Publications

The common core 1 O-glycan structure Galbeta1--> 3GalNAc-R is the precursor for many extended mucin-type O-glycan structures in animal cell surface and secreted glycoproteins. Core 1 is synthesized by the transfer of Gal from UDP-Gal to GalNAcalpha1-R by core 1 beta3-galactosyltransferase (core 1 beta3-Gal-T). Amino acid sequences from purified rat core 1 beta3-Gal-T (Ju, T., Cummings, R. D., and Canfield, W. M. (2002) J. Biol. Chem. 277, 169-177) were used to identify the core 1 beta3-Gal-T sequences in the human expressed sequence tag data bases. A 1794-bp human core 1 beta3-Gal-T cDNA sequence was determined by sequencing the expressed sequence tag and performing 5'-rapid amplification of cDNA ends. The core 1 beta3-Gal-T predicts a 363-amino acid type II transmembrane protein. Expression of both the full-length and epitope-tagged soluble forms of the putative enzyme in human 293T cells generated core 1 beta3-Gal-T activity that transferred galactose from UDP-Gal to GalNAcalpha1-O-phenyl, and a synthetic glycopeptide with Thr-linked GalNAc and the product was shown to have the core 1 structure. Northern analysis demonstrated widespread expression of core 1 beta3-Gal-T in tissues with a predominance in kidney, heart, placenta, and liver. Highly homologous cDNAs were identified and cloned from rat, mouse, Drosophila melanogaster, and Caenorhabditis elegans, suggesting that the enzyme is widely distributed in metazoans. The core 1 beta3-Gal-T sequence has minimal homology with conserved sequences found in previously described beta3-galactosyltransferases, suggesting this enzyme is only distantly related to the known beta3-galactosyltransferase family.

The O-linked oligosaccharides (O-glycans) in mammalian glycoproteins are classified according to their core structures. Among the most common is the core 1 disaccharide structure consisting of Galbeta1-->3GalNAcalpha1-->Ser/Thr, which is also the precursor for many extended O-glycan structures. The key enzyme for biosynthesis of core 1 O-glycan from the precursor GalNAc-alpha-Ser/Thr is UDP-Gal:GalNAc-alpha-Ser/Thr beta3-galactosyltransferase (core1 beta3-Gal-T). Core 1 beta3-Gal-T activity, which requires Mn2+, was solubilized from rat liver membranes and purified 71,034-fold to apparent homogeneity (>90% purity) in 5.7% yield by ion exchange chromatography on SP-Sepharose, affinity chromatography on immobilized asialo-bovine submaxillary mucin, and gel filtration chromatography on Superose 12. The purified enzyme is free of contaminating glycosyltransferases. Two peaks of core 1 beta3-Gal-T activity were identified in the final step on Superose 12. One peak of activity contained protein bands on non-reducing SDS-PAGE of approximately 84- and approximately 86-kDa disulfide-linked dimers, whereas the second peak of activity contained monomers of approximately 43 kDa. Reducing SDS-PAGE of these proteins gave approximately 42- and approximately 43-kDa monomers. Both the 84/86-kDa dimers and the 42/43-kDa monomers have the same novel N-terminal sequence. The purified enzyme, which is remarkably stable, has an apparent Km for UDP-Gal of 630 microm and an apparent Vmax of 206 micromol/mg/h protein using GalNAcalpha1-O-phenyl as the acceptor. The reaction product was generated using asialo-bovine submaxillary mucin as an acceptor; treatment with O-glycosidase generated the expected disaccharide Galbeta1-->3GalNAc. These studies demonstrate that activity of the core 1 beta1,3-Gal-T from rat liver is contained within a single, novel, disulfide-bonded, dimeric enzyme.

