Abstract
Drosophila neurotactin is a transmembrane glycoprotein with an apparent molecular mass of 135 x 103 Neurotac tin is regionally expressed at the cellular blastoderm stage; later in embryogenesis the expression of the protein becomes restricted to cells of the peripheral and central nervous system. Immunocytochemical localiz ation shows neurotactin protein at points of cell-cell contact. Using the anti-neurotactin monoclonal antibody BP-106, a neurotactin cDNA was isolated that encodes a 846 residue polypeptide. The chromosomal location of the neurotactin gene is 73C. The extracellular domain at the carboxyterminal end of the neurotactin protein shows a strong structural and sequence homology to serine esterases without retaining the amino acids forming the active center. Neurotactin therefore belongs to a growing group of proteins including Drosophila glutactin and thyroglobulins that are known to share this serine esterase protein domain motif without retaining the active center of the enzyme.
Introduction
An increasing number of membrane glycoproteins have been identified and characterized that play important roles in cell adhesion and cell interactions during Drosophila embryogenesis. A number of these pro teins, such as neuroglian and Notch, are widely expressed in a variety of different cell types during embryonic development (Johansen et al. 1989; Kidd et al. 1989; Bieber et al. 1989; Hortsch et al. 1990). Other proteins, such as fasciclin I and fasciclin III, are restricted to only a subset of cells or tissues or even cell surface subdomains (Bastiani et al. 1987; Patel et al. 1987; Zinn et al. 1988). Many of these proteins can mediate cell adhesion in a homophilic fashion (Snow et al. 1989; Elkins et al. 1990; Grenningloh et al. 1990). A few others are capable of mediating heterophilic adhesion, including the PS integrins and the Notch and Delta proteins (Leptin et al. 1987; Fehon et al. 1990).
Many of the surface glycoproteins that can mediate cell adhesion share one of several structurally related protein domains that places them into common gene families such as the immunoglobulin gene superfamily, the integrin family, or the family of molecules containing EGF-like repeats. Several of these mem brane proteins involved in cell adhesion and cell differentiation were initially identified and characterized by the use of monoclonal antibodies (mAb) (Wilcox et al. 1984; Bastiani et al. 1987; Patel et al. 1987; Bieber et al. 1989).
In the present study, we used the BP-106 mAb to characterize and clone a new membrane glycoprotein that is initially widely expressed and later restricted largely to the developing nervous system. Based on its expression at points of neuronal cell-cell contact, we propose the name ‘neurotactin’ for this protein.
Biochemical analysis shows that the neurotactin protein is a type II transmembrane protein with its carboxyterminus facing the extracellular space; the protein has an apparent molecular mass of 135 x103 The neurotactin gene was cloned by cDNA expression cloning, and the gene localized to the third chromo some at position 73C. cDNA sequence analysis of the 846 amino acid open reading frame leads to a deduced protein with a carboxyterminus extracellular domain that is structurally related to serine esterases. The active center that is necessary for enzymatic activity, however, is not conserved. Neurotactin belongs to a growing group of proteins, including the Drosophila extracellular matrix protein glutactin (Olson et al. 1990), that share this serine esterase protein domain motif but not the serine residue necessary for esterase activity. Most of the results reported here have also been obtained independently by the collaborative efforts of M. Piovant and colleagues in Marseille, and F. Jimenez and colleagues in Madrid (Barthalay et al. 1990; de la Escalera et al. 1990).
Materials and methods
Generation and screening of monoclonal antibodies and antisera
The mAb BP-106 was generated using membrane proteins from cultured embryonic Drosophila neuronal cells. The procedures for neuron culture, membrane isolation, immuniz ation and hybridoma screening have been previously de scribed by Patel et al. (1987).
A 1.9 kb KpnI-Sa/1 neurotactin cDNA fragment coding for the 419 carboxyterminal amino acid residues was subcloned into the protein A fusion protein expression vector pRIT31 (Nilsson and Abrahmsen, 1990). Protein A-neurotactin fusion protein was expressed in pop2136 cells (Haymerle et al. 1986) by heat shock for 2 h at 42°C. Fusion protein isolated from inclusion bodies was used to immunize rats. 20 μg of protein A-neurotactin fusion protein was injected s.c. at multiple sites per immunization. The first antigen adminis tration was presented in complete Freund’s adjuvant, all subsequent ones in incomplete Freund adjuvants. Immuniz ations were repeated at 2 week intervals until a high titer antiserum was obtained.
Immunocytochemistry
Imaginal discs and larval brains were collected in a mass disc isolation (Eugene et al. 1979) or hand dissected from 3rd instar larvae or white pre-pupae. Staining of whole mount embryos and imaginal discs was performed using HRP conjugated goat anti-mouse IgG secondary antibodies (Jack son ImmunoResearch Laboratories, Inc.) as outlined by Patel et al. (1989). A Zeiss Axiophot microscope with Nomarski optics was used to view and photograph mounted embryos and imaginal discs.
SDS-polyacrylamide gel electrophoresis and Western blot analysis
Microsomal membrane protein fractions from staged Dros ophila embryos and from Drosophila Schneider 1 (Sl) cells (Schneider, 1972) were prepared as previously described (Patel et al. 1987). Proteins were separated on 7.5 % SDS-polyacrylamide gels and immunoblots were performed as described by Burnette (1981). Blotted proteins were probed with antibodies and visualized using the protocol of Hortsch et al. (1985).
