The CDM Superfamily Protein MBC Directs Myoblast Fusion through a Mechanism That Requires Phosphatidylinositol 3,4,5-Triphosphate Binding but Is Independent of Direct Interaction with DCrk^,

Abstract

myoblast city (mbc), a member of the CDM superfamily, is essential in the Drosophila melanogaster embryo for fusion of myoblasts into multinucleate fibers. Using germ line clones in which both maternal and zygotic contributions were eliminated and rescue of the zygotic loss-of-function phenotype, we established that mbc is required in the fusion-competent subset of myoblasts. Along with its close orthologs Dock180 and CED-5, MBC has an SH3 domain at its N terminus, conserved internal domains termed DHR1 and DHR2 (or “Docker”), and C-terminal proline-rich domains that associate with the adapter protein DCrk. The importance of these domains has been evaluated by the ability of MBC mutations and deletions to rescue the mbc loss-of-function muscle phenotype. We demonstrate that the SH3 and Docker domains are essential. Moreover, ethyl methanesulfonate-induced mutations that change amino acids within the MBC Docker domain to residues that are conserved in other CDM family members nevertheless eliminate MBC function in the embryo, which suggests that these sites may mediate interactions specific to Drosophila MBC. A functional requirement for the conserved DHR1 domain, which binds to phosphatidylinositol 3,4,5-triphosphate, implicates phosphoinositide signaling in myoblast fusion. Finally, the proline-rich C-terminal sites mediate strong interactions with DCrk, as expected. These sites are not required for MBC to rescue the muscle loss-of-function phenotype, however, which suggests that MBC's role in myoblast fusion can be carried out independently of direct DCrk binding.

In Drosophila melanogaster, formation of multinucleate muscle fibers occurs between two distinct myoblast populations, termed founder and fusion-competent cells, and is mediated by cell adhesion molecules specific to each of these cell types (1, 3, 38, 41). Genes such as myoblast city (mbc) encode proteins that appear to function downstream of these cell surface receptors to regulate intracellular and cytoskeletal events through the activation of small GTPases (8). Embryos homozygous for mutations in the mbc locus exhibit severe defects in myoblast fusion, with a large number of unfused mononucleate myoblasts in place of multinucleate muscle fibers (14, 39).

MBC was a founding member of the CDM superfamily, which includes human Dock180 (20), Caenorhabditis elegans CED-5 (43), and almost 20 additional members in a wide variety of species (13, 17). Dock180 was originally identified as a major binding partner for the adapter protein Crk (20), while CED-5 and MBC were identified on the basis of their loss-of-function phenotypes. All three proteins are characterized by an SH3 domain at their N terminus, internal dock homology region 1 (DHR1) and dock homology region 2 (DHR2) (or Docker) domains, and proline-rich sites in their C termini that bind to the SH2-SH3 adapter protein Crk. Vertebrate Dock180 binds to Rac1 in a nucleotide-independent manner and increases the amount of GTP-bound Rac1 (24). Deletion of the Docker domain results in loss of Rac binding and abolishes the ability of Dock180 to increase GTP-bound Rac1 (6). In addition, dominant negative Rac1 suppresses Dock180-induced membrane spreading (24). Thus, members of the CDM superfamily function as guanine nucleotide exchange factors (GEFs) for Rac1, with both Rac1 binding and activation dependent on the Docker domain (6, 13, 33, 45). CDM proteins fall in a class of unconventional GEFs in that they contain neither a Dbl homology nor a pleckstrin homology (PH) domain. The function of the Dbl homology domain, which facilitates GTP-GDP exchange in the majority of Rho-family GEFs, appears to be provided by the Docker domain. The PH domain, which normally serves to bind to lipids and stabilize interactions, is provided by the associated ELMO (engulfment and cell motility) protein through its interaction with the N terminus of Dock180 (6). Together, these form a stable Dock180/Elmo/Rac1 complex (28). In similar interactions, C. elegans CED-12 (Elmo) binds directly to CED-5 (Dock180/MBC) and is found in a ternary complex with CED-10 (Rac1) (44). In cotransfected S2 cells, the SH3 domain of MBC has been shown to have a direct interaction with the PRM (proline-rich motif) domain of Drosophila Elmo (D-Elmo) (21). In a steric hindrance model, accessibility of the Dock180 Docker domain to Rac1 appears to be blocked by binding of the Dock180 N-terminal region to its own Docker domain, and this blockage is relieved by Elmo binding (29).

Dock180 and CED-5 interact directly with the small SH2-SH3-containing adapter protein Crk through PXXP sites in their C termini (20, 31). Dock180 was initially identified on the basis of a biochemical interaction with Crk (20) and is found transiently at the membrane in a complex with Crk at focal adhesions and membrane ruffles (9). Functionally, coexpression of Dock180 and Crk upregulates signaling from Crk-associated complexes that are involved in cell migration, phagocytosis of apoptotic cells, and formation of focal adhesions (9, 18, 25). In addition, the CED-12/CED-5/CED-10 complex includes the C. elegans ortholog of Crk, CED-2 (42). These data have led to a model in which Crk transduces signals from the cell surface to the small GTPase Rac1 via interaction with Dock180 and Elmo (13). More recent studies have shown that Dock180 interacts, via the DHR1 domain, with phosphatidyl-inositol 3,4,5-triphosphate [PtdIns(3,4,5)P₃] (12, 27). This interaction mediates the localization of Dock180 to sites of PtdIns(3,4,5)P₃ accumulation at the leading edge of migrating cells in culture, and the DHR1 domain is essential for this migration.

Consistent with Dock180 interactions in mammalian cells, mbc interacts genetically with rac1 in the Drosophila eye (35), and loss-of-function alleles of mbc and rac1 rac2 double mutants exhibit severe defects in embryonic myoblast fusion (19). By comparison, DCrk is expressed coincident with MBC in the mesoderm at the time of myoblast fusion (15). However, the embryonic loss-of-function phenotype of DCrk has remained elusive. The presence of significant levels of maternally provided transcript, coupled with the inability to eliminate this transcript using currently available methods, precludes its genetic analysis. In biochemical interaction assays, however, DCrk interacts with the PXXP-containing region of MBC. We therefore reasoned that MBC is likely to function in the embryonic musculature in a complex that also includes DCrk.

