HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 36, 32421-32429, September 6, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/36/32421    most recent M206213200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Schwientek, T. Articles by Clausen, H. Search for Related Content PubMed PubMed Citation Articles by Schwientek, T. Articles by Clausen, H. Social Bookmarking                      What's this?

The Drosophila Gene brainiac Encodes a Glycosyltransferase Putatively Involved in Glycosphingolipid Synthesis^*

Tilo Schwientek^§¶, Birgit Keck^§, Steven B. Levery, Mads A. Jensen, Johannes W. Pedersen, Hans H. Wandall, Mark Stroud^**, Stephen M. Cohen, Margarida Amado^§§, and Henrik Clausen^¶¶

From the  School of Dentistry, University of Copenhagen, Nørre Allé 20, 2200 Copenhagen N, Denmark, the  Department of Chemistry, University of New Hampshire, Durham, New Hampshire 03824, the ^** Northwest Biotherapeutics, Inc., Bothell, Washington 98021, and the  European Molecular Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany

Received for publication, June 21, 2002, and in revised form, July 18, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The Drosophila genes fringe and brainiac exhibit sequence similarities to glycosyltransferases. Drosophila and mammalian fringe homologs encode UDP-N-acetylglucosamine:fucose-O-Ser 1,3-N-acetylglucosaminyltransferases that modulate the function of Notch family receptors. The biological function of brainiac is less well understood. brainiac is a member of a large homologous mammalian 3-glycosyltransferase family with diverse functions. Eleven distinct mammalian homologs have been demonstrated to encode functional enzymes forming 1-3 glycosidic linkages with different UDP donor sugars and acceptor sugars. The putative mammalian homologs with highest sequence similarity to brainiac encode UDP-N-acetylglucosamine:1,3-N-acetylglucosaminyltransferases (3GlcNAc-transferases), and in the present study we show that brainiac also encodes a 3GlcNAc-transferase that uses -linked mannose as well as -linked galactose as acceptor sugars. The inner disaccharide core structures of glycosphingolipids in mammals (Gal1-4Glc1-Cer) and insects (Man1-4Glc1-Cer) are different. Both disaccharide glycolipids served as substrates for brainiac, but glycolipids of insect cells have so far only been found to be based on the GlcNAc1-3Man1-4Glc1-Cer core structure. Infection of High Five^TM cells with baculovirus containing full coding brainiac cDNA markedly increased the ratio of GlcNAc1-3Man1-4Glc1-Cer glycolipids compared with Gal1-4Man1-4Glc1-Cer found in wild type cells. We suggest that brainiac exerts its biological functions by regulating biosynthesis of glycosphingolipids.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The neurogenic Drosophila gene brainiac plays essential roles in epithelial development in the embryo and in oogenesis (1, 2). brainiac shares sequence similarity with glycosyltransferases, together with another gene, fringe (3). Recently it was demonstrated that fringe encodes a genuine glycosyltransferase (4, 5), raising the possibility that brainiac may also function as a glycosyltransferase enzyme. Fringe modulates functions of the Notch receptor by extending O-linked fucosylation sites in EGF modules of Notch. UDP-N-acetylglucosamine:Fuc1-O-Ser 1,3-N-acetylglucosaminyltransferases (O-Fuc 3GlcNAc-transferase)¹ encoded by mammalian fringe orthologs control the O-linked fucosylation pathway and allow synthesis of the sialylated tetrasaccharide NeuAc2-3Gal1-4GlcNAc1-3Fuc1-O-Ser (6). Fringe may compete with an alternate glycosylation pathway controlled by an UDP-glucose:Fuc1-O-Ser 3-glucosyltransferase (6).

In Drosophila, brainiac mutants produce defects that resemble those produced by loss of Notch function during oogenesis. Consequently, brainiac protein has been considered as a possible modulator of Notch activity (2). Brainiac activity is required in the developing germ line for proper organization of the follicle. Some of the defects associated with loss of brainiac activity in germ line cells resemble defects associated with loss of Notch activity. Recently, the role of Notch and its ligand Delta in signaling between germ line and somatic cells has been clarified (7). Delta is expressed in germ line cells and required for activation of Notch in somatic follicle cells at two stages of oogenesis. The defects associated with loss of Notch activity resemble the defects associated with loss of brainiac in some respects, but differ in other respects. By analogy to the role of fringe as a modifier of Notch, it is possible that brainiac acts in the germ line to modify Delta and contribute to signaling between germ line and somatic cells. However, it is also possible that brainiac acts differently to influence multiple interactions between germ line and somatic cells.

In the present study we demonstrate that Drosophila brainiac is a 3GlcNAc-transferase similar to fringe. However, brainiac has different acceptor substrate specificity and transfers to -linked mannose as well as -linked galactose residues. The core structure of Drosophila glycosphingolipids consists of mactosylceramide (Man1-4Glc1-Cer; MacCer), and this is extended by a 1-3 linked GlcNAc residue to the terminal mannose residue (8). It is suggested that brainiac has its primary role in glycosphingolipid biosynthesis and thereby affects glycosphingolipid mediated receptor modulation and functions mediated by lipid rafts.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Phylogenetic Analysis of the Drosophila and Mammalian 3-Glycosyltransferase Family-- tBLASTn analysis with brainiac was used to search the Drosophila melanogaster whole genome data base GadFly released by the Berkeley Drosophila Genome Project (BDGP, Release 2 and 3) (9). Computed gene sequences were manually revised using EST cDNA information and the Drosophila Gene collection (DGC Release 1 and 2 (10)). Amino acid sequences representing a central evolutionarily conserved domain of brainiac and orthologous Homo sapiens 3-glycosyltransferases were aligned with the identified D. melanogaster proteins using ClustalX 1.8 with Gonnet 250 protein weight matrix and default gap penalties (11). Multiple sequence alignments were revised for maximal residue conservation and minimal gaps in conserved amino acid sequence motifs. Distance analyses of the amino acid alignments were performed as described (12).

