The Drosophila IMP-L2 gene was identified as a 20hydroxyecdysone-induced gene encoding a membranebound polysomal transcript. IMP-L2 is an apparent secreted member of the immunoglobulin superfamily. We have used deficiencies that remove the IMP-L2 gene to demonstrate that IMP-L2 is essential in Drosophila. The viability of IMP-L2 null zygotes is influenced by maternal IMP-L2. IMP-L2 null progeny from IMP-L2+ mothers exhibit a semilethal phenotype. IMP-L2 null progeny from IMP-L2 null mothers are 100% lethal. An IMP-L2 transgene completely suppresses the zygotic lethal phenotype and partially suppresses the lethality of IMP-L2 null progeny from IMP-L2 null mothers. In embryos, IMP-L2 mRNA is first expressed at the cellular blastoderm stage and continues to be expressed through subsequent development. IMP-L2 mRNA is detected in several sites including the ventral neuroectoderm, the tracheal pits, the pharynx and esophagus, and specific neuronal cell bodies. Staining of whole-mount embryos with anti-IMP-L2 antibodies shows that IMP-L2 protein is localized to specific neuronal structures late in embryogenesis. Expression of IMP-L2 protein in neuronal cells suggests a role in the normal development of the nervous system but no severe morphological abnormalities have been detected in IMP-L2 null embryos.

Molecular recognition and intercellular adhesion are major aspects of morphogenetic processes in multicellular organisms. In particular, the development of the nervous system depends on cellular interactions mediated by adhesion molecules, signalling molecules and cell surface receptors. One aspect of nervous system development that is extensively studied at the molecular level is the outgrowth and fasciculation of axons (reviewed in Bixby and Harris, 1991). In vertebrates, a number of putative guidance molecules have been identified based on in vitro and in vivo studies using purified protein substrates (Bixby and Harris, 1991; Tomaselli et al., 1986) or antibody perturbation (Fraseret al., 1988; Landmesseret al., 1988; Silver and Rutihauser, 1984). These studies have provided some understanding of the molecular mechanisms involved in nervous system development, but the experimental approaches used do not necessarily identify the function of specific molecules. The identification of proteins in Drosophila that are homologous to vertebrate molecules (Fessler and Fessler, 1989; Grenninglogh and Goodman, 1992; Grenninglogh et al., 1991a; Hortsch and Goodman, 1991) makes possible experimental approaches that combine cellular and molecular characterization with classical genetic analysis.

Among the molecules believed to be functionally important during nervous system development are members of the immunoglobulin (Ig) superfamily. Four transmembrane Ig superfamily members, fasciclin II (Grenninglogh et al., 1991a), fasciclin III (Patel et al., 1987), neuroglian (Bieber et al., 1989) and a neurotrophin receptor homolog, Dtrk (Pulido et al., 1992), have been identified as neural cell adhesion molecules in Drosophila. In addition, at least five other Ig family proteins have been characterized in Drosophila.

In this report, we present results of our characterization of Drosophila IMP-L2 (Osterbur et al., 1988), a member of the Ig superfamily. The genomic clone containing IMP-L2 sequences was isolated in a differential screen designed to isolate genes encoding 20-hydroxyecdysone (20HE)induced secreted or membrane-bound proteins (Natzle et al., 1986). The IMP-L2 transcription unit was identified as encoding a 20HE-induced, membrane-bound polysomal RNA. In situ hybridization using antisense RNA as probe to detect IMP-L2 transcripts in frozen tissue sections of imaginal discs or metamorphosing animals indicated that the major sites of IMP-L2 expression were the peripodial epithelium of imaginal discs in prepupae and the imaginal abdominal histoblasts of early pupae (Osterbur et al., 1988). Both tissues undergo a process of hormonally induced spreading followed by fusion with adjacent imaginal tissues to form the continuous adult epithelium. These data implicated the IMP-L2 gene product in the process of epithelial spreading and fusion. We show here that IMP-L2 is a member of the Ig superfamily and is apparently a secreted protein. Like most characterized Ig genes in Drosophila, IMP-L2 is expressed in neuronal cells suggesting a role in the normal development of the nervous system. Also like other characterized Ig genes in Drosophila, no severe morphological abnormalities are detected in IMP-L2 null embryos. Genetic analysis demonstrates that IMP-L2 is an essential gene in Drosophila. Absence of zygotic IMP-L2 results in semilethality of null progeny. IMP-L2 null progeny from IMP-L2 null mothers are 100% lethal. The zygotic lethality is completely suppressed in the presence of an IMP-L2 transgene. The lethal phenotype of IMP-L2 null progeny from IMP-L2 null mothers is partially suppressed by an IMP-L2 transgene.

Drosophila stocks

Drosophila deficiency stocks used in this analysis were Df(3L)ems13 (64B1,2-64E), Df(3L)GN19 (63F9,11-64B1,2) and Df(3L)A466 (63D1,2-64B1,2). The Oregon-R stock used in this study has been continuously maintained in our laboratory since 1965. Balancer stocks and w1118; ry506 P[ry+ Δ2-3]99B are described in Lindsley and Zimm (Lindsley and Zimm, 1992).

Isolation of genomic and cDNA clones

Genomic clones were isolated from a D. melanogaster genomic library made from randomly sheared embryonic DNA (Maniatis et al., 1978). The library was screened using the 32P-labeled 3.0 kb fragment from λ64B1 carrying the 3′ portion of the IMP-L2 gene (p3.0 in Osterbur et al., 1988). The same 3.0 kb fragment was used as the probe to isolate the cDNA clone from a cDNA library derived from hormone-induced imaginal discs (Osterbur, 1986).

Hybridizations

Nucleic acid blotting and hybridizations were performed according to standard procedures (Sambrook et al., 1989). For Southern blots, genomic DNA was digested with the restiction enzyme EcoRI. The resulting products were electrophoretically separated on 0.7% agarose gels and transferred to Nytran membranes (Schleicher and Schuell). RNA samples were denatured with glyoxal, electrophoretically separated on 1.0% agarose gels and transferred to Nytran membranes. Hybridizations were performed at 65°C in 5× SSC (1× SSC is 150 mM NaCl, 15 mM sodium citrate, pH 7.0) with 0.5% SDS, 10× Denhardt’s solution, and 100 μg/ml sheared denatured salmon sperm DNA. Filters were washed at 65°C in three changes of 0.2× SSC and exposed to X-ray film.

DNA sequencing

To sequence the IMP-L2 genomic and cDNA clones, restriction fragments were cloned into the single-strand M13 phage vectors mp18 and mp19 (Norrander et al., 1983). Sequencing reactions were performed by dideoxy chain termination (Sangeret al., 1977) using Sequenase version 2.0 (US Biochemicals). Reaction products were resolved on 6% polyacrylamide/8M urea gels. Sequences were analyzed with the MacVector DNA sequence analysis program (International Biotechnologies, Inc.) Database searches were done using FASTDB (Brutlag et al., 1990) from Intelligenetics, Inc. and FASTA and TFASTA (Pearson and Lipman, 1988).

Generation of anti-IMP-L2 polyclonal antibodies

The coding region of IMP-L2 representing amino acids 22 to 263 was used to generate anti-IMP-L2 antibodies. The cDNA-containing plasmid, pL2C302, was digested with Sac II and blunt-ended with the Klenow fragment of DNA polymerase I. The DNA was purified by phenol extraction and ethanol precipitated. The DNA was resuspended and digested with Eco RI. The digestion products were electrophoretically separated on a low melting point agarose gel and the Sac II-Eco RI insert fragment isolated. The insert was ligated into the Sma I and Eco RI sites of plasmid pGEX-3X (Pharmacia-LKB) encoding glutathione-S tranferase (GST; Smith and Johnson, 1988). Recombinant plasmids were transformed into E. coli and ampicillin-resistant colonies selected. Bacteria expressing the GST/IMP-L2 fusion protein (Mr=54×103) were used for subsequent isolation of the fusion protein. The GST/IMP-L2 fusion protein was purified from induced bacterial cultures by lysing the bacteria with lysozyme and Triton X-100 followed by low speed (5000 g) centrifugation. The insoluble pellet containing the GST/IMP-L2 fusion protein was resuspended and washed several times in phosphate-buffered saline (PBS) plus 0.1% Triton X-100. The pellet was resuspended in PBS and solubilized in 0.1% SDS followed by extensive dialysis against PBS. The solubilized fusion protein (100 μg/ml in PBS) was used to immunize female Fisher 344 rats. For the initial injection antigen was emulsified in Freund’s complete adjuvant. Antigen emulsified in Freund’s incomplete adjuvant was used for booster injections. Anti-GST/IMP-L2 sera were collected and stored frozen in aliquots at −20°C. AntiGST/IMP-L2 antibodies were affinity purified by binding to GST/IMP-L2 protein immobilized on an Immunopure column (Pierce Biochemicals).

Western blotting

Protein samples dissolved in Laemmli buffer (Laemmli, 1970) were separated on 12.5% polyacrylamide/SDS gels and electrophoretically transferred to ECL nitrocellulose membranes (Amersham). Membranes were blocked in Tris-buffered saline (TBS; 150mM NaCl, 20mM Tris pH 8.0) with 3% powdered milk, 1% BSA and 0.5% Tween 20. Affinity-purified GST/IMP-L2 antibody was bound in the same blocking solution. Filters were washed 3× 15 minutes in 50 ml TBS with 0.5% Tween 20 to remove excess antibody. Filters were blocked a second time for 1 hour and horse radish peroxidase (HRP)-conjugated goat anti-rat antibody (1:1000, Cappel) was bound in blocking solution. Excess secondary antibody was removed by washing as above. HRPcatalyzed ECL reactions were done according to manufacturer’s instructions and the filters exposed to X-ray film.

