THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2004 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 279, No. 6, Issue of February 6, pp. 4346 –4357, 2004 Printed in U.S.A.
Functional Characterization of Drosophila Sialyltransferase* Received for publication, September 5, 2003, and in revised form, November 10, 2003 Published, JBC Papers in Press, November 11, 2003, DOI 10.1074/jbc.M309912200
Kate Koles‡, Kenneth D. Irvine§, and Vladislav M. Panin‡¶ From the ‡Department of Biochemistry and Biophysics, Texas A & M University, College Station, Texas 77843 and §Howard Hughes Medical Institute, Waksman Institute, and Department of Molecular Biology and Biochemistry, Rutgers University, Piscataway, New Jersey 08854
Sialylation is an important carbohydrate modification of glycoconjugates in the deuterostome lineage of animals. By contrast, the evidence for sialylation in protostomes has been scarce and somewhat controversial. In the present study, we characterize a Drosophila sialyltransferase gene, thus providing experimental evidence for the presence of sialylation in protostomes. This gene encodes a functional ␣2– 6-sialyltransferase (SiaT) that is closely related to the vertebrate ST6Gal sialyltransferase family, indicating an ancient evolutionary origin for this family. Characterization of recombinant, purified Drosophila SiaT revealed a novel acceptor specificity as it exhibits highest activity toward GalNAc1– 4GlcNAc carbohydrate structures at the non-reducing termini of oligosaccharides and glycoprotein glycans. Oligosaccharides are preferred over glycoproteins as acceptors, and no activity toward glycolipid acceptors was detected. Recombinant Drosophila SiaT expressed in cultured insect cells possesses in vivo and in vitro autosialylation activity toward -linked GalNAc termini of its own N-linked glycans, thus representing the first example of a sialylated insect glycoconjugate. In situ hybridization revealed that Drosophila SiaT is expressed during embryonic development in a tissue- and stage-specific fashion, with elevated expression in a subset of cells within the central nervous system. The identification of a SiaT in Drosophila provides a new evolutionary perspective for considering the diverse functions of sialylation and, through the powerful genetic tools available in this system, a means of elucidating functions for sialylation in protostomes.
Sialic acids compose a large family of negatively charged nine-carbon ␣-keto acids and are typically located at the nonreducing termini of glycans. In mammals, sialic acids have been implicated in a number of important biological processes, such as the regulation of turnover of circulating glycoproteins and erythrocytes, pathogen-host recognition, immune system functioning, and nervous system development (reviewed in Refs. 1 and 2). In addition, alterations in sialic acid biosynthesis often correlate with malignant transformation and tumor progression (3, 4). * This work was supported in part by the Texas Advanced Research Program from the Texas Higher Education Coordinating Board Grant 000517-0069-2001 (to V. M. P.), National Institutes of Health Grant R01-GM54594 (to K. D. I.), and start-up funds from Texas A & M University (to V. M. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ To whom correspondence should be addressed: Dept. of Biochemistry and Biophysics, Texas A & M University, TAMU 2128, College Station, TX 77843-2128. Tel.: 979-458-4630; Fax: 979-862-4718; E-mail:
[email protected].
In vertebrates, the biosynthesis of sialylated glycoconjugates includes a number of enzymatic reactions that convert the precursor sugar, uridine-diphosphate-N-acetylglucosamine, into the donor sugar, cytidine monophosphate N-acetylneuraminic acid (CMP-Neu5Ac)1 (2). This donor molecule is used by sialyltransferases for the linkage-specific sialylation of glycoconjugates. Sialyltransferases usually exhibit strict acceptor specificity, and synthesize one type of linkage between sialic acid and acceptor substrates. Their expression is often regulated, and specific patterns of sialylation may vary at different developmental stages (5). The combined characterization of the biochemical activity and the expression pattern of each sialyltransferase provide an indication of the structure and spatialtemporal distribution of sialylated glycoconjugates, and is thus essential for understanding the in vivo functions of sialic acid. Twenty different sialyltransferases have been cloned in mammals. They synthesize different linkages (␣2– 6, ␣2–3, and ␣2– 8) and differ in their acceptor specificities (5– 8). Structurally, all vertebrate sialyltransferases have a similar molecular architecture. (i) They are type II transmembrane glycoproteins that predominantly reside in the trans-Golgi compartment and have a short N-terminal cytoplasmic tail that is not important for their catalytic activity. (ii) They have a transmembrane anchor domain that contributes to Golgi retention. (iii) They include a stem region of highly variable length (from 20 to 200 amino acids) that is followed by a large C-terminal catalytic domain (5, 9). The catalytic region of all vertebrate sialyltransferases contains three conserved motifs: L (Large), S (Small), and VS (Very Small) (10, 11). These motifs are involved in substrate binding, formation of essential disulfide bonds (L and S (12, 13)), and catalysis (VS (11, 14)). In contrast to deuterostomes (vertebrates, ascidians, and echinoderms), very little is known about the presence and function of sialic acids in protostomes (annelids, arthropods, and mollusks). Data on the occurrence of this carbohydrate in protostomes has been scarce and somewhat controversial. Several efforts to detect sialyltransferase activity (15–17), sialylated glycans (17, 18), or the CMP-Neu5Ac donor sugar (16, 19) in insect cells have failed, but other papers have reported the occurrence of sialic acids or sialyltransferase activity in insects (20 –23). Experiments with insect Sf9 cells stably expressing two mammalian glycosyltransferases, 1– 4-galactosyltransferase and ␣2– 6-sialyltransferase, revealed the biochemical potential of these cells to provide a sugar donor for sialylation (24). However, endogenous sialyltransferase activity was not
1 The abbreviations used are: CMP-Neu5Ac, cytidine monophosphate N-acetylneuraminic acid; Neu5Ac, N-acetylneuraminic acid; LacNAc, Gal1– 4GlcNAc; LacdiNAc, GalNAc1– 4GlcNAc; HPTLC, high performance thin layer chromatography; PNGaseF, peptide-N4-(N-acetyl-glucosaminyl)asparagine amidase F; HA, hemagglutinin; RCA-I, R. communis agglutinin I; SiaT, ␣2– 6-sialyltransferase; LNnT, lactoN-neotetraose; LNT, lacto-N-tetraose.
4346
This paper is available on line at http://www.jbc.org
ST6Gal-related Sialyltransferase from D. melanogaster detected (25, 26). Thus, it was suggested that in insect cells the sialylation pathway may be present as a highly specialized biochemical process that occurs in a tissue- and/or developmental stage-specific manner, or perhaps that sialic acids are present in only a small number of protostome species (2, 27). The availability of complete genomic information recently made Drosophila a powerful model system for elucidating the biochemical components of insect glycosylation pathways. Searches of the Drosophila genome data base revealed the presence of several genes that encode putative orthologues of vertebrate enzymes of the sialic acid biosynthetic pathway, including Neu5Ac phosphate synthase, CMP-Neu5Ac synthase, CMP-Neu5Ac/CMP antiporter, and sialyltransferase (GadFly data base (28); CAZY data base (29); 30, 31).2 A recent study confirmed that Drosophila has a functional Neu5Ac phosphate synthase with significant homology to the human orthologue, which indicates the possibility of a vertebrate-type pathway for CMP-sialic acid biosynthesis in Drosophila (31). Evidence for the presence of sialylated structures in Drosophila has, however, only been reported in one study, and the nature of these sialylated glycoconjugates has not been analyzed further (21). In the present study, we identify and characterize the apparently single vertebrate-type sialyltransferase gene of Drosophila melanogaster, SiaT. We demonstrate that this gene encodes a functional ␣2– 6-sialyltransferase, which is structurally and functionally related to the ST6Gal family of vertebrate sialyltransferases, yet has a distinct acceptor specificity. Analysis of the Drosophila SiaT expression pattern during embryonic development reveals a distinct pattern of expression in the central nervous system, which suggests possible functions of the SiaT gene in neural development. EXPERIMENTAL PROCEDURES
Materials—Bovine ␣1-acid glycoprotein, fetuin, asialofetuin, bovine ␣-lactalbumin, bovine lactoferrin, N-acetyllactosamine, Gal1– 3GlcNAc, Gal1–3GalNAc, asialo-GM1, asialo-GM2, monosialo-GM3, type III gangliosides mixture, and lactoceramide were from Sigma. GalNAc-benzyl, lacto-N-tetraose, and lacto-N-neotetraose were from Calbiochem. Disaccharides with a hydrophobic aglycone, LacNAc-Rg and LacdiNAc-Rg (where Rg is O-(CH2)8CO2CH3 (32)), were the generous gifts from Dr. Monica Palcic (University of Alberta, Canada). GalNAc1– 4GlcNAc1–2Man␣1-O(CH2)7CH3 was from Dr. Johannis P. Kamerling (Utrecht University, The Netherlands). Drosophila mucin-D glycoprotein (33) and Drosophila glycolipid mixture (upper phase (34)) were kindly provided, respectively, by Dr. Andrei Kramerov (UCLA) and Dr. Michael Tiemeyer (Complex Carbohydrate Research Center, University of Georgia). Biotinylated Sambucus nigra (SNA), Ricinus communis agglutinin I (RCA-I), and Wistaria floribunda (WFA) lectins were from Vector Laboratories (Burlingame, CA). Merck HPTLC Silica Gel 60 F254 plates were from EM Science (Gibbstown, NJ). IgG-Sepharose 6 Fast Flow beads were from Amersham Biosciences. CMP[14C]Neu5Ac (12.0 GBq/mmol, 740 KBq/ml) was from PerkinElmer Life Sciences. PNGaseF and -N-acetylhexosaminidasef were from New England Biolabs (Beverly, MA). Purified rabbit IgG was from Jackson ImmunoResearch (West Grove, PA). Asialo-␣1-acid glycoprotein was prepared by acid hydrolysis in 0.02 N HCl for 1 h at 80 °C. Expression Constructs—Construct for the expression of HA-tagged full-length Drosophila SiaT (designated D. SiaT) protein was prepared as follows: (i) by introducing an XbaI site after the last codon of D. SiaT open reading frame in cDNA clone GH27778 by PCR; (ii) in-frame ligation of double-stranded oligonucleotide encoding HA tag (35) followed by 6 histidines and a stop codon into this XbaI site; and (iii) by subcloning the resulting DNA construct into pRmHA-3 expression vector (36). In order to make a construct expressing D. SiaT with Nterminal protein A tag, we PCR-amplified a DNA fragment encoding the IgG-binding domain of Staphylococcus aureus protein A from plasmid pRL715 (a gift from Dr. K. Severinov (37)), ligated this DNA in-frame with PCR-amplified DNA fragment of GH27778 encoding 86 – 451 amino acids of D. SiaT protein, and then inserted the resulting 2
