Helix pomatia lectin, an inducer of Drosophila immune response, binds to hemomucin, a novel surface mucin.

We describe the isolation and initial characterization of hemomucin, a novel Drosophila surface mucin that is likely to be involved in the induction of antibacterial effector molecules after binding a snail lectin (Helix pomatia A hemagglutinin). Two proteins of 100 and 220 kDa were purified from the membrane fraction of a Drosophila blood cell line using lectin columns. The two proteins are products of the same gene, as demonstrated by peptide sequencing. The corresponding cDNAs code for a product that contains an amino-terminal putative transmembrane domain, a domain related to the plant enzyme strictosidine synthase, and a mucin-like domain in the carboxyl-terminal part of the protein. The gene is expressed throughout development. In adult flies, high expression is found in hemocytes, in specialized regions of the gut, and in the ovary, where the protein is deposited onto the egg surface. In the gut, the mucin co-localizes with the peritrophic membrane. The cytogenetic location of the gene is on the third chromosome in the region 97F-98A.

Insects respond to microbial infections by the production of effector molecules, initially characterized in insects such as Hyalophora cecropia, Sarcophaga peregrina, and Apis mellifera (1)(2)(3), which, due to their size, were ideally suited to allow the biochemical characterization of these molecules. Since then, similar and new antibacterial peptides have been discovered in Drosophila melanogaster (4 -7), where the structure and function of the corresponding genes can be studied using the genetic and molecular techniques available in this organism (8,9). Much less is known, however, about the molecules involved in the initial recognition events leading to the induction of an immune response and about the regulation of hemocyte traffic in insects.
In the past few years, a number of glycoproteins collectively named surface mucins have been described in the vertebrate immune system (10). They participate in the initial attachment of leukocytes during inflammatory responses, which finally leads to the migration toward the site of inflammation (11). This function is facilitated by a specialized structure common to surface mucins, i.e., highly O-glycosylated protein domains separated by proline residues (12) that inhibit the formation of a globular structure. This leaves the sugar moieties exposed above the cell surface to interact with their counter-receptors on neighboring cells, the so-called selectins, a specialized type of lectins. Cell surface mucins, present on endothelial walls of blood vessels and high endothelial venules or on the surface of leukocytes and lymphocytes, are thus involved in the correct trafficking of blood cells. Tissue-or cell-specific glycoforms have been implied in this process as a means of directing blood cells to specific tissues (11,13).
Amongst leukocyte mucins, CD43 (sialophorin, leukosialin, and leukocyte sialoglycoprotein) was one of the first molecules to be described (14). It is the major leukocyte surface protein recognized by a snail lectin (Helix pomatia A hemagglutinin, H.p. 1 lectin) and is present on the surface of T-cells but is also found on neutrophils, monocytes, certain B-lymphocytes, and platelets (15). Incubation of T-and B-lymphocytes with antibodies against CD43 enhances in vitro proliferation (16). An important function of this mucin in the immune response is also suggested by immune deficiencies in patients suffering from the Wiskott-Aldrich syndrome, where expression of CD43 is absent or reduced (17).
Heterologous lectins have been shown to label structures on the cell surfaces of Drosophila hemocytes (18,19), giving rise to a phenotype that was described as "speckled." The number of speckled cells increased in flies after immune stimulation with heterospecific implants as well as in flies with melanotic tumors (19). These and other findings (20) suggest that the proteins that are recognized by lectins participate in immune responses.
We found that H.p lectin binds to Drosophila blood cells, thereby inducing an antibacterial response. Here we describe the isolation and characterization of a Drosophila mucin, which binds this lectin and may be involved in the induction of the immune response in this organism.  (21), and analyzed under the fluorescence microscope. In order to allow internalization of H.p. lectin, cells were incubated without sodium azide for 20 min, washed twice with PBS containing 200 mM N-acetylgalactosamin (GalNAc), and analyzed under the fluorescence microscope. Labeling of intracellular proteins was performed as described (22).

Flies-Canton
Agglutination Assays-Drosophila culture cells from a medium dense culture were incubated with H.p. lectin (Boehringer and Sigma) at a concentration of 10 g/ml and assayed after 1 h for agglutination of the cells under the microscope.
