Semaphorins are a large family of transmembrane and secreted proteins that are conserved from insects to humans. Both vertebrates and invertebrates semaphorins have been shown to play important roles in axonogenesis with the potential of acting as chemorepellents/inhibitors or attractants (for reviews see Kolodkin and Ginty, 1997; Nakamura et al., 2000; Tamagnone and Comoglio, 2000). They are also shown to function in the development of non-neuronal tissues. Studies in mice have shown that mutations in class III semaphorins can cause defects in patterning and growth of not only nerves but also bones and heart (Behar et al., 1996). In addition, recent work in Caenorhabditis elegans has shown that Semaphorin-2a is required to prevent ectopic cell contacts during epidermal morphogenesis (Roy et al., 2000). The signals generated by semaphorins are transduced into cells by their receptors, plexins, and neuropilins (Kolodkin et al., 1997; Rohm et al., 2000; Tamagnone et al., 1999; Takahashi et al., 1999; Winberg et al., 1998). Interestingly, the semaphorin receptors, plexins, also contain semaphorin sequences in their extracellular domain (Winberg et al., 1998).
The semaphorin members, discovered in animals from diverse phyla, represent seven classes, all of which contain a large conserved semaphorin domain. The conserved semaphorin domain is ∼500aa in length. It contains 14–16 cysteines, many blocks of conserved residues, and no obvious repeats (Kolodkin et al., 1992, 1993; Luo et al., 1993). In most family members, the semaphorin domain is often linked to other extracellular sequence motifs, which are thought to influence its activity potential. Classes II, III, IV, and VII semaphorins also contain a single C2-type immunoglobulin (Ig) domain (Kolodkin et al., 1992). Class V semaphorins contain seven thrombospondin type 1 repeats (Tsp) but no Ig motifs (Adams et al., 1996). The Tsp repeats were originally identified in the extracellular matrix protein (ECM) thrombospondin (TSP-1; Adams and Lawler, 1993; Bork, 1993). They include the putative melanoma cell adhesive sequence, VTCG (Tolsma et al., 1993), and a novel heparin binding sequence, WSXW (Vogel et al., 1993), that mediates binding of certain cells to thrombospondin (for review see Adams and Tucker 2000).
Mutation analyses have been described for several members of semaphorin classes, namely I, II, and III, but not for class V semaphorins. However, the expression patterns of class V semaphorins suggest they may have more general biological roles. In the mouse, semF and semG are differentially expressed in embryonic and adult tissues. During early embryogenesis, they are expressed in complementary regions, with semF detected only in mesodermal cells, whereas semG expressed exclusively in the neuroepithelium. A possible role for class V semaphorins in axonogenesis was suggested from recent work on human Cris-du-chat syndrome. It was shown that class V semaphorin, semF, accounts for 10% of the region uncovered by the deletion, causing the neurological disorder (Simmons et al., 1998). In Drosophila, the RNA expression pattern of the only class V semaphorin gene, sema 5c, has recently been reported (Khare et al., 2000). sema 5C RNA is expressed in stage 10 oocyte and in early embryos, suggesting the presence of maternal contribution. During embryogenesis, the sema 5C RNA is also expressed in mesodermal tissues.
In this study, we characterize the protein expression pattern of the invertebrate class V semaphorin, Sema 5C, and report that mutants lacking Dsema 5C are homozygous viable with no detectable phenotypes.
RESULTS AND DISCUSSION
Isolation of cDNAs Encoding Alternatively Spliced Forms of the Dsema 5C Gene
Several cDNA clones encoding the Drosophila sema 5C gene were isolated during the analysis of ∼40 kb genomic DNA from the 68F cytological region (Fig. 1A). Dsema 5C contains 12 exons and occupies ∼14 kb of genomic DNA (Fig. 1A). The complete cDNA sequence of Dsema 5C has been previously reported by Khare et al. (2000). DSema 5C is an integral membrane protein of 1093aa (Figs. 1, 2), with a typical class V semaphorin structure consisting of a signal sequence at the N-terminus (aa1-28) followed by semaphorin domain (aa65-546), seven Tsps (aa554-948), transmembrane region (aa957-979), and cytoplasmic tail at the C-terminus. With the fly genome project concluded, Dsema 5C is the only class V semaphorin in Drosophila.