V. M. Panin, unpublished data.
4347
DNA into pMT/BiP/V5-HisA vector (Invitrogen) in-frame with the BiP signal sequence (38). The final construct expressed a D. SiaT fusion protein in which the first 85 amino acids, including the endogenous signal peptide and a part of the stem region, were replaced with BiP signal peptide followed by the IgG-binding domain of S. aureus protein A. Cell Culture—Drosophila S2 cells were maintained and transfected as described (39). Spodoptera frugiperda Sf9 cells were grown as a suspension culture in Erlenmeyer flasks with orbital shaking at 110 rpm using the media described (26). Generation of recombinant baculovirus and production of recombinant protein in baculovirus-infected cells was performed using the Bac-to-Bac expression system (Invitrogen) according to the manufacturer’s protocol. Protein Purification—For purification of protein A-tagged D. SiaT protein, we collected medium from Sf9 cells 72 h after infection with recombinant baculovirus. This medium was incubated with IgG-Sepharose beads on a rotator at 4 °C overnight. Afterward, the beads were extensively washed with phosphate-buffered saline and assay buffer and used in sialyltransferase assays. The amount of purified protein was quantified by Coomassie staining of SDS gels loaded with D. SiaT and a protein standard (bovine serum albumin). Sialyltransferase Assays—Unless otherwise stated, D. SiaT activity was measured in 50 mM cacodylate buffer, pH 6.5, 1% Triton CF-54 (v/v), 3 mM oligosaccharide acceptor, 50 g of glycoprotein or 12.5 g of glycolipids, and 108 M CMP-[14C]Neu5Ac (22.7 mCi/mmol) in a total volume of 20 l at 37 °C for 2 h. Before the assays, IgG-SiaT beads were first autosialylated in the presence of 100 M CMP-Neu5Ac for 1.5 h at 37 °C in order to eliminate competition between autosialylation and the tested acceptors. During assays, IgG-SiaT beads were kept in suspension using a shaking water bath. The reactions were within linearity range for at least 3 h, and no more than 10% of CMP-Neu5Ac and substrates were consumed in any assay. Oligosaccharide samples were spotted directly on HPTLC plates and developed in 1-propyl alcohol/ aqueous ammonia/water (6:1:2.5) solvent system. Glycoproteins were analyzed by SDS-PAGE followed by autoradiography and quantification using the Storm imaging system (Amersham Biosciences). Gangliosides were desalted as described (40) and analyzed by HPTLC in a solvent system of chloroform, methanol, and 0.02% CaCl2 (55:45:10). An aliquot of Drosophila glycolipids was also treated with hydrofluoric acid to remove phosphoethanolamine residues as described (34). After the sialyltransferase assays, Drosophila glycolipids were desalted on Sephadex G-25 columns (41) and analyzed by HPTLC in a buffer system of chloroform, methanol, and 0.25% aqueous potassium chloride (10:10:3). LacNAc-Rg and LacdiNAc-Rg acceptors were also analyzed by HPTLC in the same buffer system without prior desalting. When GalNAcbenzyl was used as an acceptor, the unincorporated donor was removed by Sep-Pak C18 (Waters) column, and then incorporated radioactivity was measured by scintillation counting (32). Km and Vmax values for asialofetuin and lactoferrin were estimated based on the calculated amount of terminal LacNAc or LacdiNAc structures, i.e. 8.8 LacNAc termini per asialofetuin molecule (based on Ref. 42). Terminal LacdiNAc concentration of lactoferrin was estimated at 2.6 LacdiNAc termini per lactoferrin based on Coddeville et al. (43). Because the degree of sialylation can vary between different batches of lactoferrin, we estimated the ratio of sialylated LacdiNAc termini in our lactoferrin sample by comparing D. SiaT-mediated incorporation of [14C]Neu5Ac into Vibrio cholerae sialidase-treated and non-treated lactoferrin. We did not detect any difference in the amount of Neu5Ac incorporation between these two samples; hence we assumed that the ratio of sialylated to non-sialylated LacdiNAc in our commercial lactoferrin sample was negligible. Thus, 2.6 mol of non-sialylated LacdiNAc termini were assumed to be present on each mole of lactoferrin. Linkage Analysis—The linkage and terminal position of incorporated sialic acids were examined using a battery of exoglycosidases. After modification by D. SiaT, glycoprotein and oligosaccharide acceptors were treated with ␣2–3-sialidase (Streptococcus pneumoniae), ␣2–3,6sialidase (Clostridium perfringens), ␣2–3,6,8-specific (V. cholerae) or ␣2–3,6,8,9-sialidase (Arthrobacter ureafaciens) (all from Sigma). The position of sialic acids was also examined by 1–3,4,6-galactosidase (bovine testes, Sigma) treatment. The acceptor sugar residue on D. SiaT was determined by the combination of 1–2,3,4,6-N-acetylglucosaminidase from S. pneumoniae (Calbiochem) and 1–2,3,4,6-N-acetylhexosaminidasef (New England Biolabs) digestions. All experiments with glycosidase treatments included a control reaction (mock treatment) with identical conditions except for the absence of the corresponding glycosidase. The digestion products of oligosaccharides were analyzed by HPTLC, and those of glycoproteins were subjected to PAGE, followed by autoradiography.
