O-linked Sites Research Articles

Unmasking a new recognition signal for O-linked glycosylation in the chorionic gonadotropin β subunit

hCGβ subunit is distinguished among the other members of the family of the glycoprotein hormones by the presence of four serine O-linked oligosaccharide units in the last 25 amino acids. This carboxy terminal peptide (CTP) influences the intracellular behavior of the subunit and is important for maintaining the biological half-life of hCG. To examine how the O-linked oligosaccharides affect the metabolic behavior of hCG, we generated a CGβ mutant devoid of the native O-linked acceptor sites. An alternative site not used in the native subunit was glycosylated and the structure of this oligosaccharide differed from the wild-type O-linked carbohydrates. This glycosylation occurred at serine 130 in the CTP and in contrast to the wild type O-linked oligosaccharides, sialic acid is a major component of the alternatively linked carbohydrate. The data show that deleting the native acceptor sites exposes a new site for O-glycosylation and promotes a differential intracellular processing of the β subunit. These results support the hypothesis that the CTP participates in the folding of the newly synthesized subunit, which is manifested by the post-translational changes reported here.

Differential expression of human chorionic gonadotropin (hCG) glycosylation isoforms in failing and continuing pregnancies: preliminary characterization of the hyperglycosylated hCG epitope.

Human chorionic gonadotropin (hCG) glycoforms change as pregnancy progresses. We have developed an antibody (B152) which can measure a hyperglycosylated early pregnancy isoform of hCG. This putative hyperglycosylated form of hCG arises very early in pregnancies and is rapidly replaced by an isoform that predominates for the remainder of the pregnancy. The profiles of these hCG glycoforms are measured as a ratio of values of two immunometric assays. The profiles of these ratios differ between pregnancies which persist and those which will experience early failure. In this report, daily urine hCG isoform ratios from donor eggs (no exogenous hCG pretreatment), in vitro fertilization pregnancies were profiled and analyzed from the first day following embryo transfer (ET). Significant differences were found between continuing pregnancy and pregnancy loss throughout days 5-20 post-ET. When hCG isoform ratios were analyzed from the first day of detectable hCG, pregnancy loss could be predicted in the case of a single fetus both during the 5- to 10-day time segment (P=0.018) and the 10- to 15-day time segment (P=0.045). When single and multiple fetus pregnancies were analyzed together significance was approached in the 10- to 15-day time period (P=0.058). In a second population of pregnant women who conceived naturally, in whom urine samples were collected at approximately weekly intervals to either term birth or clinical spontaneous abortion, the ratio could discriminate between miscarriages and normal term pregnancies (P=0.043). In later pregnancy, the ratio of hCG isoforms declined more rapidly in miscarriages than in term pregnancy. Antibody B152 was produced using a choriocarcinoma-derived hCG (C5), which was hyperglycosylated at both N- and O-linked sites and was 100% nicked at position beta(47-48). Western blot analyses supported the assay results showing that early pregnancy urine does not contain nicked C5-like hCG. Also, the early pregnancy hCG appeared to be the same size as later pregnancy hCG as judged by SDS gel electrophoresis. A series of Western blot analyses and immunoassays conducted with the samples either non-reduced or reduced showed that B152 is directed to a linear epitope located in the COOH-terminal peptide region of the beta subunit. This indicated that only the O-glycan groups and not the N-linked glycans are part of the antibody epitope.

Cloning and Expression of the Bioluminescent Photoprotein Pholasin from the Bivalve Mollusc Pholas dactylus

Pholasin is the photoprotein responsible for luminescence in the bivalve Pholas dactylus and consists of a luciferin tightly bound to a glycosylated protein. It is a sensitive indicator of reactive oxygen species. A full-length clone encoding apopholasin was isolated from a P. dactylus light organ cDNA library. The unprocessed apoprotein contained 225 amino acids, starting with a signal peptide of 20 amino acids, 3 predicted N-linked glycosylation sites, 1 O-linked site, no histidines, and 7 cysteines. The recombinant apoprotein was expressed in cell extracts and insect cells. The size of the apoprotein expressed in cell extracts and the cytosol of insect cells was 26 kDa but that of the fully processed protein was 34 kDa, as was native pholasin. Both the processed and unprocessed recombinant apoproteins were recognized by a polyclonal antibody raised against native pholasin. Acid methanol extracts from Pholas added to recombinant apoprotein resulted in chemiluminescence triggered by sodium hypochlorite but not photoprotein formation. These results have important implications in understanding the molecular evolution of bioluminescence and will allow the development of recombinant pholasin as an intracellular indicator of reactive oxygen species.