In addition to the full-length sema 5C cDNA reported by Khare et al. (2000), a second alternatively spliced version appears to exist. We named this alternatively spliced form Dsema 5C1. DSema 5C1 contains sequences for 6 Tsp repeats only (Figs. 1A, 2), as the splicing event in Dsema 5C reduces Tsp repeats 2 and 3 into just one repeat. Tsp type 1 repeats were originally found in TSP-1 and TSP-2 proteins, which are components of the extracellular matrix (Adams and Lawler, 1993; Bork, 1993). In the type 1 repeats of TSP-1, the CSVTCG and WSXW motifs have been identified as being essential for the heparin and cell binding properties (Guo et al., 1992; Li et al., 1993; Prater et al., 1991; Tolsma et al., 1993; Tuszynski et al., 1993; Vogel et al., 1993). The CSVTCG motif is completely conserved in Tsp repeat 6 and only partially conserved (CSXXCG) in the other Tsp repeats of DSema 5C (Fig. 2B). The WSXW motif is also completely conserved in Tsp repeats 1, 2, 6, and 7 of DSema 5C (Fig. 2B).
Dsema 5C RNA and Protein Are Expressed in Mesodermal Precursors and Muscle Attachment Sites
The Dsema 5C RNA is maternally deposited and its zygotic expression shows a dynamic pattern starting at early stages of embryonic development (Khare et al., 2000). At embryonic stages 7–9, the RNA is detected at relatively high levels in the ventral and dorsal regions of the embryo (Fig. 3A). Weaker lateral staining can also be seen giving the appearance of stripes. RNA staining is also detected in the gut primordia. At stages 9–11, the staining becomes intense mainly in 11 segmentally reiterated mesodermal patches and the gut primordia (Fig. 3B). Double labeling experiments with anti-Engrailed showed that Dsema 5C expressing cells are partially underneath and posterior to the epidermal Engrailed stripe at this stage (Fig. 3B). At embryonic stage 12, Dsema 5C expression is found in fat body cell clusters (Dunin-Borkowski et al., 1995; Riechmann et al., 1998), visceral structures, and in rows of cells at the segment borders, which later constitute attachment sites for subsets of muscles (Volk and VijayRaghavan 1994) (Fig. 3C). In late stage embryos, strong expression of Dsema 5C RNA is observed at muscle attachment sites but not detected in the muscles as can be seen in double labeling experiments with Dsema 5C RNA and anti-MHC antibody (Kiehart and Feghali, 1986) (Fig. 3D).
Similar pattern of staining was also seen with anti-DSema 5C antibody (Fig. 3E–H). Like its RNA, the DSema 5C protein is also expressed in mesodermal precursor cells (Fig. 3E,F). At stage 11, the segmentally reiterated expression of DSema 5C becomes very prominent in a group of mesodermal cells that stretches from the tracheal pits at the dorsal side of the embryo to the ventral midline in each hemisegment (Fig. 3E,F). Similarly, at embryonic stage 12, DSema 5C expression is detected in fat body cell clusters (Figs. 3G) and double labeling experiments using anti–DSema 5C and fat body marker, anti-Serpent, showed coexpression of both proteins in these cell clusters (Fig. 4A–C). At late embryonic stages, anti–DSema 5C stains muscle attachment sites but are excluded from the muscles (Fig. 3H). Double labeling experiments with anti–DSema 5C, anti-bGal in the rp298 enhancer trap line (Nose et al., 1998), and anti-DMEF2 (Nguyen et al., 1994) muscle marker showed that DSema 5C is not present in muscle progenitors or the mature muscles (Fig. 4D–I). The specificity of anti–DSema 5C antibody was confirmed by double labeling of embryos obtained from the fly line, Df(3L)BK9/TM3Ubx-lacZ, with anti-bGal antibody. There was no anti–DSema 5C staining in embryos homozygous for the deletion (Fig. 3J). The expression data show that late in embryogenesis both RNA and protein are found at muscle attachment sites where each end of the muscle fiber is inserted and connected tightly to the specialized epidermal apodeme cell via the ECM. The absence of expression of both components from the muscles suggests DSema 5C localized on the epidermal surface of muscle attachments.