Expression of Brainiac in Insect Cells-- An expression construct of the full coding region of brainiac was prepared by PCR using D. melanogaster (Canton S) genomic DNA (CLONTECH) with the sense primer MAB1 (5'-AGCGGATCCGCCATGCAAAGTAAACACCGC-3') and the antisense primer MAB3 (5'-AGCGGATCCTGCTACGCGTAATTGGCGG-3') with BamHI restriction sites. The PCR product was cloned into the BamHI site of pVL1393 (PharMingen). Two N-terminal truncated constructs were prepared to obtain soluble secreted brainiac. One construct, pAcGP67-brainiac-sol, designed to encode amino acid residues 23-325 of brainiac was prepared by PCR using the primer pair MAB2 (5'-AGCGGATCCGACTATTGCGGCCTGCTGACC-3') and MAB3 with BamHI restriction sites. The PCR product was cloned into the BamHI site of pAcGP67B (PharMingen). A second construct, pAcGP67-brainiac-HIS-sol, designed to encode a N-terminal fusion of His₆ and T7 tags to amino acid residues 28-325 of brainiac was prepared by PCR using the primer pair TSHC273 (5'-CGAGGATCCGCTGACCCACCTGCACGAG-3') and MAB3 with BamHI restriction sites. The PCR product was fused to cDNA for six histidine residues and a T7 tag interspaced by a thrombin proteolytic site, and the expression unit was cloned into the NotI restriction site of pAcGP67A (PharMingen). pVL1393-brainiac-full, pAcGP67-brainiac-sol, and pAcGP67-brainiac-HIS-sol were co-transfected with Baculo-Gold^TM DNA (PharMingen) in Sf9 cells as described (13). Control constructs included pVL-fringe-myc and pVL-fringe-NNN-myc (4), where NNN represents an enzymatically inactive DxD motif mutant. Standard assays were performed in 50-µl total reaction mixtures containing 25 mM HEPES-KOH (pH 7.4), 10 mM MnCl₂, 0.1% n-octyl glucoside, 100 µM UDP-[¹⁴C]GlcNAc (2,300 cpm/nmol) (Amersham Biosciences), and varying concentration of acceptor substrates (Fluka, Merck, Sigma, and Toronto Research Chemicals Inc.; see Table I for structures). For microsomal preparations High Five^TM cells were lysed in 10 volumes of hypotonic lysis buffer (25 mM HEPES-KOH (pH 7.4), 10 mM MnCl₂, 1% N-octyl glucoside, 1 mM phenylmethylsufonylfluoride), and membrane pellets obtained by differential centrifugation at 15,000 × g, followed by 150,000 × g were used at 150 mg/ml. The secreted brainiac constructs were assayed with 5-20 µl of culture supernatant from infected cells. The His- and T7-tagged secreted brainiac protein was purified by nickel-nitrilotriacetic acid affinity chromatography (Qiagen) as described by the manufacturer. In addition this protein was immunoprecipitated with anti-T7 antibody agarose (Novagen) as described by the manufacturer, and enzyme assays were performed on washed beads as well as after elution at low pH. Reaction products of soluble acceptors were quantified by chromatography on Dowex 1-X8 (Sigma). Assays with glycosphingolipids included 5 mM 2-acetamido-2-deoxy-D-glucono-1,5-lactone (inhibitor of hexosaminidase activity), and products were purified on octadecyl-silica cartridges (Supelco) and analyzed by high performance thin-layer chromatography (HPTLC) and autoradiography.

Isolation of CDH and CTH from High Five^TM Insect Cells-- High Five^TM cells were grown in shaking upright roller bottles at 27 °C in serum-free medium (Invitrogen). Approximately 50 ml of packed cells were extracted in 2-propanol-n-hexane-water (55/25/20, v/v/v, upper phase removed) and subjected to Folch partition in chloroform-methanol-water (4/2/1, v/v/v). The dried lower phase glycolipids were freed from other lipids by peracetylation (pyridine-acetic anhydride 2:1 v/v), chromatography on Florisil, and base-catalyzed O-deacetylation (14) and analyzed by HPTLC. Further fractionation of wild type High Five^TM glycolipids was carried out by preparative-scale high performance liquid chromatography. The structures of the di- and triglycosylceramide fractions were determined by ¹H NMR spectroscopy and electrospray ionization mass spectrometry to be Man1-4Glc1-1Cer and Gal1-4Man1-4Glc1-1Cer, respectively.² Total CTH from control and brainiac transfected High Five cells were isolated from the crude lower phase lipids by preparative HPTLC and compared by MALDI-TOF mass spectrometry as described below.

Isolation of the Products Formed by Brainiac-- The products formed by brainiac with Man--methylumbelliferone (Man1-MeUmb) (4 mg), High Five^TM MacCer (2 mg), and human Gal1-4Glc1-Cer (2 mg) were analyzed. The substrates were glycosylated with microsomes prepared from High Five^TM cells infected with pVL-brainiac-full using thin-layer chromatography to monitor reaction progress. The reaction products were purified by application to octadecyl-silica cartridges (Bakerbond, J. T. Baker), followed by stepwise elution with increasing concentrations of methanol in water (13); the product of reaction with MacCer was further purified by preparative-scale HPTLC (chloroform-methanol-0.5% aqueous CaCl₂, 50:40:10 v/v/v), with isolation of the fraction migrating as a trihexosylceramide, prior to NMR and mass spectrometry analysis as described below.

