Aggregation and fusion of myoblasts to form myotubes is essential for myogenesis. In Drosophila the formation of syncytial myotubes is seeded by founder myoblasts. Founders fuse with clusters of fusion-competent myoblasts. The gene kin of irre (kirre), referred to frequently as dumbfounded (duf) is required for myoblast aggregation and fusion. duf encodes a member of the immunoglobulin superfamily of proteins that is an attractant for fusion-competent myoblasts. It is expressed by founder cells and serves to attract clusters of myoblasts from which myotubes form by fusion (Rúiz-Gomez, 2000).

Roughest (Rst), a paralog of Kirre/Dumbfounded, is strongly expressed in mesodermal tissues during embryogenesis, but rst null mutants display only subtle embryonic phenotypes. Evidence is presented that this is due to functional redundancy between Rst and Kirre. Both are highly related single-pass transmembrane proteins with five extracellular immunoglobulin domains and three conserved motifs in the intracellular domain. The expression patterns of kirre and rst overlap during embryonic development in muscle founder cells. Simultaneous deletion of both genes causes an almost complete failure of fusion between muscle founder cells and fusion-competent myoblasts. This defect can be rescued by one copy of either gene. Moreover, Rst, like Kirre, is a myoblast attractant (Strünkelnberg, 2001).

The somatic muscles of the Drosophila larva are laid out in a complex pattern on the body wall. Like skeletal muscles in vertebrates, these muscles consist of syncytial fibers formed by fusion of myoblasts. However, in the Drosophila larva, each muscle is a single myotube, whereas in vertebrates many myotubes are bundled together to form a single muscle. The pattern of muscles in the Drosophila larva is extremely precise, consisting of 30 myotubes in each abdominal hemisegment. Each of these fibers is a unique element in the pattern, distinguishable by its position, size, orientation, and innervation. Thus, as the Drosophila embryo develops, each hemisegment reproducibly generates a set of 30 different myotubes and provides a unique opportunity to study the control of myoblast fusion and myotube diversification (Rúiz-Gomez, 2000).

The aggregation and fusion of myoblasts to form syncytial myotubes is an integral part of myogenesis in many organisms. In vertebrates, proliferating myoblasts migrate from the somites to sites of muscle formation and fuse to form primary embryonic myotubes. Secondary myotubes are added in parallel with primary myotubes, and additional growth occurs through the fusion of satellite cells. Several different kinds of myotubes contribute to the final muscle and the characteristics of these fibers can be profoundly influenced by innervation. However, primary myotubes form independently of nerves so that other regulatory factors must control myoblast fusion and the diversification of fibers that occurs during primary myogenesis (Rúiz-Gomez, 2000).

In Drosophila, both fusion and myotube differentiation are tightly controlled so that myotubes with distinctive characteristics are generated at precise locations in the developing muscle pattern. Control is exerted in two ways: (1) by the segregation of a special class of founder myoblasts at specific points in muscle-forming mesoderm (the founders seed the formation of myotubes at these points by fusing with neighboring fusion-competent cells that constitute a different class of myoblasts); (2) fusion is regulated by the fact that there is an essential asymmetry to the process so that the two classes of myoblasts (founders and fusion-competent cells) can only fuse with each other and not with themselves. The presence of founders at specific sites thus gates myogenesis and restricts it to those locations where muscles should form. At the same time, the characteristics of the myotubes formed at these locations are dictated by transcription factors expressed by individual founders (Rúiz-Gomez, 2000 and references thererin).

Like Drosophila, vertebrate embryos produce a population of myotubes during myogenesis, rather than fusion being a generalized process recruiting cells to a single expanding syncytium. In common with other insects, where myogenesis is seeded by muscle founders or pioneers, Drosophila exemplifies one solution to the problem of recruiting cells in groups to form myotubes. However, it may be that an asymmetry to the fusion process, with some myoblasts acting as seeds and others being recruited, is common to many organisms. Because the separation of myoblasts into two classes is so central to myogenesis in Drosophila and might be a general requirement for myoblast fusion in many different organisms, it is important to identify those genes that give the two types of myoblasts their unique properties. The products of these genes will include proteins that enable founders and fusion-competent myoblasts to recognize each other as suitable partners for fusion and that are responsible for the inherent polarity of the fusion process. The cloning and functional characterization of the first such gene is described in this study. Because of its loss- and gain-of-function phenotypes, the gene has been named dumbfounded. In the absence of duf, fusion fails. In muscle-forming mesoderm, duf is expressed only in founders and their progenitors. duf encodes a putative cell adhesion protein that causes myoblasts to aggregate on founder cells prior to fusion to form a myotube (Rúiz-Gomez, 2000).