Carbonate treatment and deglycosylation of membrane proteins
Microsomal membranes from Drosophila Sl cells were carbonate washed with 0.1 M Na2CO3, pH 11, according to the method of Fujiki et al. (1982) with the modifications described by Hortsch et al. (1985). Total membrane proteins from Sl cells were deglycosylated with trifluoromethanesulfonic acid (TFMS) as described by Edge et al. (1981). Deglycosylation with endoglycosidase H was carried out according to the manufacturers specifications (Boehringer Mannheim Bio chemica). Deglycosylated proteins were analyzed by immuno blotting as outlined above.
cDNA expression cloning
A 1/ 1000 dilution of BP-106 mouse mAb ascites fluid was used to screen the Kai Zinn 9-12 h Drosophila embryo Agtll cDNA expression library (Zinn et al. 1988). Nitrocellulose filters were processed as described for Western blots using a HRP-conjugated goat anti-mouse IgG secondary antibody and diaminobenzidine for the detection of positive phages.
DNA sequencing
DNA sequencing was performed according to the dideoxy chain termination method (Sanger et al. 1977) using the Sequenase kit (United States Biochemical Corp.), 35S-dATP, and 60cm buffer gradient gels (Biggin et al. 1983). Subcloning into the Bluescript KS/+ vector (Stratagene) and Ml3 mplO has been described previously (Bieber et al. 1989).
Northern blot analysis
Total RNA was prepared from staged Drosophila embryos, separated in formaldehyde containing 1 % agarose gels (Lehrach et al. 1977) and blotted onto Genescreen Plus membranes. Hybridization was carried out according to the method of Church and Gilbert (1984) using DNA probes radiolabeled with [CY-32P]dCTP or [CY-32P]dATP by the extension of sequences randomly primed with oligonucleo tides as described by Feinberg and Vogelstein (1983).
Whole mount tissue in situ hybridization
Single-stranded, digoxigenin-labeled DNA probes were made using single primers and a modified PCR reaction (N. Patel and C. Goodman, unpublished data). Sense and anti sense probes were made from the 3.3 kb central EcoRI fragment and the 3’ 435 bp EcoRI fragment. To control for differences in sensitivity created by the differences in probe length, the reaction mix for the 3.3 kb fragment contained one-tenth the ratio of digoxigenin-dUTP:unlabeled dTTP as the mixture for the 435 bp fragment. Hybridization and detection of embryonic transcripts was done essentially according to the protocol of Tautz and Pfeifle (1989). Embryos were cleared in 70 % glycerol for microscopic analysis.
Results
The BP-106 mAb was isolated in a monoclonal antibody screen for novel antigens expressed on the surface of developing neurons in the Drosophila embryo; this was the same series of screens that led to the identification of fasciclin III (Patel et al. 1987) and neuroglian (Bieber et al. 1989; Hartsch et al. 1990). As described later in this paper, the BP-106 mAb recognizes a protein we call neurotactin based upon immunocytochemical, bio chemical, and cDNA sequence analysis; the BP-106 mAb detects the same pattern of protein expression as does a polyclonal antiserum generated against a neurotactin fusion protein. Therefore, we will refer to neurotactin rather than simply the BP-106 antigen when describing its pattern of expression.
Immunocytochemical analysis of neurotactin expression
The BP-106mAb was used to stain Drosophila em bryos, the larval central nervous system (CNS), and imaginal discs; these immunocytochemical experiments reveal that the neurotactin protein is expressed in a dynamic pattern during Drosophila development.
Neurotactin expression is first detected at the cellular blastoderm stage. The protein expression is limited to specific regions of the blastoderm. Along the ventral side, staining is seen from about 10 % to 90 % egg length (0 % is the posterior pole) and extends from O % to 15 % VD (0 % is the ventral midline). On the dorsal side, strong expression is limited to 45 % to 85 % egg length and 45 % to 100 % VD (Fig. lA). The staining seen with the antibody is clearly on the cell surface and becomes localized to the apical surfaces as gastrulation begins. At gastrulation, high levels of protein accumu late on the cells of the ventral furrow (mesoderm), dorsal folds (Fig. 2A and B) and the cephalic furrow. During the beginning of germ band extension, low levels of protein accumulate on all cells of the surface of the embryo except for the pole cells. Once the germ band is fully extended, mesoderm staining diminishes but persists longest on those mesodermal cells lying just below the midline.
As neuroblast delamination begins, higher levels of neurotactin accumulate on the neuroblasts compared to the surrounding ectodermal cells (Fig. 2C); the regions of contact between neuroblasts appear to show a marked concentration of neurotactin protein (Fig. 2D). Neurotactin protein continues to accumulate on the cells of the developing central nervous system and disappears from almost all non-neuronal tissues (excep tions are discussed below). After germ band retraction, neurotactin is strongly expressed by the neurons of the CNS, but only by a subset of the neurons of the peripheral nervous system (Fig. 3A). The expression of neurotactin in the PNS appears to be restricted to those sensory cells that send out multiple dendritic projec tions (Fig. 3B). The most intense PNS staining is seen on a small number of sensory cells of the dorsal cluster in the abdominal segments. High levels of expression are also seen on the antenna-maxillary complex and the posterior sensory cones (Fig. 3C and D). Staining of all these peripheral structures is highest on the dendrites themselves and much weaker on the actual cell bodies. Non-neuronal expression is seen on fat body cells and the dorsal vessel (Fig. 3C).
Expression of neurotactin protein and mRNA during Drosophila embryogenesis. Whole mount embryos showing localization of neurotactin protein (A,B) and transcript (C,D) in late blastoderm (A,C) and germ band shortened (B,D) embryos, as revealed by immunocytochemical localization with the BP-106 mAb, and by single-stranded, digoxigenin labeled DNA in situ hybridization. Both the neurotactin protein and transcripts are localized to two restricted regions of the early embryo. Note that the protein is localized to the cell membrane and the transcripts are within the cytoplasm of cells of the same position (arrowhead in A and C). Later in development, neurotactin transcripts and protein are restricted to the CNS, a subset of the PNS including sensory neurons in the head and posterior regions (arrowheads), and cells of the fat body (out of focus). In all photographs, anterior is to the left and dorsal is up. Scale bar: 50 μm.