¹H and ¹³C NMR Spectroscopy-- Glycosphingolipid products were deuterium exchanged by repeated addition of CDCl₃-CD₃OD 2:1, sonication, and evaporation under nitrogen, then dissolved in 0.5 ml of Me₂SO-d₆/2% D₂O (containing 0.03% tetramethylsilane as chemical shift reference) for NMR analysis. A one-dimensional ¹H NMR spectrum was acquired on the product with LacCer on a 600 MHz Varian Inova spectrometer at 35 °C, with solvent suppression by presaturation pulse. Its identity was established by comparison of the spectrum with those of relevant standards acquired under identical conditions (15). For the product with MacCer, one-dimensional ¹H, two-dimensional ¹H-¹H gCOSY, TOCSY, and NOESY NMR spectra were acquired on a Varian Inova 800 MHz spectrometer at 55 °C. The product with Man1-MeUmb eluting from the octadecyl-silica cartridge with 20 and 30% MeOH was deuterium exchanged by repeated lyophilization from D₂O and dissolved in 100% D₂O (containing a trace of acetone as chemical shift reference) for NMR analysis. One-dimensional ¹H, two-dimensional ¹H-¹H gCOSY and TOCSY, and two-dimensional ¹H-detected ¹H-¹³C gHSQC and gHMBC NMR spectra were acquired on the Varian Inova 600 MHz spectrometer at 20 °C; a directly detected one-dimensional ¹³C NMR spectrum was acquired on a Varian Inova 500 MHz spectrometer.

MALDI-TOF Mass Spectrometry-- Molecular mass profiles of glycosphingolipids were acquired on an Axima-CFR (Shimadzu/Kratos Analytical, Manchester, UK) MALDI-TOF mass spectrometer operating in positive ion reflectron mode (pulsed N₂ laser with delayed extraction; emission wavelength 337 nm; acceleration potential 5 kV). The matrix employed was 2,5-dihydroxybenzoic acid; samples were premixed with a solution of 2,5-dihydroxybenzoic acid (10 mg/ml) in acetonitrile-0.1% trifluoroacetic acid (1:1, v/v) prior to application and drying on the target. Molecular species were detected as their Na⁺ adducts; angiotensin II and Pro¹⁴-Arg were used as external mass calibration standards (monoisotopic masses 1046.5 and 1533.9 Da, respectively).

Exoglycosidase Digestion-- Brainiac products formed with lactose, Gal1-4Man, and D-mannose were prepared by incubating 1 µmol of acceptor sugars with 100 nmol of UDP-[¹⁴C]GlcNAc (3900 cpm/nmol) and brainiac microsomes in reaction buffer. The reaction products were purified on Dowex 1-X8 and octadecyl-silica cartridges and freeze-dried. The resolubilized products were digested with 10 units of -galactosidase (Escherichia coli, Sigma) for 2 h or 312 milliunits of -N-acetylglucosaminidase (jack bean, Sigma) overnight and purified by mixed bed resin chromatography (Sigma) followed by lyophilization. Resolubilized samples were analyzed by thin-layer chromatography in chloroform-methanol-water (30:60:10 v/v/v) and autoradiography.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Brainiac Encodes an UDP-GlcNAc:Man/Gal 1,3GlcNAc-transferase-- Expression of the full coding region of brainiac resulted in marked increase in GlcNAc-transferase activity using free D-mannose in a screen assay as developed for study of the activity of fringe (4). Analysis of activities with monosaccharides at varying concentrations up to 500 mM with brainiac and fringe is shown in Fig. 1. D-Mannose was the best substrate for brainiac, but at high concentrations L-fucose, and to a lesser extent D-galactose, was used as acceptor as well. The activity of brainiac with fucose was comparable with that of fringe, and fringe also appeared to have weak activity with mannose. It should be noted that the assays were quantitated for total microsomal protein, but the relative levels of enzyme expressions are unknown, and the putative products with L-fucose were not characterized. The activities were only measurable at acceptor concentrations over 50 mM. In the case of fringe it did originally indicate that this enzyme functioned with O-linked fucose (4), and later studies suggest that mammalian fringe variants lunatic and manic exhibit distinct specificities for the peptide sequence carrying O-Fuc (16). The high activity at low concentrations brainiac exhibits with other substrates than fucose clearly indicates that brainiac does not function in glycosylation of O-linked fucose. Table I summarizes activities obtained with a large panel of saccharide and saccharide derivative substrates using UDP-GlcNAc donor sugar nucleotide. No activity above background values was obtained with other donor sugar nucleotides (UDP-Glc, UDP-Gal, UDP-GalNAc, and UDP-xylose) (not shown). Analysis of substrates containing terminal -linked mannose (-Man) and -linked mannose showed strong preference for -Man with monosaccharide derivatives and near exclusive activity for -Man structures with disaccharides and larger. In agreement with free galactose serving as substrate several disaccharides with terminal -Gal were used as acceptor. Gal1-4Man, lactose, and benzyl--lactose were substrates, while related N-acetylated structures (Gal1-4ManNAc, N-acetyllactosamine, benzyl--N-acetyllactosamine) were poorly active. Analysis of apparent K_m for the most active substrates identified showed that Man1-MeUmb was the preferred acceptor substrate, and the disaccharides Gal1-4Man and Gal1-4Glc were used with significantly lower affinity (Table II).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   3GlcNAc-transferase activities of brainiac and fringe with monosaccharides. Microsomes of transfected High FiveTM cells (3 mg of protein) were used as enzyme sources. , brainiac activity with D-mannose; , fringe activity with D-mannose; , brainiac activity with L-fucose; , fringe activity with L-fucose; , brainiac activity with D-galactose; , fringe activity with D-galactose. Monosaccharides D-glucose, D-GalNAc, D-GlcNAc, and D-xylose are not included, as assays at 5, 50, and 500 mM did not detect enzyme activity. Background values obtained with identically treated microsomal fractions expressing pVL-fringe-NNN-myc were subtracted.

