Abnormal synthesis of dolichol-linked oligosaccharides in carbohydrate-deficient glycoprotein syndrome.

Carbohydrate-deficient glycoprotein syndrome (CDGS) is a rare metabolic disorder presenting in infancy with severe neurologic involvement and variable multisystemic abnormalities. Diagnosis relies upon the detection of abnormal serum glycoprotein isoforms on isoelectric focusing (IEF) gels. Carbohydrate structural analyses were performed on the N-linked oligosaccharides of serum alpha 1-antitrypsin (alpha-1AT) from two Danish children with classical type I CDGS. Following preparative gel electrophoresis of alpha-1AT isoforms, oligosaccharide charge and monosaccharide composition analyses revealed increased glycosylation heterogeneity in CDGS compared with normal alpha-1AT. CDGS alpha-1AT isoforms bore N-glycans co-migrating with monosialylated standards, while normal alpha-1AT oligosaccharides co-migrated with both mono- and disialylated standards. While the monosaccharide contents of normal alpha-1AT isoforms were relatively uniform, those of CDGS alpha-1AT isoforms varied widely, and many were relatively mannose enriched. The mannose-rich oligosaccharides of CDGS alpha-1AT were not typical oligomannose structures since they were not released by endo-beta-N-acetylglucosaminidase H (endo H) digestion. Metabolic labelling of CDGS fibroblasts with [3H]mannose showed lower than normal intracellular total mannose, free mannose and phosphorylated mannose species, as well as diminished [3H]mannose incorporation into dolichol-linked and protein-linked oligosaccharides. In addition, the glycans liberated from CDGS dolichol-linked oligosaccharides were significantly truncated compared with those from normal fibroblasts. These data suggest that our type I CDGS patients produce abnormal N-linked oligosaccharides due to impaired biosynthesis of dolichol-oligosaccharide precursors.

Malaria sporozoites and circumsporozoite protein bind sulfated glycans: carbohydrate binding properties predicted from sequence homologies with other lectins.

Circumsporozoite (CS) proteins, the major surface proteins of the sporozoites of the various malaria (Plasmodium) species, share a region of highly conserved sequence homology in common with sporozoite surface protein 2 (SSP2) and a group of proteins observed to specifically bind sulfated glycoconjugates. Recombinant P. yoelii CS protein was found to bind selectively to heparin-, fucoidan-, and dextran sulfate-Sepharose, but poorly to chondroitin sulfate A- or C-Sepharose. It also bound with lower affinity to a heparan sulfate biosynthesis-deficient mutant cell line compared with the wild-type. Likewise, P. berghei sporozoite invasion into hepatocytes was selectively inhibited by fucoidan, heparin, and dextran sulfate, and sporozoites bound specifically to sulfatide [galactosyl (3-SO4) beta 1-1 ceramide] coated surfaces. Sporozoite infectivity in mice was significantly inhibited by dextran sulfate 500,000 and fucoidan. Taken together, these data indicate that CS proteins bind selectively to certain sulfated glyconjugates and invasion of host hepatocytes by sporozoites, and sporozoite infectivity can be inhibited by such compounds.

Malaria sporozoites and circumsporozoite proteins bind specifically to sulfated glycoconjugates.

Circumsporozoite (CS) proteins, which densely coat malaria (Plasmodia) sporozoites, contain an amino acid sequence that is homologous to segments in other proteins which bind specifically to sulfated glycoconjugates. The presence of this homology suggests that sporozoites and CS proteins may also bind sulfated glycoconjugates. To test this hypothesis, recombinant P. yoelii CS protein was examined for binding to sulfated glycoconjugate-Sepharoses. CS protein bound avidly to heparin-, fucoidan-, and dextran sulfate-Sepharose, but bound comparatively poorly to chondroitin sulfate A- or C-Sepharose. CS protein also bound with significantly lower affinity to a heparan sulfate biosynthesis-deficient mutant cell line compared with the wild-type line, consistent with the possibility that the protein also binds to sulfated glycoconjugates on the surfaces of cells. This possibility is consistent with the observation that CS protein binding to hepatocytes, cells invaded by sporozoites during the primary stage of malaria infection, was inhibited by fucoidan, pentosan polysulfate, and heparin. The effects of sulfated glycoconjugates on sporozoite infectivity were also determined. P. berghei sporozoites bound specifically to sulfatide (galactosyl[3-sulfate]beta 1-1ceramide), but not to comparable levels of cholesterol-3-sulfate, or several examples of neutral glycosphingolipids, gangliosides, or phospholipids. Sporozoite invasion into hepatocytes was inhibited by fucoidan, heparin, and dextran sulfate, paralleling the observed binding of CS protein to the corresponding Sepharose derivatives. These sulfated glycoconjugates blocked invasion by inhibiting an event occurring within 3 h of combining sporozoites and hepatocytes. Sporozoite infectivity in mice was significantly inhibited by dextran sulfate 500,000 and fucoidan. Taken together, these data indicate that CS proteins bind selectively to certain sulfated glycoconjugates, that sporozoite infectivity can be inhibited by such compounds, and that invasion of host hepatocytes by sporozoites may involve interactions with these types of compounds.

