Modulation of Neural Carbohydrate Epitope Expression in Drosophila melanogaster Cells*

Neural pathways in invertebrates are often tracked using anti-horseradish peroxidase, a cross-reaction due to the presence of core α1,3-fucosylated N-glycans. In order to investigate the molecular basis of this epitope in a cellular context, we compared two Drosophila melanogaster cell lines: the S2 and the neuronal-like BG2-c6 cell lines. As shown by mass spectrometric and chromatographic analyses, only the BG2-c6 cell line expresses α1,3/α1,6-difucosylated N-glycans, a result that correlates with anti-horseradish peroxidase binding. Of all four α1,3-fucosyltransferase homologues previously identified, the core α1,3-fucosyltransferase (FucTA; EC 2.4.1.214) is expressed in the neuronal cell line as well as throughout fly development and in heads and bodies of flies of both sexes. This pattern is distinctive in comparison with the expression of the other three α1,3-fucosyltransferase homologues (FucTB, FucTC, and FucTD). Furthermore, only transfection of FucTA cDNA into S2 cells resulted in expression of the anti-horseradish peroxidase epitope, a result compatible with its substrate specificity in vitro. Finally, silencing of FucTA by RNAi in the neuronal cell line led to a significant reduction of anti-horseradish peroxidase binding. The present study, in conjunction with our previous in vitro data, thereby shows that FucTA is indispensable for expression of the neural carbohydrate epitope in Drosophila cells.

Neural pathways in invertebrates are often tracked using antihorseradish peroxidase, a cross-reaction due to the presence of core ␣1,3-fucosylated N-glycans. In order to investigate the molecular basis of this epitope in a cellular context, we compared two Drosophila melanogaster cell lines: the S2 and the neuronal-like BG2-c6 cell lines. As shown by mass spectrometric and chromatographic analyses, only the BG2-c6 cell line expresses ␣1,3/␣1,6-difucosylated N-glycans, a result that correlates with anti-horseradish peroxidase binding. Of all four ␣1,3-fucosyltransferase homologues previously identified, the core ␣1, 3- Polyclonal antibodies raised against the plant glycoprotein, horseradish peroxidase (HRP), 3 were used for many years in the study of invertebrate neurobiology, although information as to the exact structural basis for this cross-reaction was lacking (1)(2)(3)(4). More recently, neural anti-HRP staining has been suggested to be a characteristic of Ecdysozoa (5), one of the two newly redefined clades of protostomians. Thus, not only has anti-HRP been much used to track neurons in insects (especially Drosophila melanogaster), but there is also a report that 10% of the neurons of Caenorhabditis elegans are recognized by this reagent (6). The first hints as to the basis of this staining in insects were shown by its sensitivity to reagents that destroy carbohydrate moieties and its inhibition by bromelain glycopeptides (3), which carry N-glycans decorated with ␤1,2-xylose and core ␣1,3-fucose residues. The epitopes recognized by anti-HRP were partially revealed by analysis of N-glycans attached to horseradish peroxidase (7), indicating that the ␣1,6-linked mannose of the trimannosyl core and the core ␣1,3-linked fucose contribute to the recognition the most, whereas the presence of nonreducing terminal N-acetylglucosamines greatly reduces the reactivity of the anti-HRP. In this study, analyses also showed that the contribution of xylose in anti-HRP binding to HRP was minimal (7). However, the latter contrasts with more recent data demonstrating that anti-HRP binds both xylose-substituted and fucose-substituted structures to a similar extent (8). Nonetheless, the exact molecular basis of the neuronal anti-HRP staining in insects remained unclear. Thus, we began to re-examine this problem and found that only neoglycoconjugates containing native, and not defucosylated, bromelain glycopeptides inhibit this reaction, suggesting that core ␣1,3-fucose is part of the epitope recognized in flies. Furthermore, staining fly embryos with anti-bee venom, which contains antibodies directed against core ␣1,3-fucose, and not xylose, reproduces the same pattern as using anti-HRP. In addition, we detected core ␣1,3-/␣1,6-difucosylated N-glycans in adult flies and cloned a cDNA encoding an enzymatically active core ␣1,3-fucosyltransferase (FucTA) (9). Recently, we were also able to demonstrate that anti-HRP staining in C. elegans is due to the activity of FUT-1, a core ␣1,3-fucosyltransferase (10).
A number of questions, however, remained unanswered, the following in particular. Is FucTA expressed in fly neural tissue, and does it create the anti-HRP epitope in vivo? Are other ␣1,3-fucosyltransferase homologues involved in neuronal anti-HRP epitope synthesis? Are core ␣1,3-/␣1,6-difucosylated N-glycans enriched in fly neuronal tissue? Particularly the last question is problematic, since acquiring enough tissue for glycan analyses is a challenge. To circumvent this, a Drosophila third instar larval neuronal cell line (BG2-c6), previously found to bind anti-HRP (11), was considered as a suitable model for further study of the anti-HRP epitope synthesis; on the other hand, the commonly used Schneider 2 (S2) cell line was also employed in the expectation that these cells, since they are of hemocyte origin, do not bind anti-HRP. By examining N-glycans of these cell lines, performing knock-in and knock-down experiments on ␣1,3-fucosyltransferase homologues in the respective cell lines, and determining the tissue and stage specificity of the expression of all four ␣1,3-fucosyltransferase homologues, we have gained further evidence pointing to a key role for FucTA in the biosynthesis of the anti-HRP epitope in Drosophila cells and have demonstrated that the other three fucosyltransferase homologues are not responsible for the formation of anti-HRP epitopes in the neuronal cell line.