View this table: [in this window] [in a new window]   Table I Substrate specificities of brainiac 1-3-N-acetylglucosaminyltransferase

View this table: [in this window] [in a new window]   Table II Kinetic properties of brainiac 1-3-N-acetylglucosaminyltransferase

Several attempts to generate a soluble secreted brainiac protein with good catalytic activity were performed. Initially, we used an untagged construct based on amino acid residues 23-325, but no activity was found in the culture medium of infected insect cells (not shown). Second, a N-terminally tagged construct based on amino acid residues 28-325 was expressed and a specific protein reactive with an anti-HIS antibody of predicted molecular weight was found by Western blot SDS-PAGE analysis. Low activity with Man1-MeUmb acceptor substrate was detected in the culture medium (~2-3-fold over background). Purification of the His-tagged protein by nickel-nitrilotriacetic acid chromatography resulted in purification of an inactive protein (not shown). Immunoprecipitation with anti-T7 antibody resulted in specific precipitation of low activity (~3-4-fold over background), but elution inactivated the protein. We did not further pursue the enzymatic properties of the soluble brainiac protein. In our experience several of the glycosyltransferases acting exclusively in the glycolipid biosynthetic pathways are enzymatically inactive as truncated soluble recombinant proteins expressed in insect cells (17, 18).

Several parameters for the brainiac assay with microsomal fractions were analyzed for optimization. The most critical parameter was found to be the detergent solubilization. Triton X-100, Triton CF-54, and Nonidet P-40 had strong inhibiting effect on activity at 0.1%, while n-octyl glucoside at 3.4 mM (0.1%) activated the enzyme. The pH optimum of brainiac activity was neutral (pH 7.4). Addition of 5-10 mM MnCl₂ activated enzyme activity, and MgCl₂ and CaCl₂ had no effect, while the presence of EDTA destroyed the activity.

The product of brainiac with Man1-MeUmb was determined by NMR analysis to be GlcNAc1-3Man1-MeUmb, as follows. As shown in Fig. 2A, the fraction of product eluted from octadecyl-silica by 20% MeOH exhibited sets of ¹H resonances indicating a mixture of two compounds (two sets of -Man H-1 and H-2 signals in proportion ~6:4, at 5.408/4.339 and 5.436/4.242 ppm, respectively). One set clearly corresponds to unreacted starting material and the other to a product of glycosylation by -GlcNAc (additional H-1 signal at 4.757 ppm, ³J_1,2 = 8.5 Hz, NAc signal at 2.085 ppm). Following complete assignment of ¹H and ¹³C resonances from all three monosaccharide spin systems present (see Table III) by two-dimensional ¹H-¹H gCOSY and TOCSY, one-dimensional ¹³C, and two-dimensional ¹H-detected ¹H-¹³C gHSQC experiments (not shown), the connectivity between the -GlcNAc and the more abundant -Man spin system was established unambiguously as a 13 linkage by a two-dimensional gHMBC experiment. This spectrum (Fig. 2B) shows clear interglycosidic three-bond correlations between the -GlcNAc H-1 and the downfield-shifted -Man C-3 (79.61 versus 72.44 ppm), as well as between the corresponding -HexNAc C-1 (98.81 ppm) and the downfield-shifted -Man H-3 (3.986 versus 3.804 ppm). Slight upfield shifts of the corresponding -Man C-2 and C-4 resonances in the product compared with the non-glycosylated starting material (Table III) are also consistent with the 13 linkage.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   600-MHz NMR spectra (100% D2O, 20 °C) of product of brainiac with Man--methylumbelliferone. A, expansion of monosaccharide ring methine and hydroxymethyl proton region of one-dimensional 1H NMR spectrum; B, corresponding region of two-dimensional 1H-detected 1H-13C gHMBC spectrum. Interglycosidic three-bond correlations are marked by ovals in B. MU, 4-methylumbelliferone; S, resonances corresponding to starting material; P, resonances corresponding to product.

View this table: [in this window] [in a new window]   Table III 1H, 13C chemical shifts (ppm) and 3J1,2 coupling constants (Hz, in parentheses) for Man1-MeUmb substrate and biosynthetic GlcNAc1-3Man1-MeUmb product

Brainiac showed high activity with the disaccharides Gal1-4Man and Gal1-4Glc. D-Galactose and Gal-MU were poor substrates with no activity detected at 20 mM. D-Galactose in excess of 100 mM did, however, show significant activity. Since the enzyme transfers GlcNAc to both D-galactose and D-mannose monosaccharides, it was necessary to determine which sugar served as the acceptor in the disaccharide substrate Gal1-4Man. This was done by analysis of sensitivity to exoglycosidase treatment. The di- and trisaccharide products with lactose, Gal1-4Man, and D-mannose were digested by -N-acetylglucosaminidase and not by -galactosidase treatment suggesting that the structures of the brainiac products are GlcNAc1-3Gal1-4Glc, GlcNAc1-3Gal1-4Man, and GlcNAc1-3Man, respectively (Fig. 3).

View larger version (58K): [in this window] [in a new window]   Fig. 3.   Exoglycosidase digestion of brainiac products formed with Gal1-4Man, Gal1-4Glc and D-Mannose. Autoradiography of high perfomance thin-layer chromatography of brainiac products digested with jack bean -N-acetylglucosaminidase (lanes 1-6) and E. coli -galactosidase (lanes 7-12). Chromatography of purified saccharides was performed in chloroform-methanol-water (30/60/10, v/v/v). The migration of standard mono- and disaccharides is indicated. Complete digestion of disaccharide substrates by -galactosidase was observed by orcinol staining (not shown). -GlcNAcase, -N-acetylglucosaminidase.

Brainiac Functions in Glycosphingolipid Synthesis-- -Linked mannose is rare in eukaryotic glycoconjugates. The preformed dolichol-phosphate oligosaccharide precursor for N-glycosylation contains a Man1-4GlcNAc linkage, but this only serves as a substrate for -mannosyltransferases. Brainiac was not active with hen egg albumin tested as described previously (19), indicating that high mannose and hybrid-type N-glycans do not serve as substrates (data not shown).

