter (Weizmann Institute of Science, Rehovot) kindly provided the wip^D30 allele. The protein null allele vrp1^f06715 was previously described in Berger et al. (Berger et al., 2008). The sns^20-15 allele was previously described as rost¹⁵ by Paululat et al. (Paululat et al., 1995). The sns^4.3 allele has not been previously described. This EMS-induced point mutation changes V1003E. The Mef2–GAL4 driver line has been described by Ranganaykulu et al. (Ranganaykulu et al., 1998). All crosses were performed at 25°C using standard procedures.

Embryos were collected at 25°C and processed as described (Hummel et al., 1997). Drosophila S2 cells and primary myoblast cultures were fixed in 4% F-PBS at 25°C, permeabilized in 0.1% Triton X-100 in PBS for 2.5 minutes and blocked in 5% BSA in PBS for 30 minutes. As primary antibodies we used the anti-β3-Tubulin antibody (1∶10,000) (Buttgereit et al., 1996; Leiss et al., 1988), anti-Dock antibody (1∶1000) (Clemens et al., 1996), anti-β-Gal antibody (1∶3000) (Cappel Laboratories), anti-Sns antibody (1∶1000) (Haralalka et al., 2011), anti-Duf antibody (1∶1000) (Galletta et al., 2004), anti-Ladybird (1∶500) (a gift from Krzysztof Jagla, Faculté de Médicine, France) and anti-Slouch (1∶200) (a gift from Manfred Frasch, Universität Erlangen, Germany). Cy2- and Cy3-conjugated secondary antibodies were purchased from Dianova. Embryos used for phalloidin staining were fixed for 30 minutes in 4% F-PBS and devitellinized by hand after fixation. Embryos were incubated in phalloidin–TRITC (1∶500; Sigma-Aldrich), anti-β3-Tubulin and anti-β-Gal antibodies overnight. Drosophila S2 cells were stained in phalloidin–TRITC (1∶20; Sigma-Aldrich) for 15 minutes after fixation. Primary myoblasts and myosin stained embryos were incubated with primary antibody over night at 4°C and secondary antibody for 3 hours at 25°C. Fluorescence images were acquired using a Leica TCSsp2 or Zeiss LSM-510 META confocal scanning microscope.

For quantification of fusion in sns and dock recombinants, embryos were fluorescently stained for myosin heavy chain as described above. Comparable confocal Z-series were cropped to include four abdominal hemisegments between A2 and A6. Samples were de-identified and unfused myoblasts hand counted in Zeiss LSM software, as indicated by white spots in Fig. 5A–E. For statistical analysis, P-values were determined using the Student's t-test, and P-values of <0.0001 were interpreted to indicate statistically significant differences. Error was determined as standard error of the mean (s.e.m.).

All constructs used in COS7 cells studies were generated in the pcDNA3.1 vector. Dock-FLAG included full-length Dock followed by three copies of a FLAG tag. In HA-WASp, full-length WASp is preceded by one HA tag. Full-length Vrp1 was untagged. For Drosophila S2 co-immunoprecipitation studies, UAS-Sns-HA has been described (Kocherlakota et al., 2008). UAS-Sns[F14-2xPxxP]-HA combines site-directed point mutations that change all 14 cytoplasmic tyrosine residues to alanine (UAS-SnsF14-HA; (Kocherlakota et al., 2008) and site directed mutations that change the cytoplasmic consensus PxxP proline rich regions (UAS-Sns2xPxxP) (Kocherlakota et al., 2008) into one construct. For Dock constructs with a single FLAG tag, full-length Dock, the Dock SH2 domain or all three Dock SH3 domains were cloned into the Gateway pTWF vector (Invitrogen). For studies using copper-inducible genes in pRmHa3, Sns-intra-HA includes AA1108–1482 followed by a single HA tag. Sns-intra[2xPxxP]-HA was generated by replacing the region of the wild-type parent with the 2xPxxP sequence above and previously described (Kocherlakota et al., 2008). Sns-Intra[RRKK-2xPxxP]-HA was generated by site-directed mutagenesis of the previous pRmHA3 construct to change the cytoplasmic RRKK sequence to alanines. pRmHa3-Dock-2xflag was generated by PCR amplification of full-length Dock from TWF-Dock above, using oligos that added EcoRI and NheI restriction sites. This fragment was subcloned into pRmHA3. Dock-SH3(3)2xFLAG was generated by PCR amplification of the Dock SH3 domain region (amino acid 1–279) and subcloning into pRmHA3.

For studies in Fig. 2B, COS7 cells were grown in Dulbecco's Modified Eagle's Medium containing 10% fetal bovine serum and 5% CO₂ at 37°C and transiently transfected with Effectene Transfection Reagent (Qiagen) as directed by the manufacturer. After 48 hours, cells were washed with PBS and lysed by addition of lysis buffer [50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% NP40, 50 mM PMSF, 1 µg/ml leupeptin and pepstatin]. After 10 minutes, the lysate was collected, passed through a 25 gauge needle 12 times, and clarified by centrifugation at 15,000 g for 15 minutes. 800 µg of lysate was incubated with a 40 µl slurry of anti-HA or anti-FLAG beads (Sigma-Aldrich) for 5 hours at 4°C. Beads were washed 4× 10 minutes in washing buffer [20 mM Tris-HCl (pH7.5), 250 mM NaCl, 1 mM EDTA and 0.5% NP40], and then boiled for 5 minutes in 2× Laemmli buffer.

