Hibris (Hbs) is a transmembrane immunoglobulin-like protein that shows extensive homology to Drosophila Sticks and stones (Sns) and human kidney protein Nephrin. Hbs is expressed in embryonic visceral, somatic and pharyngeal mesoderm among other tissues. In the somatic mesoderm, Hbs is restricted to fusion competent myoblasts and is regulated by Notch and Ras signaling pathways. Embryos that lack or overexpress hbs show a partial block of myoblast fusion, followed by abnormal muscle morphogenesis. Abnormalities in visceral mesoderm are also observed. In vivo mapping of functional domains suggests that the intracellular domain mediates Hbs activity. Hbs and its paralog, Sns, co-localize at the cell membrane of fusion-competent myoblasts. The two proteins act antagonistically: loss of sns dominantly suppresses the hbs myoblast fusion and visceral mesoderm phenotypes, and enhances Hbs overexpression phenotypes. Data from a P-homed enhancer reporter into hbs and co-localization studies with Sns suggest that hbs is not continuously expressed in all fusion-competent myoblasts during the fusion process. We propose that the temporal pattern of hbs expression within fusion-competent myoblasts may reflect previously undescribed functional differences within this myoblast population.
INTRODUCTION
In Drosophila, each larval body wall muscle acquires a distinct set of morphological traits during embryogenesis that provides the unique identity of the muscle within the organism. These traits include the size, which is governed by the number of fusion events, the shape, which is controlled by attachment site choice and cytoskeletal dynamics, and the innervation pattern, which is guided by interactions between recognition proteins on the muscle and motoneuron surface.
Part of the diversity in muscle identity stems from specification of two sets of myoblasts. Founder myoblasts contain the information required for morphogenesis of a given muscle, while fusion-competent myoblasts are believed to be naïve myoblasts that fuse progressively to a founder cell and become entrained to a particular muscle program (Bate, 1990; Dohrmann et al., 1990). This notion is supported by analysis of fusion mutants such as myoblast city (Rushton et al., 1995; Erickson et al., 1997) in which founder cells execute aspects of their individual morphogenetic programs (e.g. attract the appropriate motoneuron), whereas fusion competent cells apoptose, suggesting that no specific developmental instructions have been received. A combination of signals from the ectoderm and mesoderm ultimately lead to segregation of both myoblast types (Frasch, 1999; Baylies and Michelson, 2001). The intersection of Wingless and Decapentaplegic activities defines a region in the dorsal mesoderm that is competent to respond to inductive signals mediated by the Ras signaling pathway. The localized activation of Ras in this region instructs clusters of myoblasts to adopt the founder cell fate (Carmena et al., 1998a; Buff et al., 1998; Wilson, 1999). This fate is restricted to a single cell within each cluster, the muscle progenitor, by a process of lateral inhibition mediated by the Notch and Delta signaling pathway (Corbin et al., 1991; Bate et al., 1993; Carmena et al., 1995). Cells that receive the Delta signal from a neighboring muscle progenitor segregate as fusion-competent cells. Muscle progenitors then divide asymmetrically to form founder cells (Ruiz-Gomez and Bate, 1997; Carmena et al., 1998b). The specific combination of signals that a given founder cell receives results in activation of a particular set of transcription factors such as Krüppel or Even-skipped (Ruiz-Gómez et al., 1997; Carmena et al., 1998a; Halfon et al., 2000), which, in turn, regulate the attributes characteristic for each muscle. Fusion-competent cells, which result from the activation of Notch, start expressing proteins such as Sticks and Stones (Sns) (Bour et al., 2000) that are required for their subsequent differentiation. However, we know little about how these transcription factors determine identity and how fusion-competent cells contribute to this process. Even though it is generally assumed that fusion-competent cells are naïve and contribute only mass to the muscle, it is formally possible that they also contribute to muscle identity in an active manner.
In Drosophila, the general framework for understanding muscle cell fusion was provided by Doberstein and colleagues (Doberstein et al., 1997) who subdivided the process into steps: cell-cell recognition, adhesion, alignment of membranes and membrane fusion. Genetic studies in Drosophila have also provided fusion mutants (Paululat et al., 1999). Essential loci for myoblast fusion include the transcriptional regulator Mef2 (Bour et al., 1995; Ranganayakulu et al., 1995), the membrane bound protein Rolling stones (Paululat et al., 1997), cytoplasmic proteins such as Blown fuse (Doberstein et al., 1997), and components of the Rac1 signaling pathway such as Myoblast city (Erickson et al., 1997; Nolan et al., 1998) and Drac1 (Luo et al., 1994). Interestingly, overexpression of weak gain-of-function Notch constructs throughout the mesoderm can completely block fusion without interfering with earlier roles for Notch (Fuerstenberg and Giniger, 1998) suggesting that Notch is also involved in muscle morphogenesis. Another family of proteins that plays crucial roles in myoblast fusion in Drosophila is the immunoglobulin (Ig) superfamily (Taylor, 2000; Frasch and Leptin, 2000; Baylies and Michelson, 2001). These Ig-containing proteins include Kirre (formerly Dumbfounded) (Ruiz-Gómez et al., 2000) and Sns (Bour et al., 2000). Although the described mutant phenotypes for kirre and sns are the same – a complete fusion block – their expression patterns in the somatic mesoderm are strikingly different. kirre is expressed exclusively in founder cells, while Sns is expressed exclusively in fusion-competent cells. Moreover, kirre is expressed during the entire fusion process and acts as an attractant for myoblasts, indicating that founder cells actively recruit fusion-competent cells until the final muscle size is achieved (Ruiz-Gómez et al., 2000). Sns, however, is a general marker for fusion-competent cells and, consistent with the segregation of these myoblasts, its expression depends on Notch signaling (Bour et al., 2000). Both Kirre and Sns are involved in the initial steps of myoblast cell-cell recognition as free myoblasts in embryos mutant for either gene do not cluster around founder cell, suggesting that there is no recognition between both types of myoblasts.