Herein, we examined further the requirement for MBC in myoblast fusion to better understand its mechanism of action and to determine whether it occurs through a conserved Crk/CED-2>CDM>DRac1/CED-10 pathway and/or involves PtdIns(3,4,5)P₃ binding. We used the binary UAS-GAL4 targeted expression system (5) to examine the importance of conserved MBC domains by assaying the ability of various transgenic mbc constructs to rescue the myoblast fusion defects of mbc mutant embryos. We describe the results of studies addressing whether MBC is required in the fusion-competent cells and which sequences within MBC are essential in these cells for myoblast fusion in the embryonic musculature. As expected, the SH3 and Docker homologies are essential. While the proline-rich sites in MBC direct its interaction with DCrk as anticipated, these regions of MBC are not needed to direct localization of MBC to the membrane or to rescue myoblast fusion. By comparison, the PtdIns(3,4,5)P₃ binding region is not essential for recruitment of MBC to the myoblast membrane but is essential to direct myoblast fusion in the embryo. Thus, MBC can function in myoblast fusion through a pathway that may involve PtdIns(3,4,5)P₃ binding but is independent of direct DCrk binding.

MATERIALS AND METHODS

Drosophila stocks.

Fly stocks were raised on standard cornmeal medium at 18°C or 25°C as needed. OregonR was used as the wild-type strain. rP298-GAL4 was provided by Devi Menon, and MBCΔPRM was provided by S. Ishimaru. GAL4 drivers (obtained from the Bloomington Stock Center) included twi-GAL4, mef2-GAL4, and 24B-GAL4. Upstream activation sequence (UAS)-MBC transgenes were generated by standard methods and by Genetic Services, Inc. Transgenes were recombined into an mbc mutant background using either mbc^D11.2, mbc^F12.7, or Df(3R)mbc⁹ (14) as indicated in the text.

Analysis of germ line clones.

The mbc^D11.2 mutation was recombined onto the FLP recombination target (FRT)-containing third chromosome of P{hs neo; ry⁺; FRT} 3R-82B Sb/TM6, Ubx. The source of transposase was P{ry⁺; FLP} 22; TM3, Sb/CxD. Males of the genotype FLP22; FRT3R-82B, P{ovo^D1}/TM3 were generated by standard crosses and mated to P{hs neo; ry⁺; FRT} 3R-82B; mbc^D11.2/TM3-lacZ-hg females. Progeny from this cross were heat shocked at 37°C for 2 h on day 4 and day 5 (late L2 or L3) essentially as described previously (10). Female progeny of the genotype FLP22/+; FRT3R-82B, P{ovo^D1}/P{hs neo; ry⁺; FRT} 3R-82B; mbc^D11.2 were recovered and crossed to mbc^D11.2/TM3-lacZ-hg heterozygous males. The muscle pattern of embryos laid by these females was then examined as described above.

Generation of plasmids and site-directed mutagenesis.

The 7,035-bp mbc cDNA was generated from overlapping partial cDNA clones, and a fragment was recovered by reverse transcription-PCR. The region from +521 to 7035 was then cloned into pUAST (5). For the hemagglutinin (HA)-tagged wild-type UAS-MBC construct, sequences encoding the HA epitope were added to the 3′ end of the mbc cDNA, with a stop codon engineered in the oligonucleotide after the tag. For MBC-CBS (Crk binding site), the mutation was introduced into the cloned mbc cDNA sequence (with or without the HA tag, as noted) via site-directed mutagenesis using the oligonucleotide CCT GCA CCG GCA GCA GCC GCA GCA CGT GAC. The MBCΔDHR1 and MBCΔ1807 constructs were generated by site-directed mutagenesis and have deletions of amino acids 441 to 678 and 1807 to 1970, respectively. The MBC-NPXXP mutation was generated using the oligonucleotide ACAAATGCAAGCTTTGCAAAAATTCGACGA. The MBC-NPXXPΔ1807 construct was generated by combining the previous two constructs using an internal restriction site. The MBC-SH3W47K mutation was generated using the oligonucleotide GAGACCACCCAAAGTACTACGGC. The mbc^F6.4 mutation was generated using the oligonucleotide GTGGATCTGGCTGTAATGGGTGGC. The wild-type region in pRmHa3 and/or pUAST vectors, as indicated, was then replaced with mutated sequences, and the final constructs were confirmed by sequence analysis of the entire coding region prior to transfection into S2 cells or injection into embryos.

Immunohistochemistry.

Embryos were collected on agar-apple juice plates and aged as needed at 25°C. They were fixed and stained colorimetrically or by indirect immunofluorescence with antibodies to muscle myosin as described previously (14). The mouse monoclonal antibody to myosin (D. Kiehart, unpublished data) was used at a dilution of 1:1,000 and detected by biotinylated secondary antibody using the Vecta stain kit (Vector Laboratories, California) according to the manufacturer's instructions. To distinguish the mutant embryos from their nonmutant siblings, embryos balanced with TM3-lacZ-hg (16) were identified by a colorimetric assay for β-galactosidase activity (26) or by indirect immunofluorescence with antibodies to β-galactosidase (Cappel).

Yeast two-hybrid assays.

Full-length MBC and mutant constructs MBC-NPXXPΔ1807 and MBC-SH3 (15 to 95 amino acids) were fused in frame to the GAL4 DNA binding domain (GAL4-BD) in the yeast two-hybrid bait vector pGBKT7 (Clontech), and the full-length DCrk cDNA was cloned in frame with the activation domain (GAL4-AD) in the prey vector pGADT7 (Clontech). Expression of each fusion protein was confirmed by Western blotting. Plasmids were cotransformed into yeast strain AH109 (Clontech) using the manufacturer's small-scale yeast transformation protocol, and transformants were plated on medium lacking leucine and tryptophan to select for both plasmids. Cotransformants were then plated on medium lacking adenine, histidine, leucine, and tryptophan to select for colonies that expressed interacting proteins. Growth rates were analyzed by drop tests using serial dilutions of mid-log-phase cultures. The interactions were quantified by assaying β-galactosidase activity in liquid cultures using o-nitrophenyl-β-d-galactopyranoside (ONPG).

Cell transfection, immunoprecipitation, fractionation, and immunoblotting.