Identifying glycoconjugate-binding domains. Building on the past.

The molecular details of how glycoconjugate-binding proteins interact with their ligands have been revealed by a variety of techniques. For example, proteases, chemical-modifying reagents and antibodies have served as effective probes of lectin functional domains. Protein crystallography has providing insight into how lectins are structured, and aided in determining which amino acids in these proteins are positioned appropriately for bond formation with glycoconjugates. In addition, the characterization and sequencing of naturally occurring, non-functional lectin variants have led to the identification of amino acids which play critical roles in a lectin's glycoconjugate-binding domain. Similarly, studies of lectin mutants produced by site-directed mutagenesis, and of synthetic peptides that mimic lectin binding properties, have demonstrated the importance of particular amino acids for glycoconjugate binding. An alternate approach to understanding lectin functional domains has been to compare the primary sequences of these proteins to reveal common sequence elements which allow them to be organized into families. For example, the discovery of amino acid homologies dispersed over long segments of the primary sequences of several lectins has suggested that many of these proteins have a related three-dimensional organization. In addition, the identification of more highly focused regions of sequence homology has indicated that many structures within the lectin glycoconjugate-binding domains themselves may be conserved. Scanning protein data banks for sequences homologous to known lectins has led to the identification of several previously unrecognized lectins, and aided in determining what portions of these proteins function in their glycoconjugate-binding domains.(ABSTRACT TRUNCATED AT 250 WORDS)

Glycosphingolipid-binding specificity of the mannose-binding protein from human sera.

Mannose-binding protein was purified from human serum to apparent homogeneity by affinity chromatography on mannose-Sepharose, followed by affinity chromatography on underivatized Sepharose. Approximately 0.4 mg protein was obtained from 1 liter serum. The glycosphingolipid-binding specificity of the purified protein was examined by chromatogram overlay and solid phase assays. It binds with high affinity to Lc-3Cer (GlcNAc beta 1-3Gal beta 1-4Glc beta 1-1ceramide) and n-Lc5Cer (GlcNAc beta 1-3Gal beta 1-4GlcNAc beta 1-3Gal beta 1-4Glc beta 1-1ceramide). It does not bind to many other glycosphingolipids without terminal N-acetylglucosamine residues that were tested. Thus, these data suggest that N-acetylglucosamine-terminated glycosphingolipids may serve as cell-surface attachment sites for mannose-binding protein in vivo. In addition, the binding specificity of the protein can be used as a sensitive probe for determining the levels of Lc3Cer and nLc5Cer in tissues, as it exhibits half-maximal binding to about 10 pmol of these lipids in solid phase assays, and detects less than 20 pmol of Lc3Cer in chromatogram overlay assays. This technique was utilized to demonstrate that one sample of chronic myeloid leukemia cells contains both Lc3Cer and nLc5Cer.

Enzymatic addition of O-GlcNAc to nuclear and cytoplasmic proteins. Identification of a uridine diphospho-N-acetylglucosamine:peptide beta-N-acetylglucosaminyltransferase.

An assay for the enzyme responsible for the addition of O-linked N-acetylglucosamine (O-GlcNAc) to proteins, a UDP-N-acetylglucosamine:peptide N-acetylglucosaminyltransferase, is reported using the synthetic peptide YSDSPSTST as the acceptor substrate. The activity is linearly dependent on time, enzyme, and substrate concentration. Replacement of the proline with a glycine in the peptide renders it ineffective as a substrate, whereas changing of the aspartic acid to a glycine has no effect. Product characterization of the glycosylated peptide demonstrates that the monosaccharide covalently attached to the peptide is N-acetylglucosamine (GlcNAc) and has not been epimerized to N-acetylgalactosamine. Mild base-catalyzed beta-elimination of the in vitro glycosylated peptide quantitatively yields GlcNAcitol, indicating that the GlcNAc is attached via an O-linkage. The transferase activity is strongly inhibited by UDP but is unaffected by GlcNAc or tunicamycin. Interestingly, EDTA only slightly inhibits activity, suggesting that the enzyme may not require divalent cations. The majority of the activity is soluble, and the remainder is lost from membranes after extracting with high salt and EDTA. Consistent with the subcellular localization of most proteins bearing O-GlcNAc, the activity appears to reside in the cytosolic portion of the cell when compared to two lumenal marker enzymes, galactosyltransferase and mannose-6-phosphatase.