Expression of neurotactin protein and mRNA during Drosophila embryogenesis. Whole mount embryos showing localization of neurotactin protein (A,B) and transcript (C,D) in late blastoderm (A,C) and germ band shortened (B,D) embryos, as revealed by immunocytochemical localization with the BP-106 mAb, and by single-stranded, digoxigenin labeled DNA in situ hybridization. Both the neurotactin protein and transcripts are localized to two restricted regions of the early embryo. Note that the protein is localized to the cell membrane and the transcripts are within the cytoplasm of cells of the same position (arrowhead in A and C). Later in development, neurotactin transcripts and protein are restricted to the CNS, a subset of the PNS including sensory neurons in the head and posterior regions (arrowheads), and cells of the fat body (out of focus). In all photographs, anterior is to the left and dorsal is up. Scale bar: 50 μm.
Expression of neurotactin protein during gastrulation and neurogenesis. During gastrulation (A,B), neurotactin expression highlights the ventral furrow (arrowhead), dorsal folds (arrow), and cephalic furrow (open arrow), all areas that are undergoing morphogenetic movement. During neurogenesis (C,D), neurotactin expression fades from the mesoderm, and becomes intense on newly formed neuroblasts and their progeny (arrowhead). The protein appears to accumulate at the junctions between neuroblasts (arrow in D; arrowhead marks midline). All embryos are viewed with anterior to the left. Dorsal is up in A and C; B and D are viewed from the ventral surface. Scale bar: (A-C) 50 μm; (D) 10μm.
Expression of neurotactin protein during gastrulation and neurogenesis. During gastrulation (A,B), neurotactin expression highlights the ventral furrow (arrowhead), dorsal folds (arrow), and cephalic furrow (open arrow), all areas that are undergoing morphogenetic movement. During neurogenesis (C,D), neurotactin expression fades from the mesoderm, and becomes intense on newly formed neuroblasts and their progeny (arrowhead). The protein appears to accumulate at the junctions between neuroblasts (arrow in D; arrowhead marks midline). All embryos are viewed with anterior to the left. Dorsal is up in A and C; B and D are viewed from the ventral surface. Scale bar: (A-C) 50 μm; (D) 10μm.
Expression of neurotactin protein during later development. Neurotactin is strongly expressed by cells in the CNS and by a subset of the PNS (arrowhead in A). In B, a higher magnification view shows the strong staining on the dendrites (arrow) of a subset of sensory neurons of the dorsal cluster. The out of focus staining (arrowhead) is from the fat body cell expression. In embryos in which the CNS is condensed (C), neurotactin is expressed in the CNS, fat body (arrow), dorsal vessel and the PNS including the sensory structures of the head and tail (arrowheads). Panel D shows the staining of the posterior sensory cone cells (arrowhead) near the anal plate. Dorsal views with anterior up in A and B; lateral view in C and dorsal view in D, in both of which anterior is to the left. Scale bar: (A) 25 μm; (B) 10 μm; (C) 50 μm; (D) 10 μm.
Expression of neurotactin protein during later development. Neurotactin is strongly expressed by cells in the CNS and by a subset of the PNS (arrowhead in A). In B, a higher magnification view shows the strong staining on the dendrites (arrow) of a subset of sensory neurons of the dorsal cluster. The out of focus staining (arrowhead) is from the fat body cell expression. In embryos in which the CNS is condensed (C), neurotactin is expressed in the CNS, fat body (arrow), dorsal vessel and the PNS including the sensory structures of the head and tail (arrowheads). Panel D shows the staining of the posterior sensory cone cells (arrowhead) near the anal plate. Dorsal views with anterior up in A and B; lateral view in C and dorsal view in D, in both of which anterior is to the left. Scale bar: (A) 25 μm; (B) 10 μm; (C) 50 μm; (D) 10 μm.
In the 3rd instar leg disc, expression can be seen by the developing chordotonal neurons (Fig. 4A). At the white pre-pupa stage, staining of the chordotonal cells is prominent (Fig. 4B). Studies with other neuronal markers indicate that at this time neurons at the distal tip are differentiated and sending axons towards the CNS (Hartsch et al. 1990), but these cells do not express the neurotactin protein. In the eye imaginal disc, neurotactin protein is expressed by the photoreceptor cells and their axons. Neurotactin expression by photoreceptor cells (Fig. 4C) begins further posterior to the morphogenetic furrow than does expression of the 22C10 antigen (Zipursky et al. 1984) or the nervous system specific form of neuroglian (Hartsch et al. 1990), but anterior to the expression of chaoptin (Fujita et al. 1982). During the third instar stage, neurotactin expression in the CNS appears only on the newly dividing neuroblasts and their lineages in subesopha geal segment 1 through abdominal segment 1 (Fig. 4D). Weak staining is also seen in a few lineages in the medial margins of the brain lobes. Previous studies have shown that many additional lineages are also generated in other portions of the brain and more posterior abdominal segments (Truman and Bate, 1988), but these neuroblasts and their progeny do not appear to express the neurotactin protein.
Neurotactin protein expression in imaginal discs and larval CNS. Leg discs from third instar (A) and pre-pupa (B), showing neurotactin expression in the chordotonal cells (arrowheads) but not in other neurons such as those at the distal tip (arrow in B). (C) In the eye disc, expression is seen on the photoreceptors and their axons. (D) In the third instar CNS, neurotactin expression is restricted to the newly proliferative neuroblasts and their lineages in Sl through Al. Anterior is up and the nerve cord is viewed from the ventral surface in D. The morphogenetic furrow is up in C.