The following sequence of events during myoblast fusion is proposed. Initially, Dumbfounded acts as an attractant for fusion-competent myoblasts. Through either direct or indirect interaction(s) between Duf and and cell adhesion protein Sticks-and-stones (Sns), fusion-competent myoblasts recognize and adhere to founder cells. In this process, Sns is localized to discrete sites in the membrane of fusion-competent myoblasts, presumably sites of cell adhesion. It is possible that Duf is also localized to discrete domains in the membrane of the founder cells. Next, within the founder cells, through interaction(s) between the cytoplasmic domain of Duf and linker protein Rolling pebbles (Rols), Rols is recruited to discrete cytoplasmic domains close to the membrane. Meanwhile, interaction between Rols and Mbc, and perhaps additional cytoskeleton-associated molecules, leads to changes in the cytoskeleton that are necessary for the proper alignment of founder cells with fusion-competent cells. This model predicts that in rols mutant embryos, despite a block of cell alignment, which requires the transmission of signals from Duf to the cytoskeleton, cell recognition and adhesion should take place normally. This is indeed what is observed. In rols mutant embryos, fusion-competent myoblasts extend filopodia toward their fusion targets. Such phenotypes are not observed in duf mutant embryos in which fusion is blocked at the cell recognition step (Ruiz-GŪmez, 2000). Taken together, the model is favored that Rols acts as a linker molecule that relays signals from the membrane receptor Duf to changes in the cytoskeleton in the founder cells (Chen, 2001).

Dumbfounded was identified in an enhancer trap screen. ß-galactosidase expression in the enhancer trap line rP298 is confined to progenitors and founders of somatic and visceral muscles in the Drosophila embryo. The question was asked whether this expression might reflect the presence of a gene essential for the development of the mesoderm near the insertion site of the P element in rP298. The insertion site was mapped to band 3C6 on the X chromosome, between roughest (rst) and Notch (N). Although no lethal gene has been described in this interval, phenotypes of deficiencies that uncover this region were examined. Out of five embryonic lethal deficiencies tested, three [Df(1)w^258-11,Df(1)vt, and Df(1)w^67k30] were found with an interesting mutant phenotype. Df(1)w^67k30 embryos are typical, with a complete lack of fusion in the somatic mesoderm and gaps in the visceral mesoderm, whereas other mesodermal derivatives such as fat body, gonads, and heart develop normally. The phenotype is manifested in those mesodermal tissues where lacZ is expressed in rP298, suggesting that the pattern of lacZ expression reflects the requirement for a gene removed by the deficiency. To map the location of this putative gene, additional deficiencies and duplications were used. Flies are viable when either the 3C2-5 region or the 3C3-6 region (which includes rst) is deleted by combining Df(1)w^67k30 and Df(1)w^258-42 or Df(1)w^67k30 and Df(1)rst², respectively. The lethality of Df(1)w^67k30 is rescued by the duplication Dp(1;3)w^VCO, which excludes the possibility that the lethality could map outside the 3C region. In addition, Df(1)N^81k1 complements Df(1)w^67k30, whereas Df(1)N⁸ does not. Furthermore, when Df(1)N⁸ is combined with cosP479BE, which rescues all known N mutations, the resulting embryos die and show a phenotype indistinguishable from that of Df(1)w^67k30 embryos. Taken together, these results identify a novel mesodermal lethal function in 3C6-7 coinciding with the insertion point of the P element in the rP298 line. Using this line as a starting point, a transcription unit close to the insertion point of the P element was identified that corresponds to a new gene. Following the characterization of the mutant phenotype of deficiencies that remove this gene, the gene was named dumbfounded (Rúiz-Gomez, 2000).

Duf acts as an attractant for myoblast aggregation. In experiments using Dll-GAL4 as a driver in a Df(1)w^67k30 background, the migration of myoblasts is strikingly redirected toward sites of ectopic duf expression. In such embryos, there is a substantial aggregation of myoblasts in the head and on the primordia of the leg discs where duf is now expressed. When Wg-GAL4 is used as a driver, unfused myoblasts distribute themselves on the inner face of the epidermis in a segmentally repeated pattern of bands at stage 12. Myoblasts are also attracted toward a restricted region of the visceral mesoderm surrounding the midgut (parasegment 8) where wg is normally expressed. Double staining such embryos with anti-myosin and anti-Wg shows that myoblasts move toward the sources of Duf. They move internally toward ps 8 of the visceral mesoderm and outward to the exterior of the embryo close to the ectodermal bands of duf expression driven by Wg-GAL4. They leave empty spaces along the anteroposterior axis on either side of the Wg domain (Rúiz-Gomez, 2000).