Human core 1 beta3-galactosyltransferase (C1beta3Gal-T) generates the core 1 O-glycan Galbeta1-3GalNAcalpha1-SerThr (T antigen), which is a precursor for many extended O-glycans in animal glycoproteins. We report here that C1beta3Gal-T activity requires expression of a molecular chaperone designated Cosmc (core 1 beta3-Gal-T-specific molecular chaperone). The human Cosmc gene is X-linked (Xq23), and its cDNA predicts a 318-aa transmembrane protein ( approximately 36.4 kDa) with type II membrane topology. The human lymphoblastoid T cell line Jurkat, which lacks C1beta3Gal-T activity and expresses the Tn antigen GalNAcalpha1-SerThr, contains a normal gene and mRNA encoding C1beta3Gal-T, but contains a mutated Cosmc with a deletion introducing a premature stop codon. Expression of Cosmc cDNA in Jurkat cells restored C1beta3Gal-T activity and T antigen expression. Without Cosmc, the C1beta3Gal-T is targeted to proteasomes. Expression of active C1beta3Gal-T in Hi-5 insect cells requires coexpression of Cosmc. Overexpression of active C1beta3Gal-T in mammalian cell lines also requires coexpression of Cosmc, indicating that endogenous Cosmc may be limiting. A small portion of C1beta3Gal-T copurifies with Cosmc from cell extracts, demonstrating physical association of the proteins. These results indicate that Cosmc acts as a specific molecular chaperone in assisting the foldingstability of C1beta3Gal-T. The identification of Cosmc, a uniquely specific molecular chaperone required for a glycosyltransferase expression in mammalian cells, may shed light on the molecular basis of acquired human diseases involving altered O-glycosylation, such as IgA nephropathy, Tn syndrome, Henoch-Schönlein purpura, and malignant transformation, all of which are associated with a deficiency of C1beta3Gal-T activity.

We report the expression of 3 well-characterized adult Schistosoma mansoni glycan antigens among molluscan stages of the parasite. These antigens are LacdiNAc (LDN; GalNAcbeta1-4GlcNAc-R), fucosylated LacdiNAc (LDNF; GalNAc[Fucal-3]beta1-4GlcNAc-R), and Lewis x (Le(x); Gal[Fucalpha1-3]beta1-4GlcNAc-R). The presence of the glycans was determined by both immunoblot and immunohistological methods using monoclonal antibodies that specifically recognize each glycan epitope. Immunoblot analyses reveal that LDN and LDNF epitopes are expressed on many different glycoproteins, including eggs, mother sporocysts, daughter sporocysts, and cercariae, although LDN expression among daughter sporocysts is greatly reduced. LDN and LDNF epitopes are localized on the tegument and in the intrasporocyst cell masses of both in vitro-derived and in vivo-derived mother sporocysts and in the daughter sporocysts derived on day 16 after infection. Unexpectedly, high levels of LDN and LDNF glycans were detected in the infected, but not in the uninfected, snail hemolymph, suggesting that the infecting larvae secrete LDN and LDNF glycoconjugates into the snail hosts. In contrast, the expression of Le(x) antigen among the molluscan stages is highly restricted. Le(x) is present on a few high-molecular weight glycoproteins in eggs and cercariae but is undetectable in mother and daughter sporocysts. Taken together with our earlier studies on vertebrate stages of S. mansoni, these results show that LDN and LDNF glycans are conserved during schistosome development. The study further extends the evidence that Le(x) is a developmentally regulated antigen in schistosomes.

Murine leukocytes are thought to express alpha2-3-sialylated and alpha1-3-fucosylated selectin ligands such as sialyl Lewis x (sLe(x)), although monoclonal antibodies (mAbs) to sLe(x) or Le(x) reportedly do not bind to murine leukocytes. We observed that P- and E-selectin bound to pronase-sensitive ligands on murine monocytic WEHI-3 cells and murine neutrophils, indicating that the ligands for both selectins are glycoproteins. CSLEX-1, HECA-452, and other widely used mAbs to sLe(x) and Le(x) did not bind to WEHI-3 cells and bound at very low levels to murine neutrophils. Only the anti-sLe(x) mAbs 2H5 and KM93, which also recognize nonfucosylated glycans, bound to WEHI-3 cells. 2H5 and KM93 bound to pronase-resistant structures, indicating that the mAbs did not identify selectin ligands. Treatment of WEHI-3 cells with glycosidases or chlorate demonstrated that sialic acid modifications, alpha1-3-galactosylation, or sulfation did not mask epitopes for mAbs to sLe(x) or Le(x). Compared to human promyelocytic HL-60 cells, WEHI-3 cells and murine neutrophils expressed low alpha1-3-fucosyltransferase activities. Consistent with very low endogenous fucosylation, forced fucosylation of intact WEHI-3 cells or murine neutrophils by exogenous alpha1-3-fucosyltransferase FTVI and GDP-fucose created many new epitopes for anti-sLe(x) mAbs such as HECA-452 and CSLEX-1. Nevertheless, forced fucosylation of intact cells did not significantly augment their ability to bind to fluid-phase P- or E-selectin or to roll on immobilized P- or E-selectin under flow. These data suggest that murine myeloid leukocytes fucosylate only a few specific glycans, which interact preferentially with P- and E-selectin.