Neurotactin protein expression in imaginal discs and larval CNS. Leg discs from third instar (A) and pre-pupa (B), showing neurotactin expression in the chordotonal cells (arrowheads) but not in other neurons such as those at the distal tip (arrow in B). (C) In the eye disc, expression is seen on the photoreceptors and their axons. (D) In the third instar CNS, neurotactin expression is restricted to the newly proliferative neuroblasts and their lineages in Sl through Al. Anterior is up and the nerve cord is viewed from the ventral surface in D. The morphogenetic furrow is up in C.
Biochemical analysis of neurotactin protein
On Western blots using membranes isolated from Drosophila Sl cells or embryos, the neurotactin protein is recognized by the BP-106 mAb (Fig. 5 and data not shown). Its apparent molecular mass was calculated to be 135 x103 Affinity purification of the neurotactin protein from Drosophila embryonic extracts on a BP-106 affinity matrix yielded only a single protein species with the same apparent molecular weight (data not shown). Immunostaining and Western blotting exper iments revealed that the Drosophila Sl cell line endogenously expresses significant levels of the BP-106 antigen. A microsomal membrane fraction from this cell line was used to biochemically characterize the neurotactin protein (Fig. 5). Identical results were obtained using membrane preparations isolated from Drosophila embryos (data not shown).
Biochemical characterization of the neurotactin protein. Membranes were isolated from Sl cells and neurotactin protein was visualized after separation on 7.5 % SDS-polyacrylamide gels and electrotransfer to nitrocellulose filters by Western blotting using a 1/500 dilution of the BP-106 mAb ascites fluid followed by horseradish peroxidase-conjugated secondary antibody and developed with diaminobenzidine. (A) Sl cell membranes (starting material, lane 1) were treated with carbonate, pH 11, as described in the Methods section and separated by centrifugation into a pH 11 soluble (lane 2) and a pH 11 insoluble (lane 3) fraction. Total membrane proteins from Sl cells were solubilized in SOS sample buffer and separated under reducing (lane 4) or nonreducing (lane 5) conditions. Blot B shows the result of a deglycosy)ation experiment. Membrane proteins were treated with TMFS (lane 2) or endoglycosidase H (lane 4). Samples shown in lanes 1 and 3 were incubated in deglycosylation buffer without TFMS or endoglycosidase H. Aliquots of the deglycosylation reactions were analyzed by Western blotting as described above.
Biochemical characterization of the neurotactin protein. Membranes were isolated from Sl cells and neurotactin protein was visualized after separation on 7.5 % SDS-polyacrylamide gels and electrotransfer to nitrocellulose filters by Western blotting using a 1/500 dilution of the BP-106 mAb ascites fluid followed by horseradish peroxidase-conjugated secondary antibody and developed with diaminobenzidine. (A) Sl cell membranes (starting material, lane 1) were treated with carbonate, pH 11, as described in the Methods section and separated by centrifugation into a pH 11 soluble (lane 2) and a pH 11 insoluble (lane 3) fraction. Total membrane proteins from Sl cells were solubilized in SOS sample buffer and separated under reducing (lane 4) or nonreducing (lane 5) conditions. Blot B shows the result of a deglycosy)ation experiment. Membrane proteins were treated with TMFS (lane 2) or endoglycosidase H (lane 4). Samples shown in lanes 1 and 3 were incubated in deglycosylation buffer without TFMS or endoglycosidase H. Aliquots of the deglycosylation reactions were analyzed by Western blotting as described above.
Alkaline washes at high pH have been used to discriminate between integral and peripheral mem brane proteins (Fujiki et al. 1982; Hortsch et al. 1985). Membranes were isolated from Sl cells or from Drosophila embryos, subjected to a high pH treatment at pH 11, and the starting material and the resulting fractions were analyzed by Western blotting for neurotactin protein (Fig. SA). The quality of the separation was judged by silver staining and immuno blotting for other Drosophila membrane proteins. Neurotactin protein was found quantitatively in the membranous pellet fraction (Fig. SA, lane 3). This indicates that the neurotactin protein is tightly associ ated or inserted in the lipid bilayer.
Under nonreducing conditions the neurotactin pro tein had a slightly higher electrophoretic mobility as compared to the completely reduced state (Fig. 5, lanes 4 and 5). Although the difference in electrophoretic mobility is very small and might also be the result of a displacement by a more abundant protein species, this observation could be indicative of one or several intramolecular disulfide bridges resulting in a more compact structure of the unreduced protein.
The deglycosylation experiment shown in Fig. SB established that the neurotactin gene product is a glycoprotein. Complete deglycosylation of the neuro tactin protein by TFMS resulted in a reduction of the molecular mass by 15 x 103 (Fig. SB, lane 2). However, the carbohydrate component of the protein was not sensitive to digestion with endoglycosidase H. It is therefore not a high mannose N-linked type sugar. In contrast, other Drosophila glycoproteins in the same membrane preparations, such as fasciclin I and neuro glian, were sensitive to endoglycosidase H (Bieber et al. 1989; Hortsch and Goodman, 1990).
Molecular analysis of neurotactin protein
The BP-106 mAb was used to screen the 9 to 12 h embryonic Drosophila Agtll cDNA library made by Kai Zinn (Snow et al. 1988). Approximately 2.lx105 recombinant phages were screened and 2 antibody positive plaques were isolated. Both phages contained identical inserts, consisting of a 5’ 3.1 kb and a 3’ 435 bp EcoRI fragment. Since sequencing of these two EcoRI fragments revealed that they did not contain the full open reading frame, the library was rescreened using a [ a-32P]dCTP labeled probe derived from the 3’ 435 bp EcoRI fragment. 12 phages, 4 of which contained longer cDNA inserts, were recovered and analyzed by restriction analysis and sequencing. The largest cDNA insert measured 4.17 kb and consisted of a 471 bp long 5’, a 3261 bp long central, and a 435 bp long 3’ EcoRI fragment (Fig. 6). The only large open reading frame lies entirely within the 3261 bp central EcoRI cDNA frag ment. It encodes for a protein of 846 amino acids with a calculated molecular weight of 92 720 and a theoretical pl of 4.95. Carbohydrate modifications as demonstrated in Fig. SB as well as the acidic pl might account for the discrepancy between the calculated and the apparent molecular weight of the neurotactin protein.