It is concluded that Duf protein acts to attract myoblasts at a distance. During normal development, this leads to the aggregation of fusion-competent myoblasts on the founders with which they will fuse. In the absence of Duf, myoblasts fail to aggregate and fusion is blocked (Rúiz-Gomez, 2000).

Since duf expression is characteristic of muscle founders, it could be that it is required for the proper specification of these cells and/or to allow them to complete myogenic differentiation. However, all the evidence indicates that in the absence of duf each founder is specified normally, expresses the appropriate set of genes, and completes myogenesis to form a properly innervated, mononucleate muscle. The crucial feature that is lacking is myoblast fusion itself (Rúiz-Gomez, 2000).

Fusion is a multistep process that depends on mature fusion-competent myoblasts recognizing an appropriate target for fusion (a founder or a myotube). This recognition step is followed by adhesion and alignment of the cells along their long axes. Once cells are closely apposed, plasma membranes start fusion events at several places, allowing communication between them and culminating in the incorporation of the myoblast into the developing syncytium. Unfused myoblasts that are just about to fuse are round cells with a single process addressed toward the founder or the myotube. Several myoblasts may be in contact with the same founder or myotube at the same time, thus leading to the formation of myoblast aggregates. Such myoblast aggregates are short lived and quickly resolved by fusion and formation of myotubes. How the small aggregates that will contribute to a myotube are formed has not been clear. A prominent feature of the phenotype in some mutations that block fusion is the appearance of obvious myoblast clusters, and it has been suggested that such embryos are blocked at the recognition step. These clusters do not form in the absence of duf. Furthermore, in duf-deficient embryos the myoblasts not only fail to cluster but are located at different levels in muscle-forming mesoderm. In wild-type embryos, founder myoblasts arise in close contact with the ectoderm, while fusion-competent myoblasts are more internal, and this arrangement persists in the absence of duf. In wild-type conditions, fusion-competent myoblasts put out filopodia that are mainly oriented toward founders, whereas in deficiencies for duf, fusion-competent myoblasts still produce filopodia but do so without any preferred orientation. Thus, the data show that the aggregation of myoblasts on founder cells prior to the formation of myotubes is an active process in which founders produce an attractant and myoblasts move toward it. This is strikingly different from the alternative, namely that myoblast clusters form by a random process of collision and recognition (Rúiz-Gomez, 2000).

Duf is a transmembrane protein whose extracellular domain contains five Ig-like repeats and is highly homologous to the Drosophila protein Roughest (Rst). Duf is sufficient to rescue the fusion defect in Df(1)w^67k30 embryos when it is reintroduced into the mesoderm, allowing such embryos to form a relatively normal pattern of syncytial muscles. Furthermore Duf acts as a signal that attracts fusion-competent myoblasts when it is expressed at ectopic sites in the mesoderm or in the ectoderm. Although the behavior of myoblasts in wild-type embryos shows that they move toward sources Duf, it is not possible to distinguish between two alternative ways in which Duf might act. Either the external part of the molecule is cleaved and diffuses away from the source or, alternatively, it remains on the founder cell membrane and is detected by the random exploration of myoblast filopodia. However, there are a number of reasons why the view that Duf acts at the membrane of the founder is favored. (1) There is the question of distance: the furthest from their normal location that myoblasts are seen aggregating is on the ventral midline when Duf is expressed ectopically with the Wg-GAL4 driver. Although this is many cell diameters from their normal location, it is well within reach of cell processes such as filopodia and cytonemes. (2) There is the question of how a diffusible signal would act in normal development over the relatively short distances between alternative sources, namely neighboring founders. It is hard to envisage a diffusible molecule acting as an attractant without interference between adjacent sources leading to locally high concentrations between founders and consequent misrouting of aggregating myoblasts. (3) There is the question of the role of Duf in the process of fusion itself. There is complete absence of fusion in Df(1)w^67k30 embryos and this is not easily explained by the simple model that Duf merely acts to attract myoblasts to founders. If this were the case, it would be expected that random contacts between founders and fusion-competent cells would lead occasionally and perhaps quite commonly to fusion events in such embryos. The absence of such events in Df(1)w^67k30 embryos and the fact that fusion is restored if Duf is reintroduced into the mesoderm suggests that Duf acts both as an attractant for myoblasts and as an essential component of the fusion process that follows (Rúiz-Gomez, 2000).