Collection and fixation of embryos

Embryos were collected from flies kept in 1/2 pint bottles maintained at 22°C. Embryos were dechorionated en masse with 50% commercial bleach and fixed in a two-phase mixture of 7.4% formaldehyde in PEM buffer (100 mM Pipes, 1 mM MgCl2, 2 mM EGTA, pH 6.9), and heptane (Karr and Alberts, 1986). Fixed embryos were devitellinized with methanol and rinsed with 100% methanol to remove residual heptane. Embryos that were not used immediately were stored in 100% methanol at −20°C.

Antibody staining of embryos

Embryos in 100% methanol were rehydrated stepwise in PBS with 0.1% TWEEN 20 and 0.1% Triton X-100. Embryos were blocked in PBS with 0.1% Tween 20, 0.1% Triton X-100, 5% normal goat serum and 1 mg/ml BSA. Affinity-purified anti-GST/IMP-L2 antibody was used at an equivalent dilution of 1:500. Whole antisera were used at a dilution of 1:500-1:1000. FITC-conjugated goat anti-rat antibody (Cappel) was used at a dilution of 1:500. To visualize the embryonic nervous system, embryos were stained with rabbit anti-horseradish peroxidase (Jackson Immunoresearch Laboratories) followed by Cy3-conjugated goat anti-rabbit antibody (Jackson Immunoresearch Laboratories). Fluorescently stained embryos were examined with a Zeiss microscope.

In situ hybridization of embryos

Embryos in 100% methanol were rehydrated and processed for in situ hybridization according to the procedures of Tautz and Pfeifle (Tautz and Pfeifle, 1989). Randomly primed, digoxigenin-labeled probe was prepared with the Genius DNA labeling and detection kit (Boehringer Mannheim) according to the manufacturer’s instructions. Digoxigenin-labeled sense and antisense DNA probes were synthesized as described by Sturzl et al., (1992). Color development of the alkaline phosphatase reaction product was monitored under a dissecting microscope. The reaction was terminated by washing the embryos extensivley in PBS. Embryos were mounted in 70% glycerol/1 × PBS.

Embryo transformations

The plasmid used for P-element mediated rescue of IMP-L2deficient flies included IMP-L2 genomic sequences extending from an Eco RV site at −1000 to a Sal I site at +4200. The IMP-L2 DNA was cloned into the P-element transformation vector CaSpeR-4 (Pirrotta, 1988). Plasmid DNA at a concentration of 500 μg/ml was injected into embryos of the genotype w; ry506 P[ry+ Δ2-3]99B. Surviving adults were crossed to w; TM3 Sb /TM6B to identify transformants.

Molecular analysis

The isolation and initial characterization of the IMP-L2-containing genomic clones λ64B1 and λ64B2 have been described previously (Natzle et al., 1986; Osterbur et al., 1988). Isolation of additional clones provided a total of 32 kb of genomic sequence representing 12 kb of upstream (centromere proximal) DNA, the IMP-L2 gene and 15 kb of downstream (centromere distal) DNA (Fig. 1A). The 3.3 kb IMP-L2 transcription unit comprises three exons and two introns (Fig. 1B) producing a single mRNA species of 2.4 kb (revised from 3.0 kb in Osterbur et al., 1988). We have found no evidence from RNA blot hybridizations for alterative RNA processing. In addition to the IMP-L2 gene, four other transcription units are represented in the cloned genomic sequences (Osterbur et al., 1988). These transcription units have not been fully characterized but only the three transcription units closest to IMP-L2 (indicated in Fig. 1A) are relevant to the genetic analysis discussed below.

Fig. 1.

(A) Molecular map of the IMP-L2 gene region. The IMP-L2 gene is located in band 64B1-2 on the left arm of chromosome three. Df(3L)ems13 breaks 12 kb distal to IMP-L2. Df(3L)GN19 breaks 5 kb proximal to IMP-L2. Df(3L)A466 breaks 8 kb distal to IMP-L2. The hatched bars indicate the extent of the three deficiencies identified by analysis of genomic Southern blots. R identifies Eco RI restriction endonuclease sites. A 10 kb transcript broken by Df(3L)A466 is indicated as are two distal transcripts deleted by all deficiencies. Further characterization of IMP-L2 genomic and cDNA clones has indicated errors in the original mapping of clones as reported in Osterbur et al. (1988). Specifically, the λ64B2 clone is wholly within the limits of λ64B1 and neither clone includes sequences from the first exon of IMP-L2. The genomic DNA represented by the additional genomic clones that have been isolated are indicated. The entire IMP-L2 gene is present in clone λ64B5. The size of the IMP-L2 transcript has been revised from 3.0 kb to 2.4 kb. A further correction concerns the sequences located immediately centromere distal to IMP-L2. The 1.7 kb transcript indicated in Osterbur et al. (1988) has not been detected using probes derived from these genomic sequences (M. Prout and J. W. Fristrom, unpublished data). (B) Molecular map of the IMP-L2 gene. The mature IMP-L2 mRNA is derived from three exons and two introns as indicated. The direction of transcription is centromere proximal to distal. The long open reading frame encoding the IMP-L2 protein begins in the second exon. The conceptual protein product is 263 amino acids long with a 23 amino acid signal sequence. Four AUG codons and three short open reading frames are found in the mRNA leader (discussed in text).

Fig. 1.

(A) Molecular map of the IMP-L2 gene region. The IMP-L2 gene is located in band 64B1-2 on the left arm of chromosome three. Df(3L)ems13 breaks 12 kb distal to IMP-L2. Df(3L)GN19 breaks 5 kb proximal to IMP-L2. Df(3L)A466 breaks 8 kb distal to IMP-L2. The hatched bars indicate the extent of the three deficiencies identified by analysis of genomic Southern blots. R identifies Eco RI restriction endonuclease sites. A 10 kb transcript broken by Df(3L)A466 is indicated as are two distal transcripts deleted by all deficiencies. Further characterization of IMP-L2 genomic and cDNA clones has indicated errors in the original mapping of clones as reported in Osterbur et al. (1988). Specifically, the λ64B2 clone is wholly within the limits of λ64B1 and neither clone includes sequences from the first exon of IMP-L2. The genomic DNA represented by the additional genomic clones that have been isolated are indicated. The entire IMP-L2 gene is present in clone λ64B5. The size of the IMP-L2 transcript has been revised from 3.0 kb to 2.4 kb. A further correction concerns the sequences located immediately centromere distal to IMP-L2. The 1.7 kb transcript indicated in Osterbur et al. (1988) has not been detected using probes derived from these genomic sequences (M. Prout and J. W. Fristrom, unpublished data). (B) Molecular map of the IMP-L2 gene. The mature IMP-L2 mRNA is derived from three exons and two introns as indicated. The direction of transcription is centromere proximal to distal. The long open reading frame encoding the IMP-L2 protein begins in the second exon. The conceptual protein product is 263 amino acids long with a 23 amino acid signal sequence. Four AUG codons and three short open reading frames are found in the mRNA leader (discussed in text).

The sequence of the IMP-L2 gene and conceptual translation product is presented in Fig. 2. The transcription start site (+1) was determined by the method of Hu and Davidson (1986; data not shown). A closely spaced TATA box begins 17 nucleotides upstream from the transcription start site. A canonical polyadenylation signal (nt +3319-3324) is present 13 nucleotides upstream from the polyadenylation site. A large open reading frame (ORF) is present in the sequence beginning at position +1684 from the transcription start site (+740 on the mRNA). Three small ORFs are located in the leader region (discussed below). The large ORF encodes a conceptual protein of 263 amino acids. At the amino terminus are 23 hydrophobic amino acids with properties characteristic of a signal sequence (von Heijne, 1983). Removal of the signal sequence would produce a protein of 240 amino acids with a calculated relative molecular mass of 27×103. No consensus N-X-S/T sites for N-linked glyco-sylation are present. The absence of a transmembrane domain in the IMP-L2 conceptual protein and the absence of hydrophobic sequences indicative of phosphoinositol linkage (Ferguson and Williams, 1988) suggest that the IMP-L2 protein product is secreted. Thirteen nucleotide differences were detected in the sequences of the genomic and cDNA for IMP-L2 (see legend, Fig. 2). Of these differences, seven are in the protein-coding region but only one results in a change of the protein sequence; a conservative substitution of isoleucine for valine at amino acid 173. We believe these differences reflect the differences between the two D. melanogaster stocks used in the construction of the genomic and cDNA libraries from which clones were isolated.

Fig. 2.

IMP-L2 DNA and conceptual protein sequence. The DNA sequence shown extends from a NaeI site at −29 from the transcription start site to the Sca I site at +3350. The TATA box (−17 to −13) is overlined and the transcription start site (+1) is indicated with an asterisk and right arrow. Three short open reading frames in the mRNA leader (discussed in text) are underlined. Intron sequences are in lowercase, the polyadenylation signal (+3329-3333) is underlined and the polyadenylation site is indicated with an up arrow. Genomic sequences before the start site and after the polyadenylation site (+3347) are italicized. The first nucleotide of the cDNA clone corresponds to nucleotide 129 of the genomic sequence. The probable signal sequence of the IMP-L2 protein is double underlined and the four cysteines of the Ig domains are indicated in boldface. There are thirteen nucleotide differences between the genomic DNA and the cDNA clone: 647:T→C; 1893:T→C; 1960:T→C; 2190:G→A; 2196:C→A; 2200:G→A; 2298:C→T; 2319:T→C; 2689→2691:CAA→ATC; and 2702→2703: AA deleted. The G→A transition at 2200 results in a conservative amino acid substitution of isoleucine for valine 173. This sequence has been submitted to the GenBank database under the accession number L23066.