For expression in S2 cells, MBC sequences and DCrkSH2L were cloned into pRmHa3 (7) and/or pUAST (5). Studies with DCrkSH2S used clones as described previously (21). Transient calcium phosphate transfections were carried out using 1 × 10⁷ Drosophila S2 cells and 15 μg of each DNA as needed. When using sequences cloned into pRmHa3, expression was induced with 0.7 mM CuSO₄ for 24 h. For immunoprecipitations, cells were lysed after 48 h in 50 mM Tris, pH 7.5, 350 mM NaCl, 1 mM EDTA, pH 8.0, 1% Triton X-100, 2 mM phenylmethylsulfonyl fluoride (PMSF), 2 μg/ml leupeptin, 2 μg/ml pepstatin. Samples were immunoprecipitated by using the following mixtures: 1.5 mg lysate and anti-HA resin (Roche), 300 μg lysate and anti-FLAG resin (Sigma), or 2 mg lysate and di-C₆ PtdIns(3,4,5)P₃ beads (Echelon). Membrane preparations of transfected S2 cells were carried out essentially as described previously for Dock180 studies with HEK293-T cells (12, 27), in which the cells were harvested in phosphate-buffered saline and resuspended in buffer A (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM NaF, 2 mM PMSF, 2 μg/ml leupeptin, 2 μg/ml pepstatin). Following a single freeze in liquid nitrogen and thaw in a water bath at 37°C, the cytosolic fraction was recovered as supernatant after centrifugation at 16,000 × g for 10 min. The pellet, or membrane fraction, was suspended in buffer A plus 1% Triton. Cell equivalents of cytosolic and membrane fractions (approximately 30 μg) were analyzed in immunoblots using anti-HA (1:4,000; Roche) or anti-FLAG (1:2,000; Sigma) using ECL plus. Anti-α-tubulin (Sigma) was used at 1:25,000 as a control for loading.

Preparation of embryo lysates.

For the determination of transgene protein expression by Western analysis, transgenic flies were crossed with the indicated GAL4 drivers. Embryos were collected, aged to 8 to 15 h after egg laying (AEL), and then dechorionated in 50% bleach and rinsed. Whole-embryo lysates were homogenized in lysis buffer (50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl, 2 mM PMSF, 2 μg/ml leupeptin, and 2 μg/ml pepstatin), and extracts were centrifuged at 20,000 × g at 4°C for 10 min. Immunoblots utilized 1 mg of total embryo extract. MBC was detected with antisera directed against the N terminus (15 to 456 amino acids) at a dilution of 1:1,000. Anti-α-tubulin (Sigma) was used at 1:25,000 as the loading control.

For the analysis of membrane proteins in embryos, MBC transgenic flies were crossed with the mef2-GAL4 driver. Embryos were collected, aged to 6 to 12 h AEL, and then dechorionated in 50% bleach and rinsed with water. The blotted wet weight of embryos was determined, and the embryos were homogenized in cold hypotonic medium (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 30 μg/ml PMSF, 2 μg/ml leupeptin, and 2 μg/ml pepstatin). The resulting extract was centrifuged at 300 × g at 4°C for 10 min. The supernatant was then overlaid onto a cushion of 0.25 M sucrose and centrifuged for 1 h at 4°C at 100,000 × g. The cytosolic fraction was recovered as the supernatant. The pellet or “membrane” fraction was homogenized in phosphate-buffered saline with PMSF. Embryo equivalents of membrane and cytosolic fractions (corresponding to approximately 2 mg wet weight of embryos) were analyzed by immunoblotting as described above.

Mapping EMS-induced mbc mutant alleles.

Stocks in which ethyl methanesulfonate (EMS)-induced mbc mutant alleles (14) were balanced with TM3 P{GAL4-twi.G}2.3, P{UAS-2xEGFP}AH2.3, Sb Ser were generated. Prior to collection, embryos were aged for 12 to 18 h AEL at 25°C and dechorionated in 50% bleach. Total RNA was isolated from these embryos by using TRIzol (Gibco/BRL). Reverse transcription was catalyzed by SuperScript II reverse transcriptase (Invitrogen) using 0.5 μg of total RNA and mbc-specific primers. Fragments of cDNA that correspond to regions +26 to +2432, +2081 to +4205, and +3932 to +6598 were amplified using PCR, and the products were sequenced by the Stowers Institute for Medical Research molecular biology core facility. All mutations were confirmed in multiple sequence reactions from independently amplified mRNA.

RESULTS

Elimination of zygotic mbc defines its loss-of-function muscle phenotype.

Previous studies have indicated that the mbc transcript is present in mRNA isolated from unfertilized eggs and may give rise to MBC protein that would be present in embryos lacking zygotic mbc. We therefore sought to determine whether this maternal product obscured or decreased the severity of the muscle defects in mbc mutant embryos, which are characterized by an apparent absence of all myoblast fusion (14, 39) (Fig. 1B and C). To address this issue, we examined the loss-of-function phenotype of embryos lacking both maternal and zygotic mbc. In short, germ line clones were generated from the mbc^D11.2 allele using the heat shock-induced autosomal FLP-FRT dominant female sterile system as described in Materials and Methods. The musculature of embryos that lacked both the maternal and zygotic components of mbc was then examined immunohistochemically. As shown in Fig. 1, there are no obvious differences between the myoblast defects seen in embryos lacking both maternal and zygotic mbc and in those that lack only zygotic mbc. Thus, mbc transcript that is provided to the egg maternally does not appear to influence the zygotic loss-of-function muscle phenotype.

FIG. 1.

Analysis of embryos lacking both maternal and zygotic mbc. The muscle pattern was visualized by immunostaining of late stage 15 or early stage 16 embryos with antisera directed against muscle myosin (myosin heavy chain). All panels are lateral views, with anterior to the left and dorsal to the top. (A) Wild-type embryo. The segmentally repeating array of multinucleate muscle fibers is apparent. (B and D) Low and high magnification views, respectively, of embryos that lack zygotic expression of mbc. As previously described (14, 39), no myoblast fusion occurs in the absence of zygotic mbc. (C and E) Low- and high-magnification views, respectively, of embryos lacking both the maternal and zygotic mbc transcripts. No myoblast fusion is observed, and the mutant phenotype is approximately as severe as that observed in panels B and D. Scale bar = 10 μm.

MBC is required in fusion-competent myoblasts.