As a putative cell adhesion protein, Duf may be required in the process of fusion to ensure a close adhesion between myoblasts and founders without which coalescence of membranes cannot occur. However, the possibility cannot be excluded that the specialized intracellular domain of the protein, which is highly divergent from that of Rst, allows signaling to occur between myoblast and founder and that it is this signaling that is essential for fusion to occur. In this view, the recognition event mediated by the binding of Duf to its (as yet unknown) partner on the myoblast would trigger the local cascade of events that allows fusion to proceed. In any event, although Duf is necessary for fusion, it is not sufficient. Duf under the control of Twi-GAL4 is present in both myoblasts and founders, but fusion still occurs only between founders and myoblasts and never among myoblasts themselves, even though these cells now express Duf and its partner. This suggests that the asymmetry of the fusion process depends not only on the selective expression of Duf in the founders but also on other specialized characteristics of founders that are not present in myoblasts generally. It may be that the intracellular domain of Duf interacts with components uniquely present in founders to initiate the fusion process. It should be noted, however, that once initiated, at an ultrastructural level at least, the events that accompany fusion are strikingly symmetrical between myoblast and founder cell (Rúiz-Gomez, 2000).

While the experiments reported here show that Duf acts nonspecifically to attract fusion-competent myoblasts to sites where it is expressed, the formation of myotubes in the Drosophila embryo is a highly regulated process that results in the fusion of specific numbers of cells to form muscles of different sizes. This might suggest that myoblasts are themselves specified to fuse uniquely with particular founders. However, there is no evidence for that kind of specificity in the myogenic pathway in Drosophila. Experimental manipulations of myogenesis in adult flies show that myoblasts are capable of fusing with any muscle that they encounter. In the embryo, muscles can be duplicated experimentally, and under these conditions, the number of cells contributing to each of the two fibers is the same as the number contributing to a single myotube in a normal embryo. Clearly, therefore, myoblasts that would not normally contribute to a developing fiber can be recruited to it if conditions change. If myoblasts are not set aside to fuse with a given myotube, what determines the ultimate size of a developing fiber? Although Duf expression could contribute to the control of size by regulating the length of time that any given founder remains an attractive target for fusion-competent myoblasts, it is certainly not the only determinant. Duf expression appears to decline early in small muscles and later in larger muscles, but in experiments, Duf expression in a subset of muscles (Ap-GAL4) or uniformly throughout the muscles (Twi-GAL4) does not cause marked aberrations in the size of the muscles that form. Once again, this suggests that there are special attributes to founder cells that contribute to the process of fusion and, in this case, set the number of fusions that are permitted for a particular myotube (Rúiz-Gomez, 2000).

Drosophila provides a model system with which to explore the essential features of myotube formation and patterning. So, for example, it is suspect that seeding events may be the key to understanding the recruitment of myoblasts to form myotubes in vertebrates as well as in flies. In both vertebrates and flies, the initial step in the formation of myotubes is aggregation and recognition. In muscle-forming mesoderm an attractant for myoblasts is selectively expressed by founder myoblasts. Thus, founders actively attract to themselves an aggregate of myoblasts with which the founders will fuse to form a myotube. The immediate task in the embryo of the fly is to identify the ligand of Duf that is expressed by fusion-competent cells and link it and Duf to the pathway of myoblast fusion. It may well be that this will reveal conserved elements in the two kinds of organisms that will clarify the formation of myotubes and the spatial organization of muscle development (Rúiz-Gomez, 2000).

GENE STRUCTURE

cDNA clone length - 3521

Bases in 5' UTR - 263

Exons - 7

Bases in 3' UTR - 378

PROTEIN STRUCTURE

Amino Acids - 959

Structural Domains

The sequence of the 3520 bp long duf cDNA reveals a single long open reading frame. A hydrophobicity plot of the putative protein reveals two prominent hydrophobic peaks. The PSORT program predicts that the first corresponds to a signal peptide, cleavable after residue 31, and that the second, including amino acids 580-596, corresponds to a transmembrane domain. The predicted mature protein thus has an extracellular domain of ~550 amino acids, a single transmembrane spanning region, and an intracellular domain of ~165 amino acids. The extracellular domain contains five immunoglobulin (Ig)-like repeats and shows extensive sequence similarity with the Drosophila protein Rst (63% identity). In contrast, BLAST searches with the intracellular domain fail to detect any similarity with any sequence in the databases. The fact that rst and duf are in the same region of DNA and encode closely related proteins probably indicates that the two are derived from a single ancestral gene by duplication. However, the expression patterns of the two genes and their postulated functions are very different (Rúiz-Gomez, 2000).

date revised: 24 December 2004

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.