Induction Assay for the Cecropin Genes-The induction of the cecropin genes was assayed by RNase protection assays as described (4). Inhibition of the induction was assayed by RNA slot blot assays using 10 g of total RNA/slot. The inhibiting sugar GalNAc (200 mM) or the water used to dissolve the sugar (see Fig. 1B, contr.) was added to 2 ml of a densely grown culture of mbn-2 cells 10 min before the addition of H.p. lectin at a concentration of 20 g/ml. After 90 min, cells were harvested, and RNA was prepared and analyzed.
Surface Labeling of mbn-2 Cells-5 ϫ 10 7 cells were washed twice in PBS and finally suspended in 1 ml of PBS. They were then mixed with 40 g of lactoperoxidase (2 mg/ml stock solution) and 1 mCi of Na 125 Iodine. H 2 O 2 was added to a final concentration of 0.001%. The cells were afterwards lysed, and proteins were isolated as described earlier (14).
DNA and RNA Blots-DNA and RNA blots were performed according to conventional protocols (24). Ribosomal RNA and a 0.24 -9.5kilobase ladder (BRL) were used as molecular weight standards for RNA blots. Poly(A) ϩ RNA was purchased from Clontech. Hybridization signals were quantitated with a PhosphorImager (Molecular Dynamics).
DNA Sequencing-The inserts of the cDNA clones were sequenced with the chain termination method, using the Sequenase kit (U. S. Biochemical Corp.) on double-stranded templates and by automated sequencing. Unidirectional deletions were made from the XbaI and HindIII sites in the pBluescript SKϪ vector using the protected SacI and KpnI sites by digestion with Exonuclease III and S1 nuclease, using the Erase-a-Base deletion kit (Promega) as recommended by the supplier. Some of the sequencing was performed using radioactive methods; most of the sequence was read on an automated sequencer using fluorescent labeled primers or synthetic primers and fluorescence-labeled nucleotides in the chain termination method (Applied Biosystems).
Affinity Purification on a Lectin Column-Cells from 500 ml of a densely grown culture were spun down, washed in PBS, and brought up in a lysis buffer (0.1 M sodium phosphate buffer, pH 7.5, containing 0.5% Nonidet P-40, 6 ϫ 10 Ϫ2 trypsin inhibiting units/ml of aprotinin and 1% epsilon amino caproic acid), lysed for 5 min on ice, and centrifuged twice for 10 min in a table top centrifuge. The supernatant was transferred to a column (1 ml of the lectin bound to CNBr-activated Sepharose according to the instructions of the supplier) and bound to the column on a tilting table. The column was washed in 40 ml of the lysis buffer, followed by a wash with 200 ml of lysis buffer without protease inhibitors. The bound proteins were eluted with a solution of 200 mM GalNAc in the last washing buffer with some traces of bromphenol blue to follow the elution process.
Preparation of Antisera with Specificity for the Lectin-binding Proteins-Antisera with specificity for the 100-and 220-kDa lectin-binding proteins were raised against the two proteins that after affinity purification were separated on a SDS-polyacrylamide gel and cut out from that gel. The gel slices were homogenized and used for immunizing rabbits according to standard procedures (21).
Electrophoretic Techniques-SDS-polyacrylamide gel electrophoresis on a Hoefer Mighty Small electrophoresis unit was essentially according to Laemmli (25). Molecular weights were determined using Bio-Rad high range or broad range molecular weight markers. The proteins were blotted to a nitrocellulose membrane (Amersham Corp.) as described (22). The blotting efficiency was determined by staining the blot with Ponceau S (26). In Western blots, the peroxidase conjugated H.p. lectin (Sigma) was used at a concentration of 50 ng/ml. In preparations used for protein sequencing, the 100-kDa protein was blotted for 8 h under the indicated conditions (26), whereas the 220-kDa protein had to be blotted much longer (approximately 60 h) in order to obtain sufficient transfer.