Isolation of Dsema 5C Mutants
Dsema 5C maps to 68F3 and several deficiencies that uncover this interval were obtained from Bloomigton stock center (see Experimental Procedures). In order to obtain specific alleles in Dsema 5C, we searched the Berkeley Genome Project Database for P-element insertions near the gene. In this search, EP(3)1049 was identified as a P-element insertion 82 bp upstream of the Dsema 5C gene (Fig. 1A). The EP-element was mobilized and 220 independent excisions were collected and analyzed on Southern blots. Two excision lines, E77 and E215, were found to delete into Dsema 5C. In E215, the genomic 1.6-kb Sal1 fragment containing exons 1, 2, and 3 is completely deleted whereas a part of the upstream adjacent 1.8-kb fragment is still present (Fig. 1A). E77 is a small deletion removing the Sal1 site located within the 5′ untranslated DNA sequence of the gene. In this mutant line, only a part of the 1.6-kb Sal1 fragment is deleted. Flies transheterozygous for these excision lines and a large deficiency, Df(3L)BK9, which removes the entire Dsema 5C gene, were also blotted and probed with the 1.6- and 1.8-kb Sal1 probes and similar results were obtained. We assume that Dsema 5CE215 is a null allele, because the mutants Dsema 5CE215/Df displayed no protein expression on Western blots (Fig. 1B). In addition, fly lines carrying the UAS-Dsema 5C transgene were generated and several Gal4 drivers were used for overexpression studies (Fig. 3I).
Dsema 5C Mutants Show No Obvious Embryonic Phenotype
The Dsema 5CE215 and Dsema 5CE77 mutant flies were viable either as homozygotes or as heterozygotes over Df(3)BK9. Homozygous stocks could be established. They did not show any readily detectable embryonic defects with the various mesodermal (anti-MHC and anti-DMEF2) and neuronal markers (22C10, BP102, anti-EVE, 1D4) used. Furthermore, overexpression experiments using various mesodermal, neuronal, and ubiquitous Gal4 drivers did not result in any obvious phenotypes. In conclusion, embryos lacking both maternal and zygotic Dsema 5C have no readily identifiable phenotypes. Double mutants carrying Dsema 5CE215 and mutations in the m-spo (Umemiya et al., 1997) and Dsema 1a (Yu et al., 1998) genes were also analysed. M-spo is an extracellular matrix protein, which, like DSema 5C, contains Tsp repeats and localises to muscle attachment sites. DSema 1a is a membrane-associated protein which, like DSema 5C, contains a large semaphorin domain and is found at muscle attachment sites. Null mutations in m-spo and Dsema 1a do not lead to any muscle attachment defects in the embryo and m-spo mutant homozygotes are viable and fully reproductive (Yu et al., 1998; Umemiya et al., 1997). The effects of the double mutant combinations on embryonic development and in particular on muscle attachment, since both proteins are localised to these sites, were studied. Double mutant embryos, lacking both maternal and zygotic m-spo and Dsema 5C, showed no muscle phenotypes. Similarly, the genetic combination with Dsema 1a did not result in any muscle attachment defects. In these double mutants, muscle attachments form normally.
The lack of muscle phenotype in Dsema 5C mutants is not surprising as muscle attachment sites are very complex structures rich in a variety of adhesion molecules. Mutations in several of these molecules result in mild or no muscle attachment phenotypes. For example, recent work describing Tiggrin, the PS2 integrin ligand, showed no abnormal muscle arrangement or structure in mutant embryos (Bunch et al., 1998). Muscle defects in tiggrin mutants were only detected in late 3rd instar larvae. In laminin A mutants, the PS1 integrin ligand, a majority of muscle fibers was inserted at their attachment sites normally, although defects were seen in the extension of some ventral muscle fibers (Henchcliffe et al., 1993). These and other studies (e.g., Fogerty et al., 1994; George et al., 1993; Saga et al., 1992; Yarnitzky and Volk, 1995) led to the suggestion that multiple components are involved in the stabilization of muscle attachments and that their overlapping functions are partially redundant.
The localization and structural features of DSema 5C are suggestive of roles for this protein in mediating cellular interactions and adhesive functions. The semaphorin domain in many semaphorin members has been shown to act as ligand for other receptors, namely plexins and neuropillins (for reviews see Nakamura et al., 2000; Tamagnone and Comoglio, 2000). The Tsp motifs have been shown in the extracellular matrix protein TSP-1 (Adams and Lawler, 1993; Bork, 1993) and other molecules to interact with a wide range of both matrix and cell surface receptors (for a recent review see Adams and Tucker, 2000). They have been shown to bind the cellular receptor CD36 (Asch et al., 1993) and integrin a3b1 (DeFreitas et al., 1995).