MATERIALS AND METHODS
Maintenance and Growth of Insect Cell Lines-The Drosophila neuronal cell line BG2-c6 was kindly provided by Kumiko Ui-Tei, Nippon Medical School, Tokyo, whereas the S2 and Sf9 cell lines were gifts from Gerald Aichinger (Intercell) and Wolfgang Ernst (Department für Biotechnologie, Universität für Bodenkultur, Wien), respectively. For expression experiments, S2 cells were grown in Schneider's Drosophila medium (Sigma) supplemented with heat-inactivated 10% fetal bovine serum (FBS), 50 units/ml penicillin G, and 50 g/ml streptomycin sulfate in 25-cm 2 cell culture flasks (Sarstedt) at 26°C. For RNAi experiments, neuronal BG2-c6 cells were grown in Shields and Sang M3 insect medium (Sigma) supplemented with heat-inactivated 10% FBS, 10 g/ml bovine insulin (Sigma) in 25-cm 2 cell culture flasks (Sarstedt) at 26°C. In order to avoid contamination from bovine serum proteins for the N-glycan analysis, both cell lines were also adapted to serum-free medium (Drosophila-SFM; Invitrogen) supplemented with 16.5 mM L-glutamine (Invitrogen) and cultured at 26°C. We found that both cell lines grew at twice the rate as in the media supplemented with 10% FBS and showed a stronger adherence to the flask surface. For general maintenance, S2 and BG2-c6 cells were split always to a density of 1 ϫ 10 6 cells/ml after they reached 0.8 -2 ϫ 10 7 cells/ml. Sf9 cells were passaged at 100% confluence using IPL-41 medium (Sigma) supplemented with 3% FBS.
N-Glycan Preparation from Drosophila Cell Lines-Typically, 500 mg (wet weight) of S2 or BG2-c6 cell lines grown in serum-free medium (Drosophila-SFM; Invitrogen) were resuspended in up to 80 mM HCl, pH 2.0 (HCl was added until pH reached ϳ2.0) and homogenized with an Ultra Turrax T25 apparatus (high speed). Pepsin (1 mg) was added, and proteolysis was allowed to proceed overnight at 37°C. The resulting extract was centrifuged, and the supernatant was applied to a Dowex AG50Wϫ2 column (1.5 ϫ 50 cm); the column was then washed with 50 ml of 2% (v/v) acetic acid. Subsequently, the glycopeptides were eluted with 0.4 M ammonium acetate, pH 6, and orcinol-positive fractions were pooled and concentrated prior to gel filtration (Sephadex G25; elution with 1% (v/v) acetic acid). The resultant glycopeptide fractions were lyophilized and dissolved in 250 mM citrate-phosphate, pH 5, and incubated at 95°C for 10 min to inactivate any remaining proteases. The N-glycans were then released using 0.375 milliunits of peptide:N-glycosidase A (37°C, 24 h). The sample was then acidified by adding two volumes of 5% (v/v) acetic acid and applied to a 2-ml Dowex AG50Wϫ2 column. The column was washed with 2% (v/v) acetic acid, and the unretained orcinol-positive fractions were passed through reverse phase clean-up columns (100 mg; Zorbax; Agilent Technologies). Lyophilized N-glycans were dissolved in water and subject to either MALDI-TOF analysis (see below) or pyridylamination followed by RP-HPLC (12)(13)(14).
Matrix-assisted Laser Desorption Ionization Mass Spectrometry-Aliquots of 0.8 l of underivatized or pyridylaminated N-glycans were applied to a flat sample platen and dried immediately under mild vacuum; subsequently, 1 l of matrix (2% (w/v) 2,5-dihydroxybenzoic acid in 30% (v/v) acetonitrile or 0.03 M 1-hydroxyisoquinoline plus 0.1 M 2,5-dihydroxybenzoic acid in 50% (v/v) acetonitrile) was added, and the samples were dried once more (15,16). MALDI-TOF mass spectra were acquired on a DYNAMO (Thermo BioAnalysis, Hemel Hempstead, UK) linear time-of-flight mass spectrometer with a dynamic extraction setting of 0.1. External mass calibration was performed with pyridylaminated N-glycans or with a partial dextran hydrolysate. On-plate ␣-fucosidase digestion was performed using 0.2 milliunits of bovine kidney ␣-fucosidase (Sigma) in 10 mM ammonium acetate, pH 5, with incubation in a "wet chamber" at 37°C for 1 h. On-plate ␣-mannosidase digestion was performed using ϳ10 milliunits of jack bean ␣-mannosidase (Sigma; repurified on Sephacryl S200) in 50 mM ammonium acetate, pH 5, with incubation in a "wet chamber" at 37°C for 5 h.
Transfection of Drosophila Schneider 2 Cells-The pIZT/V5-His vectors with native stop codons were introduced to Drosophila S2 cells using TransFectin Lipid Reagent (Bio-Rad), according to the manufacturer's protocol for adherent cells (comparable results where also obtained with the standard calcium phosphate transfection method). The cells were incubated at 26°C and harvested between the second and the fifth day of expression. A small aliquot of the cells was analyzed by confocal laser-scanning microscopy with a UV light source to confirm the presence of GFP fluorescence within the cells as an indication of a successful transfection procedure. Harvested cells were counted, spun down (1000 ϫ g for 10 min), and washed twice with PBS. Cell pellets were either processed immediately or left frozen at Ϫ80°C. Cell pellets were lysed in radioimmune precipitation buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 0.4 mM EDTA, 10% glycerol) supplemented with Complete-mini protease inhibitor mixture without EDTA (Roche Applied Science) or His tag protease inhibitor mixture (Sigma). Different volumes of radioimmune precipitation buffer were used to normalize the number of cells per microliter (normally 2.5 ϫ 10 4 cells/l). After lysis for 10 min at room temperature and 20 min on ice, cells were spun down at 14,100 ϫ g at 4°C for 30 min. Supernatants (10 l/lane) were then analyzed by Western blotting as described below.
Transfection of Sf9 Cells-pIZT/V5-His vectors carrying complete ORFs of all four Drosophila ␣1,3-fucosyltransferase homologues with either the native stop codon or in frame with V5 coding sequence were used to transfect Sf9 cells using Cellfectin reagent (Invitrogen) following the manufacturer's protocol for insect cells. A small aliquot of the cells was analyzed by confocal laser-scanning microscopy with a UV light source to confirm the presence of GFP fluorescence within the cells as an indication of successful transfection. Cells were collected 48 h posttransfection, washed once with PBS, and stored at Ϫ80°C. For assaying, cells were lysed using 50 mM Tris, pH 7.5, 150 mM NaCl, 0.5% Triton X-100 supplemented with Complete-mini protease inhibitor mixture without EDTA (Roche Applied Science) or His tag protease inhibitor mixture (Sigma) following the lysis protocol for Drosophila S2 cells. Different volumes of the buffer were used to normalize the number of cells per volume (to 1.25 ϫ 10 4 cells/l).