Protein Analysis-The band of interest was excised from the nitrocellulose and processed for internal amino acid sequence analysis as described (26) with modifications. Briefly, in situ proteolytic cleavage was done using 0.5 mg of trypsin (Promega, Madison, WI) in 25 ml of 100 mM NH 4 HCO 3 (supplemented with 0.3% Tween 80) at 37°C for 3 h. The resulting peptide mixture was reduced and S-alkylated with, respectively, 0.1% ␤-mercaptoethanol (Bio-Rad) and 0.3% 4-vinyl pyridine (Aldrich) and fractionated by reversed phase HPLC. An enzyme blank was done on an equally sized strip of nitrocellulose. HPLC solvents and system configuration were as described (27), except that a 2.1-mm Vydac C4 (214TP54) column was used with gradient elution at a flow rate of 100 ml/min. Identification of Trpcontaining peptides was done by manual ratio analysis of absorbances at 297 and 277 nm, monitored in real time using an Applied Biosystems (Foster City, CA) model 1000S diode array detector (28). Fractions were collected by hand, kept on ice for the duration of the run, and then stored at Ϫ70°C before analysis.
Peptides were analyzed by a combination of automated Edman degradation and matrix-assisted laser desorption mass spectrometry; details about this combined approach, including mass-aided post-chemical sequencing routines, can be found elsewhere (27)(28)(29). Mass analysis (on 2% aliquots) was carried out using a model Voyager RP matrix-assisted laser desorption instrument (Vestec/PerSeptive, Framingham, MA); the matrix was ␣-cyano-4-hydroxy cinnamic acid (Linear Science, Reno, NV). Every sample was analyzed twice in the presence and the absence of a calibrant (25 femtomoles of a control peptide) as described (28). Chemical sequencing (on 95% of the sample) was done using a model 477A instrument (Applied Biosystems) with "on-line" analysis (120A HPLC system with 2.1 ϫ 220 mm of phenylthiohydantoin C18 column; Applied Biosystems). Instruments and procedures were optimized for femtomole level phenylthiohydantoin amino acid analysis as described (30).
Immunocytochemistry-Adult flies were fixed in 4% paraformaldehyde for at least 3 h. Serial sections were cut at a thickness of 10 m in a Leitz cryostat. Immunostaining was performed with the peroxidaselabeled H.p. lectin at a concentration of 50 ng/ml essentially as described (31).
In Situ Hybridizations-In situ hybridizations were performed essentially as described (22) except that the probes used were labeled with digoxigenin using a digoxigenin PCR labeling kit (Boehringer Mannheim). The primers were chosen to give a probe size of approximately 200 base pairs.
Cloning of the PCR Fragments Coding for Hemomucin-Three of the peptide sequences obtained from the 100-kDa lectin-binding protein were chosen for design of primers for the PCR. The sequences of the corresponding primers were: AACAACGAGATCTACACCGG(C/T)ATC-CACGG(C/T)GG(C/T)GAGGT(C/G)ATCAA (derived from the tryptic peptide T24, see Table I), TTCGT(C/G)CACTTCGT(C/G)GG(C/T)CA-CATGGAGTCCATCACCGT(C/G)CTGGCCCC (T40), and GAGGT(C/ G)TTCGT(C/G)GACGG(C/T)CTGCCCGG(C/T)CTGCCCGACAACCTG-ACCCC (T53). For cloning the lectin-binding protein, cDNA was first synthesized using an oligo(T) primer extended by a sequence known to be rare in eukaryotic genomes (32): (T) 18 CGAAGAGCTCCTTAAGCT (EXT18). The cDNA synthesis and the PCR was based on a GeneAmp RNA PCR kit (Perkin-Elmer). The conditions used for the PCR were according to the manufacturer's instructions except for the use of the EXT18 primer and the 60-min duration for the cDNA synthesis. The PCR was performed with cycles for 1 min at 95°C, 1 min at 56°C, and 2 min at 70°C. All three combinations of the primers with the EXT18 primer were used in the PCR reaction, diluted afterwards 1:1000, and reamplified under the same condition with both of the other primer combinations. Two reactions were run for the reamplification for 25 and 35 cycles, respectively. The products, which appeared after 25 cycles, turned out to be the correct ones after subcloning into a pCR TM II vector (InVitrogen) and sequencing. Smaller bands appearing after 25 and much stronger after 35 cycles turned out to be artefacts.