Fig. 2.

IMP-L2 DNA and conceptual protein sequence. The DNA sequence shown extends from a NaeI site at −29 from the transcription start site to the Sca I site at +3350. The TATA box (−17 to −13) is overlined and the transcription start site (+1) is indicated with an asterisk and right arrow. Three short open reading frames in the mRNA leader (discussed in text) are underlined. Intron sequences are in lowercase, the polyadenylation signal (+3329-3333) is underlined and the polyadenylation site is indicated with an up arrow. Genomic sequences before the start site and after the polyadenylation site (+3347) are italicized. The first nucleotide of the cDNA clone corresponds to nucleotide 129 of the genomic sequence. The probable signal sequence of the IMP-L2 protein is double underlined and the four cysteines of the Ig domains are indicated in boldface. There are thirteen nucleotide differences between the genomic DNA and the cDNA clone: 647:T→C; 1893:T→C; 1960:T→C; 2190:G→A; 2196:C→A; 2200:G→A; 2298:C→T; 2319:T→C; 2689→2691:CAA→ATC; and 2702→2703: AA deleted. The G→A transition at 2200 results in a conservative amino acid substitution of isoleucine for valine 173. This sequence has been submitted to the GenBank database under the accession number L23066.

Similarity searches of protein and nucleic acid databases revealed significant similarity between IMP-L2 and proteins in the immunoglobulin (Ig) superfamily. Further analysis showed that two immunoglobulin domains constitute the bulk of the IMP-L2 protein. Consistent with this interpretation, the amino acid residues that characterize Ig domains are conserved in IMP-L2 and the highest level of similarity between IMP-L2 and other Ig family proteins is around the conserved cysteines. Comparison of the IMP-L2 sequence to that of other Drosophila Ig domain-containing proteins showed most similarity to Neuroglian (Nrg; Bieber et al., 1989). IMP-L2 was also found to be highly similar to the vertebrate Ig proteins mouse L1 (Moos et al., 1988) and chicken contactin/F11 (Ranscht, 1988; Brümmendorf et al., 1989). These proteins all have a similar structure and are believed to be evolutionarily related (Grenninglogh et al., 1991a). These proteins also have been identified as neural cell adhesion molecules. Less sequence similarity was found between IMP-L2 and other Drosophila Ig proteins such as Fasciclin II (Grenninglogh et al., 1991b) and the two secreted Ig molecules Amalgam (Seeger et al., 1988) and Lachesin (Karlstrom et al., 1993). The greatest degree of similarity to other Ig family members was consistently observed with the second Ig domain of IMP-L2 (Fig. 3). The identification of IMP-L2 as a member of the Ig superfamily thus suggests that IMP-L2 may be involved in some aspect of cell adhesion. However, the small size and apparent secreted character of IMP-L2 also indicate a role different from that of transmembrane cell adhesion molecules.

Fig. 3.

Alignment of the second Ig domain of IMP-L2 with similar Ig domains of Drosophila Neuroglian (Nrg; Ig-5), Amalgam (Ama; Ig3), Lachesin (Lach; Ig-3), and fasciclin II (FasII; Ig-5). Residues that are identical in three of the five sequences are indicated with an asterisk. The strongest similarity is found in the regions around the conserved cysteine residues (shown in boldface). The aligned sequences begin 10 residues before the first cysteine and end 10 residues after the second cysteine in each Ig domain. The alignment was performed with the PILEUP program from the University of Wisconsin Genetic Computer Group using the algorithm of Needleman and Wunsch [Needleman, 1970 no. 341].

Fig. 3.

Alignment of the second Ig domain of IMP-L2 with similar Ig domains of Drosophila Neuroglian (Nrg; Ig-5), Amalgam (Ama; Ig3), Lachesin (Lach; Ig-3), and fasciclin II (FasII; Ig-5). Residues that are identical in three of the five sequences are indicated with an asterisk. The strongest similarity is found in the regions around the conserved cysteine residues (shown in boldface). The aligned sequences begin 10 residues before the first cysteine and end 10 residues after the second cysteine in each Ig domain. The alignment was performed with the PILEUP program from the University of Wisconsin Genetic Computer Group using the algorithm of Needleman and Wunsch [Needleman, 1970 no. 341].

Although three additional start codons are present in the mRNA, several lines of evidence indicate that the large ORF encodes the IMP-L2 protein. First, this ORF displays codon usage consistent with that of other Drosophila genes (Ashburner, 1989) and has the proposed initiator AUG in a favorable context for translational initiation (Kozak, 1986; Cavener, 1987; Cavener and Ray, 1991;). Second, the conceptual translation product from this ORF possesses two Ig domains with sequence similarity to other Ig domain-containing proteins suggesting evolutionary conservation that would not be expected in noncoding DNA. Third, this is the only ORF that encodes a protein with a signal sequence. The presence of a signal sequence would be expected based on the isolation scheme and the association of the IMP-L2 mRNA with membrane-bound polysomes. Fourth, antisera made against a glutathione S-transferase (GST)/IMP-L2 fusion protein translated in the reading frame of this ORF recognize a Mr 32×103 protein in wild-type flies that is absent from IMP-L2-deficient flies (see below). Although the first small ORF encountered on the mRNA also has a favorable translation initiation context and appropriate codon usage for Drosophila, the sequence similarity and immunological detection of the large ORF product indicate that it is the major though not necessarily the only translation product from the IMP-L2 gene (see discussion).

Genetic analysis

Chromosomal deficiencies extending into the 64B region were characterized by Southern analysis to determine if they removed the IMP-L2 locus. Three of these, Df(3L)ems13 (64B1,2-64E), Df(3L)GN19 (63F-64B1,2) and Df(3L)A466 (63D1,264B1,2), hereafter referred to as ems13, GN19 and A466, were found to have breakpoints within 15 kb of the IMP-L2 gene as illustrated in Fig.1A. The ems13-GN19 overlap removes 25 kb of DNA and deletes the IMP-L2 gene (see Fig. 4). Three additional transcription units are deleted in ems13/GN19 heterozygotes, two encoding 1.2 kb mRNAs and one encoding a 10 kb mRNA. The ems13-A466 overlap removes 14 kb of DNA distal to the IMP-L2 gene but does not remove IMP-L2 which is still expressed in an apparently normal manner (M. Prout and J. W. Fristrom, unpublished). The two smaller transcription units are deleted in ems13/A466 heterozygotes while the 10 kb transcript is broken at its 5′ end and produces a shorter transcript of 8 kb probably from an adventitious promoter (M. Prout and J. W. Fristrom, unpublished data).

Fig. 4.

Autoradiogram showing that the IMP-L2 gene is absent in ems13/GN19 heterozygotes. Southern blots were done as described in Materials and Methods. (Lanes A,B) DNA from IMP-L2+ adults, ems13/TM6B (Lane A) and GN19/TM6B (Lane B). A single 3.7 kb band from the endogenous IMP-L2 gene is detected. (Lanes C,-E) DNA from P[w+IMP-L2+]; ems13/GN19 adults (lane C, line 337.221; lane D, line 337.481; lane E, line 337.551). Note absence of the 3.7 kb band and the presence of new bands derived from the transformed IMP-L2 gene.

Fig. 4.

Autoradiogram showing that the IMP-L2 gene is absent in ems13/GN19 heterozygotes. Southern blots were done as described in Materials and Methods. (Lanes A,B) DNA from IMP-L2+ adults, ems13/TM6B (Lane A) and GN19/TM6B (Lane B). A single 3.7 kb band from the endogenous IMP-L2 gene is detected. (Lanes C,-E) DNA from P[w+IMP-L2+]; ems13/GN19 adults (lane C, line 337.221; lane D, line 337.481; lane E, line 337.551). Note absence of the 3.7 kb band and the presence of new bands derived from the transformed IMP-L2 gene.

Unexpectedly, the GN19 deficiency chromosome was semiviable when heterozygous with the ems13 deficiency. However, the viability of ems13/GN19 F1 heterozygous progeny lacking the IMP-L2 gene was variably reduced (ranging from 0 to 60%) and was frequently less than 10% of Mendelian expectations (Table 1). The direction of the cross had no obvious effect on the viability of IMP-L2 deficient progeny (data not shown). In contrast, ems13/A466 heterozygotes were recovered at near Mendelian expectations (Table 2) suggesting that IMP-L2 is the only gene deleted in ems13/GN19 heterozygotes that is essential for full viability. The recovery of ems13/A466 heterozygotes further suggests that deficiencies in the region do not reduce F1 zygotic viability nonspecifically.

Table 1.

Viability of IMP-L2 null heterozygotes

Viability of IMP-L2 null heterozygotes
Viability of IMP-L2 null heterozygotes
Table 2.

Viability of IMP-L2+ heterozygotes

Viability of IMP-L2+ heterozygotes
Viability of IMP-L2+ heterozygotes

Examination of the ems13/GN19 heterozygote larvae and adults revealed no gross morphological defects in epidermal structures that would be expected if IMP-L2 were essential for epithelial morphogenesis. Specifically, all imaginal discderived and abdominal histoblast-derived structures were present in adults and were seen to have spread and fused normally with adjacent tissues. Furthermore, all adult cuticular structures were present and appeared in their normal positions. Similarly, examination of ems13/GN19 heterozygote embryos revealed no gross defects in development of the nervous system, the epidermis or the digestive system. More detailed morphological examinations are in progress to identify possible abnormalities in embryos that are associated with the absence of the IMP-L2 gene.