Western Blotting-Cell lysates were mixed with 2ϫ Laemmli loading buffer followed by electrophoresis on 12.5% SDS-polyacrylamide gels and transferred to nitrocellulose. The nitrocellulose sheets were reversibly stained with 0.5% Ponceau S (in 1% acetic acid) to verify that equal amounts of proteins were present in every lane. Subsequently, the membrane was blocked for 1 h with 1% bovine serum albumin in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.01% Tween 20 and incubated for 1 h at 23°C with rabbit anti-horseradish peroxidase antibody (Sigma) diluted 1:20,000 in blocking buffer. After washing the nitrocellulose membrane three times with 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.01% Tween 20, bound antibodies were detected by alkaline phosphatase-conjugated goat anti-rabbit IgG (Sigma), followed by color detection using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate as substrate. Alternatively, the expression of recombinant proteins in Sf9 cells was verified by probing the extracted proteins on Western blots with anti-V5 antibody (Invitrogen; 1:5,000), followed by anti-mouse IgG (Fc-specific, alkaline phosphatase conjugate, 1:30,000; Sigma).
Assay of Fucosyltransferase Activity-Activity of core ␣1,3-fucosyltransferase from insect cell cultures was measured as previously described (18) with modifications using 100 mM MES, pH 6.5, 10 mM MnCl 2 , 1 mM AMP, 210 M 6-acetamido-6-deoxycastanospermine, 0.4 mM GDP-fucose, and 25 M dabsylated glycopeptide, with the sequence GENR in either the GnGn-, GnGnF 6 -, MM-, or GNGN-glycoforms (see Fig. 1 for oligosaccharide structures) supplemented with Complete-mini protease inhibitor mixture (Roche Applied Science). Reaction tubes were incubated at 23°C and analyzed after 18 h. For analysis by MALDI-TOF MS, 0.5 l of a 1:10 dilution of the reaction mixture was mixed with 0.5 l of 1% ␣-cyano-4-hydroxycinnaminic acid (Fluka; in 70% acetonitrile) on a MALDI plate. An increase in glycopeptide m/z by 146.1 Da indicated the transfer of one fucose residue to the reducing terminal N-acetylglucosamine of the glycopeptide. For testing the substrate specificity of the recombinant Drosophila core ␣1,3-fucosyltransferase, a FLAG-tagged form of FucTA was expressed in Pichia pastoris under control of the AOX1 promoter using a modified form of the pPICZ␣C vector (Western blotting with anti-FLAG antibody showing a protein with apparent size of M r 75,000). After 4 days of expression at 16°C, the enzyme was concentrated 10-fold (using UltraFree centrifugal concentration devices, molecular weight cut-off 30,000) and tested with dabsylated MGn-glycopeptide, which had been prepared from dabsylated MM-glycopeptide by the activity of FLAG-tagged human GnTI expressed also in P. pastoris at 16°C, as well as with dabsylated MGnF 6 -glycopeptide that was prepared from the dabsylated MGn-glycopeptide by the activity of P. pastoris expressed C. elegans FUT-8 core ␣1,6-fucosyltransferase. The assays were performed using 0.1 mM glycopeptide, 2 mM GDP-fucose, 40 mM MES, pH 6.5, 10 mM MnCl 2 at 30°C for 4 h prior to MALDI-TOF MS analysis as described for the insect cell line assays. The enzyme was also tested using a dansylated Man 5 GlcNAc 2 -glycopeptide, which was prepared after Pronase digestion of Aspergillus oryzae amylase. The resulting dansyl-Man 5 GlcNAc 2 (0.2 mM) was incubated at 30°C in the presence of supernatant of yeast expressing human GnTI (with or without a supernatant of yeast expressing C. elegans core ␣1,6-fucosyltransferase), 1 mM UDP-Glc-NAc, 1 mM GDP-Fuc, 40 mM MES, pH 6.5, and 10 mM MnCl 2 . After 12 h, an aliquot of a supernatant of yeast expressing FLAG-tagged Drosophila FucTA was added, and the incubation continued for another 24 h. The products were analyzed and purified by RP-HPLC under isocratic conditions (9% acetonitrile, 0.05% trifluoroacetic acid), collected, and subjected to MALDI-TOF MS using 2,5-dihydroxybenzoic acid as matrix.