To study muscle identity and morphogenesis, we have devised a molecular screen to identify mRNAs that are expressed in founder or fusion-competent myoblasts (R. D. A. and M. K. B., unpublished). From this screen we isolated hibris (hbs), a novel gene that we have characterized molecular and genetically. We show that hbs is a single pass transmembrane protein belonging to the Ig superfamily and is involved in muscle morphogenesis. Phenotypic analysis of hbs and genetic interactions with sns suggests that Hbs is a dose-dependent regulator of myoblast fusion. Hbs is also a novel target of Notch and Ras signaling. In addition, we present evidence to suggest that fusion-competent cells may not be a homogeneous population over the course of muscle development.
MATERIALS AND METHODS
Drosophila genetics
Oregon R was used as the wild-type strain. Df(1)w67k30 was obtained from M. Ruiz-Gomez (Ruiz-Gomez et al., 2000); snsXB3 from S. Abmayr (Bour et al., 2000); N55e11 (Kidd et al., 1983) and UAS- Nintra from T. Lieber; and UAS-rasN17, UAS-rasV12 and twi-GAL4; Dmef2-GAL4 from A. Michelson. All other fly stocks were obtained from Bloomington, Umea or Szeged Stock Centers. Mobilization of P elements was achieved using the stable source of transposase Δ2-3 (Robertson et al., 1988). P homing was as described elsewhere (Taillebourg and Dura, 1999). Several genetic strategies were employed to mutate hbs (R. D. A. and M. K. B., unpublished). Briefly, two deficiencies, Df(2R)X28 and Df(2R)X7, were generated by reversion of the EP(2)2590 associated white+ minigene after γ-ray mutagenesis. Molecular mapping of the proximal ends of these deficiencies showed that hbs was deleted in Df(2R)X28 but not in Df(2R)X7 (Fig. 1A). We then performed EMS mutagenesis and recovered putative hbs mutations that mapped inside Df(2R)X28 but outside Df(2R)X7. We isolated seven hbs alleles, all showing a rough eye phenotype. For phenotypic studies, relevant stocks were balanced with CyO P[w+wgen11lacZ] or FM7 P[w+ ftz-lacZ].
Differential display screen for genes affecting muscle development
The details of this screen will be published elsewhere (R. D. A. and M. K. B., unpublished). Briefly, we generated two sets of embryos: one expressing an activated form of Notch (UAS-Nintra), which leads to suppression of founder cell fate (Landgraf et al., 1999), another expressing an activated form of Ras (UAS-ras1V12), which leads to enhancement of founder cell fate (Carmena et al., 1998b), under the control of twi-Gal4. In addition, these overexpression experiments were carried out in a Toll10b genetic background to reduce tissue complexity. This mutation leads to a constitutively active Toll receptor, which results in twist activation throughout the embryo (Leptin et al., 1992). Ectopic expression of Twist converts the majority of cells into mesoderm and subsequently somatic muscle. RNA from embryo collections aged 5 to 9 hours were processed for differential display using the Delta Differential Display kit (Clontech). Forty-one differentially displayed bands were selected for further study.
Molecular biology
An embryonic cDNA library (Kidd, 1992) was screened with probe P1/T9(1), which was recovered from the differential display gel (Fig. 1A), yielding partial cDNAs A (2.2 kb) and B (2.1 kb). We then screened a 0-22 hour cDNA library (Rubin et al., 2000) with cDNA A and detected 12 clones. Restriction analysis showed that five (D2, C6, C9, C1 and D1) were identical in size (4.7 kb). cDNA D1 was sequenced and translated into a predicted protein (Fig. 1D). The beginning of the transcription unit was confirmed using the FirstChoice RLM-RACE kit (Cat. No. 1700, Ambion). RNA was extracted from appropriate samples using Tri-Reagent (Cat. No. T9424, Sigma) and poly A+ RNA purified with Oligotex mRNA spin-columns from Qiagen (Cat. No. 70042). Northern blots were processed according to the manufacturer recommendations for use of DIG-labeled probes (Roche Molecular Biochemicals). DNAs were sequenced at the Cornell University DNA sequencing facility (Ithaca, NY). Protein sequence analysis was carried out using the ExPASy (Expert Protein Analysis System) proteomics server from the Swiss Institute of Bioinformatics (Appel et al., 1994).