Viable ems13/GN19 heterozygote males were sterile indicating that at least one of the deleted loci is required for male fertility. The A466 deficiency does not delete the IMP-L2 gene but ems13/A466 males were also sterile, thereby excluding IMP-L2 as a single cause of the male sterility.

Viable ems13/GN19 heterozygote females exhibit maternal effect embryonic lethality. This was initially recognized when ems13/GN19 heterozygote females were mated to IMP-L2 deficiency/balancer males (e.g. ems13/GN19 × GN19/TM6B). In these crosses, only progeny carrying the paternal TM6B chromosome were recovered (Table 3). A paternally supplied A466 chromosome (IMPL2+) rescued the maternal effect lethality as efficiently as an IMP-L2+ balancer chromosome (Table 3).

Table 3.

Viability of G2 progeny from IMP-L2 null mothers

Viability of G2 progeny from IMP-L2 null mothers
Viability of G2 progeny from IMP-L2 null mothers

To determine the lethal phase of the maternal effect, embryos from the cross ems13/GN19 ♀ ♀ × GN19/TM3 Sb Ser ♂ ♂ were examined. Data tabulated from several collections showed that only 66% of the embryos that hatched survived to eclose as adults. All eclosed adults carried the IMP-L2+ balancer chromosome. Examination of the collection plates revealed that many first instar larvae died shortly after hatching. A few additional embryos appeared to develop through stage 17 but did not hatch. In comparison, in the cross ems13/GN19 ♀ ♀ × Oregon-R ♂ ♂, 86% of hatched eggs eclosed as adults. We conclude that the major lethal phase for the ems13/GN19 progeny of ems13/GN19 mothers is during early first instar.

To verify that the observed zygotic and maternal effect embryonic lethality was due to loss of IMP-L2, cloned genomic IMP-L2 sequences were introduced into Drosophila via P-element-mediated germ-line transformation. The transformation construct included the IMP-L2 gene, 1.0 kb of upstream sequence and 1.0 kb of downstream sequence in the transformation vector CaSpeR-4 (Pirrotta, 1988). Nine independent transformants were recovered with insertion sites on the X and second chromosomes. Presence of the transposed IMP-L2 gene was confirmed by Southern blotting. The crosses indicated in Table 4 (five independent transformant lines) permitted us to determine the rescuing ability of IMP-L2 with respect to the viability of ems13/GN19 heterozygotes.

Table 4.

(A) Rescue of F1 zygotic semilethality by an IMP-L2 transgene

Table 4. (B) Rescue of lethality in G2 progeny of IMP-L2+, ems13/GN19 females

(A) Rescue of F1 zygotic semilethality by an IMP-L2 transgene
(A) Rescue of F1 zygotic semilethality by an IMP-L2 transgene

In F1ems13/GN19 heterozygotes (Table 4A), the presence of the IMP-L2 transgene resulted in full rescue of viability as determined by the number of P[w+; IMP-L2+] ems13/GN19 adults compared to P[w+; IMP-L2+] GN19/TM3. As an internal control for viability, the number of w; ems13/GN19 adults was 10% of Mendelian expectations compared to w; GN19/TM3. DNA from the P[w+; IMPL2+] ems13/GN19 progeny was analyzed by Southern blotting and confirmed the presence of both deficiency chromosomes and the P[w+, IMP-L2+] insert (Fig. 4).

To determine the rescuing ability of IMP-L2 on the maternal effect lethality, w; P[w+, IMP-L2+]; ems13/GN19 females were crossed to w; P[w+, IMP-L2+]; GN19/TM3 Sb Ser males (Table 4B). Rescue of lethality was indicated by recovery of non TM3 Sb Ser G2 progeny. No w; ems13/GN19 G2 progeny were recovered. Rescue of the maternal effect embryonic lethality by the transgene was only 10-20% efficient compared to a wild-type IMP-L2 gene. Whether this low efficiency was due to insufficient regulatory sequences for normal maternal and zygotic expression or a nonspecific effect of the mothers being heterozygous for the deficiencies is not known.

Four lines were also tested for paternal rescue of the maternal effect lethality. As shown in Table 4C, ems13/GN19 progeny from ems13/GN19 mothers can be rescued by a paternally supplied IMP-L2 transgene. The similar results in crosses where the IMP-L2 transgene was derived from both the maternal and paternal parent or only the paternal parent suggest that zygotic expression of IMPL2 provides the rescuing function in the crosses described. IMP-L2 transgenic animals transformed with a construct containing 2.5 kb of upstream sequence also showed zygotic rescue of lethality but at a lower efficiency than the transgene with 1.0 kb of upstream sequence (data not shown) suggesting the presence of negative regulatory elements in the additional upstream sequences.

To summarize, genetic analysis demonstrates that IMPL2 is an essential gene in Drosophila and that IMP-L2-deficienct embryos exhibit a lethal phenotype that is influenced by maternal IMP-L2. Rescue experiments with an IMP-L2 transgene conclusively show that the lethal phenotype is due to the loss of IMP-L2.

Expression of IMP-L2

The genetic analyses of IMP-L2 indicated that partial IMPL2 function for the developing embryo can be obtained from maternal components supplied during oogenesis. To address this point further, total RNA was isolated from embryos during the first 8 hours of development at 23°C and probed to identify IMP-L2 transcripts (Fig. 5A). RNA from embryos aged 0–2 hours had no detectable IMP-L2 mRNA and embryos aged 2-4 hours had only a barely detectable amount although the blots were intentionally overexposed to detect low levels of transcript. In contrast, embryos aged 4–6 hours and 6–8 hours had very high levels of IMP-L2 mRNA. This result indicated that the IMP-L2 function supplied to the embryo during oogenesis is not provided by IMP-L2 mRNA.

Fig. 5.

(A) Expression of IMP-L2 mRNA during early embryogenesis. Autoradiogram of northern blots done as described in Materials and Methods. Total RNA was isolated from embryos aged 0–2 hours (lane A), 2–4 hours (lane B), 4–6 hours (lane C) and 6–8 hours (lane D). No signal was detected in 0–2 hour embryo RNA and only weak signal was detected in 2–4 hour RNA. Strong signal was detected beginning in 4–6 hour and 6–8 hour RNA. The film was intentionally overexposed to detect faint signal in the 0–2 hour and 2–4 hour RNA samples. (B) Autoradiogram of western blot of proteins from IMP-L2+ and IMP-L2 prepupae. Prepupae of genotype ems13/GN19 (lane A) and Oregon-R (lanes B and C) were crushed in SDS sample buffer and the proteins separated by SDS-PAGE. Proteins were transferred to a nitrocellulose filter and probed with affinitypurified anti-IMP-L2 antibody. HRP-conjugated secondary antibody was used in the detection reaction with the Amersham ECL detection system. Note the strong signal in the Oregon-R samples for IMP-L2 (Mr=32×103) that is absent in the ems13/GN19 sample. The film was intentionally overexposed to show the absence of IMP-L2 protein in the ems13/GN19 sample.

Fig. 5.

(A) Expression of IMP-L2 mRNA during early embryogenesis. Autoradiogram of northern blots done as described in Materials and Methods. Total RNA was isolated from embryos aged 0–2 hours (lane A), 2–4 hours (lane B), 4–6 hours (lane C) and 6–8 hours (lane D). No signal was detected in 0–2 hour embryo RNA and only weak signal was detected in 2–4 hour RNA. Strong signal was detected beginning in 4–6 hour and 6–8 hour RNA. The film was intentionally overexposed to detect faint signal in the 0–2 hour and 2–4 hour RNA samples. (B) Autoradiogram of western blot of proteins from IMP-L2+ and IMP-L2 prepupae. Prepupae of genotype ems13/GN19 (lane A) and Oregon-R (lanes B and C) were crushed in SDS sample buffer and the proteins separated by SDS-PAGE. Proteins were transferred to a nitrocellulose filter and probed with affinitypurified anti-IMP-L2 antibody. HRP-conjugated secondary antibody was used in the detection reaction with the Amersham ECL detection system. Note the strong signal in the Oregon-R samples for IMP-L2 (Mr=32×103) that is absent in the ems13/GN19 sample. The film was intentionally overexposed to show the absence of IMP-L2 protein in the ems13/GN19 sample.

To characterize the distribution of IMP-L2 mRNA during embryonic development, in situ hybridizations were done on embryos ranging in age from 0 to 18 hours (i.e. stages 1-17; Fig. 6A-F). Identical results were obtained using strandspecific DNA probes (data not shown). IMP-L2 mRNA was first detected in cellular blastoderm embryos in a series of seven distinct stripes spanning 20% to 70% egg length (Fig. 6A) in a pattern similar to pair rule genes. The IMP-L2 stripes extended dorsoventrally through the neurogenic and dorsal epidermal regions. The stripes did not extend into the region of the presumptive mesoderm (Fig. 6B) and were also absent from the cells that form the amnioserosa. During germ-band elongation, the stripe pattern was modified. At stage 8, 14 broad bands of staining were localized to the neurogenic ectoderm (Fig. 6C). In stage 9 embryos, strong signal was detected surrounding the tracheal pits (Fig. 6D). During stages 15–16, IMP-L2 mRNA was detected in the pharynx, frontal sac, the esophagus, surrounding the posterior spiracles (Fig. 6E) and the lateral bipolar dendrite neurons [lbd; Bodmer and Jan, 1987 no. 250; Bodmer et al., 1989 no. 251]. By stage 17, IMP-L2 mRNA was limited to the lbd neurons (part of the transverse nerve) and other structures that show accumulation of the protein (Fig. 6F, and see below).

Fig. 6.