Formation of the body wall muscles during Drosophila embryogenesis is dependent on fusion between two cell populations, the founder and fusion-competent myoblasts. We used the UAS-GAL4 targeted expression system (5) to determine whether expression of MBC is required in the fusion-competent cell population. A single copy of the mbc cDNA under control of an upstream activating sequence was stably integrated into the genome. Its expression was then directed to all founder cells of mbc mutant embryos using rP298-GAL4 (38) and to all myoblasts using the panmesodermal drivers twi-GAL4 or 24B-GAL4. Expression directed by rP298-GAL4 occurs in founder cells over several hours, coincident with myoblast fusion. Expression directed by twi-GAL4 is observed earlier throughout the developing mesoderm, prior to the onset of myoblast fusion. 24B-GAL4 is expressed in all muscle cell types in the embryo, at a time roughly coincident with expression of muscle myosin. This approach rescued myoblast fusion in mbc mutant embryos when expression of the full-length mbc cDNA was targeted to all myoblasts using either twi-GAL4 or 24B-GAL4 (Fig. 2A or B, respectively). At least four independent embryo preparations were analyzed for each driver, in which a minimum of 25 homozygous mbc mutant embryos were identified by the absence of β-galactosidase associated with the balancer chromosome (see Materials and Methods). Upon immunohistochemical visualization of their muscle pattern with antisera directed against muscle myosin, 100% of these embryos exhibited a fairly normal pattern of multinucleate muscle fibers, with representative embryos shown in Fig. 2. By contrast, however, a comparable number of embryos in which mbc expression was targeted exclusively to founder cells appeared to be phenotypically identical to mbc mutant embryos (Fig. 2C). These results demonstrate that expression of mbc exclusively in the founder cells is inadequate to rescue myoblast fusion, whereas panmesodermal expression of MBC in both founder and fusion-competent cells under the control of twi-GAL4 or 24B-GAL4 is sufficient. We therefore conclude that MBC is essential in the fusion-competent cells.

FIG. 2.

MBC is required in the fusion-competent myoblasts. Embryos were immunostained as in Fig. 1. All panels are lateral views of stage 16 embryos, with anterior to the left and dorsal to the top. (A) mbc mutant embryo in which expression of the mbc cDNA is directed to mesodermal tissues using twi-GAL4. An apparently normal pattern of muscles is evident. (B) mbc mutant embryo in which expression of UAS-mbcHA is directed to mesodermal cells under control of 24B-GAL4. The muscle pattern is similar to that seen in the wild type (Fig. 1) and panel A. (C) mbc mutant embryo in which expression of UAS-mbcHA is provided exclusively to the founder cells under control of rP298-GAL4. Muscle myotubes are absent and have been replaced by mononucleate myoblasts. Thus, the embryo is phenotypically mutant. Scale bar = 13 μm.

The MBC SH3 and Docker domains are essential for myoblast fusion.

To address whether the other known protein-interacting or functional domains of MBC are essential to rescue the muscle defect of mbc mutant embryos, we generated mutations in the SH3 and Docker domains (Fig. 3A). The SH3 domain is critical in Dock180 for interaction with Elmo, and the conserved Docker domain interacts with Rac1 and mediates GTP-GDP exchange (6, 13, 29). The SH3 domain of MBC was disrupted by converting a conserved tryptophan (W⁴⁷) that is known to be essential in other SH3 domains to lysine (K⁴⁷), generating MBC-SH3W47K. To disrupt the Docker domain, we introduced a genetic lesion previously identified in the EMS-induced loss-of-function mbc^F6.4 allele (14) into the wild-type sequence (MBC-DockerF6.4) in pUAST (see Materials and Methods). Protein expression from the resulting transgenes was confirmed in whole-cell lysates prepared from embryos, and the absence of dominant overexpression muscle phenotypes was confirmed in embryos by myosin immunohistochemistry (data not shown). We also established that both MBC-SH3W47K and MBC-DockerF6.4 were enriched at the membrane (Fig. 3C). Thus, neither the wild-type SH3 nor the wild-type Docker domain appears to be necessary for MBC to be recruited to the myoblast membrane.

FIG. 3.

The MBC SH3 and Docker domains are essential for MBC function but not membrane localization. (A) Domains of MBC in which mutations were generated. The MBC-SH3W47K mutant has an altered consensus SH3 domain amino acid. The MBC-DockerF6.4 mutant recapitulates a mutation found in the mbc^F6.4 EMS-induced missense allele. An HA tag has been added to the C terminus of all three constructs. (B) Embryos were immunostained as in Fig. 1, with the genotypes indicated. MBC transgenes with point mutations that disrupt either the SH3 domain or the Docker domain are unable to rescue myoblast fusion in mbc mutant embryos. Scale bar = 10 μm. (C) Immunoblot from Drosophila embryo cytoplasmic (C) and membrane (M) lysates in which the MBC SH3 or Docker domain mutations were under control of the mesodermal mef2-GAL4 driver. We conclude that all forms of MBC are expressed in embryos and enriched at the membrane.

Using several independent transgenic lines, we then examined whether these altered forms of MBC are capable of rescuing the muscle defects in mbc mutant embryos. As visualized immunohistochemically using antisera against muscle myosin, MBC-SH3W47K failed to rescue the myoblast fusion defect when expression was driven by either twi-GAL4 or 24B-GAL4. A representative embryo is shown in Fig. 3B. Thus, the SH3 domain is absolutely essential for MBC function in the embryonic musculature. The importance of the Docker domain was confirmed by expressing UAS-MBC-DockerF6.4 under control of either twi-GAL4 or 24B-GAL4. As shown by a representative embryo in Fig. 3B, this protein was unable to rescue myoblast fusion, thereby validating our rescue assay by recapitulating the genetic loss-of-function phenotype observed with the mbc^F6.4 EMS-induced mutant allele.

The MBC PXXP sites direct interaction with DCrk.

Published studies with MBC and DCrkSH2L have revealed that these proteins interact biochemically in purified systems (15, 21). We have extended these studies to include both low- and high-affinity PXXP binding sites in an effort to evaluate the importance of MBC-DCrk interactions in the embryonic musculature. MBC contains three low-affinity PXXP sites and one consensus PPXLPXK Crk-binding site in the C terminus, as well as one low-affinity PXXP site in the N-terminal region. SH3 domains such as those present in DCrk can bind to their targets through interactions with multiple low-affinity PXXP sites (32). We therefore generated a series of MBC mutant constructs that altered a combination of consensus and nonconsensus proline-rich binding sites (Fig. 4A). In MBC-CBS, two conserved prolines, the leucine and the lysine residue in the consensus PPXLPXK motif, were altered to alanines. The entire C terminus of MBC was deleted in MBCΔ1807. The prolines in the N-terminal PXXP motif at amino acids 110 to 113 were converted to alanines to generate MBC-NPXXP, and this region was put into the MBCΔ1807 background to generate MBC-NPXXPΔ1807. We also examined rescue of the myoblast fusion defect using the previously described C-terminal deletion MBCΔPRM (21). This construct has a deletion of the MBC C-terminal region past amino acid 1661 and no longer binds to DCrk (21), and it should therefore behave no differently from MBCΔ1807 with respect to DCrk binding. However, in light of the large difference between these deletions, we wished to ensure that this assumption was correct. The altered sequences were cloned into appropriate vectors for the yeast two-hybrid system, for expression in cultured Drosophila S2 cells, and for expression in embryos (see Materials and Methods). For DCrk, studies have recently identified two isoforms that differ by 18 amino acids in the SH2 domain (Fig. 4A) (21). Most experiments were carried out with both the 271-amino-acid DCrkSH2L isoform and the 253-amino-acid DCrkSH2S isoform.

