Role of N-linked polymannose oligosaccharides in targeting glycoproteins for endoplasmic reticulum-associated degradation.

Misfolded or incompletely assembled multisubunit glycoproteins undergo endoplasmic reticulum-associated degradation (ERAD) regulated in large measure by their N-linked polymannose oligosaccharides. In this quality control system lectin interaction with Glc(3)Man(9)GlcNAc(2) glycans after trimming with endoplasmic reticulum (ER) alpha-glucosidases and alpha-mannosidases sorts out persistently unfolded glycoproteins for N-deglycosylation and proteolytic degradation. Monoglucosylated (Glc(1)Man(9)GlcNAc(2)) glycoproteins take part in the calnexin/calreticulin glucosylation-deglucosylation cycle, while the Man(8)GlcNAc(2) isomer B product of ER mannosidase I interacts with EDEM. Proteasomal degradation requires retrotranslocation into the cytosol through a Sec61 channel and deglycosylation by peptide: N-glycosidase (PNGase); in alternate models both PNGase and proteasomes may be either free in the cytosol or ER membrane-imbedded/attached. Numerous proteins appear to undergo nonproteasomal degradation in which deglycosylation and proteolysis take place in the ER lumen. The released free oligosaccharides (OS) are transported to the cytosol as OS-GlcNAc(2) along with similar components produced by the hydrolytic action of the oligosaccharyltransferase, where they together with OS from the proteasomal pathway are trimmed to Man(5)GlcNAc(1) by the action of cytosolic endo-beta- N-acetylglucosaminidase and alpha-mannosidase before entering the lysosomes. Some misfolded glycoproteins can recycle between the ER, intermediate and Golgi compartments, where they are further processed before ERAD. Moreover, properly folded glycoproteins with mannose-trimmed glycans can be deglucosylated in the Golgi by endomannosidase, thereby releasing calreticulin and permitting formation of complex OS. A number of regulatory controls have been described, including the glucosidase-glucosyltransferase shuttle, which controls the level of Glc(3)Man(9)GlcNAc(2)-P-P-Dol, and the unfolded protein response, which enhances synthesis of components of the quality control system.

Release of polymannose oligosaccharides from vesicular stomatitis virus G protein during endoplasmic reticulum-associated degradation.

To further explore the localization of the N-deglycosylation involved in the endoplasmic reticulum (ER)-associated quality control system we studied HepG2 cells infected with vesicular stomatitis virus (VSV) and its ts045 mutant, as in this system oligosaccharide release can be attributed solely to the VSV glycoprotein (G protein). We utilized the restricted intracellular migration of the mutant protein as well as dithiothreitol (DTT), low temperature, and a castanospermine (CST)-imposed glucosidase blockade to determine in which intracellular compartment deglycosylation takes place. Degradation of the VSV ts045 G protein was considerably greater at the nonpermissive than at the permissive temperature; this was reflected by a substantial increase in polymannose oligosaccharide release. Under both conditions these oligosaccharides were predominantly in the characteristic cytosolic form, which terminates in a single N-acetylglucosamine (OS-GlcNAc(1)); this was also the case in the presence of DTT, which retains the G protein completely in the ER. However when cells infected with the VSV mutant were examined at 15 degrees C or exposed to CST, both of which represent conditions that impair ER-to-cytosol transport, the released oligosaccharides were almost exclusively (> 95%) in the vesicular OS-GlcNAc(2) form; glucosidase blockade had a similar effect on the wild-type virus. Addition of puromycin to glucosidase-inhibited cells resulted in a pronounced reduction (> 90%) in oligosaccharide release, which reflected a comparable impairment in glycoprotein biosynthesis and indicated that the OS-GlcNAc(2) components originated from protein degradation rather than hydrolysis of oligosaccharide lipids. Our findings are consistent with N-deglycosylation of the VSV G protein in the ER and the subsequent transport of the released oligosaccharides to the cytosol where OS-GlcNAc(2) to OS-GlcNAc(1) conversion by an endo-beta-N-acetylglucosaminidase takes place. Studies with the ts045 G protein at the nonpermissive temperature permitted us to determine that it can be processed by Golgi endomannosidase although remaining endo H sensitive, supporting the concept that it recycles between the ER and cis-Golgi compartments.

Golgi apparatus immunolocalization of endomannosidase suggests post-endoplasmic reticulum glucose trimming: implications for quality control.