Leukocytic inflammation can be limited by inhibiting selectin-dependent leukocyte rolling. In spite of intensive efforts to develop small molecule selectin inhibitors with defined structure-activity profiles, inhibition of P-selectin-dependent leukocyte rolling in vivo by such a compound has yet to be described. We recently reported that glycosulfopeptides (GSP), modeled on the high affinity selectin ligand PSGL-1, inhibit leukocyte binding to P-selectin in vitro. Here, we have used intravital microscopy to investigate whether GSP can inhibit P-selectin-dependent leukocyte rolling in vivo. Surgical preparation of the mouse cremaster muscle for intravital microscopy induced P-selectin-dependent leukocyte rolling. Baseline rolling was recorded for 1 min followed by i.v. injection of GSP. 2-GSP-6 and 4-GSP-6 substantially reversed P-selectin-dependent leukocyte rolling, whereas control GSP, which are not fully glycosylated, did not. Inhibition of leukocyte rolling by 2- and 4-GSP-6 lasted 2-4 min. Clearance studies with 125I-labeled 4-GSP-6 demonstrated rapid reduction in its circulating levels concurrent with accumulation in urine. These data represent the first demonstration that a precisely defined structure based on a natural P-selectin ligand can inhibit P-selectin-dependent leukocyte rolling in vivo.

Leukocytes roll on selectins at nearly constant velocities over a wide range of wall shear stresses. Ligand-coupled microspheres roll faster on selectins and detach quickly as wall shear stress is increased. To examine whether the superior performance of leukocytes reflects molecular features of native ligands or cellular properties that favor selectin-mediated rolling, we coupled structurally defined selectin ligands to microspheres or K562 cells and compared their rolling on P-selectin. Microspheres bearing soluble P-selectin glycoprotein ligand (sPSGL)-1 or 2-glycosulfopeptide (GSP)-6, a GSP modeled after the NH2-terminal P-selectin-binding region of PSGL-1, rolled equivalently but unstably on P-selectin. K562 cells displaying randomly coupled 2-GSP-6 also rolled unstably. In contrast, K562 cells bearing randomly coupled sPSGL-1 or 2-GSP-6 targeted to a membrane-distal region of the presumed glycocalyx rolled more like leukocytes: rolling steps were more uniform and shear resistant, and rolling velocities tended to plateau as wall shear stress was increased. K562 cells treated with paraformaldehyde or methyl-beta-cyclodextrin before ligand coupling were less deformable and rolled unstably like microspheres. Cells treated with cytochalasin D were more deformable, further resisted detachment, and rolled slowly despite increases in wall shear stress. Thus, stable, shear-resistant rolling requires cellular properties that optimize selectin-ligand interactions.

A common terminal structure in glycans from animal glycoproteins and glycolipids is the lactosamine sequence Gal(beta)4GlcNAc-R (LacNAc or LN). An alternative sequence that occurs in vertebrate as well as in invertebrate glycoconjugates is GalNAc(beta)4GlcNAc-R (LacdiNAc or LDN). Whereas genes encoding beta4GalTs responsible for LN synthesis have been reported, the beta4GalNAcT(s) responsible for LDN synthesis has not been identified. Here we report the identification of a gene from Caenorhabditis elegans encoding a UDP-GalNAc:GlcNAc(beta)-R beta1,4-N-acetylgalactosaminyltransferase (Ce(beta)4GalNAcT) that synthesizes the LDN structure. Ce(beta)4GalNAcT is a member of the beta4GalT family, and its cDNA is predicted to encode a 383-amino acid type 2 membrane glycoprotein. A soluble, epitope-tagged recombinant form of Ce(beta)4GalNAcT expressed in CHO-Lec8 cells was active using UDP-GalNAc, but not UDP-Gal, as a donor toward a variety of acceptor substrates containing terminal beta-linked GlcNAc in both N- and O-glycan type structures. The LDN structure of the product was verified by co-chromatography with authentic standards and (1)H NMR spectroscopy. Moreover, Chinese hamster ovary CHO-Lec8 and CHO-Lec2 cells expressing Ce(beta)4GalNAcT acquired LDN determinants on endogenous glycoprotein N-glycans, demonstrating that the enzyme is active in mammalian cells as an authentic beta4GalNAcT. The identification and availability of this novel enzyme should enhance our understanding of the structure and function of LDN-containing glycoconjugates.