On the other hand, the core dihexosylceramide (CDH) of glycosphingolipids from flies (diptera), including D. melanogaster, and nematodes, including Caenorhabditis elegans, has been reported to be Man1-4Glc1-1Cer (MacCer), and this is extended by 1-3-linked GlcNAc in all structures characterized to date from these species (8, 20-22). We therefore isolated MacCer as well as a major glycolipid migrating as a trihexosylceramide (CTH) from High Five^TM insect cells. The latter was found to have the novel structure, Gal1-4Man1-4Glc1-1Cer, instead of the GlcNAc1-3 terminated structure. As shown in Fig. 4 brainiac uses both LacCer (lane 6) and MacCer (lane 7). The enzyme source used was a detergent-solubilized microsomal fraction, and some endogenous products were produced (lane 5). These endogenous products appear to be genuine products of brainiac as control experiments with fringe microsomes did not produce these (lane 1). Two endogenous products migrating in the CTH region and one migrating in the ceramide tetrasaccharide region were observed. The identities of these remain unknown. Similarly, the endogenous product migrating in the ceramide tetrasaccharide region is likely to be GlcNAc1-3Gal1-4Man1-4Glc1-1Cer based on the existence of Gal1-4Man1-4Glc1-1Cer in High Five^TM cells. The fastest migrating endogenous product almost co-migrated with the product formed with LacCer. This may suggest that LacCer is found in High Five^TM cells in addition to the more predominant MacCer. Co-existence of MacCer and minor amounts of LacCer was previously reported in Calliphora vicina (23). The product formed with LacCer migrated slightly faster than the product formed with MacCer.

View larger version (69K): [in this window] [in a new window]   Fig. 4.   Brainiac uses MacCer and LacCer glycosphingolipid substrates. Assays were performed with microsomal fractions of High FiveTM cells expressing full coding constructs of fringe (lanes 1-4) and brainiac (lanes 5-8). Autoradiography of high performance thin-layer chromatography of reaction products (4 h) purified by Sep-Pak C-18 chromatography. Plate was run in chloroform-methanol-water (60/38/10, v/v/v). Migration of standard glycolipids is indicated. H5CTH, High FiveTM ceramide trihexoside.

A one-dimensional ¹H NMR spectrum of the crude triglycosylceramide product formed with LacCer (not shown) exhibited resonances consistent with virtually complete conversion to GlcNAc1-3Gal1-4Glc1-1Cer, i.e. anomeric signals at 4.620, 4.264, and 4.168 ppm (³J_1,2 = 8.0, 7.3, and 7.8 Hz, respectively), corresponding to H-1 of GlcNAc1-3, Gal1-4, and Glc1-1 residues, as well as signals at 3.837 ppm, corresponding to H-4 of Gal1-4 (³J_3,4 = 2.6 Hz), and at 1.836 ppm (singlet, ³H), corresponding to NAc of GlcNAc1-3, of this glycosphingolipid (compared with published values for these resonances in an NMR study of an authentic standard: 4.621, 4.265, 4.168 ppm (³J_1,2 = 7.9, 7.3, and 7.9 Hz, respectively); 3.839 ppm (³J_3,4 = 2.4 Hz); 1.837 ppm (singlet, 3H] (15)).

The triglycosylceramide product formed with MacCer was purified by preparative HPTLC (see Fig. 5, lane 6), and confirmed by MALDI-TOF mass spectrometry and by one-dimensional ¹H and two-dimensional ¹H-¹H NMR spectroscopy to be authentic GlcNAc1-3Man1-4Glc1-1Cer (Ap₃Cer). The major ions in the mass profile (Fig. 6A) are consistent with Na⁺ adducts of a glycosphingolipid product having the glycan formula HexNAc·Hex₂ attached to ceramides composed of d14:1 and d16:1 sphing-4-enines N-acylated with 18:0, 20:0, 22:0, and 24:0 fatty acids (predominantly d14:1/20:0 and d14:1/22:0, m/z 1087.5 and 1115.6, respectively), a profile reflecting the origin of the acceptor substrate, MacCer isolated from High Five^TM cells.² NMR spectroscopic assignments for all ¹H resonances in starting material and product, derived from high resolution gCOSY and TOCSY experiments, are compiled in Table IV. As shown in Fig. 7, the one-dimensional ¹H NMR spectrum of the product (A), compared with that of the starting material (B), exhibits an additional anomeric resonance, connected to a -GlcNAc spin system (H-1 at 4.539 ppm, ³J_1,2  8 Hz, overlapping the -Man H-1 at 4.534 ppm). Although the near coincidence of the two H-1 signals in the product spectrum made it difficult to establish the linkage between the -GlcNAc and -Man by two-dimensional NOESY experiments (relevant correlation peaks were insufficiently resolved, even at 800 MHz; not shown), a comparison of the chemical shifts of all ¹H resonances for -Man in the starting material and product shows that the largest downfield glycosylation-induced shift change occurs for H-3 (3.470 versus 3.270 ppm;  = 0.20 ppm). Such a large downfield shift change is generally a reliable indication of the position of glycosylation, in the absence of unusual conformational effects or interactions between vicinally substituted residues, and in light of the linkage specificity already established for the Man1-MeUmb product, the identity of the product with MacCer appears to be confirmed.

View larger version (43K): [in this window] [in a new window]   Fig. 5.   HPTLC analysis of glycosphingolipids from High FiveTM cells infected with baculovirus expressing full coding brainiac. Compared are total Folch lower phase glycosphingolipids from High FiveTM cells infected with baculovirus expressing an irrelevant protein (lane 1), High FiveTM cells infected with baculovirus expressing full coding brainiac (lane 2), authentic Gal1-4Man1-4Glc1-1Cer (Hi5-3) isolated and previously characterized from High FiveTM cells (lane 3), total CTH fraction from High FiveTM cells infected with baculovirus expressing an irrelevant protein (lane 4), total CTH fraction from High FiveTM cells with baculovirus expressing full coding brainiac (lane 5), Ap3Cer produced by in vitro enzymatic glycosylation of High FiveTM Man1-4Glc1-1Cer (MacCer) with brainiac (lane 6). Solvent system was chloroform-methanol-0.5% aqueous CaCl2 (50:40:10 v/v/v), and plates were developed with orcinol H2SO4 staining.