Properdin binds to sulfatide [Gal(3-SO4)beta 1-1 Cer] and has a sequence homology with other proteins that bind sulfated glycoconjugates.

Properdin, which stabilizes the C3 convertase during the activation of the alternate complement pathway, contains amino acid sequence homologies with several proteins that bind sulfated glycoconjugates, including the adhesive protein thrombospondin and the leech salivary protein antistasin. This homology is based around the sequence Cys-Ser-Val-Thr-Cys-Gly-X-Gly-X-X-X-Arg-X-Arg. To determine if these homologous amino acid sequences are sulfated glycoconjugate-binding domains, purified native properdin, as well as activated properdin (a high molecular weight form of properdin), were examined for binding to various lipids in solid phase radioimmunoassays. Of the lipids tested, both native and activated properdin bind with high affinity only to sulfatide [Gal(3-SO4)beta 1-1 Cer], but not to comparable levels of cholesterol-3-SO4, or several neutral glycolipids, gangliosides, and phospholipids. Sulfatide binding by both forms of properdin is inhibited by dextran sulfate (Mr = 500,000) or fucoidan, whereas only the activated form is inhibited by dextran sulfate (Mr = 5,000) or heparin. Comparable levels of chondroitin sulfates A, B, and C, keratan sulfate, dextran (Mr = 90,000), or hyaluronic acid do not inhibit binding. Taken together, these data suggest that properdin, like antistasin and thrombospondin, binds sulfated glycoconjugates and supports the conclusion that the homologous sequences are sulfated glycoconjugate-binding domains.

Antistasin, an inhibitor of coagulation and metastasis, binds to sulfatide (Gal(3-SO4) beta 1-1Cer) and has a sequence homology with other proteins that bind sulfated glycoconjugates.

Antistasin, a 15-kDa salivary protein from the Mexican leech Haementeria officinalis, inhibits both blood coagulation and the metastasis of tumors (Tuszynski, G. P., Gasic, T. B., and Gasic, G. J. (1987) J. Biol. Chem. 262, 9718-9723). Antistasin binds to heparin-agarose, suggesting the protein interacts with sulfated glycoconjugates. The specificity of the interaction between antistasin and heparin was tested by measuring the binding of antistasin to various lipids and by comparing the ability of several charged glycoconjugates to inhibit binding. Of the lipids tested, antistasin binds with high affinity only to sulfatide (Gal(3-SO4)beta 1-1Cer) and does not bind to comparable levels of phospholipids, neutral glycosphingolipids, gangliosides, or cholesterol-3-SO4. The binding of antistasin to sulfatide is inhibited by dextran sulfate, fucoidan, and heparin, with I50 values of 1.5, 9.2, and 16 micrograms/ml, respectively. Comparable levels of chondroitin sulfates A, B, C, keratan sulfate, or hyaluronic acid do not inhibit binding. Comparisons of the amino acid sequences of antistasin and other sulfatide or heparin-binding proteins revealed a region of homology, based around the sequence Cys-Ser-Val-Thr-Cys-Gly-X-Gly-X-X-X-Arg-X-Arg, which may be a sulfated glycoconjugate binding domain. In addition, homologies were found with the alternate complement pathway protein properdin and coat proteins from malaria circumsporozoites and Herpes simplex I.

Nucleoplasmic and cytoplasmic glycoproteins.

We have described a new form of protein glycosylation in which N-acetylglucosamine is glycosidically linked to the hydroxyl of serine or threonine (O-GlcNAc). Unlike most other forms of protein glycosylation, O-GlcNAc is predominantly localized in the nuclear and cytoplasmic compartments of cells, where it occurs on important nuclear pore glycoproteins, well-characterized cytoskeletal proteins, as well as on many chromatin proteins, including factors that regulate gene transcription. Gas-phase protein sequencing of three O-GlcNAc-modified proteins has identified a common structural feature at sites of O-GlcNAc addition. An assay for UDP-GlcNAc:polypeptide O-GlcNAc transferase has been developed. The enzyme appears to be membrane-associated, its active site is cytoplasmic, and it has an absolute requirement for Mn2+. We are now purifying this glycosyltransferase, characterizing its substrate specificity, and determining the extent of elongation of attached saccharide moieties. The functions of O-GlcNAc remain largely unknown, but it may be important in blocking phosphorylation sites, it may be required for the assembly of specific multiprotein complexes, it might serve as a nuclear transport signal, or it may be directly involved in the active transport of macromolecules across nuclear pores.