Germline transformation and constructs
Hbs protein was ectopically expressed using the UAS/GAL4 system (Brand and Perrimon, 1993). The full-length hbs construct (UAS-hbsFL) was made by subcloning cDNA D1 into pUAST. Two additional constructs were made: UAS-hbsΔICD, which lacked the last 143 amino acids of the intracellular portion of the protein, and UAS-hbsΔECD, which lacked 938 amino acids of the extracellular domain but retained the transmembrane region (see Fig. 7Q). UAS-hbsΔECD was created by deleting a 2.8 kb long EagI/EcoRV fragment from cDNA D1 and blunt end religation. The resulting plasmid, retaining the wild-type 5′UTR and signal peptide, was sequenced to verify the ORF. UAS-hbsΔICD was made by purifying a 4 kb EcoRI/EclHKI fragment from cDNA D1 that was blunted and ligated to a pBluescript plasmid containing the 3′ UTR region from D1. The 3′ UTR region from cDNA D1 was subcloned into pBluescript by using specific primers containing BamHI and XhoI restriction sites. All constructs were recloned into pUAST and injected into yw embryos according to published procedures (Spradling and Rubin, 1982). For ectopic Hbs expression, flies containing the indicated UAS-hbs transgene were mated to GAL4 lines at 25°C or 29°C. For P homing, P[w+]36.1 was created using a PCR-amplified 4 kb fragment upstream the transcription unit (primers CAAGGCATTTCGCAGAAC and GTGTAGGCGGTTTCGTAGTC) cloned into a blunted NotI site in pETWnucLacZ. P element insertions in hbs were screened by PCR using primers specific for the transposable element and genomic primers.
Immunohistochemistry
The Hbs antibody was raised against a synthetic 15-mer peptide (1151-1165 in Hbs protein) supplied by Genossys. The peptide was made antigenic by coupling to the carrier tuberculin PPD (Statens Serum Institut, Denmark) using glutaraldehyde. This antigenic Hbs peptide was injected into rabbits at Pocono Rabbit Farm and Laboratory (Canandensis, PA), according to standard peptide-antigen protocols. Immunocytochemistry in embryos (Rushton et al., 1995) was performed using anti-Heavy Chain Myosin (1:1000; Kiehart and Feghali, 1986), anti-β-galactosidase (1:2000; Promega), anti-Fasciclin III (1:100; Developmental Studies Hybridoma Bank, University of Iowa) (Patel et al., 1987) and anti-Krüppel (1:2000) (provided by J. Reinitz). Hbs (1:3000) and Sns (1:100) (Bour et al., 2000) antibodies were used preabsorbed against fixed wild-type embryos in combination with the TSA system from NEN (Cat. No. NEL700A) and a hot fixation protocol (Rothwell and Sullivan, 2000). Biotinylated secondary antibodies were used in combination with Vector Elite ABC kit (Vector Laboratories, CA). Immunofluorescent signals in co-localization studies were analyzed using a Leica TCS-NT confocal microscope. hbs RNA localization was detected using digoxigenin-labeled RNA probes, as described (O’Neil and Bier, 1994).
RESULTS
Identification of hbs
To better understand muscle diversity and morphogenesis, we devised a differential display screen to identify genes specifically expressed in founder versus fusion-competent myoblasts (R. D. A. and M. K. B., unpublished; Materials and Methods). In short, we made use of the observation that muscle progenitor specification depends on Notch and Ras signaling to repress or enhance muscle progenitor fate respectively (Carmena et al., 1998b; Landgraf et al., 1999). We overexpressed activated forms of Notch and Ras in Toll10b mutant embryos (Leptin et al., 1992) to reduce tissue complexity. We have confirmed that Toll10b mutant embryos differentiate almost exclusively as somatic mesoderm.
A differentially displayed band, P1/T9(1) (Fig. 1A), was chosen for further study because it was strongly upregulated under the activated Notch conditions. Corresponding cDNAs were isolated and the gene named hbs (Materials and Methods). We have confirmed hbs dependence on Notch by Northern blot analysis. Fig. 1B compares hbs expression in Toll10b mutant embryos and Toll10b mutant embryos that expressed activated Ras or activated Notch. Our quantification of these Northern blot signals suggested that hbs expression was upregulated at least twofold upon Notch activation, and repressed at least fivefold upon Ras activation.