FIG. 4.

Interaction with DCrk is dependent on proline-rich sites in the carboxy terminus of MBC. (A) Mutations that disrupt putative DCrk-binding sites in MBC are shown in red. Also shown are the two identified isoforms of DCrk (21). The MBC-CBS construct contains a mutation in the consensus DCrk-binding site. The MBCΔ1807 construct has a deletion of amino acids 1807 to 1970. The MBC-NPXXP mutation alters the N-terminal PXXP motif. The MBC-NPXXPΔ1807 construct was generated by combining the previous two constructs. MBCΔPRM has a deletion of amino acids 1662 to 1970 (21). An HA tag has been added to the C terminus of all MBC constructs except MBC-CBS. (B) Yeast two-hybrid assay, with all MBC constructs fused in frame to the DNA binding domain (BD) and DCrk constructs fused in frame to the activation domain (AD). Growth on selective plates is shown on the right. Clearly, both DCrkSH2L and DCrkSH2S exhibit substantial interaction with full-length MBC, whereas that observed with MBC-NPXXPΔ1807 or MBCΔPRM is severely reduced. (C) Yeast two-hybrid assay with constructs described for panel B. Growth rates were analyzed by drop tests using serial dilutions of mid-log-phase cultures. Fourfold dilutions of the cell culture are shown from left to right. U, undiluted. (D) The yeast two-hybrid interactions were quantified by assaying β-galactosidase activity in liquid cultures using ONPG. β-Galactosidase activity in liquid cultures is expressed in Miller units as the mean ± standard deviation of eight independent assays.

We first used the yeast two-hybrid assay to examine direct protein-protein interactions between DCrk and MBC. These studies utilized full-length MBC or MBC-NPXXPΔ1807 expressed in frame with the yeast vector DNA binding domain and either DCrkSH2L or DCrkSH2S in frame with the yeast vector activation domain. The MBC SH3 domain alone, cloned similarly, served as a negative control. The generation of approximately equal levels of protein from all constructs was confirmed in immunoblots of yeast extract (data not shown). Growth of the yeast cells was examined on medium lacking adenine, histidine, leucine, and tryptophan to select for colonies expressing interacting proteins. As expected, a strong interaction between full-length wild-type MBC and either isoform of DCrk was obtained. Neither empty vector controls nor the SH3 domain of MBC alone exhibited detectable background binding (data not shown). Most importantly, the MBC-NPXXPΔ1807 construct, in which all proline-rich sites are mutated or deleted, exhibited only very weak growth in combination with either DCrkSH2L or DCrkSH2S (Fig. 4B). Thus, the strong interaction of MBC with either DCrk isoform was dependent on the presence of the MBC C terminus and/or N-terminal PXXP site. Growth rates between each MBC construct and DCrkSH2L were then analyzed by drop tests using serial dilutions of mid-log-phase cultures (Fig. 4C), and the β-galactosidase activity in liquid cultures was measured to quantify the two-hybrid interactions (Fig. 4D). Both of these analyses support the anticipated observation that the PXXP motifs direct interaction of MBC with DCrk.

We then assayed the ability of MBC to interact with DCrkSH2L and DCrkSH2S in cotransfected Drosophila S2 cells by immunoprecipitation (Fig. 5A). C-terminal FLAG-tagged DCrkSH2L was cotransfected with either functional full-length C-terminal HA-tagged MBC, C-terminal HA-tagged MBC-NPXXPΔ1807, or C-terminal HA-tagged MBCΔPRM (21). The immunoblot samples were normalized for equivalent amounts of immunoprecipitated FLAG-tagged DCrkSH2L and then probed with HA to detect the presence of MBC. Clearly, the ability of DCrkSH2L to immunoprecipitate MBC was severely diminished for both MBC-NPXXPΔ1807 and MBCΔPRM. Thus, deletion of the C-terminal region of MBC that contains the consensus proline-rich DCrk binding sites is essential for interaction of MBC with DCrkSH2L. We also confirmed the importance of this region for interaction of MBC with FLAG-tagged DCrkSH2S (Fig. 5B). In this assay, full-length MBC and MBC-NPXXPΔ1807 were immunoprecipitated using the HA tag at their C terminus, and equivalent amounts of each protein were analyzed in immunoblots for the ability to bring down FLAG-tagged DCrkSH2S. As anticipated, full-length MBC but not MBC-NPXXPΔ1807 was able to interact with DCrkSH2S. We therefore conclude that either isoform of DCrk, at least one of which is expressed in the mesoderm (15), can interact strongly with the MBC C terminus, and neither form is capable of binding to MBC in the absence of this region (21; this study).

FIG. 5.

The MBC C terminus mediates strong interaction with DCrk. Immunoprecipitations (IP) are from lysates of Drosophila S2 cells cotransfected with HA-tagged MBC, MBC-NPXXPΔ1807, or MBCΔPRM and FLAG-tagged DCrk. Anti-FLAG or anti-HA antibodies were used for immunoprecipitation and immunoblotting as indicated. (A) Full-length (FL) MBC, but neither MBC-NPXXPΔ1807 nor MBCΔPRM, is immunoprecipitated by interaction with DCrk. (B) DCrk is immunoprecipitated through an interaction with full-length MBC but not with MBC-NPXXPΔ1807.

The MBC proline-rich DCrk-binding regions are not required for myoblast fusion or membrane localization.

The majority, if not all, of the studies examining the interaction of MBC and DCrk have been carried out under conditions of transient transfection, in which the levels of both proteins in vast excess over those expressed normally in the cell. We therefore sought to examine MBC in the embryo using the targeted UAS-GAL4 system, which provides a more modest level of expression than that usually obtained in transfection experiments in cultured cells, since the copy number of the UAS cDNA is limited to one per cell and GAL4 can be expressed under the control of an endogenous promoter.