Characterization of the glycosylation of recombinant endopolygalacturonase I from Aspergillus niger.

The carbohydrate chains of recombinant endopolygalacturonase I (EPG I) from Aspergillus niger were characterized using a combination of mass spectrometric techniques. High performance liquid chromatography (HPLC) in conjunction with electrospray ionization mass spectrometry was used to separate the components of EPG I liberated by trypsin digestion. In-source collision-induced dissociation (CID) was utilized to fragment the digestion products entering the mass spectrometer, and the generation of carbohydrate fragment ions allowed for the identification of glycopeptides. The masses of the resulting glycans were calculated and entered into a carbohydrate database to search for possible structures. The primary sequences of the carbohydrate chains were confirmed by digesting aliquots of the intact glycopeptide with endo- and exoglycosidases and then analyzing the digestion products using matrix-assisted laser desorption/ionization mass spectrometry. These experiments demonstrated that one of the two N-linked sites of EPG I was occupied by a series of high-mannose structures, the second N-linked site was not occupied, and no O-linked sites were detected.

Comparison of CD45 extracellular domain sequences from divergent vertebrate species suggests the conservation of three fibronectin type III domains.

Mammalian CD45 is a transmembrane protein tyrosine phosphatase expressed by all nucleated cells of hematopoietic origin. In lymphocytes, CD45 is required for Ag-induced signal transduction due to its ability to positively regulate Src family members. The mechanisms by which CD45 function is regulated are unknown. Indeed, the interactions of CD45 extracellular domains are largely undefined. To gain insight into potentially important regions of the extracellular domain, we sought to identify conserved features from divergent species. cDNAs encoding the putative CD45 homologue from Heterodontus francisci (horned shark) were isolated. The cDNA sequence predicts a protein of 1200 amino acids that contains a 452-amino acid extracellular domain, a 22-amino acid transmembrane region, and a 703-amino acid cytoplasmic domain. Alignment searches revealed that the Heterodontus cytoplasmic domain sequence was most identical to mammalian CD45 and a transmembrane protein tyrosine phosphatase sequence identified from chickens, ChPTP lambda. A dendrogram with other transmembrane protein tyrosine phosphatase sequences suggest that the Heterodontus and chicken sequences represents CD45 orthologues for their respective species. Analysis of vertebrate CD45 extracellular domain sequences indicates the conservation of three structural regions: a region containing potential O-linked carbohydrate sites, a cysteine-containing region, and a region containing three fibronectin type III domains. For each vertebrate species, multiple isoforms are generated by alternative splicing of three exons that encode a portion of the region containing potential O-linked glycosylation sites. These studies provide evidence for a conservation in CD45 extracellular domain structure between divergent species and provide a basis for understanding CD45 extracellular domain interactions.

Analysis of nuclear pore protein p62 glycosylation.

Glycoprotein components of the nuclear pore are essential for nuclear transport and are modified by both glycosylation and phosphorylation. The function and control of these post-translational modifications are poorly understood. Glycosylation of the major rat nuclear pore glycoprotein, p62, was examined in vitro using recombinant p62 as a substrate. Rat p62 was expressed in Escherichia coli and purified to near homogeneity. Kinetic analysis using a partially purified mammalian transferase suggests that the recombinant protein is an excellent substrate (Km = 0.30 microM) for the transfer of GlcNAc from UDP-GlcNAc (Km = 1.8 microM). Localization of the sites of O-linked GlcNAc glycosylation of rat p62 was performed by a combination of deletion analysis of in vitro translation products and by immunoprecipitation of [14C]GlcNAc-labeled proteolytic fragments. The amino terminus of rat p62 is poorly glycosylated with no O-linked GlcNAc sites between Lys22 and Lys97; the carboxyl terminus has one known glycosylation site at Ser471. The majority of the glycosylation sites in rat p62 are likely to occur on the six clustered Ser residues in the central Ser/Thr-rich region from Ser270 to Thr294. A synthetic peptide derived from this region is a good substrate for O-GlcNAc addition (Km = 30 microM) and a potent competitive inhibitor of p62 glycosylation (Ki = 15 microM). It is proposed that this Ser/Thr-rich domain functions as a linker region between the amino-terminal beta-pleated sheet and the carboxyl terminal alpha-helical domains. O-Glycosylation and phosphorylation of this linker region could provide a dynamic means of altering the conformation of p62 during nuclear pore assembly and disassembly.