View larger version (29K): [in this window] [in a new window]   Fig. 6.   MALDI-TOF mass spectrometry of purified triglycosylceramide fractions from High FiveTM cells infected with baculovirus expressing full coding brainiac. Compared are sodiated molecular ion profiles of GlcNAc1-3Man1-4Glc1-1Cer produced by in vitro enzymatic glycosylation of: Man1-4Glc1-1Cer with brainiac (A), total CTH fraction from control-transfected High FiveTM cells (B), total CTH fraction from brainiac-transfected High FiveTM cells (C). Only monoisotopic peaks are labeled. Fractions analyzed by MALDI-TOF are the same as those analyzed by HPTLC shown in Fig. 5 (corresponding to lanes 6, 4, and 5, respectively).

View this table: [in this window] [in a new window]   Table IV 1H chemical shifts (ppm) and 3J1,2 coupling constants (Hz, in parentheses) for Man1-4Glc1-1Cer substrate and biosynthetic GlcNAc1-3Man1-4Glc1-1Cer product

View larger version (24K): [in this window] [in a new window]   Fig. 7.   Downfield region of 800-MHz 1H NMR spectrum (Me2SO-d6/2% D2O, 55 °C) of GlcNAc1-3Man1-4Glc1-1Cer produced by in vitro enzymatic glycosylation of Man1-4Glc1-1Cer with brainiac (A). A spectrum of the starting material acquired under identical conditions is reproduced in B. Arabic numerals refer to ring protons of residues designated by Roman numerals or capital letters in the corresponding structure. R refers to protons of the sphingosine backbone; cis refers to vinyl protons of unsaturated fatty-N-acyl components. Asterisks denote resonances from other lipid impurities; two asterisks appear where large interfering peaks have been truncated for clarity.

Fig. 5 presents an HPTLC comparison of Folch lower phase glycosphingolipids extracted from High Five^TM cells infected with a baculovirus expressing an irrelevant gene (lane 1) with those from High Five^TM cells infected with baculovirus containing pVL1393-brainiac-full (lane 2). In the CTH region of High Five^TM cells infected with an irrelevant virus (lane 1) the major band corresponds to Gal1-4Man1-4Glc1-1Cer (Hi5-3). A small amount of Ap₃Cer, which migrates with a relative mobility slightly higher than Hi5-3 (compare migration of authentic standards analyzed under identical conditions, lanes 6 and 3, respectively), can also be observed in the wild type profile. This is consistent with results obtained previously.² In the CTH region of the pVL1393-brainiac-full profile (lane 2), the amount of the higher R_f band is considerably greater, with intensity of staining almost equal to that of the Hi5-3 band. The identity of this higher R_f component was confirmed as follows. Total CTH fractions from control and brainiac infected High Five^TM cells lower phase glycosphingolipids were isolated by preparative HPTLC; these fractions (Fig. 5, lanes 4 and 5, respectively) were then analyzed by MALDI-TOF mass spectrometry under identical conditions (Fig. 6, B and C, respectively). The major ions in the control transfected CTH spectrum (B) are consistent with Na⁺ adducts of a glycosphingolipid with the glycan formula Hex₃Cer, having a ceramide profile qualitatively and quantitatively similar to that already described above for the biosynthetic Ap₃Cer (again predominantly d14:1/20:0 and d14:1/22:0, m/z 1046.4 and 1074.5, respectively). The presence of a small amount of Ap₃Cer in the wild type profile is indicated by a set of less abundant Na⁺ adduct ions at m/z +41 increments (compare profile in A). In the CTH spectrum from brainiac-transfected cells (C) the sets of Na⁺ adduct ions corresponding to HexNAc·Hex₂Cer and Hex₃Cer are almost equal in abundance, consistent with a substantial increase in the amount of Ap₃Cer relative to that of Gal1-4Man1-4Glc1-1Cer, as observed by HPTLC analysis.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The neurogenic gene brainiac was shown to encode a 3GlcNAc-transferase with broad acceptor substrate specificity having preference for -Man but also showing significant activity with -Gal terminating structures. Mannose-linked 1-4 is found in the core structure of insect and nematode arthroseries glycolipids as mactosylceramide (8), and brainiac was shown to have activity with this glycolipid. Brainiac also showed activity with another disaccharide glycolipid, lactosylceramide, which represents the equivalent core structure upon which vertebrate glycosphingolipids are built. With the present knowledge of structures of Drosophila glycoconjugates mactosylceramide is the only likely natural substrate identified, indicating that brainiac serves an important function in the biosynthesis of glycosphingolipids.

The acceptor substrate specificity of brainiac with various mono- and disaccharides and aglycon derivatives revealed clear preference for -linked mannose (Table I). Furthermore, brainiac showed preference for dihexosides (Gal1-4Glc and Gal1-4Man), whereas disaccharides with penultimate N-acetylglucosamine represented poor substrates (Gal1-4GlcNAc and Man1-4GlcNAc). In agreement with this brainiac used MacCer and LacCer glycolipid substrates in in vitro tests. Extended glycosphingolipids of Drosophila and other dipterans have been reported to contain two 1-3 linked GlcNAc residues (e.g. Gal1-3GalNAc1-4GlcNAc1-3Gal1-3GalNAc1-4GalNAc1-4GlcNAc1-3Man1-4Glc1-Cer) (20, 24). Since brainiac showed poor activity with disaccharide structures containing internal n-acetylhexosamine and no activity with the disaccharide Gal1-3GalNAc1-benzyl, it appears unlikely that brainiac also catalyzes the addition of the outer GlcNAc residue (Table I).