In conclusion, we have characterised Dsema 5C expression and generated mutations in this gene. The DSema 5C protein is expressed in fat body and visceral structures and localises to the epidermal apodemes of muscle attachments. The mutant analysis revealed no essential function for Dsema 5C in fly development.
Standard molecular biology techniques were carried out as described by Sambrook et al. (1989). Standard procedures involving Drosophila were performed as described by Ashburner (1989).
An EP line [EP(3)1049] located 5′ to the semaphorin gene was identified by sequence homology search. The EP fly insertion line was obtained from the fly genome project (Berkely). The EP element was mobilized using the immobile element P[ry+ Δ2-3](99B) as a transposase source (Robertson et al., 1988). Single jump starter males of the genotype EP1049/P[ry+ Δ2-3](99B) were crossed to 3rd chromosome balancer females. From each cross, one w-male was selected and crossed to 3rd chromosome balancer females. A total of 220 independent excision lines was established. These were analyzed on Southern blots using wt genomic DNA flanking the EP1049 insertion site as hybridization probes.
Fly Lines and Stocks
Two deficiencies uncovering the 68F interval, Df(3L)BK9 and Df(3L)vin7, were obtained from the Bloomington stock centre. The m-spoc26 null allele (Umemiya et al., 1997) and Dsema 1ap were used in generating double mutant stocks with Dsema 5C alleles. The Gal4 driver lines used for overexpression studies are twist-Gal4 for mesodermal driver, 24B for muscle driver, elav-Gal4 for neuronal driver, pGMR-Gal4 for eye driver, and heat shock and daughterless-Gal4 for ubiquitous drivers.
Isolation of Dsema 5C Genomic and cDNA Clones
Total genomic DNA was prepared from a fly line carrying a P-element insertion in the 68F region of the 3rd chromosome. Subcloning of the genomic DNA flanking the P-element was performed using standard plasmid-rescue techniques. The subcloned flanking DNA was used as a probe to screen an EMBL3 wt genomic library (Tamkun et al., 1992) and various cloned genomic fragments were subsequently used to screen plasmid (Brown and Kafatos, 1988) and lambda gt11 embryonic cDNA libraries. The cDNA clones were mapped with restriction enzymes and then used as probes in Northern hybridization and RNA in situ experiments.
Both strands of the cDNA clones were sequenced using universal M13 and gene-specific primers. The blast computer program was used to search the nonredundant GenBank, PDB, and SwissProt databases for sequence homology.
RNA In Situ and Immunostaining of Embryos
For RNA in situ, RNA probes were prepared using the Dig RNA labeling kit as described by the supplier (Boehringer Mannheim). Whole-mount Drosophila embryos were collected, fixed, and hybridized to the probe as described by Tear et al. (1996).
Anti-DSema 5C antibody was raised in mice using a 6His fusion protein as antigen. The fusion protein was generated in Escherichia coli by cloning a 0.6-kb Sal1/HindIII cDNA fragment carrying aa59-237 into PQR32 (QIAexpress vectors, QIAGEN). Following incubation of fixed embryos with the primary antibody and the horseradish peroxidase (HRP)-anti-mouse antibody, a modified horseradish peroxidase immunocytochemical method was used (Hsu et al., 1988). The specificity of the sera used was confirmed by showing a complete lack of staining in control animals carrying a deficiency that uncovers Dsema 5C [Df(3L)BK9]. Staging of embryos was according to Campos-Ortega and Hartenstein (1985). Photographs were taken using DIC optics with a Zeiss Axiophot microscope. For immunofluorescence, FITC- and Texas Red-conjugated anti-mouse or anti-rabbit antibodies were used. Confocal microscopy was done on a Biorad MRC 1024, using Adobe Photoshop for image processing.
We thank Kathy Mathews (Bloomington stock center) and Tod Lavorty (Fly Genome Project) for providing various fly lines, Mark Brennan (University of Louisville) for anti-Serpent antibody, Akinao Nose (Kyoto University) for providing us with reagents, HT Nguyen (Albert Einstein College of Medicine) for anti-DMEF2 antibody, and Sree Devi Menon (IMCB, Singapore) for her help in the project. We also thank Hing Fook Siong for his technical assistance.