From the two peptide sequences that are unique for the 220-kDa protein, oligonucleotides were designed as for the 100-kDa protein and the reverse complement of both primers). PCR was performed between each primer and the reverse complement of the other primer because the relative orientation was unknown. Only one combination resulted in a product of approximately 200 base pairs that was subcloned and sequenced as described for the 100-kDa LBP. One of the subcloned PCR fragments was used to screen the lambda ZAPII libraries to obtain plasmids pDMu-105 and pDMu-12 after in vivo excision.
cDNA Libraries-Two cDNA libraries were screened, one from flies that have been immunostimulated by the injection of bacteria (5) and the other a commercially available library (Clontech). Both libraries were derived from CantonS fly RNA and were constructed with the lambdaZAP II vector (33).

A Lectin Induces the Drosophila Cecropin Genes-
The Drosophila cell-line mbn-2 (34) was initially established from a mutant strain with a blood tumor phenotype and has been used as a model for the induction of the immune response (35). In addition to microbial substances (lipopolysaccharide, laminarin, and flagellin), which were shown to induce the cecropin mRNA (35), we also found H.p. lectin, a known inducer of T-lymphocytes, to induce mbn-2 cells to produce the mRNA for the antibacterial peptide, cecropin A1 (Fig. 1A). The H.p. lectindependent induction could be inhibited by GalNAc, the sugar known to inhibit H.p. lectin binding to human leukocytes (Fig.  1B), showing that the induction is not due to nonlectin contaminants in the H.p. lectin preparation. The remaining background induction seen after inhibition with GalNAc is probably due to a contamination in the sugar preparation because it was observed after the addition of the sugar alone (Fig. 1B). In addition, H.p. lectin was shown to agglutinate mbn-2 cells, suggesting the presence of lectin-binding proteins on their surface (data not shown). Like the induction, the agglutination could be inhibited by the addition of GalNAc.
Subcellular Distribution of the Lectin-binding Proteins-Most of the mbn-2 cells that were stained with the FITCconjugated lectin, showed a typical ring-like cell surface labeling. Thin, filopodia-like structures were also clearly labeled by the lectin (Fig. 2A). The intensity of the signal differed between individual cells. Because the mbn-2 cells have not been subcloned since their first isolation, the difference in lectin staining might be due to the presence of different cell-types or to cells of different maturation stage in the culture. Some cells do not show a uniform distribution of lectin binding sites on their surface; instead, they contain spots of high lectin staining (Fig.  2B, arrowhead).
When mbn-2 cells were permeabilized prior to lectin staining, a strong signal was obtained from intracellular granules (Fig. 2H). In contrast to the surface labeling, this signal was observed in all cells.
Hemocytes obtained from Drosophila larvae showed a similar staining pattern, with surface labeling on some of the cells and a staining of intracellular granules of different size in all cells (Fig. 2, C and D). To test whether the lectin can be taken up by the cells via internalization of the lectin-binding proteins, mbn-2 cells were incubated with fluorescein-conjugated lectin, with and without sodium azide, followed by washing with Gal-NAc. As seen in Fig. 2G, the homogeneous surface labeling observed in the presence of azide (Fig. 1, A and B) is almost completely abolished by GalNAc, whereas in the absence of azide, a signal was detected in some of the granules that was resistant to sugar treatment (Fig. 2E). These granules probably correspond to the large endocytic vacuoles described in mbn-2 cells (34). Thus, lectin bound to the surface is actively taken up by mbn-2 cells.
Purification and Cloning of the Lectin-binding Proteins-The major lectin-binding proteins were identified in Western blots using total extracts from mbn-2 cells as well as whole Drosophila adults. In the cells, one lectin-stained band of 100 kDa was identified. In animal lysates, an additional band of approximately 80 kDa was stained (Fig. 3A). Thus, H.p. lectin seems to be a sufficiently specific tool to allow the isolation and characterization of the glycoprotein(s) that mediate the observed immune induction.