In a developmental Northern blot, we detected a hbs transcript of 6.2 kb with high levels of expression at mid-embryogenesis (Fig. 1C). We sequenced several cDNAs that had a 1235 amino acid ORF. hbs encoded a single pass transmembrane protein belonging to the Ig superfamily and contained nine Ig-C2 type repeats and a Fibronectin type III domain adjacent to its transmembrane region (Fig. 1D). A potential signal peptide sequence that is characteristic of membrane-bound proteins was present in the N terminus. BLAST database searches (Altschul et al., 1990) with Hbs revealed Nephrin as the human ortholog (29% identity, 44% similarity). Mutations in Nephrin have been linked to the Congenital Nephrotic Syndrome of the Finnish type (Kestilä et al., 1998). Furthermore, there was a Drosophila paralog, the protein Sns, with 48% identity and 63% similarity to Hbs (compare with Fig. 4 in Bour et al. (Bour et al., 2000)). The Hbs cytoplasmic domain, however, was shorter than its equivalent region in Sns (166 versus 376 amino acids, respectively). The Hbs extracellular region contained 13 NXS/T consensus triplets for N-glycosylation, eight of which were conserved in Sns. The cytoplasmic domain of the protein showed potential target sites for cAMP and cGMP-dependent protein kinases, protein kinase C (PKC), and casein kinase II (CKII). These phosphorylation sites were not conserved in Sns, except for a CKII site and a PKC site in two regions of strong conservation (Fig. 1E). In addition, Bour et al., reported conservation of tyrosine residues between Sns and Nephrin in the intracellular domain (Bour et al., 2000), which are also present in Hbs. The Hbs cytoplasmic domain also contained an eight amino acid direct repeat, but no described conserved domains. Additional searches against the conceptual translation of predicted genes in the Drosophila genome with the intracellular region of Hbs did not find any significant match except with Sns.
Pattern of hbs expression
hbs expression was analyzed by in situ hybridization in embryos (Fig. 2A-F) and in imaginal discs (not shown). We first detected hbs transcripts at stage 8 (Campos-Ortega and Hartenstein, 1985) in precursors of the amnioserosa and in the mesectoderm, where it was maintained in progeny of these midline cells throughout embryogenesis. By stage 10, hbs was expressed in the visceral mesoderm in a pattern reminiscent of bagpipe. These cells maintained hbs expression while forming the visceral mesoderm surrounding the gut (Fig. 2B,C). By late stage 11, hbs was expressed in somatic mesoderm (Fig. 2D) where its expression peaked between stage 12 and early stage 13, and subsequently declined, disappearing by stage 14. By stage 12 there was expression in the precursors of the heart, where hbs was maintained throughout the remainder of embryogenesis. Late hbs expression (Fig. 2F) was detected in the epidermis at putative muscle attachment sites, hindgut and pharyngeal muscles. hbs was also expressed in imaginal tissues, most notably in the eye-antennal, wing, leg and haltere discs (not shown).
Immunohistochemical staining reiterated the same embryonic expression pattern of the Hbs protein. Notably, Hbs localization at the cell membrane is consistent with the protein structure predicted by the sequence data (Fig. 2K). Hbs was detected weakly in the somatic mesoderm by late stage 11. This expression increased around stage 12 and 13 (Fig. 2G-I), yet declined thereafter, and mature muscles were completely devoid of Hbs. Hbs was expressed at specific points in the myoblast membrane, as described for its paralog, Sns (Bour et al., 2000). This localized expression was detected also in the hindgut, at putative epidermal attachment sites and in the midline (Fig. 2J-N).
Transgenic flies carrying a 4 kb fragment of hbs genomic sequence fused to a lacZ reporter (Materials and Methods) reproduced aspects of hbs expression and provided an additional tool with which to study hbs. Embryos containing this reporter inserted in the hbs locus (P[w+]36.1; Fig. 1A) showed expression in visceral and somatic mesoderm, mesectoderm, pharyngeal muscles and hindgut in a pattern similar to that found with the anti-Hbs antibody. To reinforce further that P[w+]36.1 recapitulated hbs expression in the mesoderm, we simultaneously detected β-galactosidase expression and hbs transcription in P[w+]36.1 embryos, confirming there was overlap between both signals (not shown). Because of the high degree of conservation between Hbs and the kidney protein Nephrin, we analyzed hbs expression in Malpighian tubules in our reporter P[w+]36.1. Interestingly, we detected weak expression at stage 12/13 in a subset of tubule cells that, based on their morphology, appear to be stellate cells (not shown; H. Skaer, personal communication).
hbs is expressed in fusion-competent myoblasts
Fig. 1B establishes that hbs is regulated by Notch and Ras in a Toll10b mutant background. We have confirmed this regulation in vivo in wild-type embryos. We studied hbs expression in Notch and Ras loss-of-function embryos and embryos overexpressing activated forms of Notch and Ras in the mesoderm (Fig. 3). A dominant negative Ras construct activated hbs expression in the somatic mesoderm. Zygotic null Notch embryos showed lower hbs transcription. Conversely, an activated form of Notch upregulated hbs in the mesoderm, while an activated form of Ras almost completely inhibited hbs expression. These results argued that, upon stimulation, Notch activated hbs, while Ras acted as a negative signal, and predicted that hbs expression in the somatic mesoderm would be restricted to fusion-competent cells (Notch dependent) and excluded from founder cells (Ras dependent). We do not know whether this regulation is direct, that is, Notch or Ras effectors act directly on the hbs promoter, or indirect, that is, Notch/Ras converts cell fate, which in turn would lead to hbs upregulation/downregulation by some other effector.