Photomicrographs of embryos showing expression of IMP-L2 mRNA and protein during embryogenesis. Embryos were prepared for in situ hybridization (Fig. 6A-F) as described in Materials and Methods. The digoxigenin-labeled probe was detected using antidigoxigenin antibody coupled to alkaline phosphatase. To localize IMP-L2 protein (Fig. 6G,H), embryos were stained with affinitypurified anti-GST/IMP-L2 primary antibodies and FITC-conjugated goat anti-rat secondary antibodies. All embryos are oriented with anterior to the left. Lateral view of embryo at the cellular blastoderm stage (A) shows IMP-L2 mRNA in seven stripes. Ventrolateral view of embryo (B) at the cellular blastoderm stage shows IMP-L2 mRNA is not detected in the presumptive mesoderm. Lateral view of embryo after germ band extension (C). IMP-L2 mRNA is present in fourteen segmentally repeating regions of the ventral neuroectoderm. Lateral view of embryo prior to germ band retraction (D) showing IMP-L2 mRNA expression in the neuroectoderm is reduced and is replaced by expression surrounding the tracheal pits. Dorsal view of stage16 embryo (E) shows major sites of IMP-L2 mRNA expression including the pharynx (large arrow), esophagus (large arrowhead) and posterior spiracles (small arrowheads) in addition to the lateral bipolar dendrite (lbd) neurons and several other cell bodies. Lateral view of stage 17 embryo (F). IMP-L2 mRNA expression is limited to the lbd neurons in the abdomen and a pair of neurons adjacent to the pharynx. IMP-L2 protein is detected in whole-mount embryos beginning at stage 16. Lateral view of stage 16 embryo showing IMP-L2 protein on the lbd neurons in the abdomen (G, small arrowheads). Dorsolateral view of stage 16 embryo (H) shows staining of the lbd neurons, the anterior neurons adjacent to the pharynx (small arrow), and a structure near the supraesophageal ganglion (large arrow).

Fig. 6.

Photomicrographs of embryos showing expression of IMP-L2 mRNA and protein during embryogenesis. Embryos were prepared for in situ hybridization (Fig. 6A-F) as described in Materials and Methods. The digoxigenin-labeled probe was detected using antidigoxigenin antibody coupled to alkaline phosphatase. To localize IMP-L2 protein (Fig. 6G,H), embryos were stained with affinitypurified anti-GST/IMP-L2 primary antibodies and FITC-conjugated goat anti-rat secondary antibodies. All embryos are oriented with anterior to the left. Lateral view of embryo at the cellular blastoderm stage (A) shows IMP-L2 mRNA in seven stripes. Ventrolateral view of embryo (B) at the cellular blastoderm stage shows IMP-L2 mRNA is not detected in the presumptive mesoderm. Lateral view of embryo after germ band extension (C). IMP-L2 mRNA is present in fourteen segmentally repeating regions of the ventral neuroectoderm. Lateral view of embryo prior to germ band retraction (D) showing IMP-L2 mRNA expression in the neuroectoderm is reduced and is replaced by expression surrounding the tracheal pits. Dorsal view of stage16 embryo (E) shows major sites of IMP-L2 mRNA expression including the pharynx (large arrow), esophagus (large arrowhead) and posterior spiracles (small arrowheads) in addition to the lateral bipolar dendrite (lbd) neurons and several other cell bodies. Lateral view of stage 17 embryo (F). IMP-L2 mRNA expression is limited to the lbd neurons in the abdomen and a pair of neurons adjacent to the pharynx. IMP-L2 protein is detected in whole-mount embryos beginning at stage 16. Lateral view of stage 16 embryo showing IMP-L2 protein on the lbd neurons in the abdomen (G, small arrowheads). Dorsolateral view of stage 16 embryo (H) shows staining of the lbd neurons, the anterior neurons adjacent to the pharynx (small arrow), and a structure near the supraesophageal ganglion (large arrow).

Antisera against the GST/IMP-L2 fusion protein were raised as described in Materials and methods. Production of antibodies against the IMP-L2 portion of the fusion protein was verified by cross-reactivity of the antisera with a lacZ/IMP-L2 fusion protein. On western blots (see Materials and methods), affinity-purified anti-GST/IMP-L2 antibodies recognized a protein from Oregon-R white prepupae with an apparent Mr=32×103 (Fig. 5B). Preimmune serum did not recognize this protein and the protein was absent in white prepupae of ems13/GN19 heterozygotes that lacked the IMPL2 gene (Fig. 5B). The size of the presumptive IMP-L2 product as detected by western blotting of Drosophila proteins (Mr=32×103) is larger than the predicted size of the conceptual translation product (Mr=27×103). Although no sites for N-linked glycosylation are present in the conceptual IMP-L2 protein sequence, the difference between the predicted and observed size may be due to O-linked glycosylation, other post-translational modifications, or aberrant migration.

The genetic analysis and northern blot results suggested that IMP-L2 protein should be present in embryos as a maternally supplied product. A faint IMP-L2 protein signal was detected on western blots for embryos aged 2–4 hours and a stronger signal was seen for embryos aged 4–6 and 68 hours (data not shown) in agreement with the observed changes in IMP-L2 mRNA levels. However, several attempts to detect IMP-L2 protein in extracts from 0–2 hour embryos were inconclusive. IMP-L2 protein may be present in such low abundance as to be undetectable above background using the antibodies available. Alternatively, the antigenic epitopes may be blocked by modifications specific for this developmental stage.

Affinity-purified anti-GST/IMP-L2 antibody was also used to identify the sites of IMP-L2 protein accumulation during embryonic development. Although we expected the IMP-L2 antigen to be present in embryos after 4 hours of development, no localized IMP-L2 could be detected until approximately stage 16, after dorsal closure. At this stage, several sites of IMP-L2 staining appear on the lateral region of embryos in a segmentally repeating pattern (Fig. 5G, H). This staining pattern is detectable through stage 17. The lateral staining structures in the embryo abdominal region were identified as the lbd neurons (Bodmer et al., 1989; Bodmer and Jan, 1987) when embryos were double stained with anti-GST/IMP-L2 and anti-HRP, which stains all neurons (Jan et al ., 1985). The earliest stage of embryonic development at which IMP-L2 staining of the lbd neurons has been detected is when the anterior fascicle has formed and the differentiation of the sensory organs is progressing but not complete.

In the thoracic segments, similarly positioned neuronal cell bodies and their axons were also stained with the IMPL2 antibodies. The identity of these neurons is not known but they most likely correspond to multiple dendritic neurons described by Bodmer and Jan (1987). Also stained were two neurons in the anterior region of the embryo positioned adjacent to the pharynx. IMP-L2 protein was also detected in two closely positioned cells or cell clusters on the dorsolateral surface of the embryonic optic lobes (data not shown). A final staining structure was also located in the cephalic region of the embryo. This bilobed structure was positioned between the lobes of the supraesophageal ganglion and posterior to the supraesophageal commissure. This structure did not appear to stain with anti-HRP antibody but may represent specialized cells that will constitute part of the ring gland or lymph glands.

In summary, staining of whole-mount embryos with anti-IMP-L2 antibodies showed that IMP-L2 protein was localized to specific neuronal structures late in embryogenesis. All sites of IMP-L2 protein localization in stage 16-17 embryos have been identified as sites of IMP-L2 mRNA expression. This was in contrast to earlier stages of embryogenesis when IMP-L2 mRNA was abundant and IMP-L2 protein was detectable by western blotting but no protein could be localized in whole-mount embryos.

Our analysis of IMP-L2 has identified it as a new member of the immunoglobulin superfamily in Drosophila with an essential developmental role during embryogenesis. IMPL2 embryos from IMP-L2+ mothers show a partial lethal phenotype (average viability reduced to less than 10% of expected). IMP-L2 embryos from IMP-L2 mothers were 100% lethal during late embryogenesis or as early first instar larvae. Genetic analysis suggests that IMP-L2 product may be supplied to the oocyte during oogenesis. Although we have been unable to detect IMP-L2 mRNA or protein in 02 hour embryos, we suggest that IMP-L2 protein is present below the level of detection and is sufficient to permit viability of IMP-L2 null zygotes. In the absence of maternal IMP-L2, zygotically produced IMP-L2 product, synthesized after cellular blastoderm formation, rescues the lethal phenotype. An IMP-L2 transgene provided zygotic rescue at 10-20% of the efficiency of an endogenous IMP-L2 gene. The IMP-L2 transgene does not appear to provide sufficient IMP-L2 function to rescue ems13/GN19 heterozygous progeny of ems13/GN19 heterozygous mothers.

We have not identified the cause of lethality in IMP-L2 null Drosophila. The lbd neurons that accumulate IMP-L2 protein are present in both F1 and G2IMP-L2 null progeny indicating that IMP-L2 is not required for the formation of these neurons. Whether the other cells and tissues that accumulate IMP-L2 protein are also unaffected is currently being investigated. It is also possible that the lethal defect will be found at sites where IMP-L2 mRNA is expressed although the protein is not detected.

The results of our analysis raise three major issues regarding the role of IMP-L2 during development. First, the essential embryonic requirement for IMP-L2 can be partially met if the mother is IMP-L2+. This suggests that IMP-L2 is a maternal effect locus. Second, although IMP-L2 mRNA is abundant in many sites after cellular blastoderm formation, IMP-L2 protein is only localized late in development close to the stage when IMP-L2 null zygotes die. Third, like other Drosophila Ig molecules, IMP-L2 is primarily expressed in neural cells. This suggests a role for IMP-L2 in the proper development or function of the nervous system.