Two proteins were obtained from detergent lysates of the mbn-2 cells by a single affinity purification, using H.p. lectin as a ligand (Fig. 3B) and subsequent elution with GalNAc, one with the expected molecular mass of 100 kDa and a second 220-kDa protein. Although the 220-kDa protein from mbn-2 cell extracts was not detected in immunoblotting (Fig. 3A), a weak reaction with lectin could be seen with affinity purified proteins (not shown). Because both proteins were labeled after surface iodination of the cells (data not shown) and thus could have been responsible for H.p. lectin-dependent induction, we decided to obtain specific probes for both of them. Peptide sequences obtained from cyanogen bromide and tryptic fragments are shown in Table I. Surprisingly, three of the five  -100), and an antiserum specific for the 220-kDa LBP (anti-220). Immune complexes were precipitated using protein A-Sepharose. As a control, the 100-kDa LBP was immunoprecipitated from a lysate of surface iodinated cells (anti-100 iod). The arrows indicate the positions of the 220-and 100-kDa proteins.

TABLE I Peptide sequences of the two lectin-binding proteins
Amino acid sequences were obtained by sequencing cyanogen bromide (C1 and C2) and tryptic fragments (T) of the 100-kDa and the 220-kDa lectin-binding proteins.  peptide sequences obtained from the larger protein were identical to three sequences obtained from the 100-kDa protein.
The information from the peptide sequences was used to design three specific oligonucleotides. Because the relative position of the primer binding sites along the gene was not known at this stage, we used all possible combinations between the three primers and the oligo(T) primer to achieve a putative nested configuration (see "Material and Methods"). One primer combination produced a PCR fragment that was used to obtain the cDNAs pDMu-105 and pDMu-12 (see Fig. 5A), which comprise the complete open reading frame. The predicted protein sequence is shown in Fig. 5C. It contains all tryptic peptides obtained from the 100-and 220-kDa lectin-binding proteins. From this, we conclude that the 220-and 100-kDa proteins are encoded by the same gene. The difference in molecular mass is most likely due to post-translational modifications. The two proteins differ due to phosphorylation of the larger one in uninduced cells (Fig. 4). Thus the two proteins are biochemically different forms. That excludes the possibility of the larger one being a dimer of the 100-kDa protein. We decided to use the name hemomucin for this newly discovered Drosophila protein.
Structure of Hemomucin-The predicted molecular mass of hemomucin is 64 kDa. The methionine at nucleotide position 102 is likely to be the initiation codon as the surrounding nucleotides fit the requirements for an initiation codon (36) and all potential open reading frames found upstream of position 102 use different codon usage compared with the remaining part of the open reading frame, which shows a typical Drosophila codon usage. The protein consists of four domains (Fig. 5B): an amino-terminal hydrophobic sequence that may act as a signal sequence and/or a transmembrane segment, a central part with similarity to the plant enzyme strictosidine synthase, a mucin domain, and a carboxyl-terminal domain. The aminoterminal hydrophobic sequence is followed by a potential signal peptidase cleavage site (37). However, this site is obviously not cleaved in the isolated protein, as evidenced by the sequence of one of the peptide fragments (C1 in Table I). This is also in agreement with the observed localization of the protein on the cell surface, where the hydrophobic sequence is likely to serve as a membrane anchor.
A comparison with protein sequences of the data bases revealed a low but significant similarity to the plant protein strictosidine synthase (38,39), a key enzyme in the production of plant indole alkaloids. This similarity is 31% for identical amino acids alone and 79% including similar amino acids. Within this region lies one copy of a conserved sequence (YWTD) described in a number of receptors that are endocytosed and recycle to the cell surface (40 -42). In these receptors, usually several repeats of this sequence are found. They are thought to be important for the uncoupling of ligands from their receptors by the acidic pH of the endosomes.
The mucin domain consists of typical stretches of polythreonine, separated by proline residues, near the carboxyl-terminal end of the predicted protein (Fig. 5C). These stretches show  (Table I)  similarities to a number of vertebrate mucins from different tissues. Two proteins from Drosophila also contains mucin domains: the salivary glue protein, which mediates the attachment of the larvae before pupation and, interestingly, a scavenger receptor isolated from Schneider SL2 cells (43). The carboxyl-terminal domain as well as a domain amino-terminal of the mucin-domain show no obvious similarity to other proteins.