Developmental Stages and RT-PCR-Canton S or wϪ wild type D. melanogaster flies were maintained at room temperature. For different developmental stages, wϪ wild type was used; after a 1-h "precollection" phase, eggs were collected for 1.5 h on standard apple juice Petri dishes, and different developmental stages were collected. For the tissue specificity (i.e. gender-separated heads and bodies), the Canton S strain was used. Total RNAs from various stages were isolated by TRIzol reagent (Invitrogen) followed by a first strand cDNA synthesis using Superscript III reverse transcriptase (Invitrogen) and oligo (dT) 18 as primer. Expression of Drosophila fucosyltransferase genes FucTA, FucTB, FucTC, and FucTD was analyzed by performing a 32-35 cycle PCR (55-60°C for 30 min, 72°C for 2 h; 94°C for 30 min) using first strand cDNAs from staged flies and the following primers: FucTA, forward primer 5Ј-GG-CCGACATGATCCTCTAC-3Ј and reverse primer 5Ј-GTTCTTCGT-GAATGCGCTG-3Ј (58°C, 35 cycles); FucTB, forward primer 5Ј-CG-CATCACCAACAAGCGC-3Ј and reverse primer 5Ј-GACAAGGTT-GTGGAGTAG-3Ј (57°C, 35 cycles); FucTC, forward primer 5Ј-CTT-ATCGCATTGACTCGGATG-3Ј and reverse primer 5Ј-CGCGGAA-TTCTCACAAACGTATTCGGCTTTGC-3Ј (55°C, 32 cycles); Fu-cTD, forward primer 5Ј-CAATGCCGATAGACAGACTC-3Ј and reverse primer 5Ј-GTCCGGACACGCCGACG-3Ј (60°C, 35 cycles). The cDNAs were normalized against the rp49 transcript, coding for a ribosomal protein. The primers used, producing a 440-bp fragment, were as follows: rp49-fw, GACCATCCGCCCAGCATAC; rp49-rev, TCCGACCAGGTTACAAGAAC (60°C). FEBRUARY 10, 2006 • VOLUME 281 • NUMBER 6 RNA Interference (RNAi) in a Drosophila Cell Line-Primers used to create the PCR templates for double-stranded RNA production contained a T7 promoter (GAATTAATACGACTCACTATAGGGAGA) at the 5Ј-end followed by gene-specific sequences as follows: FucTA, forward primer CCCACGGGCATTCGAC and reverse primer TGACGTCCTTGCGCGC; FucTB, forward primer TACGGCAGCT-GCCTACG and reverse primer ATCAAACTCTGCGTCATTG; FucTC, forward primer GCTGGACGACGAACAAG and reverse primer GGATTGGATATCGTGGTAG; FucTD, forward primer AATTTGACAATGACCTATCG and reverse primer CCCAAGCTG-TAGGATCTC; Nervana nrv2.2 (CG9261; GenBank TM accession U22440), forward primer CGACCCCTGCCCACCCGT and reverse primer GTAGCCCTCGGAGTTCTG. Commercial double-stranded DNA (clone CG6869; Open Biosystems) was also used as template for FucTA dsRNA synthesis. The PCR products were purified over GFX columns (Amersham Biosciences) and used as templates for in vitro transcription (Megascript T7 transcription kit; Ambion, Austin, TX) to produce dsRNA. The dsRNAs were precipitated by ethanol/sodium acetate and then resuspended in 20 l of sterile deionized water. Dissolved dsRNA was incubated at 75°C for 5 min and left at 23°C to cool down to facilitate annealing of the RNA strands. To verify that the majority of the dsRNAs were intact and present as a single band, 2 g of each were electrophoresed in 1.2% agarose gel. The amount and purity of dsRNA were also estimated by measuring absorbance at 260 and 280 nm. The dsRNAs were stored at Ϫ20°C until use.

Anti-HRP Epitope of Drosophila
To silence the aforementioned genes, RNAi was essentially performed as described by Worby et al. (19). Drosophila BG2-c6 cells were diluted to a final concentration of 3-4 ϫ 10 6 cells/ml in Shields and Sang M3 medium (Sigma) lacking FBS, and 1 ml was pipetted per well of a 6-well cell culture dish (Nunc). Thereafter, each individual dsRNA was immediately added to a final concentration of 70 nM, mixed vigorously, and incubated for 45 min at room temperature, followed by the addition of 2 ml of Shields and Sang M3 medium (Sigma) containing FBS and insulin. The cells were collected after 4 days by centrifugation at 1000 ϫ g. The cell pellets were lysed in the same way as for the transfected Drosophila S2 cells.
Flow Cytometer Analysis of RNAi-treated Cells-For flow cytometer analysis, 2.25 ϫ 10 6 cells treated with dsRNA were incubated with 20 g/ml anti-HRP in PBS at 26°C for 30 min, washed once with 5 ml of PBS, and then incubated with anti-rabbit FITC (Sigma) diluted 1:100 in PBS for another 30 min. Cells were again washed with 5 ml of PBS and resuspended in PBS to around 10,000 cells/l and subject to flow cytometry using a BD Biosciences FACS Calibur with a 488-nm argon Laser (15-milliwatt output power) in the standard configuration. FITC fluorescence was measured with a 530/30 BP filter in FL1.

A Drosophila Neuronal Cell Line Is Enriched in Difucosylated
N-Glycans-Due to the challenges in collecting a sufficient amount of Drosophila heads to show by direct HPLC or MS analyses that core ␣1,3-linked fucose is enriched in the neural tissue, we decided to look for suitable Drosophila cell lines to examine the expression of this epitope. The BG2-c6 cell line isolated from third instar larval central nervous system cells has been previously shown to bind anti-HRP (11); on the other hand, readily available and commonly used Drosophila S2 cells (20) of hemocyte origin were expected not to bind anti-HRP. The complete N-glycosylation profile of neither cell line has previously been described; thus, we prepared N-glycans from these cells after culturing them in serum-free medium. The MALDI-TOF MS profiles (Fig. 2) indicate that the BG2-c6 cells have difucosylated N-glycans similar to those we found previously in the glycan profile of whole adult flies (see Fig. 1 for oligosaccharide structures). Indicative of an enrichment of these structures in neuronal cells, these glycans accounted for 20% of the total N-glycans in the BG2-c6 cells ( Fig. 2A and Table 1), as opposed to 0.8% in whole adult flies (9). On the other hand, these structures were apparently absent from S2 cells ( Fig. 2B and Table 1).
In order to prove that the difucosylated N-glycans from BG2-c6 cells are indeed carrying core ␣1,3-fucose, the complete N-glycan pool was   (12,14,21). Indeed (Fig. 3B), treatment of the difucosylated species (isolated and separated from MM structure by normal phase HPLC) using these two methods resulted in the predicted changes in chromatographic behavior, which in both cases correlated with the loss of m/z 146, equivalent to the removal of one fucose residue, as shown by MALDI-TOF MS analysis. Furthermore, reverse phase fraction 9 contained one species with an m/z value of 1507.4, corresponding to the mass of either GnMFF or MGnFF. A partial jack bean hexosaminidase digest of a GnGnF 3 standard was analyzed by RP-HPLC and compared with an ␣-fucosidase digest of fraction 9; as judged by co-elution, it was concluded that fraction 9 contained the difucosylated structure GnMF 3 F 6 with nonreducing N-acetylglucosamine linked to the ␣1,6-arm and not to the ␣1,3arm (data not shown).