Zygotic and maternal effect lethality

The developmental processes occurring during the early stages of embryogenesis are mediated by maternal products supplied during oogenesis. These maternally supplied gene products direct subsequent development by modulating the hierarchical expression of zygotic genes involved in embryonic pattern formation. Maternal effect embryonic lethal mutations are defined as those that have no obvious effect on the morphology or viability of homozygous mutant females but result in lethality of their embryos (Schüpach and Wieschaus, 1986, 1989).

Another group of maternal effect mutations are those where the gene products required for embryonic development can be supplied either by the maternal genome during oogenesis or the zygotic genome during embryogenesis. Among these mutations are maroonlike (Chovnick et al., 1969), fused (Fausto-Sterling, 1971), cinnamon (Baker, 1973), rudimentary (Fausto-Sterling, 1971), deep orange (Garen and Gehring, 1972), abnormal oocyte (Mange and Sandler, 1973) almondex (Lehmann et al., 1983), pecanex (LaBonne and Mahowald, 1985) and l(1)pole hole (Perrimon et al., 1985). Common to all of these, mutant females lay eggs that do not undergo normal embryogenesis unless the zygote receives a wild-type allele, paternally. In the case of l(1)pole hole, the severity of the lethal phenotype is substantially reduced but not eliminated by the paternally supplied allele. The zygotic and maternal effect lethal phenotype of IMP-L2 null embryos suggests that IMPL2 is a member of this class of maternal effect loci.

Post-transcriptional regulation

The second point of interest in our analysis is the apparent discrepancy between where IMP-L2 mRNA and IMP-L2 protein were expressed. As described, IMP-L2 transcripts are first observed during the cellular blastoderm stage. Subsequently, high levels of IMP-L2 mRNA accumulate in the region of the ventral neuroectoderm and around the tracheal pits. Ultimately, transcripts are detected only in the lateral bipolar neurons. In contrast, we have detected no localized IMP-L2 protein accumulation during the early stages of embryogenesis. Although it is detected in western blots, IMP-L2 protein localization is first seen only in stage 16–17 embryos. There are three possible reasons for our inability to localize IMP-L2 protein in early embryonic stages. First, the IMP-L2 protein may be secreted from the cell and rapidly diffuse into the surrounding tissues. A consequence of the IMP-L2 protein being secreted is that the phenotypic effects leading to lethality may occur in regions that show no evidence of IMP-L2 mRNA or protein synthesis. Second, turnover of the IMP-L2 protein after it is secreted may be sufficiently rapid that the protein does not accumulate to detectable levels for immunolocalization. Third, and most intriguing, IMP-L2 mRNA may be subject to some form of post-transcriptional or translational regulation mediated by sequences in the leader.

Based on the scanning model of translational initiation, Kozak has argued that most RNAs with small ORFs in the leader represent incompletely processed transcripts, not the mRNA species that are actually translated (Kozak, 1989, 1991). Indeed, for some characterized transcripts, alternatively processed mRNAs have been identified that do not have ORFs in the leader (see Kozak, 1989, 1991). Thus, it is possible that the IMP-L2 transcript represented by the 2.4 kb cDNA is an incompletely processed RNA that is not translated. This raises the possibility that during embryogenesis, a minor IMP-L2 mRNA species would be synthesized that lacks ORFs in the leader and is translated in the cells where IMP-L2 protein is detected. Although our analysis has not indicated the presence of alternatively processed IMP-L2 mRNAs, this possibility cannot be excluded.

Kozak’s arguments also raise a significant point regarding the function of IMP-L2 in Drosophila. In vertebrates, most transcripts with ORFs in the leader encode proteins that function in growth regulation or signal transduction including proto-oncogenes and transcription factors (Table II in Kozak, 1991). Overexpression of these proteins would be detrimental and incomplete RNA processing would be one mechanism to down-regulate gene expression. By analogy, IMP-L2 protein may be required in small amounts or only in specific tissues and overexpression of IMP-L2 may be detrimental or lethal. In vitro mutagenesis of the IMP-L2 leader or overexpression of the IMP-L2 protein would address this issue. At present, we do not know whether the IMP-L2 cDNA represents an incompletely processed transcript that is not translated or the normal IMPL2 mRNA with ORFs in the leader. To distinguish conclusively between these and other possibilities will require the identification of a second IMP-L2 mRNA with an ORF-free leader. However, in either situation, the predicted result would be the same as our experimental results; the amount of IMP-L2 protein detected would be lower than expected based on the amount of IMP-L2 transcript.

Functions of immunoglobulin domain-containing molecules in Drosophila

The third major point raised by our results derives from our analysis showing IMP-L2 is a member of the Ig superfamily. To speculate on the possible role of IMP-L2, it is useful to consider the apparent function of other Ig domain-containing proteins in Drosophila and the data derived from studies on Ig superfamily proteins from vertebrates. Although several Ig domain-containing proteins have been identified in vertebrates, functional analysis has been limited to experiments involving purified proteins as substrates or antibody perturbation. The Drosophila homologs of some of the vertebrate proteins have been identified, making them ideal candidates for combined molecular and genetical analysis.

At least nine Ig superfamily members have been identified and characterized in Drosophila (Hortsch and Goodman, 1991). Two of these, Neuroglian, (Bieber et al., 1989) and Fasciclin II, (Grenninglogh et al., 1991b), have been characterized as cell adhesion molecules, having extracellular regions consisting of Ig and fibronectin type III domains. The cell adhesion molecule Fasciclin III (Snow et al., 1989) has three extracellular Ig domains that are highly divergent. The Drosophila fibroblast growth factor receptor (D-FGFR; Glazer and Shilo, 1991; Klämbt et al., 1992) and neurotrophin receptor (Dtrk; Pulido et al., 1992) have cytoplasmic kinase domains. Drosophila homologs of the phosphotyrosine phosphatases DLAR and DPTP, (Streuli et al ., 1989; Tian et al., 1991) have cytoplasmic phosphatase domains. These kinase and phosphatase domains indicate roles in intercellular signalling through familiar signal transduction pathways.

The Drosophila secreted Ig family members are Amalgam, Lachesin and IMP-L2. The amalgam (ama) gene was identified as a transcription unit within the Antennape dia complex (Seeger et al., 1988). The conceptual product is a secreted protein with three Ig domains that accumulates on various mesodermal and neural cells during embryogenesis. No ama-specific null mutations are known. Analysis of deficiencies that delete ama and the adjacent zen and bicoid loci indicate no obvious neurological defects that can be attributed to the loss of ama (Seeger et al., 1988). The sequence of the Ama protein is most similar to N-CAM (Barthels et al., 1987; Hemperly et al., 1986) and Fasciclin II (Harrelson and Goodman, 1988). Lachesin was identified immunologically and is most similar to Amalgam (Karlstrom et al., 1993). No mutations in the lachesin gene have been described. IMP-L2 was identified as an ecdysoneresponsive gene in imaginal discs. IMP-L2, however, is essential for development. Embryos with IMP-L2 proteinnull genotypes die primarily in the 1st instar, and like mutants of fasciclin II and neuroglian, have no severe abnormalities. Our data clearly indicate an essential role for IMP-L2. We predict, however, that the loss of IMP-L2 function results in subtle morphological abnormalities.

Three conclusions emerge from the characterization of Ig family members in Drosophila. The first is that, among their various sites of expression, most Drosophila Ig family members are expressed in neurons. Thus, Ig family members are strongly implicated in various aspects of neural development.

The second conclusion is that loss of an Ig protein product may not result in major disruptions of the embryonic nervous system but may produce subtle defects. Loss of neuroglian results in a general loss of intercellular adhesiveness that is most evident in the PNS. Fasciclin II null embryos lack specific axons in an otherwise normal appearing CNS. These results are consistent with the possibility that these gene products have very specific roles in development, e.g. affecting particular axons but not the entire CNS or PNS. These results are also consistent with the lethal phenotypes of some of the mutants. For example, failure to hatch from the egg or death during the early first instar could easily result from the miswiring of the CNS and resultant motor defects. Finally, these results are also consistent with the reemerging view of functional redundancy, in which the same function may be mediated by different molecules, such that loss of one of these molecules will have little or no obvious effect. The data for nrg (Bieber et al., 1989) could thus be interpreted to show that, although Nrg is involved in intercellular adhesion of neural cells, other molecules (e.g., integrins; Hynes, 1992) also promote adhesion in those cells. Likewise, the function of an independent protein could compensate for the loss of ama.

The third conclusion is that secreted Drosophila Ig family members have unknown functions. This is true even though most Drosophila Ig family genes have been identified as homologs of mammalian or, in the case of the fasciclins, grasshopper genes. Both ama and IMP-L2 encode secreted proteins. Mammalian counterparts of these genes with known functions apparently have not been described. A secreted form of N-CAM has been found, but its function is unknown (Gower et al., 1988; Pizzey et al., 1989). Axonin1, a possible chicken homolog of TAG-1 and F3, is expressed by axons and contains fibronectin as well as Ig domains. In its PI anchored form, axonin-1 functions as a substrate adhesion molecule (Zuellig et al., 1992). It has been proposed (Stoekli et al ., 1991) that the secreted form of axonin-1 is an anti-adhesion molecule. Mouse F3, which also contains both Ig and fibronectin domains, may stimulate neurite outgrowth in its secreted form by binding to cell surface receptors to elicit a signal transduction response and not by acting mechanically as an adhesion molecule (Durbec et al., 1992; Saffell et al., 1992). It is not known whether the activities of axonin-1 and F3 are mediated by the fibronectin domains or the Ig domains. None the less, secreted Ig family members, by binding to cell surfaces or matrices may act at a distance from their sites of synthesis as adhesion molecules, as ligands for signal transduction, as tropic molecules (Durbec et al., 1992) or as anti-adhesion molecules. We infer that IMP-L2 protein may also bind to specific cell adhesion molecules. Because IMP-L2 is essential for development, it provides an unusual opportunity to study the developmental function of a secreted Ig family member.