Trimming of N-linked oligosaccharides by endoplasmic reticulum (ER) glucosidase II is implicated in quality control of protein folding. An alternate glucosidase II-independent deglucosylation pathway exists, in which endo-alpha-mannosidase cleaves internally the glucose-substituted mannose residue of oligosaccharides. By immunogold labeling, we detected most endomannosidase in cis/medial Golgi cisternae (83.8% of immunogold labeling) and less in the intermediate compartment (15.1%), but none in the trans-Golgi apparatus and ER, including its transitional elements. This dual localization became more pronounced under 15 degrees C conditions indicative of two endomannosidase locations. Under experimental conditions when the intermediate compartment marker p58 was retained in peripheral sites, endomannosidase was redistributed to the Golgi apparatus. Double immunogold labeling established a mutually exclusive distribution of endomannosidase and glucosidase II, whereas calreticulin was observed in endomannosidase-reactive sites (17.3% in intermediate compartment, 5.7% in Golgi apparatus) in addition to the ER (77%). Our results demonstrate that glucose trimming of N-linked oligosaccharides is not limited to the ER and that protein deglucosylation by endomannosidase in the Golgi apparatus and intermediate compartment additionally ensures that processing to mature oligosaccharides can continue. Thus, endomannosidase localization suggests that a quality control of N-glycosylation exists in the Golgi apparatus.

Sulfation of the N-linked oligosaccharides of influenza virus hemagglutinin: temporal relationships and localization of sulfotransferases.

The occurrence of sulfate substituents on several positions of glycoprotein N-linked oligosaccharides prompted us to determine the subcellular localization and temporal relationships of the addition of these anionic groups employing as a model system the hemagglutinin (HA) produced by influenza virus-infected Madin-Darby canine kidney (MDCK) cells. It became apparent from a study of the HA glycoprotein in subcellular fractions resolved by Nycodenz gradient centrifugation following pulse-chase radiolabeling that sulfation of the complex N-linked oligosaccharides occurs only after they have been processed to an endo-beta-N-acetylglucosaminidase-resistant state and have reached the medial/trans Golgi and the trans Golgi network (TGN), with the former carrying out most of the sulfation activity. Hydrazine/nitrous acid/NaBH(4) treatment of the HA from the subcellular fractions indicated that C-3 of the galactose as well as C-6 of the N-acetylglucosamine residues of the N-acetyllactosamine chains became sulfated in these post ER fractions, as did the C-6 of the outer N-acetylglucosamine of the di-N-acetylchitobiose core. Consistent with the specificities of the stepwise assembly of the N-acetyllactosamine branches, we observed that the 3'-phosphoadenosine 5'-phosphosulfate (PAPS):GlcNAc-6-O-sulfotransferase migrated in the gradient to a medial/trans Golgi position while in contrast the PAPS:Gal-3-O-sulfotransferase was found in both Golgi and TGN locations. In accordance with the concept that beta-galactosylation must precede the sulfation catalyzed by the latter enzyme, we observed the presence of UDP-Gal:GlcNAc galactosyltransferase in both these sites in the MDCK cells. The presence of the Gal-3-O-sulfotransferase in the TGN is particularly important in the influenza virus-infected cells, as it makes possible the addition of terminal anionic groups after removal of the sialic acid residues by the viral neuraminidase.

A specific structural alteration in the heparan sulphate of human glomerular basement membrane in diabetes.

AIMS/HYPOTHESIS: Heparan sulphate proteoglycan is an important component of the glomerular anionic filtration barrier and its reduced amount in diabetes contributes to glomerular dysfunction. The objective of this study was to determine if there is also an alteration in the sulphation pattern of the diabetic heparan sulphate chains. METHODS: The heparan sulphate in the glomerular basement membrane/mesangial matrix from human diabetic and nondiabetic kidneys obtained at autopsy was fragmented by a hydrazine/nitrous acid procedure and after radiolabelling with NaB[3H]4, the disaccharide products were chromatographically resolved and quantified. RESULTS: Six sulphated disaccharides were identified in both the diabetic and nondiabetic samples and the molar distribution of these was similar, with the notable exception of the iduronic acid-2-O-sulphatectl--> 4glucosamine-3-O-sulphate species which occurred in the diabetic glomeruli in less than half the amount as in the nondiabetic samples (9.0% compared to 18.7% of total sulphated disaccharides, p < 0.005). CONCLUSION/INTERPRETATION: 3-O-sulphated glucosamine is a rare constituent of heparan sulphate occurring usually in a glucuronic acidbeta1--> 4glucosamine-3-O-sulphate(+/- 6-O-sulphate) sequence within the antithrombin-binding domain. In the glomerular basement membrane where the 3-O-sulphated glucosamine is present in substantial amounts, however, it occurs exclusively in an iduronic acid-containing sequence. It is likely that the recently discovered 3-O-sulphotransferase variant which specifically acts on the iduronic acidalpha1--> 4glucosamine sequence is decreased in human diabetes and moreover that this unusual disaccharide could be a component of a specific heparan sulphate domain which interacts with bioactive proteins.