Drosophila glycosphingolipids have all been reported so far to be based on the Ap₃Cer core structure and extended as discussed above. It is noteworthy though that relatively few studies have addressed Drosophila glycosphingolipids, as well as insect glycosphingolipids in general, compared with studies of vertebrate glycosphingolipids. In the biosynthesis of vertebrate glycosphingolipids built on LacCer, an important branch point occurs at the addition of the third monosaccharide residue. This is the determining step for synthesis of different classes of glycosphingolipids, designated (neo)lactoseries (GlcNAc1-3Gal1-4Glc1-Cer), (iso)globoseries Gal1-3/4Gal1-4Glc1-Cer), and ganglioseries (GalNAc1-4Gal1-4Glc1-Cer) (25). These classes of glycolipids are differentially expressed in cell types and during cell differentiation (26-28) and have different properties and functions (25). There may be no analogous branch point in the biosynthesis of Drosophila glycosphingolipids, as only one CTH sequence, GlcNAc1-3MacCer (Ap₃Cer), has been reported from this species. On the other hand, it is possible that additional structures and pathways exist. Although Gal1-4Man1-4Glc1-Cer was found as the major CTH component in High Five^TM cells, and this structure may represent an aberrant pathway confined to cultured insect cells, its appearance implies that a Gal1-4 transferase with substrate specificity for MacCer must be present in the insect repertoire. It is also possible that the alternative pathway relates to the lepidopteran, rather than dipteran, origin of the cell line. We were not successful in establishing a stable brainiac transfectant of High Five^TM cells, but analysis of the glycosphingolipid profile of baculovirus infected cells showed that brainiac functioned and produced a significant shift in trihexoside ceramides to Ap₃Cer. Brainiac is homologous to vertebrate 3GlcNAc-transferase enzymes that control the (neo)lactoseries pathway in vertebrates by forming GlcNAc1-3Gal1-4Glc1-Cer (3, 18, 29-32), and brainiac was found to use LacCer similarly to the homologous mammalian 3GlcNAc-transferases (Fig. 4). This provides strong support for the proposed role for brainiac in glycosphingolipid biosynthesis from a functional perspective.

The proposed function for brainiac in Drosophila glycosphingolipid biosynthesis implies that brainiac mutants may lack extended glycosphingolipids. Drosophila does not appear to have close brainiac homologs, which would be predicted to have similar functions (Fig. 8). More distant Drosophila homologs group independently or with vertebrate orthologs known to represent 3-galactosyltransferases. It is therefore possible that this class of glycosphingolipids cannot be produced in brainiac mutant animals. Brainiac is required in the germ line during oogenesis and is also expressed zygotically. At present it is not technically feasible to isolate sufficient numbers of maternally and zygotically mutant animals to permit analysis of the glycosphingolipid composition. Failure to extend glycosphingolipids beyond MacCer could also lead to lack of acidic and zwitterionic glycosphingolipids in Drosophila, which contain glucuronic acid linked to galactose residues and phosphoethanolamine linked to GlcNAc residues (8, 24). Charged residues, glucuronic acid and sialic acids, of glycoconjugates are important for biological functions in vertebrates (33), and it is likely that glucuronic acid and phosphoethanolamine exert important functions in Drosophila as well.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Phylogram of 3-glycosyltransferases in D. melanogaster and H. sapiens. The consensus tree from protein distance analyses of predicted catalytic 3-glycosyltransferase domains is based on progressive sequence alignments as described under "Experimental Procedures." Bootstrap percentage values from 1000 replicates are indicated above the nodes. Putative D. melanogaster 3-glycosyltransferases are indicated by their GadFly annotation. 3GalT, 3GnT, and 3GalNAcT indicate human 3-galactosyltransferases, 3-N-acetylglucosaminyltransferases, and 3-N-acetylgalactosaminyltransferases, respectively. Phylogenetic subfamilies are indicated by alternate background shading. Sequence alignments included amino acids 78-315 of human 3GalT1 (GenBankTM accession number E07739), 151-394 of 3GalT2 (GenBankTM accession number Y15060), 78-320 of 3GalNAcT1 (3GalT3, GenBankTM accession number Y15062), 71-343 of 3GalT4 (GenBankTM accession number Y15061), 56-296 of 3GalT5 (GenBankTM accession number AB020337), 57-313 of 3GalT6 (GenBankTM accession number AY050570), 142-387 of 3GnT2 (GenBankTM accession number AB049584), 107-355 of 3GnT3 (GenBankTM accession number AB049585), 118-360 of 3GnT4 (GenBankTM accession number AB049586), 88-333 of 3GnT5 (GenBankTM accession number AB045278), and 117-367 of 3GnT6 (GenBankTM accession number AB073740). The analysis included amino acids 78-316 of D. melanogaster brainiac (GenBankTM accession number U41449), 85-342 of CG3038 (DGC accession number AY061226), 109-359 of CG8976 (DGC accession number AY071036), 99-366 of the predicted CG8734 protein sequence, 329-563 of the predicted CG8668 protein, and 134-387 of the predicted CG11357 protein. The predicted CG8673 cDNA sequence was manually revised as described (12) and found to encode a protein of 364 amino acids; amino acids 109-350 were included in the analysis.