Here we report the discovery of a unique fucosyltransferase (FT) in Caenorhabditis elegans. In studying the activities of FTs in extracts of adult C. elegans, we detected activity toward the unusual disaccharide acceptors Galbeta1-4Xyl-R and Galbeta1-6GlcNAc-R to generate products with the general structure Fucalpha1-2Galbeta1-R. We identified a gene encoding a unique alpha1,2FT (designated CE2FT-1), which contains an open reading frame encoding a predicted protein of 355 amino acids with the type 2 topology and domain structure typical of other glycosyltransferases. The predicted cDNA for CE2FT-1 has very low identity (5-10%) at the amino acid level to alpha1,2FT sequences in humans, rabbits, and mice. Recombinant CE2FT-1 expressed in human 293T cells has high alpha1,2FT activity toward the simple acceptor Galbeta-O-phenyl acceptor to generate Fucalpha1-2Galbeta-R, which in this respect resembles mammalian alpha1,2FTs. However, CE2FT-1 is otherwise completely different from known alpha1,2FTs in its acceptor specificity, since it is unable to fucosylate either Galbeta1-4Glcbeta-R or free lactose and prefers the unusual acceptors Galbeta1-4Xylbeta-R and Galbeta1-6GlcNAc-R. Promoter analysis of the CE2FT-1 gene using green fluorescent protein reporter constructs demonstrates that CE2FT-1 is expressed in single cells of early stage embryos and exclusively in the 20 intestinal cells of L(1)-L(4) and adult worms. These and other results suggest that multiple fucosyltransferase genes in C. elegans may encode enzymes with unique activities, expression, and developmental roles.

P-selectin glycoprotein ligand-1 (PSGL-1), a dimeric mucin on leukocytes, is the best characterized ligand for selectins. P-selectin binds stereospecifically to the extreme N terminus of PSGL-1, which contains three clustered tyrosine sulfates (TyrSO3-) adjacent to a Thr residue with a core 2-based O-glycan expressing sialyl Lewis x (C2-O-sLe(x)). GSP-6, a synthetic glycosulfopeptide modeled after the N terminus of PSGL-1, containing three TyrSO3- residues and a short, monofucosylated C2-O-sLe(x) bound to P-selectin with high affinity (K(d) approximately 650 nm). However, PSGL-1 from human HL-60 cells contains higher levels of O-glycans that are sialylated and polyfucosylated polylactosamines (PFPL). Furthermore, studies with fucosyltransferase-deficient mice suggest that sialylated PFPL structures contribute to binding to P-selectin. To resolve whether sialylated PFPL O-glycans participate in binding of PSGL-1 to human P-selectin, we synthesized glycosulfopeptides, designated GSP-6' and GSP-6", with three TyrSO3- residues and either difucosylated polylactosamine (C2-O-Le(x)-sLe(x)) or trifucosylated polylactosamine (C2-O-Le(x)-Le(x)-sLe(x)). Binding of the GSPs to P-selectin was measured by affinity chromatography, fluorescence solid-phase assays, and equilibrium gel filtration. Unexpectedly, both GSP-6' and GSP-6" bound to P-selectin with low affinity (K(d) approximately 37 microm for GSP-6' and K(d) approximately 50 microm for GSP-6"). Binding of GSP-6' and GSP-6" to P-selectin required fucosylation and, to a lesser extent, sialylation as well as the sulfated peptide backbone of GSP-6' and GSP-6". These results demonstrate that contrary to expectations, a core 2 O-glycan containing sialylated PFPL does not promote high affinity binding of PSGL-1 to P-selectin.