Expression of ␣1,3-Fucosyltransferase Homologues in Drosophila-In order to investigate the genetic basis for the observed core ␣1,3-fucosylated N-glycans, we initiated studies of the expression of all four ␣1,3fucosyltransferase homologues from the fly. Although previous data suggest that FucTA functions as a core ␣1,3-fucosyltransferase, it could be theoretically possible that FucTB, FucTC, and FucTD also contribute to anti-HRP epitope expression (9). Thus, in order to examine whether we could infer function from the expression pattern, we determined the transcript levels of all four ␣1,3-fucosyltransferase homologues in different stages of the fly, in male and female heads and bodies and in the BG2-c6 and S2 cell lines. If FucTA is responsible for the expression in vivo of core ␣1,3-fucosylated N-glycans in flies, then this gene should be transcribed throughout the life cycle (with the possible exception of the first 8 h before development of the neural system), in heads and bodies of both sexes (with an enrichment of neural tissue in heads) and in BG2-c6 cells; on the other hand, FucTA should be absent from S2 cells. The absence of FucTB, FucTC, and FucTD from any stage of the life cycle and their presence in S2 cells would be compatible with a potential inability to form the anti-HRP epitope.
Total RNA was prepared from heads and bodies of both male and female flies, from various developmental stages and from the two cell lines; the quantities of RNA used in each PCR were prenormalized on the basis of the levels of the transcript encoding RP49, a constitutively expressed ribosomal protein. Primers were designed across introns, so as to ensure that genomic DNA contamination could not result in a PCR product of the same size as an RT-PCR product. Consistent with our hypothesis, the results (Fig. 4) indicated that FucTA is expressed in heads and bodies of male and female flies, in all developmental stages, and in BG2-c6 cells, whereas it is apparently absent from S2 cells under the given conditions (under no conditions were we able to amplify the complete FucTA open reading frame from S2 cells). On the other hand, FucTB transcripts are present only in 16-h-old embryos, heads of both sexes, and male bodies as well as in BG2-c6 cells and S2 cells; FucTC is expressed in all stages but only in adult bodies and male heads and seemingly is not present in female heads and BG2-c6 and S2 cells; FucTD is strongly expressed in male bodies and, weakly, in the pupal stages and S2 cells.  FEBRUARY 10, 2006 • VOLUME 281 • NUMBER 6

Anti-HRP Epitope of Drosophila
Gain of Anti-HRP Staining in Schneider 2 and Sf9 Cells-Our next experiments were directed at examining which of the four ␣1,3-fucosyltransferase homologues in Drosophila (FucTA, FucTB, FucTC, and FucTD) could confer anti-HRP staining to Drosophila S2 cells. Previously, we have only demonstrated that FucTA is enzymatically active in vitro (9); using the finding that S2 cells lacked difucosylated glycans, we decided to express the complete reading frames of all four homologues in these cells. When designing the expression constructs, we considered the differences between the various sequences of each fucosyltransferase present in the data bases; these polymorphisms (whether natural or PCR-generated) could theoretically lead to acquisition or loss of enzymatic activity of the encoded proteins. For instance, our previously published FucTB sequence (AJ302046) is different from the unpublished one of Petit et al. (AY061932). Furthermore, our FucTC sequence (AJ302947) encodes a protein of 425 residues; however, the genomic sequence lacks nucleotides 530 and 531 and so would encode a protein of only 185 amino acids (see chromosome X shotgun sequence AABU01002757.2). On the other hand, the unpublished FucTD sequence of Petit et al. (AF441265) contains an Ala-Thr sequence, whereas the Drosophila Gene Collection clone encodes a Thr-Pro at this position (AY075216). Using portions of the reading frames of either Canton S-derived clones, our previous clones, or Drosophila Gene Collection clones, we generated four forms of FucTA, two forms of each FucTB and FucTC, and three forms of FucTD in order to cover most of the polymorphisms.
Two days after transfection, the acquisition of anti-HRP binding was examined by Western blotting. An extract of BG2-c6 cells was used as a positive control (Fig. 5A, lane 2). Only S2 cells transfected with FucTA gained the epitope (Fig. 5A, lane 5), whereas untransfected or mocktransfected S2 cells and S2 cells transfected with FucTB, FucTC, FucTD or empty vector showed only background staining with anti-HRP. The were analyzed by MALDI-TOF mass spectrometry. Note that the peak with m/z value of 1226.1 corresponding to difucosylated structure MMF 3 F 6 is only present in BG2-c6 cells (A). Other relevant peaks that also contain structures not present in S2 cells but found in BG2-c6 are labeled in boldface letters. See Table 1 for annotation of m/z values for [M ϩ Na] ϩ ions. The Man4 structure is shown in brackets, since the signal with m/z 1095.9 is probably primarily due to the potassium adduct of MMF structures. The structure (B, chromatogram 1) was subjected to ␣-fucosidase (which digests core ␣1,6-linked fucose) digest (B, chromatogram 2) and to hydrofluoric acid (HF) treatment (which specifically cleaves only core ␣1,3-linked fucose) (B, chromatogram 3). See also Table 1 for a summary of N-glycan composition of individual fractions. extracts of transfected cells were analyzed for fucosyltransferase activity using dabsylated GnGn-, MM-, GNGN-, and GnGnF 6 -glycopeptides (see Fig. 1 for relevant structures). All extracts transferred a fucose residue to dabsylated GnGn-glycopeptide to a similar extent, an expected result considering the lower affinity of FucTA for that particular substrate (9) and possibly due to relatively high endogenous activity of core ␣1,6-fucosyltransferase. On the other hand, only extracts of FucTAtransfected cells transferred a fucose residue to the dabsylated GnGnF 6glycopeptide (Fig. 5B). The extracts of cells expressing Drosophila fucosyltransferase homologues were further tested with dabsylated MMand GNGN-glycopeptides as substrates, but no transfer of fucose was detected.
Subsequently, constructs encoding all four fucosyltransferases with either the native stop codon or encoding C-terminally V5/His 6 -tagged forms were used to transfect Sf9 cells, which in our hands display a higher transfection efficiency than S2 cells. After 2 days of transient expression, extracts of the cells were analyzed for fucosyltransferase activity using dabsylated GnGn-, MM-, GNGN-, and GnGnF 6 -glycopeptides. All extracts transferred a single fucose residue to a dabsylated GnGn-glycopeptide to a similar extent, except that FucTA-transfected cells could transfer a second fucose to this substrate. Also, the FucTAtransfected cells fucosylated GnGnF 6 completely in 23 h, indicating a much higher activity of recombinant FucTA expressed in Sf9 cells than in S2 cells. None of the extracts of cells transfected with FucTB, FucTC, and FucTD ORFs displayed any transfer above background to this substrate (data not shown). The dabsylated MM-glycopeptide was quickly degraded under the given conditions, presumably due to an endogenous endoglycosidase H activity present in extracts of Sf9 cells, and therefore could not be used to estimate any fucosyltransferase activity. Furthermore, Western blotting confirmed that only with FucTA was there an obvious increase in anti-HRP staining upon transfection of fucosyltransferase cDNAs into Sf9 cells (Fig. 5C, lane 7). On the other hand, all four Drosophila ␣1,3-fucosyltransferase homologues were successfully expressed in Sf9 cells as judged by use of the anti-V5 antibody (Fig. 5C,  lanes 1-5).
Further Substrate Specificity Studies with FucTA-Previously, we had only tested the transfer of fucose by FucTA with MM, GnGn, and Gal-Gal substrates. However, while considering both the putative action of a processing hexosaminidase in insect cells (22) and the fact that MMF 3 F 6 , rather than GnMF 3 F 6 , is the major difucosylated species in flies, we wished to confirm that MGn, in addition to GnGn, is a substrate for the fucosyltransferase. To this end, FucTA was tested with MGn and MGnF 6 substrates. When recombinantly expressed in P. pastoris, FucTA transfers a fucose residue to dabsylated MGn-and MGnF 6glycopeptides (Fig. 6). It also transfers to a much lower extent to dabsylated GNGN-glycopeptide, but not to dabsylated GnM-or MM-glycopeptides (data not shown). Furthermore, in assays with the dansylated Man5-glycopeptide in the presence or absence of GnTI and core ␣1,6fucosyltransferase, GnTI-dependent transfer was observed, particularly when the core ␣1,6-fucosyltransferase was also present, as shown by a shift to lower retention times, thereby generating either Man5GnF 3 or Man5GnF 3 F 6 in vitro (Fig. 6E). On the basis of these data, we presume that the prior action of GnTI (23) is required for creating substrates for FucTA, whereas subsequent modification by GnTII (24) and the core ␣1,6-fucosyltransferase (25) are not required for the activity of FucTA, although prior core ␣1,6-fucosylation appears to improve the efficiency of the action of the FucTA. Whereas the action of mannosidase II may well result in more efficient transfer by core fucosyltransferases, our findings in both this and our previous studies (25) support the hypothesis that GnTI is the entry point for the generation of fucosylated paucimannosidic N-glycans in Drosophila. Furthermore, we can account for the synthesis of the major difucosylated structure in Drosophila, MMF 3 F 6 .
Reduction of FucTA Expression in a Neuronal Cell Line-The detection of core ␣1,3-fucose on its N-glycans (see above) and the anti-HRP binding characteristics of BG2-c6 cells (11) indicated that this line is a suitable model for examination of core fucosylation in vivo. Therefore, since FucTA is a proven core ␣1,3-fucosyltransferase (9), we assumed that specific targeting of FucTA transcripts by RNAi could result in reduction of binding to anti-HRP.
Double-stranded RNAs based on portions of the FucTA, FucTB, FucTC, FucTD, and Nervana genes were synthesized (Nervana being an were tested for their capability to transfer a fucose residue to dabsylated GnGnF 6 -glycopeptide using a MALDI-TOF MS-based assay. The difference of m/z 146.1 is indicative of the transfer of one fucose residue to dabsylated GnGnF 6 -glycopeptide, thus creating a dabsylated GnGnF 3 F 6 -glycopeptide. Extracts of the other S2 cells shown in A did not display this activity (data not shown). C, Sf9 cells were transfected with pIZT/V5-His constructs containing ORFs of all four Drosophila ␣1,3-fucosyltransferase homologues and harvested after 2 days. As judged by reversible Ponceau S staining, equal amounts of protein were separated on a polyacrylamide gel, transferred to a nitrocellulose membrane, and probed with either anti-V5 antibody (lanes 1-5) or anti-HRP antibody (lanes 6 -10). The anti-V5 antibody was used to demonstrate expression of recombinant Drosophila ␣1,3fucosyltransferase homologues FucTA, -B, -C, and -D (lanes 2, 3, 4, and 5, respectively), whereas no binding of the antibody to cells transfected with empty pIZT/V5-His vector was detected (lane 1). The anti-HRP antibody bound strongly only to extracts of cells transfected with FucTA (lane 7), whereas other fucosyltransferases (FucTB, FucTC, and FucTD in lanes 8, 9, and 10, respectively) did not alter anti-HRP binding to Sf9 extracts as compared with the empty vector negative control (lane 6). These results were also reproduced using other isoforms of FucTB, FucTC, and FucTD. Na ϩ /K ϩ -ATPase previously shown to carry the anti-HRP epitope). These were then incubated with BG2-c6 cells, which after 4 days of culture were then subject to Western blotting analysis using anti-HRP; as a control, a monoclonal antibody against Nervana was also employed. As shown in Fig. 7A, anti-HRP staining was reduced in BG2-c6 cells incubated with double-stranded RNA encoding part of the FucTA sequence (a similar reduction was attained when the cells were incubated with mixtures including this RNA; data not shown). However, no such diminution was observed if double-stranded RNA corresponding to any other fucosyltransferase or Nervana had been present. The same result was acquired by incubating dsRNA-treated BG2-c6 cells with anti-HRP antibodies, followed by anti-rabbit-FITC and flow cytometry; only cells treated with dsRNA encoding a part of FucTA reduced the binding of anti-HRP to the surface of intact cells (Fig. 7B). In order to verify whether the RNAi knock-down was specific for the targeted genes, the relative amounts of transcripts of the ␣1,3-fucosyltransferase homologues shown to be present in BG2-c6 cells (cf. Fig. 4; i.e. FucTA and FucTB) were estimated by performing RT-PCR from rp49-normalized cDNA. RNA was prepared from cells treated without dsRNA and from cells treated with dsRNA corresponding to ORFs of FucTA, FucTB, and nrv2.2. The cells treated with dsRNA corresponding to either FucTA or FucTB ORF contained significantly reduced amounts of the respective transcripts while leaving amounts of other transcripts unaltered, demonstrating that the RNAi knock-down was specifically affecting the targeted genes (Fig. 7C). In the case of anti-Nervana Western blots (data not shown), only the M r 37,000 band was no longer seen when Nervana double-stranded RNA was present. Thus, we believe that, since this method was successful with both a carbohydrate and a protein epitope, RNA-mediated interference is a valid method for investigation of the BG2-c6 cell line.