As these results suggested that hbs was excluded from founder cells and was present in fusion-competent cells, we performed several additional experiments to confirm this possibility. Co-localization experiments using either P[w+]36.1 that reiterated Hbs expression in the mesoderm or Hbs antibody and various founder cell markers revealed that Hbs was absent from founder cells. These experiments also suggested that Hbs was not expressed in all fusion-competent cells during myoblast fusion (Fig. 4M-O; data not shown). Therefore, we asked whether Hbs and Sns, a protein known to be expressed exclusively in fusion-competent cells (Bour et al., 2000), co-localize in fusion-competent cells. Wild-type embryos double labeled with antisera against Sns and Hbs showed several examples of co-localization of these proteins. During stage 11, when muscle progenitors are being specified and the fusion process has not begun, Hbs expression was initially more widespread in somatic and visceral mesoderm than that of Sns (Fig. 4A-C). Subsequently, as the process of muscle fusion begins, Hbs expression was less widespread than Sns. We note that at this stage, although we detected overlap in Hbs and Sns expression, we found cases of Sns and Hbs-specific expression (Fig. 4D-F). At later stages of the fusion process, we no longer detected Hbs protein in the remaining fusion-competent cells, in contrast to the continued presence of Sns (Fig. 4G-I). We also found co-localization of both Hbs and Sns in the visceral mesoderm (Fig. 4J-L). We conclude that Hbs expression initially precedes Sns, but then largely overlaps with Sns, suggesting that Hbs could modulate the fusion-promoting properties of Sns in fusion-competent cells during the course of myoblast fusion.
Isolation of hbs-specific mutations
To assay the effects of loss of hbs function, we generated both new deficiencies around region 51D7 and EMS-induced mutations (Fig. 1A; Materials and Methods). For two of the EMS-induced mutations recovered, 2593 and 459 (hereafter hbs2593 and hbs459), we identified nucleotide changes in the hbs transcription unit, therefore establishing that they were hbs alleles. hbs2593 mutation was a T to A transversion in the second nucleotide of the intron between exons 9 and 10 (Fig. 5A). The mutation leads to an aberrantly spliced transcript that retained the whole intron and would result in a prematurely truncated protein. hbs459 was an A to T transversion in exon 7 that generated a stop codon (Fig. 5B).
Immunohistochemistry provided further confirmation that these mutations led to loss of hbs function. In these experiments, hbs2593 embryos showed a dramatic decrease in Hbs (Fig. 5C,E), while hbs459 homozygous embryos showed no detectable protein (Fig. 5D,F). Allele hbs361 showed protein expression at levels indistinguishable from wild type (not shown) but behaved at least as strongly as hbs459 in our phenotypic assays.
hbs mutant embryos show a partial block in myoblast fusion and visceral mesoderm defects
Although hbs mutant flies survive to adult stages in normal numbers and show visible phenotypes such as rough eyes, reduced viability and semisterility (R. D. A. and M. K. B., unpublished), we also found clear abnormalities in muscle development in mutant embryos. Embryos heterozygous for hbs459/Df(2R)X28, and different hbs transallelic combinations, were probed with anti Myosin antibody and compared with their wild-type counterparts. In embryos lacking hbs function (Fig. 6), we found that the muscle pattern was specified, although some muscles were occasionally missing or showed abnormal morphologies (e.g. looked thinner than normal or had fewer nuclei than normal). Immunocytochemistry with antibodies directed against founder cell markers such as Krüppel, Even-skipped and Runt revealed a normal pattern (not shown). These embryos, however, reproducibly showed a partial fusion block: an increased number of free myoblasts were present when compared with wild-type embryos. We detected unfused myoblasts scattered around the muscles, surrounding the gut, heart and CNS (Fig. 6C,D). These free myoblasts showed extended processes, indicative of their competence to scan for founder cells. In these experiments, we studied stage 16 embryos to ensure that free myoblasts had the chance to fuse and that in this mutant background they failed to do so. Note, however, that by this developmental stage many unfused myoblasts might have undergone apoptosis, been phagocytosed and lost Myosin expression, decreasing the apparent number of unfused myoblasts (Bour et al., 1995). Loss of hbs function also interfered with the normal gut development as shown both by an enlargement of the first chamber (Fig. 6E-H) as well as by loss of visceral muscle progenitors at stage 12 (Fig. 6I,J).
hbs overexpression leads to a block in myoblast fusion and defects in visceral mesoderm
We used the GAL4/UAS system (Brand and Perrimon, 1993) to assess the effect of hbs overexpression. When we expressed UAS-hbsFL throughout the embryonic mesoderm driven by twist-GAL4; DMef2-GAL4, we detected defects both in the somatic and visceral mesoderm similar to loss of hbs function. In the somatic mesoderm, we found a partial fusion block with free myoblasts scattered at several locations (Fig. 7E,F). In addition, these embryos normally show a greater degree of muscle loss than hbs homozygous mutant embryos (Fig. 7A,B). Defects in the visceral mesoderm, as revealed by anti-Fasciclin III staining (Fig. 7M,N), were also found. Because delivering extra Hbs to both founder and fusion-competent myoblasts led to muscle losses, we tested whether all founder cells were correctly specified in these embryos. Antibody staining with the founder cell markers Krüppel, Even-skipped and Runt indicated that founder cells were correctly specified in embryos overexpressing Hbs (not shown). However, at later stages, we detected aberrant morphologies in the growing muscles (i.e. odd shapes; Fig. 7I,J), suggesting that prolonged Hbs expression was interfering with muscle morphogenesis rather than the initial specification.