4348
ST6Gal-related Sialyltransferase from D. melanogaster
Lectin Blotting—SDS-PAGE-separated glycoproteins were transferred to nitrocellulose membrane, blocked in PBTS (phosphate-buffered saline, 0.05% Tween, 0.05% saponin) overnight, and incubated with the corresponding biotinylated lectins in PBTS at 5 g/ml (SNA) or 10 g/ml (WFA) concentrations for 1 h at room temperature. After extensive washes, lectin binding was visualized using horseradish peroxidase-conjugated streptavidin (ABC Vectastain Elite kit, Vector Laboratories) followed by chemiluminescence detection (Pierce). Large Scale Sialyltransferase Assay and 1H NMR—For the preparative scale sialyltransferase assay, 1 mg of D. SiaT (bound to IgG beads), 100 nmol of GalNAc1– 4GlcNAc1–2Man␣1-O(CH2)7CH3, and 150 nmol of CMP-NeuAc were incubated in 50 mM cacodylate buffer, pH 6.5, at 25 °C for 72 h in 1.5-ml final volume. Every 12 h an aliquot of CMP-NeuAc (70 nmol) was added to the reaction mixture. The sialylated product was desalted on Sep-Pak C18 cartridges as described (32), and the sample was lyophilized twice from 99.8% D2O (Cambridge Isotopes Labs) and finally dissolved in 99.96% D2O. 1H NMR spectra were collected at 14.1 tesla (600-MHz proton frequency) and 300 K on a Varian Inova NMR spectrometer. Chemical shifts are referenced relative to internal acetone (␦ 2.225 ppm). NMR data were processed using a software package developed by Dr. J. A. van Kuik (Bijvoet Center, Department of Bio-organic Chemistry, Utrecht University). Immunostaining and in Situ Hybridization—Immunostaining of D. SiaT-expressing S2 cells was performed as described (44). We used primary monoclonal mouse anti-HA antibody at 1:1000 (Babco), polyclonal rabbit anti-Lva antibody at 1:5000 (a gift from Dr. John Sisson, The University of Texas, Austin), Cy3-conjugated donkey anti-mouse (1:500), and fluorescein isothiocyanate-conjugated donkey anti-rabbit (1:140) antibodies (both from Jackson ImmunoResearch). D. melanogaster Canton-S embryos were collected on apple juice plates at 25 °C as described (45). In situ hybridization of Drosophila embryos was performed using a published protocol (46) with digoxigenin-labeled RNA probe (produced with DIG RNA labeling mix from Roche Applied Science) and anti-DIG antibody conjugated with alkaline phosphatase (Roche Applied Science). Images were obtained with Axioplan 2 fluorescent microscope (Zeiss). RESULTS
Identification of a Putative Drosophila Sialyltransferase Gene—By searching the complete Drosophila genome data base with the L-sialyl motif consensus sequence (10), we identified a gene, located at 60D14 on 2R chromosome (GadFly CG4871), with homology to known sialyltransferases. Searches of the Drosophila EST data base (47) identified a cDNA clone, GH27778, with homology in the 5⬘-prime sequence to the predicted mRNA of CG4871. The sequence of this clone revealed that it included an open reading frame corresponding to the predicted open reading frame of CG4871, confirming that hypothetical gene CG4871 indeed represents an active transcription unit. This gene has been referred to as ST6Gal in the Drosophila genome data base based on its sequence similarity to vertebrate sialyltransferases, but as this name does not accurately reflect its actual enzymatic activity (see below), we instead designate it as SiaT. Similar to vertebrate sialyltransferases, this gene encodes a putative type II transmembrane protein with an N-terminally located signal peptide/anchoring domain, a stem region, and a C-terminal presumptive catalytic domain with high homology to the corresponding domain of vertebrate sialyltransferases (Fig. 1, A and B). The closest vertebrate homologues of Drosophila SiaT (designated D. SiaT) are -galactoside ␣2– 6-sialyltransferases, including human ST6Gal II (7, 8) and chick and bovine ST6Gal I proteins (48, 49) (BLAST E-values: 3e-50, 2e-47, and 2e-45, respectively). BLAST searches also revealed an Anopheles gambiae (mosquito) gene, agCG56989 (GenBank™ accession number EAA04038), encoding a putative sialyltransferase with high sequence homology to D. SiaT (E-value: 3e-89). Multiple sequence alignment with the catalytic domains of known vertebrate sialyltransferases confirmed a closer relationship of D. SiaT to ST6Gal sialyltransferases, as compared with ST3Gal, ST6GalNAc, or ST8Sia sialyltransferases (data not shown). Within a phylogenetic tree of ST6Gal sialyltransferases, the
position of D. SiaT is slightly closer to the ST6Gal II subfamily than to ST6Gal I subfamily of sialyltransferases (Fig. 1C). This closer relationship of D. SiaT to ST6Gal II sialyltransferases was further corroborated by a pairwise BLAST analysis (50) that gave the following E-values for the comparison between D. SiaT and the corresponding ST6Gal sialyltransferases: 3E-50 (hST6Gal II), 9E-45 (hST6Gal I), 9E-49 (mST6Gal II), and 1E-44 (mST6Gal I). Further searches of Drosophila genomic sequences failed to reveal any other genes encoding a protein with significant similarity to vertebrate sialyltransferases. Thus, we conclude that SiaT is apparently the sole vertebrate-type sialyltransferase gene in the Drosophila genome. Expression and Subcellular Localization of D. SiaT Protein—A full-length D. SiaT protein with a C-terminal HA tag was transiently expressed in Drosophila S2 cells. Simultaneous immunostaining of these cells for D. SiaT protein and Golgiassociated proteins, Lava lamp (Fig. 2A) and Fringe (data not shown) (51, 52), revealed that D. SiaT protein is localized mainly to the Golgi compartment. We also found that a fraction of SiaT-HA protein expressed in S2 cell culture is secreted into the cell medium in soluble form (Fig. 2B). This result is in agreement with data on the presence of the soluble secreted form of ST6Gal I sialyltransferases (53, 54). Purification of D. SiaT Protein—D. SiaT protein with an N-terminal protein A tag was transiently expressed in Drosophila S2 cells and purified from the media using IgG beads. Initial experiments confirmed its sialyltransferase activity (data not shown). As larger D. SiaT quantities than could be conveniently produced in S2 cells were required for comprehensive enzymatic characterization, D. SiaT recombinant protein was also expressed in Sf9 cells using the baculovirus expression system (Invitrogen). The ProtA-SiaT fusion protein was expressed in these cells at high level and efficiently secreted into the culture medium. We routinely purified 2–5 g of recombinant D. SiaT protein from 1 ml of culture medium (Fig. 2C). Sialyltransferase Activity and Substrate Specificity of D. SiaT—Sialyltransferase activity of purified D. SiaT protein was revealed in a series of sialyltransferase assays. To determine the substrate specificity of D. SiaT, a panel of different oligosaccharide, glycoprotein, and glycolipid acceptors was assayed (Table I). Among oligosaccharide acceptors, D. SiaT exhibited highest activity toward GalNAc1– 4GlcNAc-Rg (LacdiNAc-Rg), whereas lower activity was detected toward Gal1– 4GlcNAc (LacNAc or type II) and Gal1–3GlcNAc (type I) disaccharide acceptors, as well as toward lacto-N-neotetraose (LNnT) and lacto-N-tetraose (LNT) acceptors. No activity was detected toward Gal1–3GalNAc or GalNAc␣1-benzyl acceptors. Among the glycoproteins tested, bovine lactoferrin and ␣-lactalbumin were the best acceptors. With relatively low efficiency, D. SiaT could also utilize asialofetuin and asialo-␣1-acid glycoprotein as acceptors. No activity toward glycolipid acceptors was detected (Table I). Effects of Temperature, pH, and Divalent Cations on D. SiaT Activity—When the activity of D. SiaT was compared at different temperatures, an ⬃30% increase in activity was observed after raising the temperature from 25 to 37 °C (Fig. 3A). Similar effects have been described for a number of other Drosophila enzymes (55, 56). We also analyzed the dependence of D. SiaT activity on pH and found that pH between 6.0 and 6.5 was optimal (Fig. 3B). This is close to the optimal pH (6.0) reported for ST6Gal I sialyltransferases (57). Next, we examined the dependence of D. SiaT activity on the presence of metal cations. None of the tested cations potentiated D. SiaT activity, whereas some of them (Cu2⫹, Ni2⫹, and
ST6Gal-related Sialyltransferase from D. melanogaster
4349
FIG. 1. D. SiaT protein is structurally related to ST6Gal sialyltransferases. A, alignment of protein sequences of Drosophila SiaT (D. SiaT), human ST6Gal II sialyltransferase (hST6Gal-II), and chick ST6Gal I sialyltransferases (gST6Gal-I). GenBankTM accession numbers are as follows: AF397532, XM_038616, and Q92182, respectively. For D. SiaT sequence, a putative transmembrane anchoring domain is indicated by a horizontal bracket (predicted by DAS server (86)); a potential signal peptide cleavage site is indicated by an arrowhead (predicted using SignalP-NN program (87)); and asterisks indicate potential N-linked glycosylation sites. Regions of sialyl motifs are underlined (L, S, and VS). Alignment was performed using ClustalW algorithm (88). The D. SiaT protein sequence corresponds to the conceptual transcript CG5871-RA listed in the GadFly data base (GenBankTM accession number AAF47256). B, hydropathy plot of D. SiaT protein sequence (according to Kyte and Doolittle (89)). Hydropathy analysis revealed the presence of a prominent N-terminal hydrophobic region corresponding to a putative signal peptide/ anchoring domain, which suggests a type II transmembrane topology for D. SiaT protein. C, dendrogram of Drosophila (D. SiaT), putative mosquito (A. SiaT), and ST6Gal sialyltransferases. Amino acid sequences of presumptive catalytic domains of sialyltransferase proteins were aligned, and the phylogenetic tree was constructed using ClustalW server (EMBL-EBI, www.ebi.ac.uk/clustalw). The scale bar shows relative units of evolutionary distances. The dendrogram positions of insect sialyltransferases are slightly closer to ST6Gal II as compared with ST6Gal I sialyltransferases (e.g. D. SiaT is closer to mouse ST6Gal II than to mouse ST6Gal I by 0.009 units). GenBankTM accession numbers of ST6Gal I proteins are as follows: NM_173216 (human), CAA75385 (bovine), P13721 (rat), and Q64685 (mouse). The accession number of mST6Gal II protein is BAC87752.