Effect of proteasome inhibitors on the release into the cytosol of free polymannose oligosaccharides from glycoproteins.

Prompted by previous observations which suggested that the release of polymannose oligosaccharides shortly after the cotranslational N-glycosylation of proteins is a function of the ER-associated quality control system (Moore and Spiro (1994) J. Biol. Chem., 269, 12715-12721), we evaluated the effect which proteasome inhibitors have on the appearance of these free saccharide components. Employing as a model system castanospermine-treated BW5147 mouse T-lymphoma cells in which accelerated degradation of the T-cell receptor (TCR) alpha subunit takes place (Kearse et al. (1994) EMBO J., 13, 3678-3686), we noted that both lactacystin and N-acetyl-L-leucyl-L-leucyl-L-norleucinal, but not leupeptin, brought about a rapid and substantial reduction in the release of free polymannose oligosaccharides into the cytosol during pulse-chase studies, while the oligosaccharides in the intravesicular compartment remained unchanged, as measured by streptolysin O permeabilization. This inhibition was furthermore selective in that it affected solely the components terminating in a single N-acetylglucosamine residue (OS-GlcNAc(1)) and not the oligosaccharides terminating in a di-N-acetylchitobiose sequence (OS-GlcNAc(2)), which reside primarily in the intravesicular compartment. Despite the quantitative effect of the proteasome inhibitors on the cytosolic oligosaccharides, the molar distribution of the triglucosyl OS-GlcNAc(1) species was unaffected. The decrease in cytosolic oligosaccharides brought about by proteasome inhibition was reflected in a pronounced increase in the stability of the TCRalpha subunit. Our findings suggest that the N-deglycosylation and proteasome mediated degradation are coupled events. On the basis of our data and those of others we propose that the quality control mechanism involves proteasomes associated with the cytosolic side of the endoplasmic reticulum acting in concert with a membrane situated N-glycanase. Such a complex by removing the carbohydrate units could facilitate the retrograde ER to cytosol translocation of glycoproteins.

Use of recombinant endomannosidase for evaluation of the processing of N-linked oligosaccharides of glycoproteins and their oligosaccharide-lipid precursors.

Although glucose residues in a triglucosyl sequence are essential for the N-glycosylation of proteins and in their monoglucosyl form have been implicated in lectin-like interactions with chaperones, their removal is required for the formation of mature carbohydrate units and represents the initial steps in the glycoprotein processing sequence. In order to provide a probe for the glucosylation state of newly synthesized glycoproteins obtained from normal or altered cells, we have evaluated the usefulness of recombinant endo-alpha-mannosidase employing sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) to monitor the change in molecular mass brought about by the release of glucosylated mannose (Glc(1-3)Man). With this approach the presence of two triglucosylated-N-linked oligosaccharides in vesicular stomatis virus (VSV) G protein formed by castanospermine-treated CHO cells or the glucosidase I deficient Lec23 mutant could be clearly demonstrated and an even more pronounced change in migration was observed upon endomannosidase treatment of their more heavily N-glycosylated lysosomal membrane glycoproteins. Furthermore, the G protein of the temperature sensitive VSV ts045 mutant was found to be sensitive to endomannosidase, resulting in a change in electrophoretic mobility consistent with the presence of mono-glucosylated-N-linked oligosaccharides. The finding that endomannosidase also acts effectively on oligosaccharide lipids, as assessed by SDS-PAGE or thin layer chromatography, indicated that it would be a valuable tool in assessing the glucosylation state of these biosynthetic intermediates in normal cells as well as in mutants or altered metabolic states, even if the polymannose portion is truncated. Endomannosidase can also be used to determine the glucosylation state of the polymannose oligosaccharides released during glycoprotein quality control and when used together with endo-beta-N- acetylglucosaminidase H can distinguish between those terminating in a single N-acetylglucosamine or in a di-N-acetylchitobiose sequence.

Immunohistochemical evaluation of endomannosidase distribution in rat tissues: evidence for cell type-specific expression.