Nucleotide and deduced amino acid sequence of a neurotactin cDNA clone. The nucleotide sequence of a 4.2 kb cDNA clone and the translated sequence of the Drosophila neurotactin protein sequence are presented. The putative transmembrane domain is double underlined. Cysteine residues, that are at homologous sites to cysteine residues in the acetylcholine esterase protein known to form disulfide bridges (MacPhee-Quigley et al. 1986), are connected by single lines. Possible sites for N-linked glycosylation (-N-X-s/,) in the extracellular domain are indicated by -CHO-. Two polyadenylation signals (AATAAA) are underlined in the 3’ untranslated region of the nucleotide sequence.
Nucleotide and deduced amino acid sequence of a neurotactin cDNA clone. The nucleotide sequence of a 4.2 kb cDNA clone and the translated sequence of the Drosophila neurotactin protein sequence are presented. The putative transmembrane domain is double underlined. Cysteine residues, that are at homologous sites to cysteine residues in the acetylcholine esterase protein known to form disulfide bridges (MacPhee-Quigley et al. 1986), are connected by single lines. Possible sites for N-linked glycosylation (-N-X-s/,) in the extracellular domain are indicated by -CHO-. Two polyadenylation signals (AATAAA) are underlined in the 3’ untranslated region of the nucleotide sequence.
Four lines of evidence demonstrate that the cDNA that we have cloned encodes the neurotactin protein. First, the deduced aminoterminal amino acid sequence from our cDNA (Met-Gly-Glu-Leu-Glu-Glu-Lys-Glu Thr-Pro-Pro-Thr-Glu-) is identical to the aminotermi nal protein microsequence data from affinity-purified neurotactin protein (X-Glu-Leu-Glu-Glu-Lys-Glu-Thr Pro-Pro-Thr-Glu-) (M. Piovant, personal communi cation; de la Escalera, 1990). Second, a polyclonal antisera raised against a fusion protein using part of the neurotactin cDNA stains the same embryonic pattern as does the BP-106mAb (see below). Third, the cloned cDNA maps to position 73C on the polytene chromo some, and embryos homozygous for deficiencies that remove this region show no staining with the BP-106 mAb (see below). Fourth, whole mount in situ hybridizations with the cDNA clone show a pattern of transcript accumulation that is consistent with the pattern of protein expression as revealed by the BP-106 mAb (see below; Fig. 1).
The observation that the first 12 amino acids of the open reading frame are identical to the aminoterminal protein sequence derived from the affinity-purified mature neurotactin protein (M. Piovant, personal communication) indicate the absence of an aminotermi nal cleavable signal sequence. A hydrophobicity analy sis reveals only one hydrophobic, potential membrane spanning segment of 22 amino acids dividing the neurotactin protein into a 324 amino acid aminotermi nal domain and a 500 amino acid carboxyterminal domain (double underlined in Fig. 6). This hydro phobic segment probably acts as an internal signal sequence (Wickner and Lodish, 1985) and mediates the SRP-dependent insertion of the neurotactin protein into microsomal membranes in an in vitro transcription-translation assay system as described by Hortsch and Meyer (1988) (data not shown).
Fusion protein constructs using parts of the neurotac tin cDNA were used to map the epitope of the BP-106 mAb to within the first 280 aminoterminal amino acid residues (data not shown). Since the BP-106 antibody reacts with neurotactin-expressing cells or tissues only after detergent treatment, we presume this to be the cytoplasmically exposed part of the protein. Supporting this assumption, an antiserum raised against a fusion protein containing the carboxyterminal portion of the neurotactin protein reacts with neurotactin expressing tissues without detergent pretreatment. Therefore the neurotactin molecule presents a type II integral membrane protein with its aminoterminus facing the cytoplasm and its carboxyterminus pointing towards the extracellular space (Wickner and Lodish, 1985).
The characteristics of the cytoplasmic and the extracellular portion of the neurotactin molecule are very different. The cytoplasmic part of the neurotactin protein is extremely rich in charged amino acids representing 41 % of all amino acid residues. In contrast the content of charged amino acid residues in the extracellular domain is only 20 % . This unusual amino acid composition of the cytoplasmic domain might also explain our observation that it is highly susceptible to proteolytic degradation and the finding by other investigators of multiple degradation products copurify ing with the intact neurotactin protein on affinity matrices (Piovant and Lena, 1988; M. Piovant, personal communication).
The first 300 amino acid residues of the extracellular portion of the neurotactin protein exhibit a strong similarity to serine esterases and to one domain of thyroglobulin. This domain is the only part of the neurotactin molecule that shows a significant sequence homology to known proteins. Further towards the carboxyterminus this sequence homology is less pronounced and barely significant. Fig. 7 shows an alignment of the neurotactin protein sequence to Drosophila glutactin, rat thyroglobulin, and acetylchol inesterase from Torpedo californica. Since neurotactin, like glutactin and thyroglobulin, does not have the serine residue that is the landmark of the active center of serine esterases (MacPhee-Quigley et al. 1985), an enzymatic function of the neurotactin molecule seems unlikely. Although all four proteins are from phylogen etically very distant species and functionally very different, they appear to share a structurally conserved protein domain motif. Four regions in the protein sequences are especially well conserved between them and could therefore be critical for the correct folding of the polypeptide chain. These regions extend from residues 380 to 391, 435 to 442, 490 to 500, and 531 to 544 of the neurotactin sequence (boxed regions in Fig. 7). In addition, many more amino acid residues are shared by all four proteins. Another feature of the neurotactin sequence, underscoring its structural simi larity to the esterase proteins, is the finding of six cysteine residues at roughly equivalent positions to cysteine residues in the acetylcholinesterase protein that are engaged in intracellular disulfide bridges (MacPhee-Quigley et al. 1986). These cysteine residues and their putative disulfide linkages are indicated in Fig. 6 and outlined in Fig. 7. The sequence of the extracellular Drosophila neurotactin domain contains 6 potential N-linked glycosylation sites (Fig. 6). The potential glycosylation site at position 410 is conserved in Drosophila glutactin as well as in acetylcholinester ase of Torpedo californica.