We thank Mike Simon, Steve Wasserman and Rob Jackson for Drosophila stocks used in this study and Aki Nose for assistance in neuron identification. We also thank the members of the Fristrom lab for their comments and suggestions. J. C. G. was the recipient of a USPHS postdoctoral fellowship. This work was supported in part by USPHS grant GM19937 to J. W. F.

Ashburner
,
M.
(
1989
).
Drosophila, a laboratory handbook
.
Cold Spring Harbor, New York
:
Cold Spring Harbor Laboratory Press
.
Baker
,
B. S.
(
1973
).
The maternal and zygotic control of development by cinnamon. A new mutant in Drosophila melanogaster
.
Dev. Biol
.
33
,
429
440
.
Barthels
,
D.
,
Santoni
,
M. J.
,
Wille
,
W.
,
Ruppert
,
C.
,
Chaix
,
J. C.
,
Hirsch
,
M. R.
,
Fontecilla-Camps
,
J.
and
Goridis
,
C.
(
1987
).
Isolation and nucleotide sequence of mouse NCAM cDNA that codes for a Mr 79,000 polypeptide without a membrane spanning region
.
EMBO J
.
6
,
907
914
.
Bieber
,
A. J.
,
Snow
,
P. J.
,
Hortsch
,
M.
,
Patel
,
N. H.
,
Jacobs
,
J. R.
,
Traquina
,
Z. R.
,
Schilling
,
J.
and
Goodman
,
C. S.
(
1989
).
Drosophila neuroglian: a member of the immunoglobulin superfamily with extensive homology the the vertebrate neural adhesion molecule L1
.
Cell
59
,
447
460
.
Bixby
,
J. L.
and
Harris
,
W. A.
(
1991
).
Molecular mechanisms of axon growth and guidance
.
Annu. Rev. Cell Biol
.
7
,
117
159
.
Bodmer
,
R.
,
Carretto
,
R.
and
Jan
,
Y. N.
(
1989
).
Neurogenesis of the peripheral nervous system in Drosophila embryos: DNA replication patterns and cell lineages
.
Neuron
3
,
21
32
.
Bodmer
,
R.
and
Jan
,
Y. N.
(
1987
).
Morphological differentiation of the embryonic peripheral neurons in Drosophila
.
Roux’s Arch. Dev. Biol
.
196
,
69
77
.
Brümmendorf
,
T.
,
Wolff
,
J. M.
,
Frank
,
R.
and
Rathjen
,
F. G.
(
1989
).
Neural cell recognition molecule F11: homolgy with fibronectin type III and immunoglobin type C domains
.
Neuron
2
,
1351
1361
.
Brutlag
,
D. L.
,
Dautricourt
,
J. P.
,
Maulik
,
S.
and
Pelph
,
J.
(
1990
).
Improved sensitivity of biological sequence database searches
.
J. Computer Appl. Biosciences
6
,
237
245
.
Cavener
,
D. R.
(
1987
).
Comparison of consensus sequences flanking translational start sites in Drosophila and vertebrates
.
Nucleic Acids Res
.
15
,
1353
1361
.
Cavener
,
D. R.
and
Ray
,
S. C.
(
1991
).
Eukaryotic start and stop translation sites
.
Nucleic Acids Res
.
19
,
3185
3192
.
Chovnick
,
A.
,
Finnerty
,
V.
,
Schalet
,
A.
and
Duck
,
P.
(
1969
).
Studies on genetic organization in higher organisms: I Analysis of a complex gene in Drosophila melanogaster
.
Genetics
62
,
145
160
.
Durbec
,
P.
,
Gennarini
,
G.
,
Goridis
,
C.
and
Rougon
,
G.
(
1992
).
A soluble form of the F3 neural cell adhesion molecule promotes neurite outgrowth
.
J. Cell Biol
.
117
,
877
887
.
Fausto-Sterling
,
A.
(
1971
).
On the timing and place opf action during embryogenesis of female-sterile mutants fused and rudimentary Drosophilamelanogaster
.
Dev. Biol
26
,
452
463
.
Ferguson
,
M.
and
Williams
,
A. F.
(
1988
).
Cell-surface anchoring of proteins via glycosyl-phosphatidylinositol structures
.
Ann. Rev. Bioch
.
57
,
285
320
.
Fessler
,
J. H.
and
Fessler
,
L. I.
(
1989
).
Drosophila extracellular matrix
.
Annu. Rev. Cell Biol
.
5
,
309
339
.
Fraser
,
S. E.
,
Carhart
,
M. S.
,
Murray
,
B. A.
,
Chuong
,
C. M.
and
Edelman
,
C. M.
(
1988
).
Alterations in the Xenopus retinotectal projection by antibodies to Xenopus N-CAM
.
Dev. Biol
.
129
,
217
230
.
Garen
,
A.
and
Gehring
,
W.
(
1972
).
Repair of the lethal developmental defect in deep orange embryos of Drosophila by injection of normal egg cytoplasm
.
Proc. Nat. Acad. Sci. USA
69
,
2982
2985
.
Glazer
,
L.
and
Shilo
,
B.
(
1991
).
The Drosophila FGF-R homologue is expressed in the embryonic tracheal system and appears to be required for directed tracheal cell extension
.
Genes Dev
.
5
,
697
705
.
Gower
,
H. J.
,
Barton
,
C. H.
,
Elsom
,
V. L.
,
Thompson
,
J.
,
Moore
,
S. E.
,
Dickson
,
G.
and
Walsh
,
F. S.
(
1988
).
Alternative splicing generates a secreted form of N-CAM in muscle and brain
.
Cell
55
,
955
964
.
Grenninglogh
,
G.
,
Bieber
,
A. J.
,
Rehm
,
E. J.
,
Snow
,
P. M.
,
Traquina
,
Z. R.
,
Hortch
,
M.
,
Patel
,
N. H.
and
Goodman
,
C. S.
(
1991a
).
Molecular genetics of neuronal recognition in Drosophila: evolution and function of immunoglobulin superfamily cell adhesion molecules
.
Cold Spring Harbor Symp. Quant. Biol
.
55
,
327
340
.
Grenninglogh
,
G.
,
Rehm
,
E. J.
and
Goodman
,
C. S.
(
1991b
).
Genetic analysis of growth cone guidance in Drosophila: fasciclin II functions as a neural recognition molecule
.
Cell
67
,
45
57
.
Grenninglogh
,
G.
and
Goodman
,
C. S.
(
1992
).
Pathway recognition by neuronal growth cones: genetic analysis of neural cell adhesion molecules in Drosophila
.
Curr. Opin. Neurobiol
.
2
,
42
47
.
Harrelson
,
A. L.
and
Goodman
,
C. S.
(
1988
).
Growth cone guidance in insects: fasciclin II is a member of the immunoglobulin superfamily
.
Science
242
,
700
708
.
Hemperly
,
J. J.
,
Murray
,
B. A.
,
Edelman
,
G. M.
and
Cunningham
,
B. A.
(
1986
).
Sequence of a cDNA clone encoding the polysialic acid rich and cytoplasmic domains of the neural cell adhesion molecule N-CAM
.
Proc. Natl. Acad. Sci. USA
83
,
3037
3041
.
Hortsch
,
M.
and
Goodman
,
C. S.
(
1991
).
Cell and substrate adhesion molecules in Drosophila
.
Annu. Rev. Cell Biol
.
7
,
505
557
.
Hu
,
M.
and
Davidson
,
N.
(
1986
).
Mapping transcriptional start points on cloned genomic DNA with T4 DNA polymerase: a precise and convenient technique
.
Gene
42
,
21
29
.
Hynes
,
R. O.
(
1992
).
Integrins: versatility, modulation, and signalling in cell adhesion
.
Cell
69
,
11
25
.
Jan
,
Y. N.
,
Ghysen
,
A.
,
Christoph
,
I.
,
Barbel
,
S.
and
Jan
,
L. Y.
(
1985
).
Formation of neuronal pathways in the imaginal discs of Drosophila melanogaster
.
J. Neurosci
.
5
,
2453
2464
.
Karlstrom
,
R. O.
,
Wilder
,
L. P.
and
Bastiani
,
M. J.
(
1993
).
Lachesin: an immunoglobulin superfamily protein whose expression correlates with neurogenesis in grasshopper embryos
.
Development
118
,
509
522
Karr
,
T. L.
and
Alberts
,
B. M.
(
1986
).
Organization of the cytoskeleton in early Drosophila embryos
.
J. Cell Biol
.
102
,
1494
1509
.
Klämbt
,
C.
,
Glazer
,
L.
and
Shilo
,
B.
(
1992
).
breathless, a Drosophila FGF receptor homolog, is essential for migration of tracheal and specific midline glial cells
.
Genes and Dev
.
6
,
1668
1678
.
Kozak
,
M.
(
1986
).
Point mutations define a sequence flanking the AUG initiator codon that modulates translation by eukaryotic ribosomes
.
Cell
44
,
283
292
.
Kozak
,
M.
(
1989
).
The scanning model for translation: an update
.
J. Cell Biol
.
108
,
229
241
.
Kozak
,
M.
(
1991
).
An analysis of vertebrate mRNA sequences: intimations of translational control
.
J. Cell Biol
.
115
,
887
903
.
LaBonne
,
S. G.
and
Mahowald
,
A. P.
(
1985
).
Partial rescue of embryos from two maternal effect neurogenic mutants by transp[lantation of wild type ooplasm