To dissect which parts of the protein are important for Hbs function, two additional constructs were tested. The constructs either deleted the extracellular (UAS-hbsΔECD) or the intracellular (UAS-hbsΔICD) region of the protein, but maintained the transmembrane domain such that the truncated proteins remain membrane anchored (Fig. 7Q). Overexpression of UAS-hbsΔECD in the mesoderm resulted in detectable phenotypes, indicating that the intracellular domain of Hbs alone can interfere with muscle development. Specifically, we detected a greater degree of fusion block and muscle loss as well as defects in the late pattern of founder cell markers compared with overexpression of the full-length Hbs (Fig. 7C,G,K). This construct also caused greater defects in visceral mesoderm, leading to a greater reduction of visceral muscle progenitors (Fig. 7O). Altogether, these results suggested that the cytoplasmic domain of Hbs is required for hbs function in the mesoderm. By contrast, overexpression of UAS-hbsΔICD did not affect somatic muscle development appreciably but blocked visceral mesoderm development (Fig. 7D,H,L,P).
hbs interacts with sns in the somatic and visceral with mesoderm
As Sns, Kirre and Roughest are all Ig-C2-containing proteins implicated in myoblast fusion (Bour et al., 2000; Ruiz-Gómez et al., 2000; Strunkelnberg et al., 2001), we investigated their possible functional relationship with hbs by testing for genetic interactions. For this analysis we used first Df(1)w67k30, which removes both kirre and roughest, both of which are suggested to be partially redundant in the fusion process (Strunkelnberg et al., 2001). We found that hbs and Df(1)w67k30 did not interact genetically, either in loss- or gain-of-function hbs backgrounds (see Fig. 9I; data not shown). In addition, overexpression of hbs did not rescue any aspect of the Df(1)w67k30 phenotype (not shown) (Ruiz-Gómez et al., 2000).
As described above, Hbs and Sns co-localized at the cell membrane at discrete points, which would be consistent with both proteins working together in the fusion process. Fig. 8, Fig. 9 show the result of manipulating the dose of sns and hbs. Stage 16 hbs homozygous mutant embryos characteristically showed groups of free myoblasts throughout the embryo (Fig. 8D,G), and a gut phenotype revealed early by anti-Fasciclin III (Fig. 8F,I) or late by anti-Myosin staining (Fig. 8E,H). hbs-null embryos in which the sns dose was halved showed dominant suppression of both the partial fusion block and gut phenotype (Fig. 8J-L). When we analyzed hbs and sns double mutant embryos, we did not detect a noticeable change in the sns phenotype. Thus, sns and hbs appeared to act antagonistically.
Consistent with the hypothesis that hbs and sns act antagonistically rather than cooperatively in myoblast fusion, sns mutations dominantly enhance hbs overexpression phenotypes (Fig. 9). In the somatic muscles, there was an increase in the number of free myoblasts (Fig. 9C,E). This interaction also led to a greater loss in visceral muscle progenitors, as measured by Fasciclin III expression (Fig. 9D,F). To quantify these observations, we used the number of gaps in Fasciclin III expression in stage 12 embryos as a measurement of genetic interaction between hbs and sns. The result of this quantification was consistent with our qualitative analysis (Fig. 9I). We found that the dominant increase in Fasciclin III expression gaps when sns dose was lowered was statistically significant at the P<0.0001 level, indicating that a synergistic, rather than an additive, effect between hbs overexpression and heterozygosity for sns accounts for the increase in Fasciclin III expression gaps observed. As an additional test to determine the relationship between hbs and sns, we attempted to rescue the sns loss-of-function phenotype by overexpression of hbs. In this experiment, Hbs did not rescue the sns loss-of-function phenotype in the somatic muscles (not shown) in agreement with the genetic data presented above. Altogether, the results of these experiments suggested that hbs antagonizes sns function during mesoderm development, but cannot differentiate whether sns is downstream hbs or both genes are working in parallel pathways.
DISCUSSION
Hbs is an Ig domain protein involved in signal transduction
Hbs, Sns and Nephrin, in addition to their homologs in rat, mouse and C. elegans, establish a new subgroup in the Ig superfamily. These proteins show a characteristic run of 7-9 Ig-C2 domains in the extracellular domain followed by a single Fibronectin type III domain closest to the transmembrane region and no known motifs in their cytoplasmic tail. In agreement with the functions of the Ig superfamily, Hbs, Sns and Nephrin have been implicated in cell adhesion and/or cell signaling.