Alignment of homologous amino acid residues for Drosophila neurotactin, Drosophila glutactin, rat thyroglobulin, and acetylcholinesterase from Torpedo californica. Deduced amino acid sequences for Drosophila glutactin (Glu) (Olson et al. 1990), rat thyroglobulin {Thy) (DiLauro et al. 1985), and acetylcholinesterase from Torpedo californica (Ace) (Schuhmacher et al. 1986) were compared to the Drosophila neurotactin sequence (Neu). Amino acid identities to the neurotactin sequence are marked by vertical lines. The active l,ite of acetylcholinesterase is underlined and the serine residue essential for its enzymatic activity is marked by an asterisk. Areas that are especially well conserved between all four protein sequences are indicated by boxes. Those cysteine residues are outlined that are at homologous sites to cysteine residues that are involved in disulfide bridges in the acetylcholinesterase protein (MacPhee-Quigley et al. 1986).
Alignment of homologous amino acid residues for Drosophila neurotactin, Drosophila glutactin, rat thyroglobulin, and acetylcholinesterase from Torpedo californica. Deduced amino acid sequences for Drosophila glutactin (Glu) (Olson et al. 1990), rat thyroglobulin {Thy) (DiLauro et al. 1985), and acetylcholinesterase from Torpedo californica (Ace) (Schuhmacher et al. 1986) were compared to the Drosophila neurotactin sequence (Neu). Amino acid identities to the neurotactin sequence are marked by vertical lines. The active l,ite of acetylcholinesterase is underlined and the serine residue essential for its enzymatic activity is marked by an asterisk. Areas that are especially well conserved between all four protein sequences are indicated by boxes. Those cysteine residues are outlined that are at homologous sites to cysteine residues that are involved in disulfide bridges in the acetylcholinesterase protein (MacPhee-Quigley et al. 1986).
In situ hybridization to polytene chromosomes indicated the presence of a single gene for Drosophila neurotactin on the third chromosome at position 73Cl/2. Drosophila embryos with homozygous de letions [Df(3L)st103 (73A3-4 -74A6); Df(3L)tra (73A4 -74Al-2); Df(3L)stdll (73All-Bl -73Dl-2)) cover ing this region of the genome show no staining with the BP-106 antibody as indicated by a lack of staining of 25 % of the embryos collected from balanced adults (data not shown). The available deletions [Df(3L)st103, Df(3L)tra, and Df(3L)stdll] covering the neurotactin gene are homozygous lethals and extend also over the nearby ab! and dab loci (Henkemeyer et al. 1987; Gertler et al. 1989). Since double mutants in these loci exhibit a strong effect on axonal organization (Gertler et al. 1989) we are presently unable to address the question of a possible nervous system phenotype for neurotactin mutations.
From the neurotactin gene at least three different major mRNA species are transcribed that have dis tinctly different expression profiles during Drosophila embryogenesis (Fig. 8). During the first six hours of Drosophila development the major mRNA species detected by a neurotactin probe derived from the open reading frame, is the smallest of the transcripts and has a size of 3.2 kb (Fig. 8, lanes 1 and 2). This neurotactin transcript is not expressed at later stages of embryogen esis. A second neurotactin mRNA species of 3.65 kb that is also expressed during later stages of develop ment appears to be most abundant between 6 and 15 h of embryonic development (Fig. 8, lanes 3 to 5). The third major neurotactin gene product is the largest of the three transcripts and is about 4.25 kb in size. It is first detectable by Northern blotting after 9 h of embryonic development (Fig. 8, lanes 4 to 7) and is also expressed during the adult stage (data not shown). The size of this mRNA species correlates well with the size of the cDNA presented in Fig. 6. Only the 4.25 kb transcript was recognized on a developmental Northern blot by a [CY-32P)dCTP labeled probe derived from the 3’ 435 bp long EcoRI fragment (Fig. 8B). The cDNA sequence shown in Fig. 6 therefore corresponds to this 4.25 kb late embryonic transcript. Independently from us the group of Dr F. Jimenez has isolated a cDNA clone coding for the medium size mRNA of 3.65 kb (de la Escalera et al. 1990).
Developmental Northern blots for Drosophila neurotactin mRNAs. Total RNA was prepared from staged Drosophila embryos. Aliquots (30 μg) were separated on formaldehyde containing 1 % agarose gels (lanes 1, 0-3h; lanes 2, 3-6h; lanes 3, 6-9 h; lanes 4, 9-12 h; lanes 5, 12-15 h; lanes 6, 15-18; lanes 7, 18-21 h) and transferred to nylon filters. The Northern blot shown in A was probed with a 32P labeled 3.3 kb co RI cDNA fragment covering the entire open reading frame of the neurotactin cDNA. Blot (B) was hybridized to a [cr-32P]dCTP labeled probe derived from the 3’ 435 bp EcoRI fragment that encodes part of the 3’ untranslated region.