Asparagine-linked oligosaccharides of glycoproteins are subject to a series of trimming reactions by glucosidases and mannosidases in the endoplasmic reticulum which result in the removal of all three glucose residues and several of the nine mannose residues. At present, endomannosidase represents the only processing enzyme which cleaves internally and provides an alternate deglucosylation pathway. However, in contrast to the endoplasmic reticulum residential proteins glucosidase I and II, endomannosidase is primarily situated in the Golgi apparatus of rat liver hepatocytes and hepatocyte cell lines. We have performed a confocal immunohistochemical study to investigate endomannosidase in various rat tissues and used a monoclonal antibody against Golgi mannosidase II as a marker for the Golgi apparatus. Although immunofluorescence for both endomannosidase and Golgi mannosidase II was detectable in the epithelia of many tissues, renal proximal tubular cells, cortex and medulla of adrenal gland, gastric mucosa, and Leydig cells of testis were unreactive for endomannosidase. Furthermore, the endothelia in all studied tissues were unreactive for endomannosidase but positive for Golgi mannosidase II. It is concluded that by immunohistochemistry endomannosidase exhibits a cell type-specific expression in rat tissues.

Identification of a glycoprotein from rat liver mitochondrial inner membrane and demonstration of its origin in the endoplasmic reticulum.

Employing antisera against various subfractions of rat liver mitochondria (mitoplast, inner membrane, intermembrane, and matrix) as well as metabolically radiolabeled BRL-3A rat liver cells, we undertook a search for the presence of glycoproteins in this major cellular compartment for which little information in regard to glycoconjugates was available. Subsequent to [35S]methionine labeling of BRL-3A cells, a peptide:N-glycosidase-sensitive protein (45 kDa) was observed by SDS-polyacrylamide gel electrophoresis of the inner membrane immunoprecipitate, which was reduced to a molecular mass of 42 kDa by this enzyme. The 45-kDa protein was readily labeled with [2-3H]mannose, and indeed the radioactivity of the inner membrane immunoprecipitate was almost exclusively present in this component. Moreover, antisera directed against mitochondrial NADH-ubiquinone oxidoreductase (complex I) or F1F0-ATPase (complex V) also precipitated a 45-kDa protein from BRL-3A cell lysates as the predominant mannose-radiolabeled constituent. Endo-beta-N-acetylglucosaminidase completely removed the radiolabel from this glycoprotein, and the released oligosaccharides were of the partially trimmed polymannose type (Glc1Man9GlcNAc to Man8GlcNAc). Cycloheximide as well as tunicamycin resulted in total inhibition of radiolabeling of the inner membrane glycoprotein, and moreover, pulse-chase studies employing metrizamide density gradient centrifugation demonstrated that the glycoprotein was initially present in the endoplasmic reticulum (ER) and subsequently appeared in a mitochondrial location. Early movement of the glycoprotein to the mitochondria after synthesis in the ER was also evident from the limited processing undergone by its N-linked oligosaccharides; this stood in contrast to lysosomal glycoproteins in which we noted extensive conversion to complex oligosaccharides. Our findings suggest that the 45-kDa glycoprotein migrates from ER to mitochondria by the previously observed contact sites between the two organelles. Furthermore, the presence of this glycoprotein in at least two major mitochondrial multienzyme complexes would be consistent with a role in mitochondrial translocations.

Processing of viral envelope glycoprotein by the endomannosidase pathway: evaluation of host cell specificity.

Endo-alpha-D-mannosidase is an enzyme involved in N-linked oligosaccharide processing which through its capacity to cleave the internal linkage between the glucose-substituted mannose and the remainder of the polymannose carbohydrate unit can provide an alternate pathway for achieving deglucosylation and thereby make possible the continued formation of complex oligosaccharides during a glucosidase blockade. In view of the important role which has been attributed to glucose on nascent glycoproteins as a regulator of a number of biological events, we chose to further define the in vivo action of endomannosidase by focusing on the well characterized VSV envelope glycoprotein (G protein) which can be formed by the large array of cell lines susceptible to infection by this pathogen. Through an assessment of the extent to which the G protein was converted to an endo-beta-N-acetylglucosaminidase (endo H)-resistant form during a castanospermine imposed glucosidase blockade, we found that utilization of the endomannosidase-mediated deglucosylation route was clearly host cell specific, ranging from greater than 90% in HepG2 and PtK1 cells to complete absence in CHO, MDCK, and MDBK cells, with intermediate values in BHK, BW5147.3, LLC-PK1, BRL, and NRK cell lines. In some of the latter group the electrophoretic pattern after endo H treatment suggested that only one of the two N-linked oligosaccharides of the G protein was processed by endomannosidase. In the presence of the specific endomannosidase inhibitor, Glcalpha1-->3(1-deoxy)mannojirimycin, the conversion of the G protein into an endo H-resistant form was completely arrested. While the lack of G protein processing by CHO cells was consistent with the absence of in vitro measured endomannosidase activity in this cell line, the failure of MDBK and MDCK cells to convert the G protein into an endo H-resistant form was surprising since these cell lines have substantial levels of the enzyme. Similarly, we observed that influenza virus hemagglutinin was not processed in castanospermine-treated MDCK cells. Our findings suggest that studies which rely on glucosidase inhibition to explore the function of glucose in controlling such critical biological phenomena as intracellular movement or quality control should be carried out in cell lines in which the glycoprotein under study is not a substrate for endomannosidase action.