Clues to how Hbs functions during myogenesis have been revealed in our overexpression experiments. These data show that hbs overexpression gives a similar phenotype to hbs loss and suggest that either Hbs transduces a signal or our constructs behave as dominant negatives. In support of Hbs acting as a signal transducer, we find that overexpression of UAS-hbsΔECD resulted in a partial block in somatic muscle development, indicating that the cytoplasmic domain of Hbs mediates the activity of the protein. A similar dependence on the cytoplasmic domain of Notch in the embryonic nervous system (Lieber et al., 1993) or Echinoid in the eye imaginal discs (Bai et al., 2001) has been used as support for these proteins acting as signal transducers. Alternatively, our overexpression data with both the full-length and intracellular domain constructs could be interpreted as Hbs exerting a dominant negative effect through the titration of another crucial component, such as another membrane protein or cytoskeletal component. This key protein could be a target of Hbs without Hbs actually transducing a signal in the classical sense. However, as the behavior of mutant combinations of sns and hbs is different – Hbs loss of function is suppressed by sns mutations whereas hbs overexpression is enhanced by sns mutations – we favor the possibility that Hbs transduces a signal rather than behaves as a dominant negative in the overexpression studies (Fig. 9). We would expect sns heterozygosity to suppress both the hbs loss and overexpression phenotypes if hbs overexpression was acting as a dominant negative. Our data also indicate that the extracellular domain of hbs does not behave as a dominant negative construct either, at least in the somatic mesoderm, because overexpression of this construct does not cause a detectable phenotype in the somatic musculature (Fig. 7D,H,L).
Given the lack of enzymatic features in the cytoplasmic domain, we propose that Hbs associates with adapter proteins to connect to the cytoskeleton, cytoplasmic kinases or other transmembrane receptors. Interestingly, the intracellular domain contains motifs that are conserved in Sns. These motifs could potentially serve as interfaces for adaptor proteins. Work with Nephrin provides a putative candidate, the mouse CD2-associated protein (CD2AP). CD2AP, an SH3-containing cytoplasmic protein, has been shown to interact with the intracellular domain of Nephrin and has been proposed to anchor Nephrin to the actin cytoskeleton at the slit diaphragm in the kidney (Shih et al., 1999) and to participate in setting up the immunological synapse in T cells (Dustin et al., 1998). The Drosophila genome contains a CD2AP homolog, which is being analyzed in our laboratory.
Overexpression of UAS-hbsΔECD or UAS-hbsFL throughout the mesoderm resulted in a partial fusion block, but with greater muscle loss and morphological defects when compared with the loss of function phenotype. This increased muscle loss could reflect that, in our overexpression experiments, we express Hbs in both fusion-competent as well as founder cells, thereby breaking the expression asymmetry that is found under normal situations. However, we could equally suppose that prolonged and higher levels of Hbs block muscle fusion and disrupt later events in muscle morphogenesis. We note that both loss and gain of hbs result in a fusion phenotype. Although it is obvious why gain of a negative regulator could result in a fusion block, it is less clear why loss results in a similar phenotype. We speculate that hbs highlights a novel aspect of the regulated events in the fusion process.
Ig-containing proteins as adhesion receptors in fusion
Cell recognition and adhesion between myoblasts is the first step of the fusion process (Doberstein et al., 1997). Kirre has been shown to be expressed in founder cells and behave as a myoblast attractant. By contrast, Sns is expressed in fusion competent myoblasts exclusively. A simple model can be proposed with these data: Kirre interacts at the cell surface with Sns to allow recognition between founder and fusion-competent cells (Taylor, 2000; Frasch and Leptin, 2000). Subsequently, this recognition would signal the start of an orderly fusion process that requires such proteins as Rac, Blown fuse and Rolling stone. Later, myotubes continue to ‘grow’ by this process of attraction, recognition and fusion in a directed manner, sending growth cone-like structures toward specific attachment sites (Bate, 1993). Our analysis suggests that Hbs acts at the first step of this process as free myoblasts in hbs mutant embryos do not cluster nearby muscles (like sns) (Bour et al., 2000). In addition, Hbs serves as a negative regulator of Sns activity. This negative regulation is necessary to complete the fusion process as in hbs mutant embryos we find unfused myoblasts.
There are several ways in which Hbs could antagonize Sns function at the molecular level. Firstly, Hbs could compete for a putative Sns ‘ligand’. Given the similarity (69%) of the extracellular domains of Hbs and Sns, this appears reasonable. However, we show that overexpression of the extracellular domain of Hbs alone can not block myoblast fusion, while the full-length construct does. Moreover, our data support a requirement for activity of the intracellular domain of Hbs. Secondly, Hbs and Sns could form a receptor complex which switches the positive Sns signal to a negative signal. An example of this type of regulation is seen with the Netrin receptors, UNC-40/Deleted in Colorectal Cancer (DCC) and UNC-5 (Hong et al., 1999). Whereas axons expressing UNC-40/DCC are attracted to the ligand source positively, axons expressing both UNC-40/DCC and UNC-5 now are repelled and move away from the source of ligand. A third possibility would have Hbs and Sns function independently of one another but converge intracellularly on downstream proteins involved in fusion. For example Echinoid, an Ig domain protein resembling Hbs in overall structure, has been shown to antagonize the Ras pathway in the eye, not at the cell membrane but intracellularly at the level of transcription of a target gene, tramtrack (Bai et al., 2001). Hbs, Sns and Kirre may also be functioning together similarly in the generation of the visceral muscles, as recent data indicate that visceral muscles are not mononucleated, as previously reported, but syncytial (Klapper et al., 2001; San Martin et al., 2001).