Developmental Northern blots for Drosophila neurotactin mRNAs. Total RNA was prepared from staged Drosophila embryos. Aliquots (30 μg) were separated on formaldehyde containing 1 % agarose gels (lanes 1, 0-3h; lanes 2, 3-6h; lanes 3, 6-9 h; lanes 4, 9-12 h; lanes 5, 12-15 h; lanes 6, 15-18; lanes 7, 18-21 h) and transferred to nylon filters. The Northern blot shown in A was probed with a 32P labeled 3.3 kb co RI cDNA fragment covering the entire open reading frame of the neurotactin cDNA. Blot (B) was hybridized to a [cr-32P]dCTP labeled probe derived from the 3’ 435 bp EcoRI fragment that encodes part of the 3’ untranslated region.
Single-stranded, digoxigenin-Iabeled, DNA probes indicate a pattern of transcript accumulation in the embryo that corresponds very well with the protein patterns detected by the BP-106 mAb. Transcripts are first detected just prior to cellularization in the same regions of the blastoderm that are stained by the BP-106 mAb (Fig. 1). Neurotactin transcripts accumulate to high levels in the ventral furrow and dorsal folds during gastrulation. At about 5 h of development, during neurogenesis, hybridization becomes restricted to the developing CNS. After germ band retraction, transcripts are detected in the CNS, a subset of PNS neurons and head and tail sensory structures in close correlation with the patterns seen for the neurotactin protein (Fig. 1). Transcript accumulation is also seen in the fat body and dorsal vessel cells. Hybridization with probes derived from the 435 bp EcoRI fragment specific for the 4.25 kb late embryonic transcript reveals no transcripts in embryos younger than 5 h. Hybridization with this probe is seen in the developing nervous system at the time of neuroblast formation and continues to be localized in the nervous system during later stages of development.
Discussion
In this paper we describe the immunocytochemical localization, biochemical characterization, molecular cloning and sequence analysis of Drosophila neurotac tin, a cell surface glycoprotein with a serine esterase protein domain motif. The neurotactin protein is dynamically expressed during Drosophila embryogen esis. It is first detectable in two restricted regions of the blastoderm embryo. During gastrulation, expression is prominent on the cells of the ventral furrow and dorsal folds. At the time of neurogenesis it is strongly expressed by neuroblasts and other neuronal cells. Later in development, neurotactin is found at high levels in the CNS and a subset of the PNS; at these stages, the protein subsequently becomes largely confined to the nervous system. Immunocytochemical localization shows neurotactin protein at points of membrane apposition, particularly at points of cell-cell contact between neighboring neuroblasts.
During the course of our work, it became clear that the protein that we were studying was identical to the protein initially described by Piovant and Lena (1988). Aminoterminal protein sequence data (M. Piovant, personal communication) confirms that the protein described by Piovant and Lena (1988) is identical to the neurotactin protein described here. Using a mAb (Mab ElC) that is different from our BP-106 mAb, Piovant and Lena (1988) showed a similar expression pattern of the neurotactin protein in Drosophila embryos. There are some minor differences, however, between the two reports. For example, Piovant and Lena (1988) describe a much more uniform expression pattern in the late blastoderm embryo and the PNS, whereas we observe a more localized distribution for both the protein (Fig. lA) and transcript (Fig. lC) at the late blasto derm stage, and for the protein (Fig. 3A and B) and transcript (data not shown) in the PNS. Piovant and Lena (1988) also reported the neurotactin molecule to be a glycoprotein with an apparent molecular mass of 135 x 103 However, we were unable to confirm their finding of other glycoproteins with apparent molecular weights of 100 and 80 ( x 103) that copurify with the intact neurotactin molecule. Since the cytoplasmic domain of Drosophila neurotactin is highly susceptible to proteolytic degradation, it is appears that the lOOx 103 and the 80x 103 protein species observed by Piovant and Lena (1988) represent neurotactin degradation products that were generated during the isolation procedure (M. Piovant, personal communication).
The same authors also reported a crossreactivity of the neurotactin protein with a mAb against human insulin receptor (Piovant and Lena, 1988). The protein sequence determined by us and others (this paper and de la Escalera et al. 1990), however, reveals no similarity between the neurotactin protein and the human insulin receptor (Ullrich et al. 1985; Ebina et al. 1985) or the insulin receptor homolog from Drosophila that was described by Petruzzelli et al. (1986) and Nishida et al. (1986). Why the anti-human insulin receptor mAb B6 displays a crossreactivity to the neurotactin protein remains unknown.
Several laboratories independently generated mAbs recognizing the neurotactin protein (Piovant and Lena, 1988; de la Escalera et al. 1990; this report), and two groups have independently cloned the gene by cDNA expression cloning (de la Escalera et al. 1990; this report). Northern blot analysis reveals three different neurotactin mRNAs: a 3.2kb mRNA which predomi nates during the first six hours of embryogenesis, a second 3.65 kb and a third 4.25 kb mRNA that are also expressed in the adult animal. F. Jimenez and colleagues have isolated a cDNA for the 3.65 kb neurotactin mRNA (de la Escalera et al. 1990), whereas the cDNA reported here is for the 4.25 kb neurotactin mRNA.
With the exception of 4 single nucleotide substi tutions, both the Jimenez group cDNA sequence and our cDNA sequence are colinear up to nucleotide 3540, where their shorter cDNA ends with a poly(A) tail. Our longer transcript continues for another 627 nucleotides until it too is terminated by a poly(A) stretch. Thus, the major difference between the two transcripts appears to be the developmentally dependent usage of two different polyadenylation signals, both of which are underlined in Fig. 6. Three of the 4 nucleotide substitutions between our sequence and the one reported by de la Escalera et al. (1990) are in the coding region resulting in a glycine to leucine exchange at amino acid residue 250, a serine to threonine change at position 556, and a change from phenylalanine to tyrosine at residue 568. The fourth nucleotide differ ence is a T to A exchange at nucleotide position 3187 in the 3’ untranslated region of the cDNA. Since the size and the amino acid sequences of the proteins encoded by the two different cDNAs are almost identical, we assume that these differences represent polymorphisms between the two fly strains used for the construction of the two different cDNA libraries.