O-glycosylation mimics N-glycosylation in the 16-kDa fragment of bovine pro-opiomelanocortin. The major O-glycan attached to Thr-45 carries SO4-4GalNAc beta 1-4GlcNAc beta 1-, which is the archetypal non-reducing epitope in the N-glycans of pituitary glycohormones.

The NH2-terminal domain of pro-opiomelanocortin, designated as the 16-kDa fragment, is highly conserved throughout the vertebrate family and is likely therefore to have an important functional role. Bovine 16-kDa fragment is a 77- residue glycopeptide, which has been found to be glycosylated at threonine 45 and asparagine 65. Available evidence suggests that glycoforms lacking glycans at the O-linked site are processed in the intermediate pituitary at -Arg49-Lys50- to give the residue 1-49 amino-terminal peptide and a carboxyl-terminal glycopeptide referred to as Lys1 gamma 3-melanotropin. Glycoforms carrying O-glycans remain unprocessed in the intermediate pituitary. Thus O-glycosylation is likely to play an important role in controlling the fate of the NH2-terminal portion of pro-opiomelanocortin, thereby affecting the biological events that are influenced by peptides and glycopeptides derived from this domain. In a recent study (Siciliano, R. A., Morris, H. R., McDowell, R. A., Azadi, P., Rogers, M. E., Bennett, H. P. J., and Dell, A. (1993) Glycobiology 3, 225-239), we sequenced the N-glycans attached to Asn-65 of bovine 16-kDa fragment and demonstrated that the acidic components contain, in addition to neutral antennae, a single SO4-4GalNAc beta 1-4GlcNAc beta 1- antenna, which is characteristic of the pituitary glycohormone N-glycans (Baenziger, J. U., and Green, E. D. (1988) Biochim. Biophys. Acta 947, 287-306). We now report the structural characterization of the O-linked oligosaccharides found in bovine 16-kDa fragment. The major component, which constitutes about 80% of the O-glycan population, is a novel sulfated tetrasaccharide, which carries the same sulfated epitope as the N-glycans. This is the first time that the SO4-4GalNAc beta 1-4GlcNAc beta 1- moiety has been observed in O-glycans, and it raises the interesting possibility that the beta-N-acetylgalactosaminyltransferase responsible for the addition of N-acetylgalactosamine to the pituitary glycohormones (Smith, P. L., and Baenziger, J. U. (1988) Science, 242, 930-933) might be capable of glycosylating both N- and O-linked acceptors.

Identification of O-linked oligosaccharide chains in the activation peptides of blood coagulation factor X. The role of the carbohydrate moieties in the activation of factor X.

Conversion of factor X to factor Xa results in release of a heavily glycosylated activation peptide. Analysis of protease-digested glycopeptides derived from the activation peptides of bovine and human blood coagulation factor X allowed the identification of sites of the O-linked oligosaccharide chains in these peptides. Glycopeptides were prepared from the activation peptides by digestion with chymotrypsin or Staphylococcus aureus V8 protease. By combined analysis of amino acid sequence and sialic acid content, we found that bovine factor X had an O-linked oligosaccharide chain linked to Thr26, and human factor X had four carbohydrate-attachment sites, namely, O-glycosidic linkages to Thr17 and Thr29, respectively, and N-glycosidic linkages to Asn39 and Asn49, respectively, in their activation peptides. The O-linked carbohydrate-attachment sites were identified since the yields of phenylthiohydantoin derivatives of amino acids that corresponded to their residues were increased during amino acid sequencing after deglycosylation of the glycopeptides with sialidase and O-glycanase. The effect of deglycosylation of bovine factor X1 was investigated with factor-X-activating enzyme from Russell's viper venom or extrinsic Xase (factor VIIa/tissue factor/phospholipid) by examining the activation rates of derivatives of factor X prepared using O-glycanase, sialidase, and/or N-glycanase. The removal of O-linked carbohydrate resulted in a decrease in the rate of activation. It appears that carbohydrate residues in factor X play an important role in the activation of the zymogen.