Localization of the Gene Coding for Hemomucin on the Drosophila Genome-A hemomucin-specific cDNA fragment was used as a probe to cytogenetically localize the corresponding gene on polytenic chromosomes. Only one signal was obtained per Drosophila chromosome set in corresponding chromosome positions, suggesting that a unique single copy gene is coding for hemomucin. The signal was localized to the right arm of the second chromosome at the position 97F-98A (Fig. 6).
Expression of Hemomucin-To investigate the developmental expression of hemomucin, hybridization experiments were performed on Northern blots containing mRNA of embryonic, larval, and adult RNA (Fig. 7). The gene coding for hemomucin is expressed at all developmental stages, at levels similar to a housekeeping gene, the ribosomal protein gene Rp49. This indicates a basic function for hemomucin throughout all developmental stages.
Tissue Distribution of Hemomucin-In order to investigate the expression of hemomucin in adult Drosophila tissues, sections of flies were stained with H.p. lectin, and in parallel, in situ hybridization to mRNA was performed using probes specific for the hemomucin gene. Only tissues that gave rise to a signal with both methods are shown ( Fig. 8 and 9). Using H.p. lectin as a probe, additional signals were obtained in the Malpighian tubules, most likely due to the additional band seen in Western blots of whole fly lysates (Fig. 3A). Immunocytochemistry was also performed with an antiserum against hemomucin. Essentially, the same tissues were stained with both methods, although results with the lectin gave stronger signals in both Western blots and tissue sections and are shown here.
In sections of adult flies, H.p. lectin bound to individual cells associated with the fat body, presumably hemocytes (Fig. 8A) because similar signals have been observed with Drosophila calpain, another protein known to be expressed in hemocytes (22). In addition, strong staining was observed in the gut, namely in the cardia (Fig. 8B), and in some sections of the midgut where the inner surface of the ventriculus epithelium was labeled (Fig. 8C). This staining pattern corresponds to the location of the peritrophic membrane and could be confirmed by in situ hybridization in which the transcripts were also de-tected in the cardia (Fig. 9A), where the peritrophic membrane is synthesized, and in specific regions of the ventriculus (Fig.  9B), including the part posterior to the cardia (Fig. 9A). Individual sections of the ventriculus showed different intensity in expression of the transcript (Fig. 9B).
A strong signal was also observed in the ovary, which appears to be associated with the egg shell because freshly laid eggs are stained on the surface with H.p. lectin (Fig. 8D). In situ hybridizations show a signal in the follicle cells (Fig. 9C) that are producing the chorion layer on the egg surface. DISCUSSION We describe here the isolation and initial characterization of hemomucin, a novel cell surface molecule from a Drosophila hemocyte cell line. Hemomucin was isolated using H.p. lectin, a snail lectin that enhances induction of T-lymphocytes and induces the antibacterial response in mbn-2 cells. The induction is lectin-dependent because it could be inhibited by addition of the H.p. lectin-specific sugar. Although hemomucin is the only hemocyte protein that could be detected in Western blots and affinity purification using H.p. lectin, minor undetected glycoproteins or glycolipids might have led to the observed effects. Nevertheless, hemomucin is the major target for H.p. lectin on the cell surface and a likely candidate for mediating the lectin-dependent induction.
The isolation and characterization of the corresponding cDNA predicts a protein that contains typical mucin domains (10 -12, 43). Hemomucin shows no sequence similarity to CD43, the major surface mucin on human T-lymphocytes, which is recognized by H.p. lectin (44). The cellular location of hemomucin in mbn-2 cells and the presence of an amino-terminal hydrophobic domain in the mature protein suggest that it is attached to the membrane of these cells by this domain, which serves as a transmembrane anchor.