The most interesting feature of the neurotactin protein sequence is the homology of the extracellular protein domain to serine esterases and thyroglobulins. The sequence similarity between serine esterases and thyroglobulin has been observed and discussed by a number of authors (Schuhmacher et al. 1986; Swillens et al. 1986; Lockridge et.al. 1987) and has been explained by the incorporation of a redundant copy of the acetylcholinesterase gene into an archaic thyroglobulin precursor gene (Swillens et al. 1986; Mori et al. 1987). The incorporation of the serine esterase protein domain into two different insect proteins, neurotactin and glutactin, must have been an unrelated event to the generation of the thyroglobulin proteins. Thyroid organs developed first in fishes 400 to 450 million years ago, long after the split of vertebrates and invertebrates (Fujita, 1980; Mori et al. 1987). Furthermore, compari son of the Drosophila neurotactin protein sequence to acetylcholinesterase sequences from different species indicates that the extracellular domain of neurotactin is not more closely related to several serine esterases from Drosophila (Hall and Spierer, 1986) than to those from other species such as Torpedo californica, an electric ray (Schuhmacher et al. 1986). This suggests that the neurotactin gene is not the result of a recent gene duplication and fusion event but rather has accumu lated mutational amino acid changes over a long period of time.
It has been suggested that thyroglobulin has only retained the three-dimensional structure of the serine esterase protein domain but not its enzymatic activity (Swillens et al. 1986; Mori et al. 1987). All serine esterases share a conserved string of amino acids that form the active center of the enzyme and includes the serine residue that is essential for the esterase activity (Sikorav et al. 1987). Neither neurotactin, glutactin nor thyroglobulin has retained either the crucial serine residue or the block of amino acid residues constituting the active side (Fig. 7). However, certain features important for the three-dimensional structure of the protein domain such as disulfide bridges, glycosylation sites and the hydrophobicity-hydrophilicity profile are well conserved between all four proteins (MacPhee Quigley et al. 1986; Swillens et al. 1986; Olson et al. 1990; this report). The reason for the structural conservation of this protein domain is currently unknown.
It is interesting to note, however, that acetylcholin esterase as well as thyroglobulin form homo-dimers and -multimers (Massoulie and Bon, 1982; Lissitzky et al. 1975). It is therefore tempting to speculate that neurotactin might also form homo- or hetero-mul timers. If these interactions could take place between neurotactin molecules and/or their putative ligand molecules expressed on different cells, neurotactin might act as a cell recognition and adhesion molecule. The cell surface expression of the neurotactin molecule, especially at regions between cells and at points of cell-cell contact, would support this hypothesis. Cell transfection experiments (Hartsch et al. unpublished data; Barthalay et al. 1990) using the Drosophila S2 cell line (that does not endogenously express the neurotac tin protein) and a neurotactin cDNA gave no indication for a homophilic adhesion function for the neurotactin protein. Preliminary evidence from the two groups (Hartsch et al. unpublished data; Barthalay et al. 1990) suggests, however, that neurotactin-expressing cells may bind to a heterologous ligand expressed by many primary Drosophila embryonic cells. Drosophila neuro tactin might, for example, interact with a different molecule having a serine esterase like protein domain. In such a scenario, neurotactin might be a potential cellular receptor for the recently discovered glutactin molecule. Glutactin is a sulfated extracellular matrix glycoprotein with an apparent molecular weight of 155x 103 (Olson et al. 1990). It consists of an aminoter minal domain of approximately 600 amino acids that is structurally and on the sequence level related to serine esterases, thyroglobulin and the extracellular domain of neurotactin (see Fig. 7) and a highly charged carboxy terminal domain of about 400 amino acids that is extremely rich in glutamine and glutamic acid residues. Glutactin was initially isolated from the Drosophila Kc cell line and is found in the Drosophila embryo in the basement membrane surrounding the gut, somatic muscles, brain, sensory organs and especially the ventral nerve cord (Olson et al. 1990). Further experiments will be necessary to test whether the neurotactin and the glutactin proteins can directly interact with one another.
It is intriguing that the serine esterase protein domain was adopted by functionally very different proteins during evolution. Glutactin, neurotactin and thyro globulins may not be the only members of this group of proteins that share this novel serine esterase protein domain motif. Common to all proteins is the extracellu lar location of the serine esterase domain and the association of the protein either with the nervous or the endocrine system. For example, antigens that cross react with anti-thyroglobulin antibodies have been detected in the nervous systems of other species such as earthworms (Marcheggiano et al. 1985). Once more proteins with this structural protein domain are discovered, it might be better understood why this domain was incorporated into such different molecules such as cell surface membrane proteins, extracellular matrix proteins, hormone precursors, and serine esterases, and moreover, what the functional capacities of this new structural protein motif are.
ACKNOWLEDGEMENTS
We thank Ursula Weber for the chromosome in situs, Karen Jepson-Innes for tissue in situs, Brigitte Altenberg (EMBL, Heidelberg) for help with sequence comparison programs, Dianne Fristrom for help with the imaginal disc procedure, Marek Mlodzik for providing the eye disc shown in Fig. 4C, and Rolf Bodmer for helpful discussions concern ing the PNS pattern of expression. We give special thanks to Michel Piovant and colleagues (Marseille, France), and Fernando Jimenez and colleagues (Madrid, Spain), for sharing their unpublished results with us. Supported by NATO and DFG postdoctoral fellowships to M. H., NIH postdoctoral fellowship to A.J.B., and NIH grant NS18366 to C.S.G. who is an investigator with the Howard Hughes Medical Institute.
References
Note added in proof
The EMBL Data Bank Library accession number for the neurotactin mRNA from Drosophila melanogaster is X54999.