For studies in Fig. 4C, Drosophila S2 cells were cultured in Drosophila Schneider's medium (Invitrogen) containing 10% fetal bovine serum at 25°C, and transiently transfected using TransFectin Lipid Reagent (Bio-Rad). Transfected S2 cells were centrifuged at 13,000 g for three minutes at 4°C. Cell pellets were resuspended in lysis buffer. Cells were centrifuged at 13,000 g for 10 minutes and pellet was collected. 50 µl of 2× Laemmli buffer was added to the 50 µl of the supernatant and boiled for 10 minutes, this was used as Input. For HA-Hbs^intra and Dock-FLAG co-IP, 50 µl of anti-HA (Sigma-Aldrich) agarose beads were taken and washed twice with 750 µl of IP buffer and the rest of the supernatant was added to the anti-HA agarose beads for 2–3 hours at 4°C. The mixture was spun down for 30 seconds at 2000 rpm and washed in cold IP buffer and boiled in 2× Laemmli buffer.

For studies in Fig. 6, Drosophila S2 cells were cultured in SFX medium (serum-free insect cell culture medium, HyClone) at 25°C, and transiently transfected using the Effectene Transfection Reagent (Qiagen). For Sns/Dock interaction studies genes were cloned into the pUAST vector. After transfection cells were washed with PBS and resuspended in crosslinking solution [1.5 mM dithiobis-succinimidyl propionate (DSP) in PBS] for 60 minutes. The samples were rewashed with PBS. Cell pellets were resuspended in lysis buffer [250 mM NaCl, 20 mM Tris-HCl (pH7.5), 0.5% NP40, 2 mM EDTA, 1× protease inhibitor cocktail (cOmplete, EDTA-free, Roche) and Na₃VO₄]. Lysate was passed through a 25 gauge needle 10 times, centrifuged at 13,000 g for 15 minutes at 4°C and the supernatant collected. For each sample, 500–1000 µg of cell lysate was incubated with 20 µl of 50% slurry of anti-HA beads (Sigma-Aldrich) overnight at 4°C. Beads were washed with Buffer A [20 mM Tris-HCl (pH 8.0), 250 mM NaCl, 1 mM EDTA, 0.5% NP40, protease inhibitor cocktail and Na₃VO₄] 2× 10 minutes and washed 2× 10 minutes with Buffer B [20 mM Tris-HCl (pH 8.0), 50 mM NaCl, 1 mM EDTA, 0.5% NP40, protease inhibitor cocktail and Na₃VO₄] for 10 minutes. Beads were resuspended in loading buffer and processed as above. For transfections involving genes cloned into the RmHa3 vector (Fig. 6B), expression was induced with 0.5 mM CuSO₄ 20–24 hours after transfection. Cells were washed in PBS and pellets were resuspended in lysis buffer, and passed through a 25 gauge needle 10 times. Cells were centrifuged at 13,000 g for 15 minutes at 4°C. Equal volumes of supernatant and 2× Laemmli buffer were boiled for 5 minutes and used as input. 10 µl anti-FLAG resins (EZview; Sigma-Aldrich) was mixed with 1 mg of lysate for 4 hours at 4°C. The beads were washed with lysis buffer 4× 10 minutes at 4°C, and processed in 2× Laemmli buffer as above. Proteins were analyzed in immunoblots using anti-HA (1∶5000; Roche) or anti-FLAG (1∶5000; Sigma) using ECL plus.

The yeast two-hybrid assay was performed using the Matchmaker LexA two-hybrid system (Clontech) according to the manufacturer's instructions. scar full-length, scar containing the proline-rich region (aa 388–491), vrp1 full-length, vrp1 lacking the proline-rich region (aa 348–635) and only containing the proline-rich region as well as wasp full-length, wasp lacking the proline-rich region (aa 310–415) and only containing the proline-rich region were amplified by PCR and cloned into pGilda. As PCR templates the scar cDNA SD02991, vrp1 cDNA GH25793 and the wasp cDNA RE12101 were used. Furthermore, the intracellular domains of Duf and Rst were amplified from the duf cDNA (Galletta et al., 2004) and the rst cDNA RE01586 (DGRC) and cloned into pGilda. dock and dock deletions were amplified from the LD42588 cDNA and cloned into pB42AD.

We thank Renate Renkawitz-Pohl and Christine Dottermusch-Heidel for fruitful discussions. We are grateful to Sabina Huhn, Tim Fries and Helga Kisselbach-Heckmann, Claude Shelton, Shufei Zhuang, Elspeth Pearce and Erin Katzfey for technical advice and assistance.