Prior to examining the ability of MBC-CBS, MBCΔ1807, MBC-NPXXP, and MBC-NPXXPΔ1807 to rescue the muscle defects of mbc mutant embryos, we confirmed the absence of dominant overexpression phenotypes from these transgenes by myosin staining (data not shown). Protein expression was then confirmed in immunoblots of whole-cell lysates prepared from embryos and probed with antisera against the HA tag (data not shown). We then compared the levels of expression of MBC-NPXXPΔ1807 and MBCΔPRM directed by 24B-GAL4 and twi-GAL4 in whole-embryo lysate immunoblots probed with antisera against the N terminus of MBC. The full-length GAL4-directed MBC protein migrates at the same position as endogenous MBC in this assay, while the truncated forms migrate at a lower position. As shown in Fig. 6A, the overall level of expression driven by 24B-GAL4 is somewhat lower than that driven by twi-GAL4 for all three transgenes. In addition, the total amount of truncated MBC visible in lanes 2, 3, and 5 is lower than the total amount of the endogenous protein. Although the level of truncated MBC per myoblast likely exceeds the endogenous level of MBC severalfold, since endogenous MBC is more broadly expressed, this level is significantly closer to the endogenous level than that obtained in standard transient transfections. We also examined whether the truncated proteins were enriched at the membrane using crude membrane preparations from embryo lysates. We anticipated that, if DCrk was necessary to recruit MBC to the membrane as expected, the truncated proteins would be enriched in the cytoplasmic fractions while full-length MBC would be enriched in the membrane fraction. Surprisingly, immunoblots of embryo lysates probed with antisera against the HA tag revealed that a significant amount of the truncated forms of MBC was retained in the membrane fraction relative to the α-tubulin control (Fig. 6B). We conclude from these analyses that the transgenes are expressed and that the proline-rich regions are not necessary for them to become associated with the membrane.

FIG. 6.

MBC C-terminal truncations are expressed and present in the embryo membrane. (A) Immunoblot of whole-cell lysates from embryos aged 8 to 15 h AEL, in which the HA-tagged full-length (FL) MBC or MBC-NPXXPΔ1807 and MBCΔPRM truncations were under control of the later mesodermal 24B-GAL4 driver (lanes 1 to 3) or the early mesodermal driver twi-GAL4 (lanes 4 to 6). Lanes 1 and 4, both endogenous and GAL4-directed MBC migrate at the same position and are detected by antisera directed against the MBC N terminus; lanes 2, 3, 5, and 6, truncated MBC proteins as indicated correspond to the lower migrating form in each lane, while the endogenous protein corresponds to the more slowly migrating form. While there are some differences in signal intensity between the samples from 24B-GAL4 (lanes 1 to 3) and those from twi-GAL4 (lanes 4 to 6), the level of the truncated MBC proteins is lower than endogenous full-length MBC when 24B-GAL4 is used. (B) Immunoblot of embryo cytoplasmic (C) and membrane (M) lysates in which the HA-tagged MBC-NPXXPΔ1807 or MBCΔPRM truncations were under control of the mesodermal mef2-GAL4 driver. Truncated proteins were detected by anti-HA antisera and correspond to the anticipated sizes. All forms of MBC are present at the membrane, with a significant membrane enrichment of full-length MBC and MBC-NPXXPΔ1807. Cytoplasmic α-tubulin served as a control.

We then examined the ability of these transgenes to rescue the muscle defects of embryos lacking mbc. As described above, at least three experiments were carried out for each transgene, in which a minimum of 25 homozygous mbc mutant embryos were identified by the absence of β-galactosidase associated with the balancer chromosome (see Materials and Methods). For each transgene, a minimum of 20 rescued embryos was analyzed. In all cases, only modest variations were observed in the extent of rescue, with representative embryos shown in Fig. 7. The UAS-MBC-CBS transgene, with mutations in the consensus PPXLPXK Crk-binding site of MBC, rescued myoblast fusion to a nearly normal pattern of muscle fibers (Fig. 7A and B). Thus, the consensus Crk-binding site in MBC is not required for MBC to function in the mesoderm when a single-copy construct is driven by a single copy of either twi-GAL4 or 24B-GAL4. Expression of UAS-MBCΔ1807, UAS-MBC-NPXXP, UAS-MBC-NPXXPΔ1807, or UAS-MBCΔPRM was also sufficient to rescue myoblast fusion (Fig. 7C to J), demonstrating that none of the proline-rich Crk-binding sites are necessary for MBC to direct embryonic myoblast fusion in this assay.

FIG. 7.

The MBC DCrk-binding sites are not essential for myoblast fusion. Embryos were immunostained as in Fig. 1. All panels are lateral views of stage 16 embryos, with genotypes as indicated. In panels A, C, E, G, and I, expression of the indicated MBC truncations is driven early in muscle development using twi-GAL4. In panels B, D, F, H, and J, expression of the indicated MBC truncations is driven later and at lower levels during muscle formation using 24B-GAL4. (A and B) The MBC consensus DCrk-binding site is not essential to rescue the mbc mutant muscle phenotype. (C and D) The MBC region carboxy terminal to amino acid 1807 is not essential to rescue the mbc mutant muscle phenotype. (E and F) The amino-terminal DCrk-binding site is not essential to rescue the mbc mutant muscle phenotype. (G and H) Neither the amino- nor carboxy-terminal DCrk-binding sites direct binding of DCrk to MBC. (I and J) The MBCΔPRM C-terminal truncation described by Ishimaru et al. (21), which does not interact with DCrk, is able to rescue the MBC mutant muscle phenotype. Scale bar = 10 μm.

In summary, the yeast two-hybrid results and immunoprecipitation assays described above demonstrate that DCrk does not interact with MBC in the absence of the proline-rich Crk-binding sites. However, functional data in embryos suggest that the evolutionarily conserved interaction of DCrk with MBC is not required to rescue MBC-associated defects in myoblast fusion. While the role, if any, of DCrk in myoblast fusion remains elusive, our data clearly indicate that MBC can function through a pathway that is independent of its interaction with DCrk in both myoblast cell types.

The DHR1 region binds to PtdIns(3,4,5)P₃ and is essential for myoblast fusion.

The DHR1 region of Dock180 has recently been identified as a novel PtdIns(3,4,5)P₃ binding module that localizes Dock180 to the leading edge of elongating and migrating mammalian cells in culture (12, 27). To examine whether the analogous MBC domain is also capable of binding to PtdIns(3,4,5)P₃ and plays a critical role in embryonic myoblast fusion, we generated a truncated form of MBC similar to the Dock180 DHR1 deletion described by Cote et al. and Kobayashi et al. (12, 27), shown in Fig. 8A. Following cotransfection into S2 cells (see Materials and Methods), lysates were analyzed for the ability of full-length MBC and MBCΔDHR1 to be immunoprecipitated by PtdIns(3,4,5)P₃-coupled beads. Like Dock180, full-length MBC is capable of binding to PtdIns(3,4,5)P₃ through an interaction mediated by the DHR1 domain (Fig. 8B).