Characterization of a spleen sulphotransferase responsible for the 6-O-sulphation of the galactose residue in sialyl-N-acetyl-lactosamine sequences.

An enzyme which catalyses the transfer of sulphate from 3'-phosphoadenosine 5'-phosphosulphate (PAPS) to C-6 of galactose in the NeuAcalpha2-3Galbeta1-4GlcNAc (3'SLN) sequence has been found in rat spleen microsomes and its specificity indicates that it is well suited to participate in the assembly of 3'-sialyl-6'-sulpho-LacNAc [NeuAcalpha2-3Gal(6-SO4)beta1-4GlcNAc] and 3'-sialyl-6'-sulpho-LewisX [NeuAcalpha2-3Gal(6-SO4)beta1-4(Fucalpha1-3)GlcNAc] saccharide groups which have been implicated as selectin ligands. This sulphotransferase has a strict requirement for oligosaccharide acceptors which are capped by an alpha2-3-linked sialic acid residue, although GlcNAc in 3'SLN can be substituted by Glc, and Galbeta1-4GlcNAc can be replaced by Galbeta1-3GlcNAc without loss of activity. The finding that 3'-sialyl LewisX was inert as an acceptor suggested that fucosylation, in contrast with sialylation, follows the addition of the sulphate group. Since fetuin glycopeptides containing the NeuAcalpha2-3Galbeta1-4GlcNAc sequence had a similar affinity for the enzyme as the unattached 3'SLN, it would appear that the acceptor determinants reside primarily in the peripheral trisaccharide constellation. The position of the sulphate on C-6 of galactose was elucidated by Smith periodate oxidation, hydrazine/nitrous acid/NaBH4 treatment and elder (Sambucus nigra) bark lectin chromatography of the desialylated [35S]sulphate-labelled products of the enzyme. Assays carried out with 3'SLN as acceptor indicated that the sulphotransferase had a pH optimum between 6.5 and 7.0 and a dependence on a bivalent cation best met by Mn2+ (12-25 mM); Triton X-100 (0.02 to 0.35%) brought about maximal stimulation. Tentative Km values determined for this enzyme were 4.7 microM for PAPS, and 0.72 mM and 1.16 mM for 3'SLN and fetuin glycopeptides respectively. A survey of several rat organs indicated that the PAPS:3'SLN-6-O-sulphotransferase is selectively distributed with maximal activity occurring in spleen which was substantially greater than thymus or lymph nodes. In contrast, other enzymes (i.e. PAPS:Gal-3-O-and GlcNAc-6-O-sulphotransferases) involved in the sulphation of sialyl-lactosamine and lactosamine sequences, which in the sulphated form are believed to also be selectin ligands, were more evenly distributed in lymphoid tissues. Relatively high activities for all three enzymes were found in brain.

Sulphation of N-linked oligosaccharides of vesicular stomatitis and influenza virus envelope glycoproteins: host cell specificity, subcellular localization and identification of substituted saccharides.

The presence of sulphate groups on various saccharide residues of N-linked carbohydrate units has now been observed in a number of glycoproteins. To explore the cell specificity of this post-translational modification, we evaluated sulphate incorporation into virus envelope glycoproteins by a variety of cells, since it is believed that assembly of their N-linked oligosaccharides is to a large extent dependent on the enzymic machinery of the host. Employing the vesicular stomatitis virus (VSV) envelope glycoprotein (G protein) as a model, we noted that the addition of [35S]sulphate substituents into its complex carbohydrate units occurred in Madin-Darby canine kidney (MDCK), Madin-Darby bovine kidney, LLC-PK1 and BHK-21 cell lines but was not detectable in BRL 3A, BW5147.3, Chinese hamster ovary, HepG2, NRK-49F, IEC-18, PtK1 or 3T3 cells. The sulphate groups were exclusively located on C-3 of galactose [Gal(3-SO4)] and/or C-6 of N-acetylglucosamine [GlcNAc(6-SO4)] residues in the N-acetyllactosamine sequence of the branch chains. Moreover, we observed that the pronounced host-cell-dependence of the terminal galactose sulphation was reflected by the 3'-phosphoadenosine 5'-phosphosulphate:Gal-3-O-sulphotransferase activity assayed in vitro. Comparative studies carried out on the haemagglutinin of the influenza virus envelope formed by MDCK and LLC-PK1 cells indicated that sulphate in this glycoprotein was confined to its complex N-linked oligosaccharides where it occurred as Gal(3-SO4) and GlcNAc(6-SO4) on peripheral chains as well as on the mannose-substituted N-acetylglucosamine of the core. Since sulphation in both internal and peripheral locations of the virus glycoproteins was found to be arrested by the alpha1-->2 mannosidase inhibitor, kifunensine, as well as by the intracellular migration block imposed by brefeldin A, it was concluded that this modification is a late biosynthetic event which most likely takes place in the trans-Golgi network.