Structure and Localization of O- and N-Linked Oligosaccharide Chains on Basement Membrane Protein Nidogen

The carbohydrate content of mouse nidogen predicts the occupation of two N- and about seven O-linked acceptor sites. The corresponding oligosaccharides were examined by sequential exoglycosidase digestions. The data indicate N-linked substitutions by several bi-, tri- and tetraantennary complex types of oligosaccharides which are further modified by additional lactosamines and terminal α-galactose and/or sialic acid. Mannose-rich oligosaccharides were of low abundance. O-linked structures included a di- and tetrasaccharide core structure that were in addition sialylated and may be similar to structures found in fetuin. Evidence is provided that the two sequence-predicted asparagine acceptors are almost fully substituted. Sequence analysis of tryptic peptides identified Thr-271, Ser-303, Thr-309, Thr-317, Thr-320, Thr-892 and Thr-905 as the most likely sites for galactosamine substitutions. These residues are located in the flexible link connecting the N-terminal globular domains G1 and G2 of nidogen and at the border between the rod and the C-terminal globe G3. Four of them showed Pro in the −1 or + 3 position. All these Ser, Thr and Pro residues but not the N-linked attachment sites are identical in human nidogen.

Glycosylation sites identified by solid-phase Edman degradation: O-linked glycosylation motifs on human glycophorin A.

The human red blood cell sialoglycoprotein, glycophorin A (GpA), contains a 'mucin-like' extensively O-glycosylated extracellular domain which carries the MN blood group antigens. We have revised the sites of O-glycosylation in the extracellular domain of GpA by automated solid-phase Edman degradation, which allowed positive identification and quantitation of O-glycosylated Ser and Thr residues, as well as the single N-glycosylation site. One N-linked and 16 O-linked sites were identified. Carbohydrate was absent on Ser1, Ser14, Ser15, Ser23, Thr28 and Thr58 in GpA. We propose that the glycosyltransferases present in erythrocytes recognize specific flanking sequences around potential O-glycosylation sites. All 16 O-glycosylation sites are explained on the basis of four motifs. Three motifs are associated with Thr-glycosylation: Xaa-Pro-Xaa-Xaa where at least one Xaa = Thr; Thr-Xaa-Xaa-Xaa where at least one Xaa = Thr; Xaa-Xaa-Thr-Xaa where at least one X = Arg or Lys. The fourth motif is associated with Ser-glycosylation: Ser-Xaa-Xaa-Xaa where at least one Xaa = Ser. These simple rules explain the glycosylation (or lack of it) on 21 of 22 Ser/Thr in the extracellular domain of GpA.

Herpesviral Fc receptors and their relationship to the human Fc receptors.

Nearly two decades ago, it was observed that cells infected with herpes simplex virus (HSV) acquired an IgG Fc binding activity. The properties of the viral Fc receptor (FcR) have now been characterized by several laboratories. The Fc binding activity appears on the surface of the infected cell prior to formation of progeny virions. The FcR induced by HSV has been identified as the HSV glycoprotein, gE. When HSV gE forms a complex with a second HSV glycoprotein, gI, the receptor binds IgG with higher affinity. Varicella-zoster virus (VZV), which is closely related to HSV, has also been shown to induce an FcR. Like the HSV FcR, the FcR specified by VZV possesses characteristics common to viral glycoproteins. VZV encodes two glycoproteins, gpI and gpIV, which are the homologs of HSV gE and gI. The VZV glycoproteins have many properties common to cell surface receptors, including O-linked glycans and phosphorylation sites. However, extensive computer-assisted analyses of the amino acid sequences of VZV gpI and gpIV did not uncover regions of homology to the human cellular Fc receptors for IgG.

Recombinant human thrombomodulin. Regulation of cofactor activity and anticoagulant function by a glycosaminoglycan side chain.