Are all fusion-competent myoblasts functionally equivalent?
We find that Hbs and Sns co-localize at the cell membrane, which could correspond to focal adhesion points between founder cell and fusion-competent cells or among fusion competent myoblasts. However, Hbs precedes Sns temporally in the somatic and visceral mesoderm at stage 11, prior to the initiation of fusion. The two expression patterns largely overlap during stage 12-13 as fusion begins, with some Sns and Hbs-specific foci of expression, and by late stage 14 unfused myoblasts are basically devoid of Hbs but maintain Sns as fusion continues towards completion. We note that these differences in the pattern of expression of Sns and Hbs are meaningful as, when we genetically remove hbs such that fusion competent cells express Sns exclusively, we detect a muscle fusion phenotype. Hbs could, therefore, be revealing different potentials for the fusion-competent cells during muscle formation. Initially, cells that fail to become muscle progenitors during the first round of segregation (e.g. C2/P2 segregation) (Carmena et al., 1995) may be included in a new equipotential cluster. These clusters appear largely to overlap and form in rapid sequence. Early Hbs expression could participate in maintaining myoblasts that fail to become progenitors in the first round of selection uncommitted for successive rounds of progenitor segregation. This provides a reasonable explanation for occasional muscle losses in hbs mutant embryos. Subsequently, upon adoption of fusion-competent cell fate and expression of Sns, Hbs could regulate early events in fusion, maintaining an orderly process. During later events in the fusion process (i.e. while myotubes are already growing towards their specific attachment sites), when there are fewer free fusion-competent cells, Hbs may no longer be necessary to regulate or orient fusion. We thus propose three ‘fusion-competent cell’ stages:
(1) Pre-fusion stage. Early somatic myoblasts respond first to a laterally inhibiting Notch signaling event by expressing Hbs and subsequently have another chance at progenitor cell fate.
(2) Early fusion stage. Once the progenitor developmental option passes, the Hbs-expressing cells opt for fusion-competent fate. Expression of both Hbs and Sns regulates early steps in recognition and fusion of founders to fusion competent cells.
(3) Late fusion stage. Fusion-competent cells no longer require Hbs expression but do require Sns.
As founder cell markers such as Krüppel and Even-skipped define subsets of founders, we envisage that additional markers will be found that serve to characterize these different fusion-competent cell stages. Moreover, our hbs enhancer trap provides a novel tool to recognize subsets of fusion-competent cells without the technical difficulties associated with a punctate, membrane-bound signal.
Pleiotrophy of Hbs and links to Notch
Clearly, Sns is not the only potential partner for Hbs. Although Sns is expressed in the visceral mesoderm and muscle attachments, it is not expressed in heart, midline, hindgut, Malpighian tubules, imaginal discs or adult tissues where we find hbs expression and phenotypes. Either Hbs is acting alone in these tissues or it is interacting with other Ig domain proteins. Similarly, Sns does not always act with Hbs. Sns, like Kirre, is expressed in garland cells, yet Hbs is not.
Although Hbs operates in multiple places, we do find one unifying theme to hbs pleiotrophy – an association with Notch signaling. Our evidence demonstrates that hbs is a novel target of Notch activity in the somatic musculature. There are also several compelling similarities between the hbs and Notch activities over the course of development. Weak gain-of-function Notch constructs block myoblast fusion, one of the aspects of hbs phenotype (Fuerstenberg and Giniger, 1998). Notch is involved in planar polarity in the eye and both hbs loss- and gain-of-function lead to polarity defects (N. Paricio, personal communication). Preliminary results suggest that the hbs semisterility phenotype is due to failure to specify stalk cells in the ovarioles (N. Matova, personal communication), a phenotype described for Notch loss-of-function mutations. We also find that hbs overexpression in adults show misplacement of bristles and that hbs and Notch interact weakly in the wings, resulting in a measurable increase in wing nicks. Perhaps like the Enhancer of split complex, hbs could mediate part of Notch signaling. Future work will be directed to uncover the interaction between these two genes throughout development.
Acknowledgements
We thank the BDGP consortium for LD cDNA library; Bloomington and UMEA stock centers for fly strains; Patrick Morcillo and Kathryn Anderson for genetic advice; Lanre Adedokum, Jason Kass, Vidya Kishore, Maia Dorsett and Sabino Guzman for technical assistance; and H. Thaler for assistance with the statistical analysis. We thank A. Martinez-Arias, A. Michelson, M. Ruiz-Gomez, K. Anderson and members of our laboratory for comments on the manuscript. We also thank M. Chakravarti from S. Abmayr’s laboratory for providing purified Sns antibody necessary to complete this work. The project described was made possible in part by funds granted by the Charles H. Revson Foundation to R. D. A., the Muscular Dystrophy Association, The Society of Memorial Sloan Kettering Cancer Center, The New York Academy of Science – Speaker’s Fund for Biomedical Research Toward the Science of Patient Care and National Institutes of Health (GM 56989) to M. K. B.