Molecular cloning and expression of rat liver endo-alpha-mannosidase, an N-linked oligosaccharide processing enzyme.

A clone containing the open reading frame of endo-alpha-D-mannosidase, an enzyme involved in early N-linked oligosaccharide processing, has been isolated from a rat liver lambdagt11 cDNA library. This was accomplished by a strategy that involved purification of the endomannosidase from rat liver Golgi by ligand affinity chromatography (Hiraizumi, S., Spohr, U., and Spiro, R. G. (1994) J. Biol. Chem. 269, 4697-4700) and preparative electrophoresis, followed by sequence determinations of tryptic peptides. Using degenerate primers based on these sequences, the polymerase chain reaction with rat liver cDNA as a template yielded a 470-base pair product suitable for library screening as well as Northern blot hybridization. EcoRI digestion of the purified lambda DNA released a 5.4-kilobase fragment that was amplified in Bluescript II SK(-) vector. Sequence analysis indicated that the deduced open reading frame of the endomannosidase extended from nucleotides 89 to 1441, encoding a protein of 451 amino acids and corresponding to a molecular mass of 52 kDa. Data base searches revealed no homology with any other known protein. When a vector coding for this protein fused to an NH2-terminal peptide containing a polyhistidine region was introduced into Escherichia coli, high levels of the enzyme were expressed upon induction with isopropyl-beta-D-thiogalactoside. Purification of the endomannosidase to electrophoretic homogeneity from E. coli lysates was accomplished by Ni2+-chelate and Glcalpha1-->3Man-O-(CH2)8CONH-Affi-Gel ligand chromatographies. Polyclonal antibodies raised against this protein reacted with Golgi endomannosidase. By both immunoblotting and silver staining, the purified E. coli-expressed enzyme was approximately 8 kDa smaller than anticipated from the open reading frame; timed induction studies indicated that this was due to scission of the enzyme's COOH-terminal end by host cell proteases. All rat tissues examined demonstrated mRNA levels (4.9-kilobase message) for the endomannosidase that correlated well with their enzyme activity.

Structure of the O-linked oligosaccharides from a major thyroid cell surface glycoprotein.

A major glycoprotein at the surface of calf thyroid cells, GP-3 (M(r) 20,000), contains the I-antigenic activity of calf thyroid which has been attributed to its poly-N-acetyllactosamine N-linked saccharide chains (Edge, A. S. B., and Spiro, R. G., J. Biol. Chem. 260, 15332-15338, 1985). The present study demonstrated that alkaline borohydride treatment of GP-3 results in the release of five neutral and five acidic saccharides that were found to represent over 30% of the saccharides of this carbohydrate-rich glycoprotein. Three of the oligosaccharides contained terminal alpha1 --> 3-linked galactose residues which accounted for their affinity toward Bandeiraea simplicifolia I lectin. The saccharides could be grouped into several distinct categories on the basis of their internal sequence. A novel tetrasaccharide in GP-3 was shown to have the structure: Gal alpha1 --> 3Gal beta1 --> 6(Gal beta1 --> 3)GalNAcH2. An unsubstituted N-acetylgalactosamine unit and a Gal beta1 --> 3GalNAc disaccharide were prominent O-linked constituents, with the disaccharide serving as a core unit for the attachment of sialic acid residues to form tetra- and trisaccharides. A branched core structure, Gal beta1 --> 3(GlcNAcbeta1 --> 6)GalNAcH2 was shared by 4 of the 10 saccharides, the most complete of which was assigned the sequence NeuAc alpha2 --> 3Gal beta1 --> 3(Gal alpha1 --> 3Gal beta1 --> 4GlcNAc beta1 --> 6)GalNAcH2.

Phylogenetic survey of endomannosidase indicates late evolutionary appearance of this N-linked oligosaccharide processing enzyme.