Co2⫹) significantly inhibited the activity of the enzyme (Fig. 3C). EDTA at 15 mM concentration (in the absence of added ions) had no effect on D. SiaT activity (data not shown). Linkage Specificity of D. SiaT—We analyzed the linkage and
attachment site of D. SiaT-incorporated sialic acid by NMR spectroscopy. To this end, a large scale sialyltransferase assay was performed using GalNAc1– 4GlcNAc1–2Man␣1O(CH2)7CH3 as an acceptor. The choice of the acceptor was
4350
ST6Gal-related Sialyltransferase from D. melanogaster
FIG. 2. Subcellular localization and purification of D. SiaT protein. A, immunofluorescent staining of Drosophila S2 cells expressing D. SiaT-HA protein reveals its co-localization with Golgi resident protein (Lva): red (Cy3), D. SiaT; green (FITC), Lva. Right panel shows the overlay of red and green channels. B, Western blot analysis with anti-HA antibody revealed the presence of D. SiaT-HA protein in the cell medium and cell pellet fractions of transiently transfected Drosophila S2 culture cells. C, Western blot and Coomassie staining of purified protein A-tagged D. SiaT protein. Protein A-SiaT fusion protein was expressed in Sf9 cells using baculovirus system, purified using Sepharose-IgG beads, separated on SDS-PAGE gel, and analyzed by Western blotting (probed with purified rabbit IgG) and Coomassie staining. Beads from mock purification (Sf9 cells infected with irrelevant baculovirus) were used as a negative control. Based on Coomassie staining, the purity of IgG-immobilized D. SiaT protein was estimated to be ⱖ98%. TABLE I Acceptor substrate specificity of D. SiaT Oligosaccharides were assayed at 3 mM, gangliosides at 0.5 mg/ml, and glycoproteins at 2.5 mg/ml concentration. R represents the remainder of the N-linked glycan; RCer indicates the remainder of the glycolipid structure; PEtn is phosphoethanolamine. Acceptors
Oligosaccharides Type I LacNAc (type II) Type III Lacto-N-tetraose Lacto-N-neotetraose LacdiNAc-Rg GalNAc-benzyl Glycoproteins Fetuin Asialofetuin ␣1-acid glycoprotein Asialo-␣1-acid glycoprotein Bovine lactoferrm Bovine ␣-lactalbuminc Drosophila mucin-Dd Glycolipids GM1a Lactosylceramide Asialo-GM2 Monosialo-GM3 Gangliosides from bovine brain (mixture) Drosophila glycolipids Same as above. PEtn residues removed
Representative carbohydrate structuresa
Gal1–3GlcNAc Gal1–4GlcNAc Gal1–3GalNAc Gal1–3GlcNAc1–3Gal1–4Glc Gal1–4GlcNAc1–3Gal1–4Glc GalNAc1–4GlcNAc-Rg GalNAc␣1-benzyl NeuAc␣2–3Gal1–3GalNAc-O-Ser/Thr NeuAc␣2–3Gal1–3(NeuAc␣2–6)GalNAc-O-Ser/Thr NeuAc␣2–6(3)Gal1–4GlcNAc-R
Activity
Relative activityb
nmol/mg/h
%
6.1 16.8 ⬍0.27 5.1 13.3 31.9 0
36 100 ⬍1.6 30 79 190 0
0.04
0.24
GalNAc1–4GlcNAc-R (NeuAc␣2–6), GalI-4GlcNAc-R GalNAc1–4GlcNAc-R Gal1–3GalNAc-O-Ser/Thr
0.09 ⬍0.02 0.04 0.23 0.13 ⬍0.02
0.54 ⬍0.12 0.24 1.4 0.77 ⬍0.12
Gal1–3GalNAc1–4(NeuAc␣2–3)Gal1–4Glc1-Cer Gal1–4Glc1-Cer GalNAc1–4Gal1–4Glc1-Cer NeuAc␣2–3Gal1–4Glc1-Cer See Itoh et al. (92) for details GalNAc1–4(PEtn-)GlcNAc1-RCer, etc.e GalNAc1–4GlcNAc1-RCer, etc.e
⬍0.01 ⬍0.01 ⬍0.01 ⬍0.01 ⬍0.01 ⬍0.01 ⬍0.01
⬍0.05 ⬍0.05 ⬍0.05 ⬍0.05 ⬍0.05 ⬍0.05 ⬍0.05
NeuAc␣2–6(3)Gal1–4GlcNAc-R
a Carbohydrate structures were inferred from published data for the following glycoproteins: fetuin (42); ␣1-acid glycoprotein (91); bovine lactoferrin and bovine ␣-lactalbumin (62); Drosophila mucin-D (33). b Relative rates were calculated in comparison to the activity toward LacNAc (100%). c Only 3–10% of ␣-lactalbumin pool is glycosylated (62). d Drosophila mucin-D was assayed at 0.5 mg/ml concentration. e Upper phase, for details see Seppo et al. (34). Individual glycosphingolipids of this mixture were present at approximately 50 M concentration.
based on the fact that this oligosaccharide has a terminal LacdiNAc structure, the preferred acceptor of D. SiaT (Table I). First, the 1H NMR spectrum of the non-sialylated acceptor was acquired (Table II) and found to be in agreement with data published previously (58). Next, we analyzed the structure of GalNAc1– 4GlcNAc1–2Man␣1-O(CH2)7CH3 after sialylation with D. SiaT (Fig. 4). We noted that the sialylation reaction was ⱖ95% efficient because we could not detect any non-sialylated acceptor in the NMR spectrum after the sialyltransferase assay. Comparison of the structural reporter group chemical shift values to published data (58 – 61) indicated that sialic acid was ␣2– 6-linked to the terminal -linked GalNAc residue. Specifically, the appearance of an N-acetyl signal at ␦ 2.030 and
the equatorial (␦ 2.662) and axial (␦ 1.717) H-3 signals of NeuAc indicated the ␣2– 6 linkage of sialic acids (59). Furthermore, an upfield shift of the GalNAc H-1 from ␦ 4.516 to ␦ 4.500, of GlcNAc H-1 from ␦ 4.558 to ␦ 4.586, and the downfield shift of the GlcNAc N-acetyl signal from ␦ 2.045 to ␦ 2.069 upon the addition of an ␣2– 6-linked Neu5Ac to GalNAc is in agreement with the chemical shift changes described between non-sialylated and sialylated LacdiNAc residues of structures N1.3 and N2.1, respectively (60) (see also N2/B and N1/B-6⬘ in Ref. 61). In addition, because we used a trisaccharide acceptor, we noted that the Man H-1 signal also shifts upfield from ␦ 4.854 to ␦ 4.879 (Table II). Together, these results established that D. SiaT attaches ␣2– 6-linked sialic acid to the terminal GalNAc
ST6Gal-related Sialyltransferase from D. melanogaster
4351
FIG. 3. Temperature (A), pH (B), and ion (C) dependence of D. SiaT activity in vitro. The activities were measured using LacNAc as acceptor. C, the effect of metal cations was assessed at 15 mM concentration in an otherwise standard reaction mixture. No ions indicates the control assay without added metal cations. 1
TABLE II H chemical shifts of structural reporter group protons of the constituent monosaccharides of unmodified and sialylated acceptor Chemical shift
Proton
Residue
GalNAc1–4GlcNAc1–2Man␣1– O(CH2)7CH3
H-1
GlcNAc GalNAc Man Man GalNAc NeuAc NeuAc GlcNAc GalNAc NeuAc
4.558 4.516 4.854 4.03 3.942
Neu5Ac␣2–6GalNAc1–4GlcNAc1– 2Man␣1–O(CH2)7CH3 ppm
H-2 H-3a H-3e NAc
a
2.045 2.067
4.586 4.500 4.879 NDa ND 1.717 2.662 2.069 2.069 2.030
ND, not determined.
residue of LacdiNAc acceptor structure. To confirm the linkage of D. SiaT-mediated sialylation of other oligosaccharide acceptors (type I, LacNAc, LNnT, and LNT), the sialylated oligosaccharides were digested with linkage-specific sialidases. Sensitivity of incorporated [14C]Neu5Ac to ␣2–3,6- but not to ␣2–3sialidase indicated that all these acceptors were modified with ␣2– 6-linked sialic acid (Fig. 5A). Therefore, by analogy with LacdiNAc, we concluded that D. SiaT modifies the terminal Gal residue on LacNAc, type I, LNnT, and LNT oligosaccharides with ␣2– 6-linked sialic acid. This was also confirmed by comigration of authentic Neu5Ac␣2– 6Gal1– 4GlcNAc standard trisaccharide with D. SiaT-sialylated LacNAc acceptor on HPTLC (data not shown). Additional experiments were performed to analyze the linkage of sialic acids on glycoprotein acceptors. Digestion of labeled asialofetuin and asialo-␣1-acid glycoprotein with linkagespecific sialidases confirmed that sialic acids were incorporated into these glycoproteins via an ␣2– 6 linkage (Fig. 5B). Moreover, degalactosylated asialofetuin did not serve as an acceptor for D. SiaT (data not shown), which implied that D. SiaT modified the terminal Gal residues of asialofetuin glycans. Glycoproteins were also treated with PNGaseF to determine whether the incorporated sialic acids were on N-linked (PNGaseF-sensitive) or O-linked (PNGaseF-insensitive) glycans. In all cases, PNGaseF treatment led to the removal of incorporated radioactivity, indicating that the sialic acids were present on N-linked structures (Fig. 5C). Thus, these data indicated that D. SiaT produced Sia␣2– 6Gal1– 4GlcNAc termini on N-linked glycans of asialofetuin and asialo-␣1-acid glycoprotein (Table I). Unlike asialofetuin, lactoferrin was not modified by D. SiaT on its terminal 1– 4-linked Gal residues, because galactosidase pretreatment had no detectable effect on its ability to serve
as acceptor for D. SiaT (Fig. 6A). Bovine lactoferrin was previously shown to contain only N-linked glycans which, in addition to LacNAc, also carry LacdiNAc termini (Table I) (43). Therefore, we reasoned that D. SiaT might have modified terminal -linked GalNAc residues on this acceptor. Indeed, pretreatment of lactoferrin with -N-acetylhexosaminidasef that removes terminal 1– 4,6-linked GalNAc and GlcNAc residues abolished D. SiaT-mediated Neu5Ac incorporation, confirming that sialic acids were attached to LacdiNAc termini (Fig. 6A). This result correlates well with D. SiaT activity detected toward LacdiNAc-containing oligosaccharide acceptors (Table I and Fig. 4) and toward bovine ␣-lactalbumin, the major glycoform of which carries a biantennary structure with LacdiNAc termini (62). To confirm that lactoferrin is modified with ␣2– 6-linked sialic acids, we treated the [14C]Neu5Ac-sialylated lactoferrin with linkage-specific sialidases. Interestingly, the radiolabeled sialic acids could only be removed from lactoferrin by V. cholerae ␣2–3,6,8-sialidase but not C. perfringens ␣2–3,6sialidase (Fig. 6B). Sialylated LacdiNAc oligosaccharide revealed a similar pattern of sensitivity to the sialidase digests, as it was also resistant to C. perfringens ␣2–3,6-sialidase treatment (Fig. 6C). A resistance of ␣2– 6-sialylated fluorophorelabeled GalNAc to C. perfringens ␣2–3,6-sialidase has been noted previously (Glyko, manufacturer’s protocol). Considering that the linkage created by D. SiaT on LacdiNAc oligosaccharide is clearly ␣2– 6 (Fig. 4 and Table II), we concluded that the LacdiNAc termini of lactoferrin were modified by D. SiaT with ␣2– 6-linked sialic acids but that the product was resistant to cleavage by C. perfringens ␣2–3,6-sialidase as shown for model structures (Fig. 6C). Interestingly, there is also prior evidence that -linked GalNAc modified with ␣2–3-sialic acid is resistant to cleavage by certain ␣2–3-sialidases (63). Thus, sialylated
4352
ST6Gal-related Sialyltransferase from D. melanogaster
FIG. 4. 600-MHz 1H NMR spectrum of Neu5Ac␣2– 6GalNac1– 4GlcNAc1–2Man␣1-O(CH2)7CH3 at 300 K. Only the relevant regions of the spectrum are shown. Asterisk indicates the acetone peak.
-linked GalNAc structures appear to have an intrinsic resistance to certain sialidases. Kinetic Analysis of D. SiaT—We also determined the kinetic parameters of D. SiaT for certain acceptors. The apparent Km value of D. SiaT for LacNAc (Gal1– 4GlcNAc) non-reducing termini of asialofetuin glycans was 1.4 mM, and Vmax was estimated as 0.4 nmol/h/mg (relative Vmax/Km ⫽ 0.3). We could not determine the apparent Km value for the LacNAc disaccharide acceptor, because no signs of saturation of the sialyltransferase reaction could be reached at the tested concentrations of this acceptor. Nevertheless, the reaction velocity obtained for a range of acceptor concentrations indicated that LacNAc disaccharide was a better acceptor for D. SiaT than the LacNAc termini of asialofetuin glycans (Fig. 7). Similarly, LacdiNAc-Rg disaccharide (Km ⫽ 8.0 mM, Vmax ⫽ 139 nmol/h/mg, relative Vmax/Km ⫽ 17.3) was a better acceptor than the LacdiNAc termini of lactoferrin (Km ⫽ 0.5 mM, Vmax ⫽ 1.9 nmol/h/mg, relative Vmax/Km ⫽ 3.8). Moreover, the kinetic parameters clearly indicate that LacdiNAc is preferred over LacNAc, both in the context of glycoprotein and disaccharide acceptors (Fig. 7). Autosialylation of D. SiaT on LacdiNAc Termini of N-Linked Glycans—In the course of our sialyltransferase assays, we noticed that D. SiaT protein could sialylate itself. This autosialylation occurred on N-linked glycans, as revealed by PNGaseF digestion of [14C]Neu5Ac-labeled D. SiaT protein (Fig. 8A). Several observations indicate that the incorporated sialic acid
was attached to terminal LacdiNAc structures. Similar to sialylated lactoferrin, the product of D. SiaT autosialylation was resistant to C. perfringens ␣2–3,6-sialidase but could be removed by V. cholerae ␣2–3,6,8-sialidase (Fig. 8B). Pretreatment of D. SiaT with -N-acetylhexosaminidasef, but not with bovine testes -galactosidase or S. pneumoniae -N-acetylglucosaminidase (which specifically removes terminal -linked GlcNAc residues), abolished the autosialylation of D. SiaT (Fig. 8C), indicating that sialic acids were probably linked to terminal -linked GalNAc residues. This conclusion was confirmed by lectin blotting using WFA, a lectin that specifically recognizes terminal GalNAc residues (64). Only sialidase-treated D. SiaT protein, and not fully autosialylated D. SiaT, is recognized by WFA (Fig. 8D), thus indicating that terminal GalNAc residues were masked on autosialylated D. SiaT protein by the attached sialic acids. So far, to the best of our knowledge, terminal -linked GalNAc on N-glycans has only been described in the context of LacdiNAc and Sda (Neu5Ac␣2–3[GalNAc1– 4]Gal1-) structures (65, 66). To discriminate between these two possibilities, we analyzed -N-acetylhexosaminidasef-treated D. SiaT protein by lectin blotting using RCA-I, which recognizes terminal -linked galactose residues (67). Untreated D. SiaT protein showed significant reactivity to RCA-I, presumably because of the presence of terminal -linked GalNAc residues that can also interact with RCA-I (Vector Laboratories, manufacturer’s information). At the same time, hexosaminidasef-treated D.
ST6Gal-related Sialyltransferase from D. melanogaster
4353
FIG. 5. Analysis of D. SiaT linkage specificity. Autoradiographs of acceptors sialylated by D. SiaT with [14C]Neu5Ac, separated on HPTLC plate (oligosaccharides) or SDS-PAGE gel (glycoproteins), and exposed to a PhosphorImager screen. A, sialylated oligosaccharides were mocktreated or treated with ␣2–3-sialidase (S. pneumoniae) or ␣2–3,6-sialidase (C. perfringens). In the control treatment (data not shown), the same amount of ␣2–3-sialidase completely removed sialic acid from [14C]NeuAc-labeled type I disaccharide standard that was prepared with commercial ␣2–3-sialyltransferase (Sigma). B, sialylated glycoproteins, asialofetuin and asialo-␣1-acid glycoprotein, were mock-treated, treated with ␣2–3sialidase or ␣2–3,6-sialidase. In the control treatment (data not shown), the same amount of ␣2–3-sialidase completely removed incorporated sialic acid from [14C]NeuAc-labeled ␣2–3-sialylated asialofetuin (prepared with commercial ␣2–3-sialyltransferase (Sigma)). C, PNGaseF treatment of sialylated asialofetuin and asialo-␣1-acid glycoprotein. *, autosialylated D. SiaT protein.
FIG. 6. Analysis of D. SiaT linkage specificity with lactoferrin and LacdiNAc-Rg acceptors. Autoradiographs of acceptors sialylated by D. SiaT with [14C]Neu5Ac, separated on HPTLC plate (LacdiNAc-Rg) or SDS-PAGE gel (lactoferrin), and exposed to a PhosphorImager screen. A, prior to sialylation, lactoferrin was mock-treated or treated with -galactosidase (-gal-ase) or -N-acetylhexosaminidasef (Hex-asef). B, sialylated lactoferrin was mock-treated or treated with ␣2–3-sialidase, ␣2–3,6-sialidase, or ␣2–3,6,8-sialidase. C, sialylated LacdiNAc-Rg was mock-treated or treated with ␣2–3-sialidase, ␣2–3,6-sialidase, or ␣2–3,6,8-sialidase. Note that the structure is partially resistant to ␣2–3,6-sialidase treatment.
SiaT protein had significantly decreased RCA-I reactivity, thus indicating that the penultimate sugar is not Gal (Fig. 8E). The presence of unchanged IgG band (explained by the presence of terminal galactose on IgG glycans (68)) in both treated and untreated samples served as a control indicating the absence of proteolytic and galactosidase activity (Fig. 8E). These results are compatible with the presence of LacdiNAc structure and confirm the absence of Sda-like structure on D. SiaT glycans. Thus, these data, together with the observations that (i) terminal LacdiNAc structure has been described previously (69, 70) on insect N-glycans, (ii) lepidopteran cells used for D. SiaT expression possess the 1– 4GalNAc-transferase activity required for the synthesis of LacdiNAc termini (71), and that (iii) LacdiNAc structure was identified as a preferred acceptor for
D. SiaT in our sialyltransferase assays, indicate that sialylation of D. SiaT N-linked glycans most likely occurs on the -linked GalNAc residues of LacdiNAc termini. In Vivo Activity of D. SiaT—In order to examine if autosialylation occurs during the expression of D. SiaT in Sf9 cells, the purified sialyltransferase was probed with SNA that specifically recognizes ␣2– 6-linked sialic acids (72). Strong lectin reactivity was observed for the purified protein (Fig. 9, lane 1). Importantly, SNA reactivity could be eliminated by sialidase treatment and restored by in vitro autosialylation (Fig. 9). Nonetheless, the capacity for in vitro autosialylation, even without sialidase treatment (e.g. Fig. 8A, 1st lane), indicates that D. SiaT is produced by these cells as a partially sialylated pool. Similar results were obtained with D. SiaT protein ex-
4354
ST6Gal-related Sialyltransferase from D. melanogaster
FIG. 7. D. SiaT sialyltransferase activity (nanomoles of Neu5Ac transferred per mg/h) as the function of acceptor concentration plotted for oligosaccharides (A) or glycoprotein acceptors (B). All assays with oligosaccharides were done in duplicate (S.D. error bars are shown). For glycoproteins, acceptor concentration is expressed in concentration of corresponding acceptor termini, LacNAc for asialofetuin and LacdiNAc for lactoferrin (see “Experimental Procedures”).