DISCUSSION
Twenty years after the first description of the use of anti-HRP for the staining of neural tissue in Drosophila (1), we determined the structural basis for this cross-reaction by demonstrating the presence of core ␣1,3fucosylated N-glycans and of a relevant fucosyltransferase gene (FucTA) (9). However, as indicated by antibody-binding, mass spectrometric, and HPLC data, we only had in vitro evidence that FucTA was capable of synthesizing the anti-HRP epitope. Identifying ␣1,3-fucosyltransferase FUT-1 as the enzyme responsible for anti-HRP staining in C. elegans (10) also supported the role of core ␣1,3-fucosyltransferases in the neuronal anti-HRP epitope synthesis in invertebrates in general. In the present study, we made use of two different Drosophila cell lines in order to investigate the biological role of FucTA in vivo and to show that, of the four ␣1,3-fucosyltransferase homologues, only FucTA is indispensable in synthesis of the anti-HRP neuronal epitope in these cells.
First of all, it was necessary to show that the two cell lines, BG2-c6 and S2 cells were relevant models (positive and negative, respectively). This necessitated the analysis of the N-glycans of both cell lines; indeed, this is the first time that the complete N-glycan profile of any Drosophila cell line has been examined. The results are generally in agreement with the data obtained with other insect cell lines in that oligomannose and core ␣1,6-fucosylated N-glycans were found. Furthermore, in the neuronal cell line, previously found to express the anti-HRP epitope (11), difucosylated N-glycans carrying both ␣1,3and ␣1,6-linked fucose residues were detected at a level enriched as compared with whole adult flies, these structures being apparently absent from S2 cells, which do not express this epitope as judged by our Western blot data. This is also in agreement with the glycan structures found on two recombinant glycoproteins expressed in S2 cells (26). Thus, there appears to be a correlation between expression of the anti-HRP epitope and the presence of core ␣1,3-linked fucose. This is, perhaps, not surprising, since horseradish peroxidase is a plant glycoprotein, on which around 80% of the glycans carry core ␣1,3-fucose (7) and since core ␣1,3-fucosylated glycoconjugates inhibit the binding of anti-HRP to embryonic neural tissue (9). Various insect lines were previously shown to express glycoproteins containing core ␣1,3-linked fucose; some have a low degree of ␣1,3fucosylation (Sf-21 and Bm-N cells), whereas up to 30% of the glycans of the IZD-Mb-0503 cell line are ␣1,3-fucosylated glycans, also predominantly in difucosylated form (27). However, our data suggest that S2 cells are suitable for expression of proteins lacking the immunogenic ␣1,3-fucose epitope. The presence of MMF 3 F 6 structures in the neuronal cell line, taking into account that recombinantly expressed FucTA was previously shown to act on GnGn and GalGal structures (9), raised a question as to how this structure is actually created in the cell line. Our new finding that FucTA acts on MGn (and MGnF 6 ) but not on GnM structures is in agreement with the evidence of an N-glycan processing hexosaminidase in insects (22) and with the previously proposed glycosylation order in invertebrates (25), where it is expected that MGn is modified by the action of ␣1,6-fucosyltransferase and FucTA, yielding MGnF 3 F 6 , which is, in turn, processed by the hexosaminidase, finally yielding the MMF 3 F 6 structure. It is also noteworthy that all substrates demonstrably utilized by FucTA (MGn, MGnF 6 , Man5Gn, Man5GnF 6 , GnGn, GnGnF 6 , GalGal, and GNGN) contain the residue transferred by GnTI. On the other hand, MM, GnM, and Man5 (the former two requiring the prior action of GnTI and hexosaminidase during their biosynthesis) were not detectably substrates, although, of course, an extremely low transfer in vivo by FucTA or other FucTs to such glycans cannot be ruled out. Interestingly, this substrate specificity is completely the opposite of that of the other proven invertebrate core ␣1,3-fucosyltransferase, C. elegans FUT-1 (10).
Another line of evidence to suggest that FucTA is a mediator of this epitope's expression comes from our RT-PCR data. The particular pattern of presence and absence of transcripts under given conditions would strongly suggest that expression of FucTA correlates with the appearance of the anti-HRP epitope. The data from the present study are comparable with the rather limited data available from other sources. The data from the EST data base (NCBI) indicates the presence of Drosophila FucTA in 0 -24 h old embryos. In situ RNA hybridization analysis of Drosophila FucTA expression in embryos shows that expression of this gene takes place in the central nervous system throughout the last few stages of embryonic development 4 (stages 13-16 equivalent to 10 -15-h-old embryos) (28). These data are in agreement with the expression pattern acquired in this study; furthermore, they also show that the expression of Drosophila FucTA correlates spatially and temporally with expression of the anti-HRP epitope in flies (1).
In contrast, FucTB is, according to the microarray data, only expressed in the first 12 stages (although we also see low levels of transcript in the later developmental stages); the seemingly stronger expression of FucTB in male bodies is compatible with the presence of many corresponding adult testes ESTs in the data base. Interestingly, FucTB ESTs containing one intron have been isolated from adult head, but the FucTB fragments we detect in heads by RT-PCR are of the correctly spliced size. It also appears that FucTB is expressed more in male flies and predominantly in the testis, since other recent data indicate 14.5 times stronger expression in testis over ovaries and 8.9 times stronger expression in testis over samples of male flies without gonads (29). Other data from the EST data base are compatible with the expression panel, indicating expression of FucTB in testis, adult heads, and 0 -24h-old embryos.