Partial cDNA sequence encoding a nuclear pore protein modified by O-linked N-acetylglucosamine.

The nuclear pore complex contains a family of proteins ranging in molecular mass from 35 to 220 kDa that are glycosylated with O-linked N-acetylglucosamine (GlcNAc) residues. We sought to determine the primary sequence of a nuclear pore protein modified by O-linked GlcNAc. The major (62 kDa) nuclear pore glycoprotein (np62) was purified from rat liver nuclear envelopes by immunoaffinity chromatography and preparative gel electrophoresis. After CNBr fragmentation, a glycopeptide was isolated and microsequenced. An oligonucleotide probe based on this sequence information was used to screen a lambda gt11 cDNA library constructed from poly(A) mRNA of the rat thyroid cell line FRTL-5. A clone (B5) was isolated and shown to hybridize to a single 2.5-kilobase species in poly(A) mRNA from rat liver and FRTL-5. This insert was sequenced and found to contain a 691-base-pair cDNA encoding a 155-amino acid open reading frame. This open reading frame contained a CNBr fragment identical to the original glycopeptide sequence and a second CNBr fragment corresponding to a nonglycosylated peptide that was also isolated from the purified pore glycoprotein. The B5 cDNA produced a beta-galactosidase fusion protein of the size predicted by the open reading frame. Analysis of the residues making up a presumptive glycosylation site suggests that the sequence is unlike any known sites for enzymatic N- or O-linked glycosylation. The partial sequence of the 62-kDa nuclear pore glycoprotein shows little similarity to other characterized proteins and elucidates structural features of a member of the family of nuclear pore glycoproteins.

Erythrocytes contain cytoplasmic glycoproteins. O-linked GlcNAc on Band 4.1.

Previously we reported that the novel protein-saccharide linkage, O-linked N-acetylglucosamine (GlcNAc), is found in abundance on proteins associated with the cytoplasmic and nucleoplasmic faces of the nuclear pore complex. Here we demonstrate that O-GlcNAc moieties are also added to human erythrocyte cytoplasmic proteins. Intact or permeabilized erythrocytes, as well as subcellular fractions, were labeled with bovine milk galactosyltransferase and UDP-[3H] galactose. The proportion of the incorporated label found on O-GlcNAc was determined by a variety of chemical and enzymatic techniques. The bulk of the O-GlcNAc residues are found in the cytoplasm of erythrocytes, the majority of which are on an as yet unidentified 65-kDa protein. In addition, we have determined that Band 4.1, a protein which serves as a bridge joining the cytoskeleton to the inner surface of the plasma membrane in erythrocytes, also contains O-GlcNAc moieties. One of the sites of O-GlcNAc addition has been localized to the last 117 amino acids of the carboxy terminus of Band 4.1.

Nuclear pore complex glycoproteins contain cytoplasmically disposed O-linked N-acetylglucosamine.

A novel form of protein-saccharide linkage consisting of single N-acetylglucosamine (GlcNAc) residues attached in O-linkages directly to the polypeptide backbone has been described (Holt, G. D., and G. W. Hart, 1986, J. Biol. Chem., 261:8049-8057). This modification was found on proteins distributed throughout the cell, although proteins bearing O-linked GlcNAc moieties were particularly abundant in the cytosolic and nuclear envelope fractions of rat liver. In the accompanying article (Snow, C. M., A. Senior, and L. Gerace, 1987, J. Cell. Biol., 104: 1143-1156), the authors describe monoclonal antibodies directed against eight proteins localized to the nuclear pore complex. These proteins occur on the cytoplasmic and nucleoplasmic (but not lumenal) sides of nuclear membranes. In this report, we demonstrate that all members of this group of pore complex proteins bear multiple O-linked GlcNAc residues. Further, we show that the O-linked GlcNAc moieties are linked via serine (and possibly threonine) side chains to these proteins. Perturbing the O-linked GlcNAc residues either by covalently attaching galactose to them or by releasing them with beta-N-acetylglucosaminidase strongly diminishes the immunoreactivity of the proteins with all of the monoclonal antibodies. However, the O-linked GlcNAc moieties are only part of the epitopes recognized, since O-GlcNAc-containing limit pronase fragments of nuclear pore complex proteins cannot be immunoprecipitated by these antibodies. These findings, taken together with those in the accompanying article, are a direct demonstration that proteins of the cytoplasm and nucleoplasm bear O-linked GlcNAc residues.

The subcellular distribution of terminal N-acetylglucosamine moieties. Localization of a novel protein-saccharide linkage, O-linked GlcNAc.

Previously our laboratory reported the discovery of a novel protein-saccharide linkage in which single N-acetylglucosamine (GlcNAc) residues are attached in O-linkages to protein (Torres, C. R., and Hart, G. W. (1984) J. Biol. Chem. 259, 3308-3317). This linkage was first found on plasma membrane proteins of living cells by galactosylation with bovine milk galactosyltransferase. Here we report the distribution of O-linked GlcNAc in highly enriched rat liver subcellular organelles. Nonidet P-40 solubilized organelles were labeled by galactosyltransferase with UDP-[3H]galactose, and the amount of radiolabel occurring on GlcNAc residues in O-linkages was assessed by its sensitivity to beta-elimination and by its resistance to deglycosylation with endo-beta-N-acetylglucosaminidase F. The presence of galactose-labeled O-linked GlcNAc residues was confirmed by high voltage paper electrophoresis. There is a 17-fold range per mg of protein in the amount of galactosylatable terminal GlcNAc residues found in the various organelles, as well as a wide range in the organelles' apparent content of O-linked GlcNAc residues. Nuclei and the soluble fraction of rat liver cells are particularly enriched with proteins bearing O-linked GlcNAc residues, although these residues are demonstrable in virtually all organelles tested. Furthermore, examination by sodium dodecyl sulfate-polyacrylamide gel electrophoresis reveals that many different organelle-specific proteins are glycosylated with O-linked GlcNAc residues. Because of the wide occurrence of this unique linkage, these data suggest that glycosylation with O-linked GlcNAc residues is not an exclusive marker for a particular organelle. In addition, we have surveyed the organelles for their content of glycoproteins bearing GlcNAc-terminated N-linked oligosaccharides. Our data demonstrate that there are significant amounts of these oligosaccharides in rough and stripped microsomes, nuclei, and nuclear envelopes. In light of evidence that terminal GlcNAc transferases are localized to the Golgi complex, these data suggest that there are glycoproteins which enter into the Golgi for processing and then are transported back into the rough endoplasmic reticulum, and possibly the nucleus.

Murine Ia-associated invariant chain's processing to complex oligosaccharide forms and its dissociation from the I-Ak complex.

The processing of murine invariant chain (Ii) to a cell surface form bearing complex N-linked oligosaccharides has been demonstrated in the B cell lymphoma, AKTB-1b. In addition, the rate of processing of pulse-labeled Ii has been determined relative to its rate of dissociation from the alpha/beta complex of I-Ak. Ii, alpha-, and beta-chains were immunoprecipitated with anti-I-Ak or anti-Ii monoclonal antibodies. The heretofore uncharacterized complex oligosaccharide form of Ii (Ii-c) was identified in gel-purified immunoprecipitates by peptide mapping with reverse-phase HPLC. Ii-c is resistant to deglycosylation by Endo H, which is specific for high-mannose N-linkages, but can be digested with Endo F, a glycosidase capable of cleaving both complex and high-mannose N-linked oligosaccharides. Immunoprecipitation of surface iodinated cells indicates that Ii-c is expressed on the plasma membrane. Pulse-chase metabolic labeling data show that the processing of Ii to Ii-c occurs with a t1/2 of about 120 min. In contrast, the processing of both alpha- and beta-chains of I-Ak to complex forms occurs with a t1/2 of 15 to 20 min. Our data show that Ii-hm begins to dissociate rapidly from the I-Ak complex after 100 to 120 min of chase. Only a small amount (less than 5% on a per mole basis) of Ii-c was found associated with the I-Ak complexes after 300 min of continuous metabolic labeling. These results are consistent with Ii serving as a carrier for Ia antigens as they are transported to the cell surface. In addition, they suggest that the processing of Ii to Ii-c, or a late processing event of the alpha- and beta-chains, such as their sialylation, may be a possible mechanism for inducing the dissociation of Ii from the I-Ak complex.