Two glycoforms of a secretable human thrombomodulin mutant [TMD1-105 and TMD1-75; Parkinson, Grinnell, Moore, Hoskins, Vlahos & Bang (1990) J. Biol. Chem. 265, 12602-12610] were expressed in human 293 cells and used to study the role of glycosylation in the functions of this endothelial-cell thrombin receptor. Carbohydrate content analysis and intrinsic labelling with [3H]glucosamine and [35S]sulphate showed that TMD1-105 contained a chondroitin sulphate whereas TMD1-75 did not. Other than chondroitin sulphate, the carbohydrate contents of the two glycoforms were identical, indicating similar glycosylation patterns at other O-linked and N-linked sites in the two glycoforms. The properties of TMD1-105 were converted into those of TMD1-75 by chondroitin ABC lyase digestion. Trypsin digestion of labelled TMD1-105 permitted isolation of two overlapping peptides that contained chondroitin sulphate, spanned the entire O-glycosylation domain and had O-glycosylation sites at Ser-492, Ser-498, Thr-500, Thr-504 and Thr-506. The chondroitin sulphate-attachment site was assigned to Ser-492 as this residue is conserved in mouse and bovine thrombomodulin and lies within a sequence Ser-Gly-Ser-492-Gly-Glu-Pro, which has strong similarity to chondroitin sulphate attachment sites in other proteoglycans. Five peptides with N-linked carbohydrate were also isolated and contained glycosylation sites in the lectin-like domain (Asn-47, Asn-115, Asn-116) and in the fourth (Asn-382) and fifth (Asn-409) epidermal growth factor domains. The role of N-linked and simple O-linked carbohydrates in the functions of human thrombomodulin remain unclear. The present studies demonstrate, however, that the presence of chondroitin sulphate in human thrombomodulin has profound effects on all of the anticoagulant properties of this important anticoagulant thrombin receptor.

Identification of the O-linked glycosylation site of the human transferrin receptor.

The human transferrin receptor is a glycoprotein containing three N-linked and one O-linked glycosylation sites. Tryptic digestion of the receptor, followed by chromatography on BioGel P-2 and reverse-phase HPLC, yields a glycopeptide (amino acids 101-120) containing the O-linked site. Amino acid sequence analysis reveals that the site of O-glycosylation is Thr-104. Mass spectral analysis is consistent with the presence of a Gal-GalNAc core with predominantly two sialic acid residues.

Structure of N-linked Oligosaccharide Chains in the Triple-Helical Domains of Human Type VI and Mouse Type IV Collagen

Asparagine-linked oligosaccharides were liberated from pepsin-treated type VI collagen and the 7S domain of type IV collagen by hydrazinolysis and their structures analysed by exoglycosidase treatment. The major component in both proteins was complex biantennary oligosaccharide being partly modified by the addition of fucose and sialic acid residues. The 7S domain contained in addition distinct amounts of truncated biantennary structures lacking one or two (β-galactose residues and a minor triantennary structure. Carbohydrate analysis indicated that all of the N-linked and 80–90% of the O-linked acceptor sites are occupied. The lack of galactosamine content in both collagens showed the 0-linked oligosaccharides were only those attached to hydroxylysine and not to serine or threonine. The high carbohydrate density along both triple helical domains is discussed with regard to their limited ability to form lateral aggregates.

Cloning of the alpha chain of human platelet glycoprotein Ib: a transmembrane protein with homology to leucine-rich alpha 2-glycoprotein.

Glycoprotein Ib is a surface membrane glycoprotein of platelets that functions as a receptor for von Willebrand factor. It is a heterodimer composed of an alpha and a beta chain linked by a disulfide bond(s). A phage lambda gt11 cDNA expression library prepared from mRNA from a human erythroleukemia cell line, HEL, was screened using an affinity-purified antibody to the glycocalicin portion of the alpha chain of glycoprotein Ib. Eleven positive clones were isolated and plaque-purified. The largest cDNA insert was 2420 nucleotides in length and coded for a leader sequence of 16 amino acids, a mature protein of 610 amino acids, and a stop codon. It also contained 42 nucleotides of 5' noncoding sequence and 497 nucleotides of 3' noncoding sequence, including a poly(A) tail. The amino acid sequence of the alpha chain of GPIb predicted from the cDNA agreed completely with the sequence of 156 amino acids that was determined by Edman degradation of peptides isolated from human platelet glycocalicin after digestion with trypsin or Staphylococcus aureus V8 protease. The extracytoplasmic domain of the alpha subunit of GPIb contains several noteworthy structural features, including a region of seven tandem repeats of 24 amino acids that are homologous with those present in leucine-rich alpha 2-glycoprotein. The extracytoplasmic domain also contains two hydrophilic regions, one rich in charged amino acids and a second rich in serine and threonine residues. The region rich in serine and threonine includes five repeats of nine amino acids as well as the majority of the O-linked carbohydrate sites present in the molecule. The extracytoplasmic domain is followed by a potential transmembrane segment of approximately 29 amino acids and a potential intracellular domain of approximately 100 amino acids located at the carboxyl end of the molecule.