FIG. 8.

The MBC DHR1 domain binds PtdIns(3,4,5)P₃ (PIP3) and is essential for MBC function in myoblasts. (A) Schematics of DHR1 deletions in MBC. (B) Immunoprecipitations from lysates of Drosophila S2 cells transfected with HA-tagged MBC (lanes 1 and 6), MBC-NPXXPΔ1807 (lanes 2 and 7), MBCΔPRM (lanes 3 and 8), MBCΔDHR1 (lanes 4 and 9), and MBC-NPXXPΔDHR1:1807 (lanes 5 and 10) as described in Materials and Methods. Only those forms of MBC that lacked the DHR1 domain were not immunoprecipitated by interaction with PtdIns(3,4,5)P₃ (lanes 9 and 10). (C) Immunoblot from Drosophila embryo cytoplasmic (C) and membrane (M) lysates in which expression of the MBCΔDHR1 deletion alone or the DHR1 deletion in a background lacking DCrk-binding sites was under control of the mesodermal mef2-GAL4 driver. The top panel was probed with anti-HA to detect the tagged MBC protein, and the bottom panel provides a control for loading and membrane fractionation. We conclude that all forms of MBC are expressed in embryos and present at the membrane. (D) Embryos were immunostained as in Fig. 1, and the genotypes are indicated. MBC transgenes lacking the PtdIns(3,4,5)P₃ binding region DHR1 are unable to rescue myoblast fusion in mbc mutant embryos.

Transgenic flies containing UAS-MBCΔDHR1 were also generated, and the expression of this gene was directed to the musculature under GAL4 control. Both transgene expression and the absence of dominant overexpression muscle phenotypes were confirmed as described above (data not shown). We then examined whether HA-tagged MBC lacking the PtdIns(3,4,5)P₃ binding DHR1 region was enriched in crude membrane fractions from the myoblasts in which it was expressed. As shown in Fig. 8C, MBCΔDHR1 is clearly associated with the membrane at levels comparable to its full-length counterpart. Thus, the DHR1 domain does not appear to be essential for recruitment of MBC to the myoblast membrane. Moreover, it does not function redundantly with DCrk in membrane recruitment, since deletion of both of these domains does not dramatically alter membrane enrichment of MBC-NPXXPΔDHR1:1807.

We then examined whether MBCΔDHR1 is capable of rescuing the muscle defects in mbc mutant embryos by immunostaining as described above. In brief, two independent transgenic lines failed to rescue the myoblast fusion defect when expression was driven by either twi-GAL4 or 24B-GAL4. A representative embryo is shown in Fig. 8D. Clearly, the DHR1 domain is essential for MBC function in the embryonic musculature. We infer from these studies that PtdIns(3,4,5)P₃ binding by MBC is important in myoblast fusion independently of any role that it might have in membrane recruitment.

Analysis of MBC through EMS-induced loss-of-function mutations.

As a final approach to examine regions of MBC necessary to rescue the mutant phenotype, we identified the sequence lesions in 16 independent EMS-induced mutations (Fig. 9). These mutant alleles were isolated in genetic F₂ adult lethal screens (14, 39, 40), and the mutant flies exhibit similar defects in the embryonic musculature (data not shown). All of the nonsense mutations result in translation stops prior to, or within, the Docker domain, and four missense mutations are distributed throughout the Docker domain (Fig. 9A and B). We examined the corresponding regions of CDM family members to determine whether these amino acid residues are conserved. Shown in Fig. 9C is an alignment of the four corresponding regions of the Docker, or “CZH2,” domain of the Dock180-related subfamily (34). The alanine at position 1281, which is highly conserved among subfamily members, has been changed to a threonine in the mbc^O146 mutant. In the mbc^ZZ260 mutant allele, the negatively charged glutamate at position 1381 has been changed to a positively charged lysine. While the glutamate residue is not conserved in all subfamily members, 10 of these 14 contain a negatively charged amino acid in this position, and only Neurospora Dock is predicted to have a positively charged amino acid in this position. The mbc^26.2 mutant contains an arginine-to-tryptophan mutation. This change is just 3 amino acids downstream of the conserved ISP residues shown by Brugnera et al. to eliminate membrane ruffling and Rac-GTP loading by Dock180 (6). While most subfamily members contain an asparagine in this position, the other Drosophila ortholog Sponge (Dock4) contains a tyrosine. It is interesting that a tryptophan in this position renders MBC inactive, yet a tyrosine residue is normally found in this position in Sponge/Dock4, perhaps indicating that the Drosophila MBC and Sponge proteins utilize this region in quite different interactions. Finally, a proline-to-leucine change at position 1579 of the mbc^F6.4 mutant, which we have recapitulated in our UAS-GAL4 rescue assay, renders mbc nonfunctional, yet several other family members, including C. elegans CED-5, contain a leucine in this position. The relevance of these amino acid changes to the activity and interactions of the corresponding family members will be fully understood only when the interactors are known and crystal structures have been obtained.

FIG. 9.

Sequence lesions in EMS-induced mbc mutant alleles. A total of 17 EMS-induced mbc mutant alleles were isolated in previously published studies (14), obtained in other screens, or generously provided by other investigators. The sequence lesions in these mutant alleles were determined as described in Materials and Methods. (A) Schematic representation of the MBC protein, with protein domains and the relative positions of EMS-induced mutations indicated. (B) The mutant allele, the amino acid (aa) affected by the mutation, and the nature of the change are listed. (C) Homology alignment of the locations of the four Docker domain missense mutations within the Dock180-related subgroup of CZH Rho-GEFs described by Meller et al. (34). Alignment was carried out using Vector NTI 9.0.0 (Informax) and shaded using Boxshade. The following are GenBank GenInfo Identifier numbers: for D. melanogaster MBC, 7511969; for human Dock180/Dock1, 4503355; for human Dock2, 31377468; for human Dock3, 23297197; for human Dock4, 29568109; for human Dock5, 45439362; for D. melanogaster Sponge/Dock4, 28381487; for C. elegans CED-5, 7511497; for Dictyostelium discoideum (Dicty) DocA, 66801748; for Dictyostelium discoideum DocB, 60474615; for Dictyostelium discoideum DocC, 66801673; for Dictyostelium discoideum DocD, 66809741; for Neurospora crassa Dock, 32418746; and for Saccharomyces cerevisiae Dock, 6323454.