Endo-alpha-D-mannosidase is a processing enzyme which in contrast to other glycosidases involved in the trimming of N-linked oligosaccharides of glycoproteins acts at an internal position by cleaving the linkage between the glucose-substituted mannose and the internal portion of the polymannose unit and thereby provides an alternate deglucosylating pathway. In order to evaluate at what stage in evolution this unusual enzyme first emerged, we have carried out a phylogenetic survey of its distribution among a broad group of eukaryotes ranging from unicellular organisms to highly developed animals and plants, all of which are known to have the capacity to N-glycosylate proteins and subsequently trim the nascent glucosylated polymannose oligosaccharides. It became evident from enzyme assays and in vivo studies that endomannosidase is limited in its distribution to members of the chordate phylum, including placental and marsupial mammals, birds, reptiles, amphibians, and fish, with the single except of the Mollusca in which it was detected in three distinct classes. The enzyme's absence in all other invertebrates examined as well as in yeast, various protozoa and higher plants, stands in contrast to glucosidase II and alpha 1,2-mannosidase which were found to be present in all eukaryotes studied. The observation that endomannosidase activity was not present in insects was confirmed by radiolabeling experiments with Sf9 cells in culture. These cells, which are widely employed for the expression of mammalian genes, were in distinction to mouse cells unable to circumvent a castanospermine (CST)-induced glucosidase blockade. Moreover we observed that Tetrahymenae, which synthesize glycoproteins with truncated N-linked oligosaccharides, could not process these beyond the Glc3Man5GlcNAc2 stage in the presence of CST. The late appearance of endomannosidase during evolution suggests a need for an alternate deglucosylation route in higher animals which parallels the development of elaborate complex N-linked oligosaccharides. Such carbohydrate units are believed to carry out vital biological functions and deglucosylation is a prerequisite to the further processing steps required for their formation.

Demonstration of a peptide:N-glycosidase in the endoplasmic reticulum of rat liver.

Prompted by previous observations that polymannose oligosaccharides are released from newly synthesized glycoproteins [Anumula and Spiro (1983) J. Biol. Chem. 258, 15274-15282], we examined rat liver endoplasmic reticulum (ER) for the presence of endoglycosidases that could be involved in an event presumed to be a function of the protein quality control machinery. Our investigations indicated that a peptide:N-glycanase (PNGase) is present in ER membranes that has the capacity to release from radiolabelled glycopeptides glucosylated as well as non-glucosylated polymannose oligosaccharides terminating at their reducing end in a di-N-acetylchitobiose sequence (OS-GlcNAc2). This enzyme, which was found to be luminal in orientation, was most active in the pH range 5.5-7.0 and although it had no exogenous bivalent-cation requirements it was inhibited by EDTA. Detailed studies with Man9GlcNAc2-peptides demonstrated that in addition to the free oligosaccharide (Man9GlcNAc2) an additional neutral product characterized as Man9GlcNAc2 linked to an as yet unidentified aglycone was released in a manner that suggests its role as an intermediate. Our observation that ER, in contrast with cytosol, had no endo-beta-N-acetylglucosaminidase activity would indicate that oligosaccharides terminating in a single GlcNAc residue (OS-GlcNAc1), which have been noted to appear in the extravesicular compartment shortly after N-glycosylation [Moore and Spiro (1994) J. Biol. Chem. 269, 12715-12721] are released from the protein as OS-GlcNAc2 and undergo an ER-to-cytosol translocation in that form before undergoing cleavage of their chitobiose core.

Evaluation of the early processing routes of N-linked oligosaccharides of glycoproteins through the characterization of Man8GlcNAc2 isomers: evidence that endomannosidase functions in vivo in the absence of a glucosidase blockade.

Since it has become apparent that the early processing of the N-linked oligosaccharides of glycoproteins can proceed by several routes, we undertook to determine whether the isomeric nature of Man8GlcNAc2, which is the first intermediate with the potential for structural diversity, can provide information relating to the pathways utilized in various intact cultured cells as well as in the total membrane fraction derived from these cells (BW5147.3, HepG2, HL60, F-9, and MDCK). With the use of kifunensine (KIF) to block processing by Golgi mannosidase I, it could be shown that a substantial amount of Man8GlcNAc2 components in which the terminal mannose is missing in the alpha 1,3-linked and alpha 1,6-linked chain (isomers A and C, respectively) are produced, although in the absence of the inhibitor only the B-isomer, in which the mannose of the middle chain has been excised, was apparent. Our findings in vivo and in vitro suggest that the distinctive Man8GlcNAc2 product of endomannosidase (isomer A) and of ER mannosidase II (isomer C) are not evident in the absence of KIF, since they are rapidly degraded by Golgi mannosidase I, which is located in an intracellular compartment distal to the other two enzymes and itself exclusively generates the Man8GlcNAc2 isomer B. Investigations carried out in HepG2 cells indicated that glycoproteins with N-linked oligosaccharides whose processing has been blocked by KIF at the Man8GlcNac2 isomer A and C stage can nevertheless be effectively secreted. The observation that isomer A of Man8GlcNAc2 is a specific product of endomannosidase action made it possible to demonstrate the action of this enzyme in vivo without employing a glucosidase blockade and to show that a substantial amount of the deglucosylation of N-linked oligosaccharides is carried out by this enzyme.