The large vertebrate 3-glycosyltransferase family homologous to brainiac (29) (Fig. 8) has been extensively characterized within the last few years. Functional subgroups with considerable apparent redundancies have been identified, and many of these have been assigned important roles in the biosynthesis of all glycosphingolipid classes in mammals. One group is represented by UDP-Gal:GlcNAc 3-galactosyltransferases, 3Gal-T1, -T2, and -T5, which are predicted to control synthesis of type 1 chain lactoseries structures on glycolipids and N- and O-linked glycoproteins (18, 34, 35). All three function in vitro with glycolipids, whereas only 3Gal-T2 has activity with N-linked glycoproteins, and only 3Gal-T5 functions with O-linked core 3 structures (29). Note that murine 3Gal-T3 was originally erroneously proposed to function in lactoseries synthesis (36); however, 3Gal-T3, renamed as 3GalNAc-T1, is unique and functions in globoseries glycolipid biosynthesis forming GalNAc1-3Gal1-4Gal1-4Glc1-Cer (37). Surprisingly, a recent report indicated that this gene was essential in mice (38). However, the orthologous gene in man is inactivated in healthy individuals of the rare P^k blood group (39). 3Gal-T4 is also unique and functions in ganglioseries glycolipid biosynthesis forming Gal1-3GalNAc1-4Gal1-4Glc1-Cer (18, 40). Again, 3Gal-T6 was originally erroneously reported as 3GnT with a 3GlcNAc-transferase activity similar to brainiac (41); however, a recent report shows that this gene encodes the Gal-I enzyme involved in the proteoglycan core region synthesis (Gal1-3Gal1-4Xyl1-O-Ser) (42). A single Drosophila ortholog (CG8734) is predicted to have similar enzymatic functions. The human core 1 3Gal-T (Gal1-3GalNAc1-O-Ser/Thr) is only distantly related and groups in an independent clade with two Drosophila orthologs (not shown) (43).

The dendrogram in Fig. 8 based on protein distance analyses of the putative catalytic units of the 3-glycosyltransferase family depicts brainiac in a subfamily with five mammalian orthologs, which are all known to function as 3GlcNAc-transferases. 3GnT2 functions in poly-N-acetyllactosamine synthesis (GlcNAc1-3Gal1-4Glc[NAc]) of glycoproteins and glycolipids (30, 44). 3GnT3 was shown to function as a core 1 extension enzyme (GlcNAc1-3Gal1-3GalNAc1-O-Ser/Thr) (45). The function of 3GnT4 may be related to the function of 3GnT2, although only low activity has been demonstrated thus far (30). 3GnT5 also has similar functions, and it may have a primary function in glycosphingolipid biosynthesis (GlcNAc1-3Gal1-4Glc1-Cer) (31). Finally, the most distant of the close 3GlcNAc-T brainiac orthologs, 3GnT6, was recently shown to represent a core 3 enzyme (GlcNAc1-3GalNAc1-O-Ser/Thr) (32). The mammalian 3GnTs thus all use Gal or GalNAc as acceptor sugar, while brainiac uses both terminal Gal and Man. Human 3GnT2 in contrast to brainiac does not function with MacCer.³

The phenotypes associated with brainiac mutations have led to the proposal that it might modulate the activities of several signaling pathways, including Delta/Notch and TGF/EGF (1, 2, 46). Could the diversity of effects observed in these mutants be due to an influence of glycosphingolipids on signaling? In vertebrates, glycosphingolipids are known to modulate receptor function and signaling pathways through direct interaction with receptors (47, 48) and through formation of lipid rafts which provide structurally distinct membrane domains for localization of receptors and ligands (47, 49). Drosophila lipid rafts have been identified and shown to be enriched with MacCer (50). It is therefore proposed that brainiac exerts its biological functions in Drosophila by directing glycosphingolipid biosynthesis, rather than by directly modifying receptor molecules as established for fringe. brainiac mutants may therefore provide a unique genetically tractable system to study the biological role of glycosphingolipids in vivo.

	ACKNOWLEDGEMENTS

We thank Heidi Geiser (Department of Chemistry, University of New Hampshire) for her expert help in acquisition of MALDI-TOF mass spectra; the University of Georgia Complex Carbohydrate Research Center for use of their Varian 500, 600, and 800 MHz NMR spectrometers; and Dr. John Glushka for providing invaluable assistance with NMR data acquisition.

	FOOTNOTES

^* This work was supported by The Danish Cancer Society, the Velux Foundation, the Danish Medical Research Council, the Lundbeck Foundation, the National Institutes of Health Resource Center for Biomedical Complex Carbohydrates (NIH P41 RR05351), and Biological Research Infrastructure Network-Center for Structural Biology (NIH P20 RR16459).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ These authors contributed equally and should be considered joint first authors.

^¶ Present address: Inst. of Biochemistry II, University of Cologne, Joseph-Stelzmann-Str. 52, 50931 Köln, Germany.

^§§ Present address: Dept. of Molecular Biology, MEM-L71, The Scripps Research Inst., 10550 North Torrey Pines Rd., La Jolla, CA 92037.

^¶¶ To whom correspondence should be addressed: School of Dentistry, Nørre Alle 20, DK-2200 Copenhagen N, Denmark. Tel.: 45-35326835; Fax: 45-35326505; E-mail: henrik.clausen@odont.ku.dk.

Published, JBC Papers in Press, July 18, 2002, DOI 10.1074/jbc.M206213200

² M. Fuller, T. Schwientek, H. Clausen, and S. B. Levery, manuscript in preparation.

³ T. Schwientek, unpublished data.

	ABBREVIATIONS

The abbreviations used are: 3GlcNAc-transferase, UDP-N-acetylglucosamine:acceptor 1,3-N-acetylglucosaminyltransferase; Cer, ceramide; CDH, ceramide dihexoside; CTH, ceramide trihexoside; LacCer, lactosylceramide; MacCer, mactosylceramide; TOCSY, total correlation spectroscopy; gCOSY, gradient-enhanced correlation spectroscopy; NOESY, nuclear overhauser effect spectroscopy; gHSQC, gradient-enhanced heteronuclear single quantum correlation; gHMBC, gradient-enhanced heteronuclear multiple bond correlation; MeUmb, 4-methylumbelliferone; 3Gal-T, UDP-galactose: acceptor 1,3-galactosyltransferase; HPTLC, high performance thin-layer chromatography; MALDI-TOF, matrix-assisted laser desorption ionization time-of-flight; Ap₃Cer, GlcNAc1-3Man1- 4Glc1-1Cer.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES