Van Den Eijnden Research Articles

Hedgehog-stimulated Phosphorylation of the Kinesin-related Protein Costal2 Is Mediated by the Serine/Threonine Kinase Fused

The Hedgehog (Hh) signaling molecule is required for the development of numerous tissues in Drosophila. Within the cell, Hh signal transduction utilizes a large protein complex consisting of the Fused (Fu), Costal2 (Cos2), and Cubitis interruptus (Ci) proteins, but the functional interactions between these proteins are still largely uncharacterized. Using a baculovirus system, we demonstrate that the serine/threonine kinase Fu phosphorylates the kinesin-like protein Cos2 when coexpressed with Cos2. Coexpression of Cos2 and a kinase-inactive version of Fu eliminates the majority of Cos2 phosphorylation. We then show that the primary Fu-induced phosphorylation site of Cos2 is serine 572, whereas serine 931 is phosphorylated to a lesser extent. Mutation of serine 572 to alanine eliminates most, but not all, specific phosphopeptides of Cos2 when coexpressed with Fu. We also demonstrate that the phosphorylation pattern of Cos2 produced by baculovirus coexpression with kinase-dead Fu is almost identical to the phosphorylation pattern of Cos2 isolated from unstimulated S2 cells. Finally, the phosphorylation pattern of Cos2 produced by baculovirus coinfection with wild-type Fu is almost identical to that of Cos2 isolated from S2 cells stimulated by Hh, indicating that phosphorylation of serines 572 and 931 is a genuine Hh signaling event. This study clarifies the unique functions of Fu and Cos2 in Hh signal transduction and identifies only the second known phosphorylation site of a kinesin-like molecule.

Identification and Characterization of Three Novel β1,3-N-Acetylglucosaminyltransferases Structurally Related to the β1,3-Galactosyltransferase Family

We have isolated three types of cDNAs encoding novel beta1,3-N-acetylglucosaminyltransferases (designated beta3Gn-T2, -T3, and -T4) from human gastric mucosa and the neuroblastoma cell line SK-N-MC. These enzymes are predicted to be type 2 transmembrane proteins of 397, 372, and 378 amino acids, respectively. They share motifs conserved among members of the beta1,3-galactosyltransferase family and a beta1,3-N-acetylglucosaminyltransferase (designated beta3Gn-T1), but show no structural similarity to another type of beta1,3-N-acetylglucosaminyltransferase (iGnT). Each of the enzymes expressed by insect cells as a secreted protein fused to the FLAG peptide showed beta1,3-N-acetylglucosaminyltransferase activity for type 2 oligosaccharides but not beta1,3-galactosyltransferase activity. These enzymes exhibited different substrate specificity. Transfection of Namalwa KJM-1 cells with beta3Gn-T2, -T3, or -T4 cDNA led to an increase in poly-N-acetyllactosamines recognized by an anti-i-antigen antibody or specific lectins. The expression profiles of these beta3Gn-Ts were different among 35 human tissues. beta3Gn-T2 was ubiquitously expressed, whereas expression of beta3Gn-T3 and -T4 was relatively restricted. beta3Gn-T3 was expressed in colon, jejunum, stomach, esophagus, placenta, and trachea. beta3Gn-T4 was mainly expressed in brain. These results have revealed that several beta1,3-N-acetylglucosaminyltransferases form a family with structural similarity to the beta1,3-galactosyltransferase family. Considering the differences in substrate specificity and distribution, each beta1,3-N-acetylglucosaminyltransferase may play different roles.

Control of Bisecting GlcNAc Addition to N-Linked Sugar Chains

In the present study, experimental control of the formation of bisecting GlcNAc was investigated, and the competition between beta-1,4-GalT (UDP-galactose:N-acetylglucosamine beta-1, 4-galactosyltransferase) and GnT-III (UDP-N-acetylglucosamine:beta-d-mannoside beta-1, 4-N-acetylglucosaminyltransferase) was examined. We isolated a beta-1,4-GalT-I single knockout human B cell clone producing monoclonal IgM and several transfectant clones that overexpressed beta-1,4-GalT-I or GnT-III. In the beta-1,4-GalT-I-single knockout cells, the extent of bisecting GlcNAc addition to the sugar chains of IgM was increased, where beta-1,4-GalT activity was reduced to about half that in the parental cells, and GnT-III activity was unaltered. In the beta-1,4-GalT-I transfectants, the extent of bisecting GlcNAc addition was reduced although GnT-III activity was not altered significantly. In the GnT-III transfectants, the extent of bisecting GlcNAc addition increased along with the increase in levels of GnT-III activity. The extent of bisecting GlcNAc addition to the sugar chains of IgM was significantly correlated with the level of intracellular beta-1,4-GalT activity relative to that of GnT-III. These results were interpreted as indicating that beta-1, 4-GalT competes with GnT-III for substrate in the cells.

Another note on forced burgers turbulence

The power law range for the velocity gradient probability density function in forced Burgers turbulence has been an issue of intense discussion recently. It is shown in E and Vanden Eijnden, Phys. Rev. Lett. 83, 2572 (1999) that the negative exponent in the assumed power law range has to be strictly larger than 3. Here we give another direct argument for that result, working with finite viscosity. At the same time we discuss viscous corrections to the power law range. This should answer the questions raised in Kraichnan, Phys. Fluids 11, 3738 (1999) regarding the results of E and Vanden Eijnden, Phys. Rev. Lett. 83 2572 (1999).

The effect of increasing nucleotide-sugar concentrations on the incorporation of sugars into glycoconjugates in rat hepatocytes.

Treatment of rat hepatocytes with 0.5 mM concentrations of uridine and cytidine results in increased cellular concentrations of UTP, UDP-sugars and CTP, whereas that of CMP-N-acetylneuraminate remained unchanged [Pels Rijcken, Overdijk, Van den Eijnden and Ferwerda (1993) Biochem. J. 293, 207-213]. The incorporation of radioactivity from 3H-labelled sugars into the cell-associated and secreted glycoconjugate fraction was influenced by these altered cellular concentrations of the nucleotides. For [3H]glucosamine, pretreatment with uridine resulted in a reduction of the glycosylation in both fractions. Increases in the secreted fractions were observed for fucose with both uridine and cytidine and for N-acetylglucosamine with uridine only. With [3H]N-acetylglucosamine, similar specific radioactivities for UDP-N-acetylhexosamine and CMP-N-acetylneuraminate were found, regardless of the pretreatment conditions. With [3H]N-acetylmannosamine, the specific radioactivity of CMP-N-acetylneuraminate showed an almost 2-fold increase on pretreatment. The latter increase did not result in an increased incorporation of radioactivity into the glycoconjugates. It was estimated that, in untreated cells, the ratio of radioactivity incorporated from [3H]glucosamine into glycoconjugate-bound N-acetylhexosamine and N-acetylneuraminate amounted to 2:3. In pretreated cells this ratio changed to approx. 2:1. Overall, the data show that pretreatment resulted in an increased incorporation of N-acetylhexosamine into cell-associated and secreted glycoconjugates, accompanied by a reduction in sialylation. It was concluded that an increased availability of UDP-N-acetylhexosamine caused the increased incorporation of N-acetylhexosamine. The elevated cytosolic level of UDP-N-acetylhexosamine (and of compounds like CMP) is suggested to impair the transport of CMP-acetylneuraminate to the Golgi, resulting in reduced sialylation. This study demonstrates that protein glycosylation can be regulated at the level of the availability of the various nucleotide-sugars in the Golgi lumen.

Murine alpha 1,3-galactosyltransferase. A single gene locus specifies four isoforms of the enzyme by alternative splicing.

We have reported the characterization of a cDNA (clone 31A) encoding bovine alpha 1,3-galactosyltransferase (alpha 1,3-GT) (Joziasse, D. H., Shaper, J. H., Van den Eijnden, D. H., Van Tunen, A. J., and Shaper, N. L. (1989) J. Biol. Chem. 264, 14290-14297). With the goal of isolating a full-length cDNA encoding murine alpha 1,3-GT we screened a cDNA library with clone 31A and isolated a 3.4-kilobase (kb) alpha 1,3-GT clone (4A). The murine coding sequence is 78% similar to that of the bovine alpha 1,3-GT cDNA, but the "stem" region (defined as the region that links the single transmembrane domain to the catalytic domain) of the murine alpha 1,3-GT encoded by clone 4A, is 31 amino acids shorter than the corresponding region of the bovine alpha 1,3-GT. To screen for heterogeneity in the murine alpha 1,3-GT transcripts, we carried out a polymerase chain reaction (PCR) analysis on mouse C127 cDNA. Four distinct transcripts were detected, which predict four isoforms of the alpha 1,3-GT polypeptide that differ only in the length of their stem region. To determine how the four different transcripts are generated from a single gene, we have established the genomic organization for murine alpha 1,3-GT. The full-length mRNA spans at least 35 kb of genomic DNA and is distributed over nine exons that range in size from 36 base pairs (bp) to approximately 2600 bp. The protein coding region is distributed over six exons, and the 5'-untranslated sequence is distributed over three exons. Comparison of the genomic DNA sequence with that of the four different mRNAs indicates that these transcripts are produced by alternative splicing of the murine pre-mRNA according to a cassette model. A tissue survey using RNA-PCR revealed the presence of four different alpha 1,3-GT transcripts in all mouse tissues and cell lines examined to date, with the notable exception of male germ cells. Additionally, although alpha 1,3-GT levels increased upon thioglycollate-induced activation of mouse peritoneal macrophages, the ratio of the alpha 1,3-GT isoforms was essentially unchanged. Similar results were obtained upon retinoic acid-induced differentiation of murine F9 teratocarcinoma cells. Lastly, a similar PCR analysis of bovine cDNA produced only a single DNA fragment, corresponding to bovine cDNA clone 31A.

Characterization of an alpha 1—-3-galactosyltransferase homologue on human chromosome 12 that is organized as a processed pseudogene.

UDP-Gal:Gal beta 1----4GlcNAc alpha 1----3-galactosyltransferase is a terminal glycosyltransferase that is widely expressed in a variety of mammalian species, with the notable exception of man, apes, and Old World monkeys. We recently reported the isolation of a bovine cDNA clone that contains the complete coding sequence for this enzyme (Joziasse, D. H., Shaper, J. H., Van den Eijnden, D. H., Van Tunen, A. J., and Shaper, N. L. (1989) J. Biol. Chem. 264, 14290-14297). Using this cDNA as a probe, we have demonstrated that, although transcripts cannot be detected in a variety of established human cell lines by Northern blot analysis, homologous sequences are present in human genomic DNA. To establish that these sequences represent a human homologue of alpha 1----3-galactosyltransferase, we have used the bovine cDNA as a probe to isolate two nonoverlapping clones (HGT-2 and HGT-10) from a human genomic DNA library. Clone HGT-2 contains a 1.5-kilobase uninterrupted linear sequence similar to bovine alpha 1----3-galactosyltransferase that is organized as a processed pseudogene. This sequence, flanked by Alu type repeats, contains a short 5'- and 3'-untranslated region and a complete recognizable coding region that is 81% similar at the nucleotide level to bovine alpha 1----3-galactosyltransferase. This putative coding region contains multiple frameshift mutations and nonsense codons in all three reading frames which precludes the synthesis of a functional enzyme. Nevertheless, after optimal alignment, translation predicts a polypeptide that is 68% similar at the amino acid level to the bovine enzyme. Based on Southern analysis and limited sequence analysis, clone HGT-10 contains coding sequences similar to the NH2-terminal region of bovine alpha 1----3-galactosyltransferase. By analysis of panels of human-rodent somatic cell hybrids we have established that the nonfunctional, processed pseudogene and the human homologue represented by HGT-10 are located on human chromosomes 12 and 9, respectively. Interestingly, a comparison of the predicted amino acid sequence of the carboxyl-terminal two-thirds of human alpha 1----3-galactosyltransferase, with the corresponding region of the human blood group A, UDP-GalNAc:[Fuc alpha 1----2]Gal beta 1----4GlcNAc alpha 1----3-GalNAc-transferase (Yamamoto, F., Marken, J., Tsuji, T., White, T., Clausen, H., and Hakomori, S. (1990a) J. Biol. Chem. 265, 1146-1151), reveals a significant similarity (39%) suggesting that these two enzymes may have arisen from the same ancestral gene as a result of gene duplication and subsequent divergence.

Regulation of the expression of Gal alpha 1-3Gal beta 1-4GlcNAc glycosphingolipids in kidney.

Previous studies (Galili, U., Clark, M. R., Shohet, S. B., Buehler, J., and Macher, B. A. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 1369-1373; Galili, U., Shohet, S. B., Korbrin, E., Stults, C. L. M., and Macher, B. A. (1988) J. Biol. Chem. 263, 17755-17762) have established that there is a unique evolutionary distribution of glycoconjugates carrying the Gal alpha 1-3Gal beta 1-4GlcNAc epitope. These glycoconjugates are expressed by cells from New World monkeys and non-primate mammals, but not by cells from humans, Old World monkeys, or apes. The lack of expression of this epitope in the latter species appears to result from the suppression of gene expression for the enzyme UDP-galactose:nLc4Cer alpha 1-3-galactosyltransferase (alpha 1-3GalT) (Joziasse, D. H., Shaper, J. H., Van den Eijnden, D. H., Van Tunen, A. J., and Shaper, N. L. (1989) J. Biol. Chem. 264, 14290-14297). Although many non-primate species are known to express this carbohydrate epitope, the nature (i.e. glycoprotein or glycosphingolipid) of the glycoconjugate carrying this epitope is only known for a few tissues in a few animal species. Furthermore, it is not known whether all animal species express this epitope in the same tissues. We have investigated these questions by analyzing the glycosphingolipids in kidney from several non-primate animal species. Immunostained thin layer chromatograms of glycosphingolipids from sheep, pig, rabbit, cow, and rat kidney with the Gal alpha 1-3Gal beta 1-4GlcNAc glycosphingolipid-specific monoclonal antibody, Gal-13, demonstrated that kidney from all of these species except rat contained Gal alpha 1-3Gal beta 1-4GlcNAc neutral glycosphingolipids. A lack of expression of Gal alpha 1-3Gal beta 1-4GlcNAc glycosphingolipids in rat may be due to the lack of expression of the enzyme (alpha 1-3GalT) which catalyzes the formation of the Gal alpha 1-3Gal nonreducing terminal sequence of these compounds or to the lack of expression of glycosyltransferases which are necessary for the synthesis of the neolacto core structure of these compounds. These possibilities were evaluated in two ways. First, the three enzymes (UDP-N-acetylglucosamine:LacCer beta 1-3-N-acetyl-glucosaminyltransferase, UDP-galactose:Lc3Cer beta 1-4-galactosyltransferase, and alpha 1-3GalT) involved in the synthesis of the Gal alpha 1-3Gal beta 1-4GlcNAc glycosphingolipids were assayed using an enzyme-linked immunosorbent assay-based assay system and carbohydrate sequence-specific monoclonal antibodies. Second, TLC immunostaining was done to determine if the glycosphingolipid precursors (i.e. Lc3Cer and nLc4Cer) are expressed in rat kidney. Interestingly, rat kidney had a relatively high level of alpha 1-3GalT activity compared with the other animals tested.(ABSTRACT TRUNCATED AT 400 WORDS)

Identification of a region of UDP-galactose:N-acetylglucosamine beta 4-galactosyltransferase involved in UDP-galactose binding by differential labeling.

The location of regions in the primary structure of UDP-galactose:N-acetylglucosamine beta 4-galactosyl-transferase (GT) that are involved in binding UDP-galactose has been investigated by differential chemical modification with two different reagents in the presence and absence of UDP-galactose. Treatment with periodate-cleaved UDP and NaCNBH3 resulted in a loss of 80% of GT activity, which was largely prevented by UDP-galactose. Stoichiometry of labeling and peptide maps of the modified enzyme samples indicated partial labeling at many sites. A major site of reaction in the absence of UDP-galactose that was essentially unmodified in its presence was found to correspond to Lys341 in the cDNA sequence of GT. As a second approach, the reactivities of the amino groups of GT were compared in the presence and absence of saturating levels of UDP-galactose by trace acetylation with [3H]acetic anhydride. UDP-galactose binding was found to perturb the reactivities of a number of lysines in the C-terminal region of GT, the most pronounced effect being a reduction in the reactivity of Lys351. The two procedures thus identified a region between residues 341 and 351 as being associated with UDP-galactose binding. This region overlaps a small section in the sequence of GT that was previously noted to be similar to part of bovine alpha-1,3-galactosyltransferase (Joziasse, D. H., Shaper, J. H., Van den Eijnden, D. H., Van Tunen, A. J., and Shaper, N. L. (1989) J. Biol. Chem. 264, 14290-14297). Sequence comparisons indicate that extended regions at the C terminus of each enzyme encompassing this area may represent homologous UDP-galactose-binding domains.

α1 → 3‐Galactosyltransferase: the use of recombinant enzyme for the synthesis of α‐galactosylated glycoconjugates

We have reported the isolation and characterization of a bovine cDNA clone containing the complete coding sequence for UDP-Gal:Gal beta 1----4GlcNAc alpha 1----3-galactosyltransferase [Joziasse, D. H., Shaper, J. H., Van den Eijnden, D. H., Van Tunen, A. J. & Shaper, N. L. (1989) J. Biol. Chem. 264, 14290-14297]. Insertion of this cDNA clone into the genome of Autographa californica nuclear polyhedrosis virus (AcNPV) and subsequent infection of Spodoptera frugiperda (Sf9) insect cells with recombinant virus, resulted in high-level expression of enzymatically active alpha 1----3-galactosyltransferase. The expressed enzyme accounted for about 2% of the cellular protein; the corresponding specific enzyme activity was 1000-fold higher than observed in calf thymus, the tissue with the highest specific enzyme activity reported to date. The recombinant alpha 1----3-galactosyltransferase could be readily detergent-solubilized and subsequently purified by affinity chromatography on UDP-hexanolamine-Sepharose. The recombinant alpha 1----3-galactosyltransferase showed the expected preference for the acceptor substrate N-acetyllactosamine (Gal beta 1----4GlcNAc), and demonstrated enzyme kinetics identical to those previously reported for affinity-purified calf thymus alpha 1----3-galactosyltransferase [Blanken, W. M. & Van den Eijnden, D. H. (1985) J. Biol. Chem. 260, 12927-12934]. In pilot studies, the recombinant enzyme was examined for the ability to synthesize alpha 1----3-galactosylated oligosaccharides, glycolipids and glycoproteins. By a combination of 1H-NMR, methylation analysis, HPLC, and exoglycosidase digestion it was established that, for each of the model compounds, the product of galactose transfer had the anticipated terminal structure, Gal alpha 1----3Gal beta 1----4-R. Our results demonstrate that catalysis by recombinant alpha 1----3-galactosyltransferase can be used to obtain preparative quantities of various alpha 1----3-galactosylated glycoconjugates. Therefore, enzymatic synthesis using the recombinant enzyme is an effective alternative to the chemical synthesis of these biologically relevant compounds.

Biosynthesis of polylactosaminoglycans. Novikoff ascites tumor cells contain two UDP-GlcNAc:beta-galactoside beta 1----6-N-acetylglucosaminyltransferase activities.

Novikoff ascites tumor cells contain a UDP-GlcNAc:beta-galactoside beta 1----6-N-acetylglucosaminyltransferase (beta 6-GlcNAc-transferase B) that acts on galactosides and N-acetylgalactosaminides in which the accepting sugar is beta 1----3 substituted by a Gal or GlcNAc residue. Characterization of enzyme products by 1H-NMR and methylation analysis indicates that an R beta 1----3(GlcNAc beta 1----6)Gal- branching point is formed such as occurs in blood-group-I-active substances. The enzyme does not show an absolute divalent cation requirement and 20 mM EDTA is not inhibitory. The activity is strongly inhibited by Triton X-100 at concentrations of greater than or equal to 0.2%. Competition studies suggest that a single enzyme acts on Gal beta 1----3Gal beta 1----4Glc, GlcNAc beta 1----3Gal beta 1----4GlcNAc and GlcNAc beta 1----3GalNAc alpha-O-benzyl (Km values 0.71, 0.83 and 0.53 mM, respectively). Gal beta----3Gal beta 1----4Glc as an acceptor substrate for beta 6-GlcNAc-transferase B does not inhibit the incorporation of GlcNAc in beta 1----6 linkage to the terminal Gal residues of asialo-alpha 1-acid glycoprotein catalyzed by a beta-galactoside beta 1----6-N-acetylglucosaminyltransferase (beta 6-GlcNAc-transferase A) previously described in Novikoff ascites tumor cells [D. H. Van den Eijnden, H. Winterwerp, P. Smeeman & W.E.C.M. Schiphorst (1983) J. Biol. Chem. 258, 3435-3437]. Neither is Triton X-100 at a concentration of 0.8% inhibitory for the activity of beta 6-GlcNAc-transferase A. This activity is absent from hog gastric mucosa microsomes, which has been described to contain high levels of beta 6-GlcNAc-transferase B. [F. Piller, J. P. Cartron, A. Maranduba, A. Veyrières, Y. Leroy & B. Fournet (1984) J. Biol. Chem. 259, 13,385-13,390]. Our results show that Novikoff tumor cells contain two beta-galactoside beta 6-GlcNAc-transferases, which differ in acceptor specificity and tolerance towards Triton X-100. A role for these enzymes in the synthesis of branched polylactosaminoglycans and of O-linked oligosaccharide core structures having blood-group I activity is proposed.

Purification and enzymatic characterization of CMP-sialic acid: beta-galactosyl1----3-N-acetylgalactosaminide alpha 2----3-sialyltransferase from human placenta.

A CMP-NeuAc:Gal beta 1----3GalNAc-R alpha 2----3-sialyltransferase has been purified over 20,000-fold from a Triton X-100 extract of human placenta by affinity chromatography on concanavalin A-Sepharose and CDP-hexanolamine-Sepharose in a yield of 10%. Sodium dodecyl sulfate-gel electrophoresis under reducing conditions revealed that the enzyme consists of a major polypeptide species with a molecular weight of 41,000 and some minor forms with molecular weights of 40,000, 43,000, and 65,000, respectively, which can be resolved partially by gel filtration on Sephadex G-100. Isoelectric focusing revealed that the enzyme occurs in a major and a minor charged form with pI values of 5.0-5.5 and 6.0, respectively. Acceptor specificity studies indicated that the enzyme catalyzes the incorporation of sialic acid from CMP-NeuAc into glycoproteins, glycolipids, and oligosaccharides which possess a terminal Gal beta----3GalNAc unit. Analysis of the structure of the product chain by high-pressure liquid chromatography and thin layer chromatography as well as methylation analysis revealed that a NeuAc alpha 2----3Gal beta 1----3GalNAc sequence is elaborated. The best glycoprotein acceptors are antifreeze glycoprotein and porcine submaxillary asialo/afucomucin. The disaccharide Gal beta 1----3GalNAc-Thr shows values for Km and V which are close to those of the latter glycoprotein. Lactose as well as oligosaccharides in which galactose is linked beta 1----3 or beta 1----4 to N-acetylglucosamine are less efficient acceptors. Of the glycolipids tested only gangliosides GM1 and GD1b served as an acceptor. The enzyme does not show an absolute aglycon specificity, and attaches sialic acid regardless the anomeric configuration of the N-acetylgalactosaminyl residue in the accepting Gal beta 1----3GalNAc unit. By use of specific acceptor substrates it could be demonstrated that the purified enzyme is free from other known sialyltransferase activities. Studies with rabbit antibodies raised against a partially purified sialyltransferase preparation indicated that the enzyme is immunologically unrelated to a Gal beta 1----4GlcNAc-R alpha 2----3-sialyltransferase, which previously had been identified in human placenta (Van den Eijnden, D.H., and Schiphorst, W. E. C. M. (1981) J. Biol. Chem. 256, 3159-3162). Initial-rate kinetic studies suggest that the sialyltransferase operates through a mechanism involving a ternary complex of enzyme, sugar donor, and acceptor. This is the first report on the extensive purification and characterization of a sialyltransferase from a human tissue.

Branch specificity of bovine colostrum CMP-sialic acid: N-acetyllactosaminide alpha 2—-6-sialyltransferase. Interaction with biantennary oligosaccharides and glycopeptides of N-glycosylproteins.

By use of 500-MHz 1H NMR spectroscopy, the branch specificity of bovine colostrum CMP-NeuAc:Gal beta 1----4GlcNAc-R alpha 2----6-sialyltransferase towards a biantennary glycopeptide and oligosaccharides of the N-acetyllactosamine type, differing in completeness and structure of their core portion, was investigated. In agreement with earlier reports (Van den Eijnden, D. H., Joziasse, D. H., Dorland, L., Van Halbeek H., Vliegenthart, J. F. G., and Schmid, K. (1980) Biochem. Biophys. Res. Commun. 92, 839-845), it appears that the enzyme strongly prefers the galactosyl residue at the Man alpha 1----3Man branch of the biantennary glycopeptide for attachment of the first sialic acid residue. This branch specificity is fully preserved with the structure (formula; see text) Reduction of the reducing N-acetylglucosaminyl residue in this structure, however, leads to a decreased branch specificity, whereas removal of this residue results in a random attachment of sialic acid to the galactoses at both branches. The decrease in branch specificity is accompanied by a reduction in the rate of sialic acid transfer to the galactose at the alpha 1----3 branch. Our results indicate that the presence of the aforementioned N-acetylglucosaminyl residue is a minimal structural requirement for branch specificity of the sialyltransferase. We propose that in the interaction of the sialyltransferase with its substrates, this N-acetylglucosaminyl residue functions as a recognition site mediating the correct positioning of the substrate on the enzyme.