FIG. 8. Autosialylation of D. SiaT. For autoradiography analysis, D. SiaT protein was autosialylated with CMP-[14C]Neu5Ac, separated on SDS-PAGE gel, and exposed to a PhosphorImager screen (A–C). For lectin blot analysis, D. SiaT was autosialylated with non-radiolabeled donor (D and E). Note: the presence of IgG band on the lectin blots (D and E and Fig. 9) resulted from IgG leaching from beads and the reactivity of IgG glycans with corresponding lectins (68, 90). A, PNGaseF treatment of autosialylated D. SiaT. The absence of protease activity during PNGaseF treatment was confirmed by Western blotting (data not shown). B, sialidase treatments of autosialylated D. SiaT: mock, ␣2–3-sialidase, ␣2–3,6-sialidase, and ␣2–3,6,8-sialidase treatments. C, pre-treatments of D. SiaT before autosialylation with -N-acetylhexosaminidasef (Hexasef), S. pneumoniae -N-acetylglucosaminidase (GlcNAc-ase), or -galactosidase (-gal-ase). Mock pretreatments (controls) are shown on the left for each digest. D, WFA lectin blot analysis of D. SiaT: lane 1, purified D. SiaT protein fully autosialylated in vitro; lane 2, same sample treated with ␣2–3,6,8-sialidase; lane 3, same sample treated with both ␣2–3,6,8-sialidase and -N-acetylhexosaminidasef. E, RCA-I lectin blot analysis of D. SiaT. Purified D. SiaT was mock-treated (Mock) or treated with -N-acetylhexosaminidasef (Hex-asef) prior the lectin blot.
pressed in S2 cells (data not shown). Because neither S2 cells, Sf9 cells, nor Sf9 cells infected with an irrelevant baculovirus possess any detectable sialyltransferase activity (data not shown and see Refs. 25 and 26), we conclude that D. SiaT can sialylate its own glycans in vivo. SiaT Expression during Drosophila Development—As the expression pattern of a gene might shed light on its function, in situ hybridization was performed to examine D. SiaT expression during embryonic development. No maternally provided
SiaT mRNA was detected at the beginning of embryogenesis, and no significant expression was revealed until stage 16 (data not shown). Beginning at stage 16 (13–14 h of embryonic development), a prominent expression of SiaT was detected in the developing embryonic central nervous system. At this stage, SiaT is expressed in a segmentally repeated pattern of two rows of cells laterally located within the embryonic central nervous system (Fig. 10A). At the end of embryogenesis (stage 17, 16 h of embryonic development), more central nervous
ST6Gal-related Sialyltransferase from D. melanogaster system cells started to contribute to this pattern of SiaT expression (Fig. 10B). Expression was also detected in some regions of the embryonic brain (Fig. 10C) and, at significantly lower levels, in the ectoderm. DISCUSSION
An Ancient Origin for Sialylation—As shown in this study, Drosophila encodes a functional sialyltransferase that is related to the ST6Gal family of sialyltransferases. D. SiaT is the first characterized sialyltransferase in the protostome lineage of animals. An insect homologue of D. SiaT, with 47% amino acid sequence identity, was also found in the mosquito genome. Sequence analysis revealed a closer relationship of these insect proteins to the ST6Gal II than to the ST6Gal I vertebrate sialyltransferases. These data suggest that vertebrate ST6Gal sialyltransferases and insect sialyltransferases probably have a common ancestral gene and that ST6Gal I subfamily possibly evolved from the more ancient ST6Gal II-like sialyltransferases after the separation of deuterostome lineage. Also, this indicates the possibility that ST6Gal family represents the most ancient type of animal sialyltransferases that gave rise to other vertebrate sialyltransferase families during evolution. Comparison of D. SiaT to Vertebrate Sialyltransferases—The enzymatic characterization of purified D. SiaT confirmed a general relationship between D. SiaT and vertebrate ST6Gal sialyltransferases. At the same time, our assays and kinetic analysis also revealed a distinct substrate specificity of D. SiaT, as it prefers LacdiNAc termini over LacNAc termini on both oligosaccharide and glycoprotein acceptors in in vitro assays (Table I and Fig. 7). Bovine and rat ST6Gal I sialyltransferases have been shown to utilize LacdiNAc as an acceptor in vitro (49, 73, 74), and the existence of sialylated LacdiNAc on mammalian glycoproteins suggests that these sialyltransferases, in addition to LacNAc, may also sialylate LacdiNAc termini in vivo. However, all previously characterized ST6Gal I sialyltransferases preferred LacNAc-terminating acceptors to other
FIG. 9. In vivo autosialylation of D. SiaT. SNA lectin blot analysis of D. SiaT protein: lane 1, D. SiaT expressed in Sf9 cells and purified on IgG beads (“in vivo” autosialylation); lane 2, same sample treated with ␣2–3,6,8-sialidase (control for the lectin staining specificity); lane 3, partial sialidase treatment of purified D. SiaT protein; lane 4, same sample as in (lane 3) subsequently subjected to in vitro autosialylation. The positions of D. SiaT and IgG bands are indicated.
4355
structures (57), including LacdiNAc (49, 73, 74). The recently characterized hST6Gal II also exhibited a preference for LacNAc, but its activity toward LacdiNAc structures was not examined (7, 8). Based on the sequence similarity between D. SiaT and hST6Gal II, it can be expected that LacdiNAc might be an acceptor for hST6Gal II. In contrast to the tight specificity shown by mammalian sialyltransferases, we found that D. SiaT can also modify type I disaccharide and LNT, a tetrasaccharide containing type I structure at its non-reducing end (Table I and Fig. 7). This activity is lower than that toward LacNAc-containing oligosaccharides; nevertheless, it is still unusually high considering that type I disaccharide is a very poor acceptor for ST6Gal sialyltransferases (7, 8, 48, 53, 74, 75). The exception represents data on the relatively high activity of mouse and human ST6Gal I toward LNT (7, 76), which, however, was not confirmed by another study (8). Thus, D. SiaT exhibits broader substrate specificity than its vertebrate homologues. Besides its broader substrate utilization, D. SiaT has several important characteristics in common with previously characterized ST6Gal sialyltransferases. First, we found that D. SiaT can modify LacNAc termini of oligosaccharides with relatively high efficiency (Table I and Fig. 7). Second, D. SiaT does not have any detectable in vitro activity toward glycolipids (Table I), including Drosophila glycolipids (with or without phosphoethanolamine groups) that are known to contain terminal LacdiNAc residues (34). This result correlates with the fact that sialylated glycolipids were not detected in Drosophila (34). In this respect, D. SiaT is similar to the characterized ST6Gal enzymes which, unlike some ST3Gal or ST6GalNAc sialyltransferases (5), do not utilize glycolipid acceptors (7, 48). Third, similar to hST6Gal II (7, 8), the closest characterized mammalian homologue of the Drosophila enzyme, D. SiaT prefers oligosaccharide acceptors to glycoproteins (Table I and Fig. 7). Based on this acceptor preference, a role in the in vivo synthesis of sialylated oligosaccharides (e.g. milk oligosaccharides) was suggested for hST6Gal II (7, 8). However, nothing is known about the existence of such oligosaccharides in Drosophila. Alternatively, it is also possible that as yet undefined recognition determinants (e.g. a particular glycan or protein structural motifs (77, 78)) are required by D. SiaT for the recognition of specific glycoprotein acceptors, and that these determinants are absent on the tested glycoproteins. We found that D. SiaT activity was not potentiated by divalent metal cations and not inhibited by EDTA (Fig. 3C). In this respect, D. SiaT is also similar to the ST6Gal sialyltransferases, activities of which (unlike for some other sialyltransferases, e.g. ST6GalNAc III (79)) do not require the presence of metal cations (57, 75). Insect Cells Can Synthesize Sialylated LacdiNAc—The anticipated in vivo activity of D. SiaT is of particular interest. It is plausible that the high in vitro activity toward LacdiNAccontaining structures reflects the in vivo substrate preference of the enzyme. Although LacdiNAc termini are present on
FIG. 10. D. SiaT expression pattern during embryonic development as revealed by in situ hybridization. In all images, anterior is to the left. A, stage 16, ventral view. SiaT is expressed in two rows of cells within central nervous system (arrows). B, stage 17, ventral view. More central nervous system cells begin to express SiaT. C, stage 17, lateral view. Expression of SiaT in the embryonic brain (arrowheads).