DISCUSSION

Dock180, the most extensively studied CDM family member, is a nonconventional GEF that facilitates GTP loading on Rac1 (24). PXXP sites near its C terminus mediate interaction with the small SH2-SH3 adapter protein Crk, forming a complex that can increase GTP exchange on Rac1. These interactions form the basis for the Crk-Dock180-Rac1 signaling pathway. Dock180 has also been found in a trimolecular complex which, along with CrkII, contains the PH domain-containing protein Elmo (18). The primary role of Crk appears to be in recruiting these nonconventional GEFs to target sites at the cell membrane, thereby facilitating interaction with downstream effectors that regulate the cytoskeleton. Crk also appears to regulate the assembly and function of the Dock180/Elmo complex itself (2). Orthologs of these molecules have been identified in C. elegans, and genetic studies support a mechanism in which they function together in cell engulfment (36). In Drosophila, mutants with loss-of-function mutations in mbc, dominant negative and constitutively active forms of Rac1, and loss-of-function mutations in both rac1 and rac2 all exhibit defects in embryonic myoblast fusion (14, 19, 30, 35). DCrk was isolated in a biochemical screen for molecules that interact strongly with the MBC C terminus, but efforts to determine its embryonic loss-of-function phenotype have been hindered by the presence of high levels of maternally provided transcript (15). Moreover, its location on the fourth chromosome effectively eliminates the possibility of generating germ line clones to uncover a role for DCrk during embryonic development. Notwithstanding these obstacles, or perhaps because of them, more recent studies have made use of RNA interference (RNAi) technology and insertional mutations to examine adult thorax closure, where maternally provided transcripts are not present to complicate the analysis (21). These studies have revealed modest genetic interactions between a DCrk insertional mutation, DCrk RNAi, D-Elmo RNAi, or mbc RNAi with loss-of-function alleles for Pvr, which encodes the Drosophila ortholog of the platelet-derived growth factor/vascular endothelial growth factor receptor. Genetic interactions between pathway members have also been observed in the Drosophila eye, where mutations in mbc suppress the effects of ectopic Rac1 (35). Together, these observations support conservation of the Crk-CDM-Rac1 pathway in Drosophila.

The studies presented herein were designed to examine MBC in the embryonic musculature. Along with the analysis of traditional EMS-induced point mutations in MBC, we examined rescue of the loss-of-function embryonic muscle phenotype. In this assay, a single copy of the mbc transgene is integrated into the genome, and expression is under the control of a single-copy muscle-specific promoter driving the yeast transcriptional activator GAL4. We determined that the SH3 domain, the PtdIns(3,4,5)P₃ binding DHR1 domain, and the Docker domain of MBC are all essential. The latter requirement is consistent with the rac1-associated myoblast fusion phenotype and supports the interpretation that, as in other systems, the Docker domain interacts with Rac1. Perhaps most interesting is the implication that phosphoinositol signaling is essential for myoblast fusion in a mechanism that appears to involve PtdIns(3,4,5)P₃ binding to MBC.

We also addressed the in vivo relevance of the CDM-Crk biochemical association in Drosophila by examining the ability of mutant forms of MBC to rescue its loss-of-function muscle phenotype. Unexpectedly, in contrast to working models (4, 21, 37), the direct interaction of MBC with DCrk does not appear to be an essential part of the pathway through which MBC regulates myoblast fusion. Biochemical and yeast two-hybrid interactions confirm that the Crk-binding sites in MBC are active and are the only regions of MBC to directly interact with Crk. Nevertheless, MBC appears to be fully functional and present at the membrane despite the absence of these sites. This finding is reminiscent of results in cultured cells transiently transfected with Dock180, in which it cooperates with Elmo to stimulate migration in a transwell assay in the absence of the Crk-binding sites (17, 22). Our results extend this analysis to MBC in the musculature of the intact embryo and establish that no other regions of MBC directly interact with DCrk in the absence of the C terminus.

It remains to be determined how MBC is localized to the membrane, assuming that this is an important step in the pathway. Notably, simple membrane targeting of full-length MBC via a myristoylation signal does not rescue myoblast fusion (see Fig. S1 in the supplemental material). It is a formal possibility that DCrk directs membrane association of MBC through a novel mechanism that does not involve its direct binding to MBC or that different domains of MBC function redundantly with DCrk in this capacity. In mammalian cells, the SH3 domain of Dock180 can target it to the membrane through its interaction with Elmo and, in turn, Elmo's interaction with RhoG (17, 22, 23). In addition, interaction of the highly conserved DHR1 domain with PtdIns(3,4,5)P₃ directs Crk-independent membrane localization of Dock180 (12, 27). By comparison, MBC is present at the membrane when the PtdIns(3,4,5)P₃ binding domain is deleted and when mutations rendering the SH3 domain nonfunctional are present. Thus, none of these domains are exclusively essential for membrane targeting of MBC in the somatic mesoderm. Despite the possibility that they act redundantly with DCrk in directing membrane localization of MBC, however, the SH3, DHR1, and Docker domains clearly play essential roles independent of this process in the ability of MBC to direct myoblast fusion.

Interestingly, a role in recruitment of MBC to target sites at the membrane in the embryonic musculature has been suggested for the founder myoblast-specific Ants/Rols protein (8). This protein links an N terminus-containing fragment of MBC to the cytoplasmic domain of the founder-specific cell adhesion molecule Duf/Kirre (8). Studies have not yet addressed whether the domains that direct these interactions are essential or whether similar interactions occur between Ants/Rols and IrreC/rst, which functions redundantly with Duf/Kirre in the musculature (1). Moreover, our data suggest that MBC expression is also required in the fusion-competent cells, which lack Ants/Rols. One might anticipate the presence of a molecule similar to Ants/Rols that recruits MBC to the cytoplasmic domain of the SNS cell adhesion molecule in these myoblasts. In a variation of this model, PXXP sites in the SNS cytoplasmic domain may interact directly with the SH3 domain of MBC, thereby recruiting it to the membrane of the fusion-competent cells. In summary, it is becoming increasingly apparent that the well-characterized interactions of CDM proteins with Crk, Elmo, and Rac1 represent only a subset of their potential partners. Future studies will likely reveal a larger spectrum of proteins through which they act.