On the other hand, the Berkeley data do not include information on FucTC (since this was not a predicted reading frame, so no primers were designed); in any case, if in the isogenic y; cn bw sp strain (that used for sequencing the genome (30)) and in the Canton S strain 5 the reading frame is interrupted and if, as we find, transcripts of this gene are not present in female heads and neuronal cell line, then one may conclude that FucTC cannot be relevant to anti-HRP epitope biosynthesis in these fly strains. We also detected FucTC transcripts by in situ hybridization in primordial gut. 6 Indicative, however, of a conserved pattern of expression among insects for this gene is the isolation of ESTs of a FucTC homologue from Glossina gut (GenBank TM accession numbers BX552226, BX552225, BX558576, and BX563797). The expression of the Drosophila FucTC in all developmental stages cannot be confirmed by data from the EST data base; to date, FucTC ESTs were only found in adult heads.
Our data, indicating the absence of FucTD transcripts from early stages of fly development, agree with the Berkeley microarray data; also, the large number of FucTD ESTs originate from adult testes and only one from adult heads, which partially concurs with our finding that FucTD is expressed in male bodies but not in male heads or females. In theory, FucTD could therefore be responsible for the previously found anti-HRP epitope expression in male reproductive tissue (1, 4); indeed, Drosophila FucTD is found to be predominantly expressed in testis (14.3 times more than in ovaries and 13.8 times more than in samples of male flies without gonads) (29). However, the finding that FucTD does not confer anti-HRP binding to S2 cells (neither at the level we find by RT-PCR to be naturally expressed in these cells nor after overexpression) would argue against such a biochemical function. The expression pattern acquired during the present study for this gene is, however, compatible with expression in testis, since we found transcripts in male bodies (as the nearest "equivalent" to testis). The expression is also seemingly taking place in later developmental stages and in S2 cells. Further indirect evidence that FucTD is not required for anti-HRP epitope expression is that homologues of this gene have as yet not been found in species that do not belong to the Drosophila genus (with 20 -40% identity on protein level), whereas both FucTA and FucTC homologues are also found in other insects (with 40 -80% identity on the protein level). On the other hand, FucTB orthologues from Dro-sophila species and Bombyx mori appear to be most related to mammalian FucT-X and FucT-XI. Furthermore, it should be noted that FucTD is missing part of a motif conserved seemingly in all known animal ␣1,3-fucosyltransferases: instead of the DY(I/V)TEK motif (31,32), FucTD has DYIPPQ at positions 328 -333. Additionally, the hydrophobicity plot of FucTD indicates that, for a type II membrane protein, it has an unusually long cytosolic tail (the predicted transmembrane domain covers amino acids 89 -119), showing another difference between FucTD and other core ␣1,3-fucosyltransferase homologues/enzymes. Interestingly, the tollo gene shown to rescue the epitope in Drosophila mutants that lack the staining (33), was also shown to be expressed throughout the Drosophila life cycle, even in the S2 cell line lacking the staining (data not shown). The limited data from the EST data base show that Tollo is expressed in 0 -24-h-old embryos and in the mbn2 cell line (of hemocyte origin consisting of tumorous blood cells) and possibly in imaginal disks and adult heads. Although suggestive of Tollo not directly taking part in the expression of the anti-HRP epitope, the presence of the Tollo transcripts in the cell line lacking the staining might be misleading; the regulation of the gene product synthesis and activity might be on the translational level or be on the level of the gene product itself. Due to the demonstration that incubation of asialoagalactotransferrin with FucTA results in creation of an anti-HRP epitope in vitro (9) and that this enzyme has a role in the synthesis of this epitope in Drosophila neuronal cells in vivo, the exact role of Tollo in anti-HRP epitope expression remains unclear; one of the possibilities is that the Tollo functions as an upstream regulator of FucTA or that it is involved in biochemical processes that precede the actual transfer of fucose to glycoproteins by FucTA. The role of Tollo in anti-HRP epitope expression and its relation to FucTA are subjects of ongoing research.
Subsequently, we tested all four Drosophila ␣1,3-fucosyltransferase homologues for the ability to confer anti-HRP staining to the otherwise nonstaining S2 cells and for their ability to increase the anti-HRP binding to normally low staining Sf9 cells. Considering that in Sf9 cells we could show that recombinant forms of all four fucosyltransferase homologues were indeed expressed, it was demonstrated that, of these four, only FucTA is involved in synthesis of the anti-HRP epitope in vivo; this is consistent with our previous in vitro and expression panel data. Furthermore, encouraged by the N-glycan analyses and the expression panel results, we used RNA-mediated interference of FucTA expression in the neuronal BG2-c6 cell line using double-stranded RNA with no homology to any other gene; the specificity of the RNAi experiments for the targeted ␣1,3-fucosyltransferase genes transcribed in the BG2-c6 neuronal cell line was verified by measuring relative amounts of respective transcripts in dsRNA-treated cells. As judged by Western blots, the expression of the epitope was significantly reduced in the neuronal cell line when using the FucTA double-stranded RNA, whereas other double-stranded RNAs encoding other ␣1,3-fucosyltransferase homologues (FucTB, FucTC, and FucTD) and Nervana had no discernable effect. Furthermore, analysis of RNAi-treated cells by flow cytometry shows a reduction of anti-HRP binding to the surface of cells treated with FucTA dsRNA, suggesting that the cell surface epitopes are also created by the addition of a fucose residue by FucTA. Therefore, both knock-down of Drosophila core ␣1,3-fucosyltransferase homologues in an anti-HRP-positive cell line (BG2-c6) and knock-in of Drosophila core ␣1,3-fucosyltransferase homologues in an anti-HRP negative cell line (S2) demonstrated that FucTA is required for expression of the anti-HRP epitope in these two cell lines. Our data are also in agreement with FucTA being the only enzyme in Drosophila known to date with core ␣1,3-fucosyltransferase activity (9).
In summary, the present study is the first to indicate a specific role for the core ␣1,3-fucosyltransferase FucTA in the synthesis of the neural anti-HRP epitope in insects. It is certainly of interest to follow up these studies with larger scale screens to uncover how its expression is controlled as well as to examine the effect of targeted reduction of FucTA transcription in whole animal models. Understanding the biosynthesis of this tissue-specific glycosylation event as well as of a recently identified relevant endogenous C-type lectin (34) opens up the possibility of pursuing new ways toward revealing the function of carbohydrate recognition networks in the fly.