Two proteins of different molecular mass (100 and 220 kDa) were recognized by the lectin, similar to other surface mucins that exist in different glycoforms (45). Because the two proteins contained identical tryptic peptides, we conclude that they are encoded by the same gene and that the size difference is due to post-translational modifications. One such difference between the two proteins was shown to be a phosphorylation of the larger protein. Additional modifications, such as glycosylation, may account for the apparent size difference. The 100-kDa protein might be a precursor of the 220-kDa protein, which could be produced in a similar way as described for episialin, another surface mucin, which is modified by constitutive endocytosis (see below) and concomitant glycosylation in the cell (45). Incorporation of phosphorylated sugars might explain the appearance of phosphate in the 220-kDa protein. The absence of this protein in some of the immunoblots is most likely due to ineffective transfer during the blot (see "Material and Methods").
Hemomucin appears to be stored in intracellular granules that are detected in all mbn-2 cells as well as in all hemocytes in blood smears. However, only a subfraction of hemocytes displayed a typical cell surface labeling, whereas some cells had no detectable hemomucin on their surface. These findings suggest that although hemomucin is produced in all hemocytes, the location on the cell surface is regulated in some way and restricted to specific cells or specific cellular differentiation stages. In mbn-2 cells, the surface label is transported to endocytic vacuoles after 30 min (Fig. 2E) that appear to increase in size after 60 min (not shown). In this regard, hemomucin behaves similar to episialin (45).
In sections of adult flies, hemomucin is localized in other tissues apart from hemocytes, namely the gut and the ovary. The protein in the gut appears to be synthesized by transcripts in the cardia, a specialized structure of the midgut that is known to produce the peritrophic membrane (46) and by more posterior parts of the midgut. In the ventriculus, the lectin signal co-localizes with the peritrophic membrane where hemomucin might play a similar role as the mucous layer that protects the vertebrate gut epithelium from bacterial (47) and helminth (48,49) invasion. In the ovary, hemomucin is found in mature eggs. The signal is associated with the outermost layer covering the egg as shown in Fig. 8D, which indicates that the protein is located on the chorion surface. Follicle cells that produce the chorion in the egg chamber give rise to a signal in in situ hybridizations (Fig. 9C). Both the signal in the gut and in the ovary suggest that hemomucin exists in a secreted form, in addition to its cellular form. Two mechanisms might account for the presence of the protein in different subcellular compartments: 1) the signal peptidase acts on hemomucin that is secreted in all tissues except hemocytes where the signal peptide remains covalently bound to the protein and serves as a membrane anchor, and 2) the signal peptide is not cleaved in which case hemomucin is a class II membrane protein (50). The latter explanation is more compatible with our results. In addition, two prolines at positions Ϫ1 and Ϫ2, which are present in hemomucin, are rarely found in functional cleavage sites (37). A mechanism that may account for the secretion of the protein is the formation of so-called microparticles described in vertebrate platelets (51,52) leading to the shedding of membrane proteins (53). The aggregates seen in Fig. 2B might correspond to such microparticles.
The similarity of part of the hemomucin protein sequence to a known enzyme that catalyses a condensation reaction in plants is significant, although the functional relevance is not clear. The two substrates of the plant enzyme belong to the monoterpenes and the indole amines, respectively. Because no substances that potentially interact with the mucin in Drosophila are known, the possibility of an enzymatic activity remains to be determined.
Interestingly, a scavenger receptor (dSR-CI), which was recently isolated from Drosophila Schneider cells also contains a mucin-like domain (43). dSR-CI has a broad polyanionic binding specificity that is likely to be involved in a number of hemocyte functions including immune reactions (43). Similarly our results obtained with hemomucin are compatible with a function in the induction of immune responses in insects. This induction may require a soluble lectin, the function of which was mimicked by H.p. lectin in our studies. In this model, Drosophila lectin would mediate attachment of microorganisms that invade the hemolymph by cross-linking them to the hemocyte surface using hemomucin as a hemocyte receptor. dSR-CI and hemomucin might be members of a diverse group of cell surface molecules mediating immune recognition in Drosophila either by directly interacting with foreign objects or by binding soluble recognition molecules. Further studies will determine the role of hemomucin in the protection against microbial and parasite invasion of gut and hemolymph using a genetic approach.