4356
ST6Gal-related Sialyltransferase from D. melanogaster
glycoconjugates from a variety of animal species (65), they are thought to be more common in invertebrates, whereas LacNAc structure is more prevalent in vertebrates (65, 80). Thus, the detected shift in substrate specificity between D. SiaT and related ST6Gal sialyltransferases may reflect the evolution of animal glycomes. In mammals, LacdiNAc structure is less common than LacNAc. Nevertheless, it is present on numerous glycoproteins and sometimes is found to bear different modifications, including ␣2– 6-sialylation of the terminal GalNAc residue (65). Interestingly, sialylated (and/or fucosylated) LacdiNAc structures were suggested to mediate immunosuppression and to block sperm-zona pellucida binding in humans (81). Besides the potential involvement in specific adhesion processes, a role for sialylated LacdiNAc in secretion of mammalian glycoproteins has also been suggested (82). In insects, the function of LacdiNAc has not yet been assessed. To the best of our knowledge, this structure has only been reported on the antennae of honeybee venom glycoproteins (69, 70). A 1– 4GalNAc-transferase activity responsible for the synthesis of LacdiNAc termini has been detected in lepidopteran cells (71), but the identity of glycoproteins with this determinant is still unknown. We found that recombinant D. SiaT protein is produced by insect cells with sialylated -linked GalNAc termini on N-linked glycans. These data provide the first evidence for the presence of a sialylated LacdiNAc structure on an insect glycoprotein. Although the structure of D. SiaT glycans under native conditions may not correlate accurately with glycosylation observed in cell culture, our results suggests that endogenously expressed D. SiaT might also be autosialylated. We are currently investigating this possibility. Potential Biological Functions of Sialylation in Drosophila— Our in situ hybridization experiments revealed a restricted pattern of SiaT expression during Drosophila embryonic development. The strongest expression was detected during late embryonic stages in a subset of cells, presumably neurons, within the developing central nervous system. This pattern of SiaT expression is similar to the previously reported pattern of cytochemical staining using Limax flavus lectin that specifically recognizes sialic acids (21). Some difference between SiaT expression determined in our experiments and the reported lectin staining at the early stages (e.g. SiaT expression was not detected in the pole cells and at the cellular blastoderm stage) might be explained by a maternal supply of sialylated glycoconjugates. The immunochemical detection of polysialic acid by Roth et al. (21) would suggest the existence of a polysialyltransferase in Drosophila. However, exhaustive searches of the complete Drosophila genome have only revealed the presence of one sialyltransferase, SiaT, although it cannot be excluded that other non-vertebrate-type sialyltransferase(s) might also be present. The expression of D. SiaT in developing central nervous system is particularly interesting. Developmentally regulated glycosylation of cell adhesion molecules has been implicated in sensory afferent targeting and the establishment of the sensory nervous system architecture in the leech (83). Similarly, dynamic expression of SiaT may provide a basis for temporally and spatially regulated sialylation that could play a role in Drosophila nervous system development. Our preliminary data indicate that the position of SiaT-expressing cells within central nervous system map to the region of specified motoneurons (84). Thus, it is conceivable that D. SiaT functions within these cells to modulate the adhesive properties of motoneuronal processes by sialylating cell-surface glycoproteins. Alternatively, D. SiaT-mediated sialylation may be important for the establishment of neuromuscular junctions (e.g. the activity of a mus-
cle-specific receptor tyrosine kinase, a key mediator of neuromuscular synapse formation, was found to be modulated by sialylated N-linked glycans (85)). It is interesting to note that the expression of hST6Gal II sialyltransferase, the human homologue of D. SiaT, is significantly elevated in fetal brain (7, 8), which may reflect an evolutionarily conserved molecular mechanism underlying the function of these enzymes in neural development. Further studies are underway to map precisely SiaT-expressing cells and to reveal the function of SiaT expression in developing central nervous system. Acknowledgments—We thank Monica Palcic, John A. F. Joosten, and Johannis P. Kamerling for synthetic oligosaccharides and for valuable discussions during this work; John Sisson for anti-Lva antibody; Konstantin Severinov for protein A-encoding plasmid; Tetsuya Okajima for the glycolipid assay protocol; Andrei Kramerov for Drosophila mucin-D; Michael Tiemeyer for Drosophila glycolipids and helpful suggestions; Donald Jarvis for protocols on baculovirus system; and Robert Haltiwanger for helpful discussions. We are grateful to Ioannis Vakonakis for help with the NMR instrumentation. We thank Pamela Stanley, Monica Palcic, Robert Haltiwanger, and Konstantin Severinov for comments on the manuscript. The NMR instrumentation in the Biomolecular NMR Laboratory at Texas A & M University was supported by National Science Foundation Grant DBI-9970232. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.
9. 10. 11. 12. 13. 14. 15. 16. 17. 18.
19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.
30. 31. 32.
33.
34. 35.
Schauer, R. (2000) Glycoconj. J. 17, 485– 499 Angata, T., and Varki, A. (2002) Chem. Rev. 102, 439 – 469 Fukuda, M. (1996) Cancer Res. 56, 2237–2244 Kim, Y. J., and Varki, A. (1997) Glycoconj. J. 14, 569 –576 Harduin-Lepers, A., Vallejo-Ruiz, V., Krzewinski-Recchi, M. A., Samyn-Petit, B., Julien, S., and Delannoy, P. (2001) Biochimie (Paris) 83, 727–737 Takashima, S., Ishida, H. K., Inazu, T., Ando, T., Ishida, H., Kiso, M., Tsuji, S., and Tsujimoto, M. (2002) J. Biol. Chem. 277, 24030 –24038 Takashima, S., Tsuji, S., and Tsujimoto, M. (2002) J. Biol. Chem. 277, 45719 – 45728 Krzewinski-Recchi, M. A., Julien, S., Juliant, S., Teintenier-Lelievre, M., Samyn-Petit, B., Montiel, M. D., Mir, A. M., Cerutti, M., Harduin-Lepers, A., and Delannoy, P. (2003) Eur. J. Biochem. 270, 950 –961 Paulson, J. C., and Colley, K. J. (1989) J. Biol. Chem. 264, 17615–17618 Datta, A. K., and Paulson, J. C. (1997) Indian J. Biochem. Biophys. 34, 157–165 Geremia, R. A., Harduin-Lepers, A., and Delannoy, P. (1997) Glycobiology 7, v–vii Datta, A. K., and Paulson, J. C. (1995) J. Biol. Chem. 270, 1497–1500 Datta, A. K., Sinha, A., and Paulson, J. C. (1998) J. Biol. Chem. 273, 9608 –9614 Kitazume-Kawaguchi, S., Kabata, S., and Arita, M. (2001) J. Biol. Chem. 276, 15696 –15703 Butters, T. D., Hughes, R. C., and Vischer, P. (1981) Biochim. Biophys. Acta 640, 672– 686 Hooker, A. D., Green, N. H., Baines, A. J., Bull, A. T., Jenkins, N., Strange, P. G., and James, D. C. (1999) Biotechnol. Bioeng. 63, 559 –572 Lopez, M., Tetaert, D., Juliant, S., Gazon, M., Cerutti, M., Verbert, A., and Delannoy, P. (1999) Biochim. Biophys. Acta 1427, 49 – 61 Marz, L., Altmann, F., Staudacher, E., and Kubelka, V. (1995) in Glycoproteins (Montreuil, J., Schachter, H., and Vliegenthart, J. F., eds) pp. 543–563, Elsevier Science Publishers B.V., Amsterdam Tomiya, N., Ailor, E., Lawrence, S. M., Betenbaugh, M. J., and Lee, Y. C. (2001) Anal. Biochem. 293, 129 –137 Hiruma, K., and Riddiford, L. M. (1988) Dev. Biol. 130, 87–97 Roth, J., Kempf, A., Reuter, G., Schauer, R., and Gehring, W. J. (1992) Science 256, 673– 675 Vadgama, M. R., and Kamat, D. N. (1969) Histochemie 19, 184 –188 Malykh, Y. N., Krisch, B., Gerardy-Schahn, R., Lapina, E. B., Shaw, L., and Schauer, R. (1999) Glycoconj. J. 16, 731–739 Hollister, J. R., and Jarvis, D. L. (2001) Glycobiology 11, 1–9 Hollister, J. R., Shaper, J. H., and Jarvis, D. L. (1998) Glycobiology 8, 473– 480 Jarvis, D. L., and Finn, E. E. (1996) Nat. Biotechnol. 14, 1288 –1292 Marchal, I., Jarvis, D. L., Cacan, R., and Verbert, A. (2001) Biol. Chem. 382, 151–159 FlyBase (2003) Nucleic Acids Res. 31, 172–175 Coutinho, P. M., and Henrissat, B. (1999) in Recent Advances in Carbohydrate Bioengineering (Gilbert, H. J., Davies, G., Henrissat, B., and Svensson, B., eds) pp. 3–12, Royal Society of Chemistry, Cambridge, UK Angata, T., and Varki, A. (2000) J. Biol. Chem. 275, 22127–22135 Kim, K., Lawrence, S. M., Park, J., Pitts, L., Vann, W. F., Betenbaugh, M. J., and Palter, K. B. (2002) Glycobiology 12, 73– 83 Hindsgaul, O., Kaur, K. J., Srivastava, G., Blaszczyk-Thurin, M., Crawley, S. C., Heerze, L. D., and Palcic, M. M. (1991) J. Biol. Chem. 266, 17858 –17862 Kramerov, A. A., Arbatsky, N. P., Rozovsky, Y. M., Mikhaleva, E. A., Polesskaya, O. O., Gvozdev, V. A., and Shibaev, V. N. (1996) FEBS Lett. 378, 213–218 Seppo, A., Moreland, M., Schweingruber, H., and Tiemeyer, M. (2000) Eur. J. Biochem. 267, 3549 –3558 Niman, H. L., Houghten, R. A., Walker, L. E., Reisfeld, R. A., Wilson, I. A.,
ST6Gal-related Sialyltransferase from D. melanogaster 36. 37. 38. 39. 40. 41. 42. 43.