.
Dev. Biol
.
110
,
264
267
.
Laemmli
,
U. K.
(
1970
).
Cleavage of structural proteins during the assembly of the head of bacteriophage T4
.
Nature
227
,
680
685
.
Landmesser
,
L.
,
Dahm
,
L.
,
Schultz
,
K.
and
Rutihauser
,
U.
(
1988
).
Distinct roles for adhesion molecules during innervation of embryonic chick muscle
.
Dev. Biol
.
130
,
45
70
.
Lehmann
,
R.
,
Jimenez
,
F.
,
Dietrich
,
U.
and
Campos-Ortega
,
J. A.
(
1983
).
On the phenotype and development of mutants of early neurogenesis in Drosophila melanogaster
.
Wilhelm Roux’s Arch. Dev. Biol
.
192
,
62
74
.
Lindsley
,
D. L.
and
Zimm
,
G. G.
(
1992
).
The Genome of Drosophila melanogaster
.
San Diego, CA
:
Academic Press
.
Mange
,
A. P.
and
Sandler
,
L.
(
1973
).
A note on the maternal effect mutants daughterless and abnormal oocyote in Drosophila melanogaseter
.
Genetics
73
,
73
86
.
Maniatis
,
T.
,
Hardison
,
R. C.
,
Lacy
,
E.
,
Lauer
,
J.
,
O’Connell
,
C.
,
Quon
,
D.
,
Sim
,
G. K.
and
Efstratiadis
,
A.
(
1978
).
The isolation of structural genes from libraries of eucaryotic DNA
.
Cell
15
,
687
701
.
Moos
,
B.
,
Tacke
,
R.
,
Scherer
,
H.
,
Teplow
,
D.
,
Fruh
,
K.
and
Schachner
,
M.
(
1988
).
Neural cell adhesion molecule L1 is a member of the immunoglobulin superfamily with binding domains similar to fibronectin
.
Nature
334
,
701
703
.
Natzle
,
J. E.
,
Hammonds
,
A. S.
and
Fristrom
,
J. W.
(
1986
).
Isolation of genes active during ecdysone-induced imaginal disc morphogenesis in Drosophila imaginal discs
.
J. Biol. Chem
.
261
,
5575
5583
.
Norrander
,
J.
,
Kempe
,
T.
and
Messing
,
J.
(
1983
).
Construction of improved M13 vectors using deoxyribonucleotide-directed mutagenesis
.
Gene
26
,
101
106
.
Osterbur
,
D. L.
, (
1986
).
Ecdysteroid action in Drosophila melanogaster: receptors and genes
.
Ph.D Thesis
,
University of California
,
Berkeley
Osterbur
,
D. L.
,
Fristrom
,
D. K.
,
Natzle
,
J. E.
,
Tojo
,
S. J.
and
Fristrom
,
J. W.
(
1988
).
Genes expressed during imaginal disc morphogenesis. IMP-L2, a gene associated with epithelial fusion
.
Dev. Biol
.
129
,
439
448
.
Patel
,
N. H.
,
Snow
,
P. M.
and
Goodman
,
C. S.
(
1987
).
Characterization and cloning of fasciclin III: a glycoprotein expressed on a subset of neurons and axon pathways in Drosophila
.
Cell
48
,
975
988
.
Pearson
,
W. R.
and
Lipman
,
D. J.
(
1988
).
Improved tools for biological sequence comparison
.
Proc. Natl. Acad. Sci. USA
85
,
2444
2448
.
Perrimon
,
N.
,
Engstrom
,
L.
and
Mahowald
,
A.
(
1985
).
A pupal lethal mutation with a paternally influenced maternal effect on embryonic development in Drosophila melanogaster
.
Dev. Biol
.
110
,
480
491
.
Pirrotta
,
V.
(
1988
).
Vectors for P-element transformation in Drosophila
.
In A Survey of Molecular Cloning Vectors and their Uses
. (ed.
R. L.
Rodrighuez
and
D. T.
Denhardt
). pp
437
456
.
Boston, MA
:
Butterworth
Pizzey
,
J. A.
,
Rowett
,
L. H.
,
Barton
,
C. H.
,
Dickson
,
G.
and
Walsh
,
F. S.
(
1989
).
Intercellular adhesion mediated by human muscle neural cell adhesion molecule: effects of alternate exon use
.
J. Cell Biol
.
109
,
3465
3476
.
Pulido
,
D.
,
Campuzano
,
S.
,
Koda
,
T.
,
Modolell
,
J.
and
Barbacid
,
M.
(
1992
).
Dtrk, a Drosophila gene related to the trk family of neurotrophin receptors, encodes a novel class of neural cell adhesion molecule
.
EMBO J
.
11
,
391
404
.
Ranscht
,
B.
(
1988
).
Sequence of contactin, a 130-kD glycoprotein concentrated in areas of interneural contact, defines a new member of the immunoglobin supergene family in the nervous system
.
J. Cell Biol
.
107
,
1561
1573
.
Saffell
,
J. L.
,
Walsh
,
F. S.
and
Doherty
,
P.
(
1992
).
Direct activation of second messanger pathways mimics cell adhesion molecule-dependent neurite outgrowth
.
J. Cell Biol
.
118
,
663
670
.
Sambrook
,
J.
,
Fritsch
,
E. F.
and
Maniatis
,
T.
(
1989
).
Molecular Cloning: a Laboratory Manual
..
Cold Spring Harbor, New York. USA
:
Cold Spring Harbor Laboratory Press
.
Sanger
,
F.
,
Nicklen
,
S.
and
Coulsen
,
A. F.
(
1977
).
DNA sequencing with chain terminating inhibitors
.
Proc. Natl. Acad. Sci. USA
74
,
5463
5467
.
Schüpach
,
T.
and
Wieschaus
,
E.
(
1986
).
Germline autonomy of maternal-effect mutations altering the embryonic body pattern of Drosophila
.
Dev. Biol
.
113
,
443
448
.
Schüpach
,
T.
and
Wieschaus
,
E.
(
1989
).
Female sterile mutations on the second chromosome of Drosophila melanogaster. I. Maternal effect mutations
.
Genetics
121
,
101
117
.
Seeger
,
M. A.
,
Haffley
,
L.
and
Kaufman
,
T. C.
(
1988
).
Characterization of amalgam: a member of the immunoglobulin superfamily from Drosophila
.
Cell
55
,
589
600
.
Silver
,
J.
and
Rutihauser
,
U.
(
1984
).
Guidance of optic axons in vivo by a preformed adhesive pathway on neuroepithelial endfeet
.
Dev. Biol
.
106
,
485
499
.
Smith
,
D. B.
and
Johnson
,
K. S.
(
1988
).
Single-step purification of polypeptides expressed in Escherichia coli as fusions with glutathione-Stranferase
.
Gene
67
,
41
40
.
Snow
,
P. M.
,
Bieber
,
A. J.
and
Goodman
,
C. S.
(
1989
).
Drosophila fasciclin III: a novel homophilic adhesion molecule
.
Cell
59
,
313
323
.
Stoekli
,
E. T.
,
Kuhn
,
T. B.
,
Duc
,
C. o.
,
Ruegg
,
M. A.
and
Sonderegger
,
P.
(
1991
).
The axonally secreted protein axonin-1 is a potent substratum for neurite outgrowth
.
J. Cell Biol
.
112
,
449
455
.
Streuli
,
M.
,
Krueger
,
N. X.
,
Tsai
,
A. Y. M.
and
Saito
,
H.
(
1989
).
A family of receptor-linked protein tyrosine phosphatases in humans and Drosophila
.
Proc. Natl. Acad. Sci. USA
86
,
8698
8702
.
Sturzl
,
M.
,
Oskoui
,
K. B.
and
Roth
,
W. K.
(
1992
).
‘Run-off’ polymerization with digoxygenin labelled nucleotides creates highly sensitive and strand specific DNA hybridization probes: synthesis and applications
.
Molec. Cell. Probes
6
,
107
114
.
Tautz
,
D.
and
Pfeifle
,
C.
(
1989
).
A non-radioactive in situ hybridization method for the localization of specific RNAs in Drosophila embryos reveals translational control of the segmentation gene hunchback
.
Chromosoma
98
,
81
85
.
Tian
,
S.
,
Tsoulfas
,
P.
and
Zinn
,
K.
(
1991
).
Three receptor-linked proteintyrosine phosphatases are selectively expressed on central nervous system axons in the Drosophila embryo
.
Cell
67
,
675
685
.
Tomaselli
,
K. J.
,
Reichardt
,
L. F.
and
Bixby
,
J. L.
(
1986
).
Distinct molecular interactions mediate neuronal process outgrowth on nonneuronal cell surfaces and extracellular matrices
.
J. Cell Biol
.
103
,
2659
2672
.
von Heijne
,
G.
(
1983
).
Pattern of amino acids near signal-sequence cleavage sites
.
Eur. J. Biochem
.
133
,
17
21
.
Zuellig
,
R. A.
,
Rader
,
C.
,
Schroeder
,
A.
,
Kalousek
,
M. B.
,
Von Bolen und Halbach
,
F.
,
Osterwalder
,
T.
,
Inan
,
C.
,
Stoekli
,
E. T.
,
Affolter
,
H. U.
and
Fritz
,
A.
(
1992
).
The axonally secreted cell adhesion molecule, axonin-1. Primary structure, immunoglobulin-like and fibronectin-typeIII-like domains and glycosyl-phosphatidylinositol anchorage
.
Eur. J. Biochem
.
204
,
453
463
.