Characterization of a rat liver Golgi sulphotransferase responsible for the 6-O-sulphation of N-acetylglucosamine residues in beta-linkage to mannose: role in assembly of sialyl-galactosyl-N-acetylglucosamine 6-sulphate sequence of N-linked oligosaccharides.

Rat liver Golgi membranes were found to contain an enzyme that can transfer sulphate from 3'-phosphoadenosine 5'-phosphosulphate (PAPS) to C-6 of the terminal GlcNAc in beta-linkage to mannose and has properties indicating that it is involved in the synthesis of the NeuAc alpha 2-3(6)Gal beta 1-4GlcNAc(6-SO4) sequences observed in the N-linked carbohydrate units of various glycoproteins. Assays performed with [35S]PAPS (Km 0.67 microM) and GlcNAc beta 1-6Man alpha 1-O-Me (GnMaMe) acceptor (Km 0.71 mM) indicated that the sulphotransferase had a pH optimum of approx. 7.0 and is markedly stimulated by Mn2+ ions (maximum approx. 15 mM) and Triton X-100 (0.05-0.1%). Hydrazine/nitrous acid/NaBH4 treatment of the 35S-labelled product yielded radiolabelled 2,5-anhydromannitol(6-SO4). The sulphated GnMaMc product of the GlcNAc-6-O-sulphotransferase could be galactosylated by a rat liver Golgi enzyme that was shown to have the same properties as the UDP-Gal:GlcNAc beta-1,4-galactosyltransferase from bovine milk. Competition studies performed with GlcNAc and GlcNAc-6-SO4 furthermore indicated that the same liver enzyme acted on both acceptors to produce Gal beta 1-4GlcNAc and Gal beta 1-4GlcNAc(6-SO4) with Km values of 1.04 and 1.68 mM respectively. Because the sulphated N-acetyl-lactosaminc could in turn serve as an acceptor for rat liver sialyltransferase, it seems that this enzyme, together with the Golgi galactosyltransferase and the GlcNAc-6-O-sulphotransferase, could act in concert in assembling the NeuAc alpha 2-3(6)Gal beta 1-4GlcNAc(6-SO4) branches of complex N-linked oligosaccharides.

Definition of the lectin-like properties of the molecular chaperone, calreticulin, and demonstration of its copurification with endomannosidase from rat liver Golgi.

Calreticulin was identified by immunochemical and sequence analyses to be the higher molecular mass (60 kDa) component of the polypeptide doublet previously observed in a rat liver Golgi endomannosidase preparation obtained by chromatography on a Glc alpha 1 --> 3Man-containing matrix. The affinity for this saccharide ligand, which paralleled that of endomannosidase and was also observed with purified rat liver calreticulin, suggested that this chaperone has lectin-like binding properties. Studies carried out with immobilized calreticulin and a series of radiolabeled oligosaccharides derived from N-linked carbohydrate units revealed that interactions with this protein were limited to monoglucosylated polymannose components. Although optimal binding occurred with Glc1Man9GlcNAc, substantial interaction with calreticulin was retained after sequential trimming of the polymannose portion down to the Glc1Man5GlcNAc stage. The alpha 1 --> 6-mannose branch point of the oligosaccharide core, however, appeared to be essential for recognition as Glc1Man4GlcNAc did not interact with the calreticulin. The carbohydrate-peptide linkage region had no discernible influence on binding as monoglucosylated oligosaccharides in N-glycosidic linkage interacted with the chaperone to the same extent as in their unconjugated state. The immobilized calreticulin proved to be a highly effective tool for sorting out monoglucosylated polymannose oligosaccharides or glycopeptides from complex mixtures of processing intermediates. The copurification of calreticulin and endomannosidase from a Golgi fraction in comparable amounts and the strikingly similar saccharide specificities of the chaperone and the processing enzyme have suggested a tentative model for the dissociation through glucose removal of calreticulin-glycoprotein complexes in a post-endoplasmic reticulum locale; in this scheme, deglucosylation would be brought about by the action of endomannosidase rather than glucosidase II.