44. 45.
46. 47. 48. 49.
50. 51. 52. 53. 54. 55. 56. 57.
58. 59. 60. 61. 62. 63.
Hogle, J. M., and Lerner, R. A. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 4949 – 4953 Bunch, T. A., Grinblat, Y., and Goldstein, L. S. (1988) Nucleic Acids Res. 16, 1043–1061 Opalka, N., Mooney, R. A., Richter, C., Severinov, K., Landick, R., and Darst, S. A. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 617– 622 Kirkpatrick, R. B., Ganguly, S., Angelichio, M., Griego, S., Shatzman, A., Silverman, C., and Rosenberg, M. (1995) J. Biol. Chem. 270, 19800 –19805 Panin, V. M., Shao, L., Lei, L., Moloney, D. J., Irvine, K. D., and Haltiwanger, R. S. (2002) J. Biol. Chem. 277, 29945–29952 Williams, M. A., and McCluer, R. H. (1980) J. Neurochem. 35, 266 –269 Dahms, N. M., and Schnaar, R. L. (1983) J. Neurosci. 3, 806 – 817 Green, E. D., Adelt, G., Baenziger, J. U., Wilson, S., and Van Halbeek, H. (1988) J. Biol. Chem. 263, 18253–18268 Coddeville, B., Strecker, G., Wieruszeski, J. M., Vliegenthart, J. F., van Halbeek, H., Peter-Katalinic, J., Egge, H., and Spik, G. (1992) Carbohydr. Res. 236, 145–164 Panin, V. M., Papayannopoulos, V., Wilson, R., and Irvine, K. D. (1997) Nature 387, 908 –912 Rothwell, W. F., and Sullivan, W. (2000) in Drosophila Protocols (Sullivan, W., Ashburner, M., and Hawley, R. S., eds) pp. 141–157, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY Lehmann, R., and Tautz, D. (1994) Methods Cell Biol. 44, 575–598 Rubin, G. M., Hong, L., Brokstein, P., Evans-Holm, M., Frise, E., Stapleton, M., and Harvey, D. A. (2000) Science 287, 2222–2224 Kurosawa, N., Kawasaki, M., Hamamoto, T., Nakaoka, T., Lee, Y. C., Arita, M., and Tsuji, S. (1994) Eur. J. Biochem. 219, 375–381 Mercier, D., Wierinckx, A., Oulmouden, A., Gallet, P. F., Palcic, M. M., Harduin-Lepers, A., Delannoy, P., Petit, J. M., Leveziel, H., and Julien, R. (1999) Glycobiology 9, 851– 863 Tatusova, T. A., and Madden, T. L. (1999) FEMS Microbiol. Lett. 174, 247–250 Sisson, J. C., Field, C., Ventura, R., Royou, A., and Sullivan, W. (2000) J. Cell Biol. 151, 905–918 Bruckner, K., Perez, L., Clausen, H., and Cohen, S. (2000) Nature 406, 411– 415 Paulson, J. C., Beranek, W. E., and Hill, R. L. (1977) J. Biol. Chem. 252, 2356 –2362 Weinstein, J., Lee, E. U., McEntee, K., Lai, P. H., and Paulson, J. C. (1987) J. Biol. Chem. 262, 17735–17743 Garcia, J. V., Fenton, B. W., and Rosner, M. R. (1988) Biochemistry 27, 4237– 4244 Nakamura, Y., Haines, N., Chen, J., Okajima, T., Furukawa, K., Urano, T., Stanley, P., and Irvine, K. D. (2002) J. Biol. Chem. 277, 46280 – 46288 Hamamoto, T., and Tsuji, S. (2002) in Handbook of Glycosyltransferases and Related Genes (Taniguchi, N., Honke, K., and Fukuda, M., eds) pp. 295–300, Springer-Verlag, Tokyo Joosten, J. A. F., Evers, B., van Summeren, R. P., Kamerling, J. P., and Vliegenthart, J. F. (2003) Eur. J. Org. Chem. 2003, 3569 –3586 Vliegenthart, J. F., Dorland, L., and van Halbeek, H. (1983) Adv. Carbohydr. Chem. Biochem. 41, 209 –374 Bergwerff, A. A., Van Oostrum, J., Kamerling, J. P., and Vliegenthart, J. F. (1995) Eur. J. Biochem. 228, 1009 –1019 Weisshaar, G., Hiyama, J., Renwick, A. G., and Nimtz, M. (1991) Eur. J. Biochem. 195, 257–268 Slangen, C. J., and Visser, S. (1999) J. Agric. Food Chem. 47, 4549 – 4556 Toivonen, S., Aitio, O., and Renkonen, O. (2001) J. Biol. Chem. 276,
4357
37141–37148 64. Torres, B. V., McCrumb, D. K., and Smith, D. F. (1988) Arch. Biochem. Biophys. 262, 1–11 65. Van den Eijnden, D. H., Bakker, H., Neeleman, A. P., Van den Nieuwenhof, I. M., and Van Die, I. (1997) Biochem. Soc. Trans. 25, 887– 893 66. Hard, K., Van Zadelhoff, G., Moonen, P., Kamerling, J. P., and Vliegenthart, F. G. (1992) Eur. J. Biochem. 209, 895–915 67. Kobata, A., and Yamashita, K. (1993) in Glycobiology: A Practical Approach. (Fukuda, M., and Kobata, A., eds) pp. 103–125, Oxford University Press, New York 68. Raju, T. S., Briggs, J. B., Borge, S. M., and Jones, A. J. (2000) Glycobiology 10, 477– 486 69. Kubelka, V., Altmann, F., Staudacher, E., Tretter, V., Marz, L., Hard, K., Kamerling, J. P., and Vliegenthart, J. F. (1993) Eur. J. Biochem. 213, 1193–1204 70. Kubelka, V., Altmann, F., and Marz, L. (1995) Glycoconj. J. 12, 77– 83 71. van Die, I., van Tetering, A., Bakker, H., van den Eijnden, D. H., and Joziasse, D. H. (1996) Glycobiology 6, 157–164 72. Shibuya, N., Goldstein, I. J., Broekaert, W. F., Nsimba-Lubaki, M., Peeters, B., and Peumans, W. J. (1987) J. Biol. Chem. 262, 1596 –1601 73. Nemansky, M., and Van den Eijnden, D. H. (1992) Biochem. J. 287, 311–316 74. Wlasichuk, K. B., Kashem, M. A., Nikrad, P. V., Bird, P., Jiang, C., and Venot, A. P. (1993) J. Biol. Chem. 268, 13971–13977 75. Weinstein, J., de Souza-e-Silva, U., and Paulson, J. C. (1982) J. Biol. Chem. 257, 13845–13853 76. Takashima, S., Tsuji, S., and Tsujimoto, M. (2003) J. Biochem. (Tokyo) 134, 287–296 77. Joziasse, D. H., Schiphorst, W. E., van den Eijnden, D. H., van Kuik, J. A., van Halbeek, H., and Vliegenthart, J. F. (1985) J. Biol. Chem. 260, 714 –719 78. Nelson, R. W., Bates, P. A., and Rutishauser, U. (1995) J. Biol. Chem. 270, 17171–17179 79. Sjoberg, E. R., Kitagawa, H., Glushka, J., van Halbeek, H., and Paulson, J. C. (1996) J. Biol. Chem. 271, 7450 –7459 80. Kawar, Z. S., Van Die, I., and Cummings, R. D. (2002) J. Biol. Chem. 277, 34924 –34932 81. Dell, A., Morris, H. R., Easton, R. L., Panico, M., Patankar, M., Oehniger, S., Koistinen, R., Koistinen, H., Seppala, M., and Clark, G. F. (1995) J. Biol. Chem. 270, 24116 –24126 82. Ohkura, T., Seko, A., Hara-Kuge, S., and Yamashita, K. (2002) J. Biochem. (Tokyo) 132, 891–901 83. Tai, M. H., and Zipser, B. (1999) Dev. Biol. 214, 258 –276 84. Schmid, A., Chiba, A., and Doe, C. Q. (1999) Development 126, 4653– 4689 85. Watty, A., and Burden, S. J. (2002) J. Biol. Chem. 277, 50457–50462 86. Cserzo, M., Wallin, E., Simon, I., von Heijne, G., and Elofsson, A. (1997) Protein Eng. 10, 673– 676 87. Nielsen, H., Engelbrecht, J., Brunak, S., and von Heijne, G. (1997) Protein Eng. 10, 1– 6 88. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994) Nucleic Acids Res. 22, 4673– 4680 89. Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biol. 157, 105–132 90. Aoki, N., Matsuda, T., Sakiyama, T., Iwatsuki, K., and Furukawa, K. (1997) Biochim. Biophys. Acta 1334, 207–213 91. van Halbeek, H., Dorland, L., Vliegenthart, J. F., Montreuil, J., Fournet, B., and Schmid, K. (1981) J. Biol. Chem. 256, 5588 –5590 92. Itoh, M., Fukumoto, S., Baba, N., Kuga, Y., Mizuno, A., and Furukawa, K. (1999) Glycobiology 9, 1247–1252
