PARTICULATE AND SOLUBILIZED FUCOSYL TRANSFERASES FROM MOUSE BRAIN

The transfer of [¹⁴C]fucose from GDP-[U-¹⁴C]fucose to endogenous and exogenous acceptors by particulate and solubilized preparations from mouse brain is described. Suspensions of brain microsomes incorporated [¹⁴C]fucose into a heterogenous group of glycoprotein products, which have a distribution on gel electrophoresis similar to those synthesized in vivo. Fucosyl transferase, extracted from brain microsomes by Triton X-100, transferred [¹⁴C]fucose from GDP-[U-¹⁴C]fucose to terminal galactose residues exposed by mild acid hydrolysis of porcine plasma glycoprotein. Comparison of the specific activities of the solubilized fucosyl transferase from a number of organs showed that, in the presence of the exogenous acceptor which was used, the transferase of brain was more active than the transferases from all other organs tested, with the exception of kidney. Examination of subcellular fractions of brain, with endogenous and exogenous acceptors, showed that activity was limited to fractions containing microsomal membranes, whereas synaptosomal and other fractions were virtually inactive.     

Membrane glycoprotein transferases: Time courses of activity changes in a synchronized cell population and enzyme activity half-lives in an asynchronous population

The time course of activity changes of five membrane glycoprotein transferases: the glycoprotein:galactosyl, the fetuin:fucosyl, the PSM:fucosyl, the fetuin:N-acetylglucosaminyl, and the polypeptide: N-acetylgalactosaminyl was determined using defined acceptors and conditions in synchronized L5178Y cells. In addition, time course of activity changes of these transferases was determined for the endogenous acceptors present in the cell extract. Each of the membrane glycoprotein transferases was an S peak enzyme in the L5178Y cell. Activity with endogenous acceptors, in general, was constant throughout the cell cycle indicating that acceptor availability may, in part, control glycoprotein synthesis in vivo.     

Synthesis of an O-sulfo Lewis analog as glycolipid antigen

The title compound containing dihydroceramide as a ligand for CD1d was accomplished using the mannosyl, glucosaminyl, and fucosyl donors, and a sphinganine analogue, as suitable building blocks. The 2-O-unprotected mannosyl donor was coupled effectively with the sphinganine analog to afford the mannnosyl sphinganine derivative. The coupling of the glucosaminyl donor with the mannnosyl sphinganine acceptor required triflic acid as a promoter and the promoter change to silver triflate led to the undesired glycal production. The reduction of azide group using Zn powder was the key process, in which the amount of acetic acid was restricted to avoid the benzoyl migration and N-trichloroacetyl deprotection. The trisaccharide glycolipid was sulfonated at the 3-position of fucose moiety.     

Involvement of dolichol phosphates as intermediates in the mannosyl and galactosyl transferases of rat testicular germ cell Golgi apparatus membranes

Isolated Golgi apparatus membranes from the germinal elements (spermatocytes and early spermatids) of rat testis were examined for their ability to incorporate [14C]mannose and [14C]galactose into glycolipid and glycoprotein fractions. Transfer of mannose from GDP-[14C]mannose into a Lipid I fractions (GPD:MPP mannosyl transferase activity), identified as mannosyl phosphoryl dolichol, showed optimal activity at 1.5 mM manganese and at pH 7.5. Low concentrations of Triton X-100 (0.1%) stimulated transferase activity in the presence of exogenous dolichol phosphate (Dol-P); however, inhibition occurred at Triton X-100 concentrations greater than 0.1%. Maximal activity of this GDP:MPP mannosyl transferase occurred at 25 microM Dol-P. Activity using endogenous acceptor was 2.34 pmole/min/mg, whereas in the presence of 25 microM Dol-P the specific activity was 284 pmole/min/mg, a stimulation of 125-fold. Incorporation of mannose into a Lipid II (oligosaccharide pyrophosphoryl dolichol) and a glycoprotein fraction was also examined. In the absence of exogenous Dol-P, rapid incorporation into Lipid I occurred with a subsequent rise in Lipid II and glycoprotein fractions suggesting precursor-product relationships. Addition of exogenous Dol-P to galactosyl transferase assays showed only a minor stimulation, less than twofold, in all fractions. Over the concentration range of 9.4 to 62.5 micrograms/ml Dol-P, only 1% of radioactive product accumulated in the combined lipid fractions. These observations suggest that the mannose transfer involves Dol-P intermediates and also that spermatocyte Golgi membranes may be involved in formation of the oligosaccharide core as well as in terminal glycosylations.     

Biosynthesis of polyuronides inMucor rouxii: Partial characterization of fucosyl transferase

Most of the fucosyl transferase activity fromMucor rouxii was detected in a crude membrane fraction. The enzyme transferredl-fucose from GDP-fucose to endogenous and exogenous acceptors. When crude membrane fractions were treated with neutral detergents such as Trition X-100 or Brij 36 T enzyme activity became dependent on exogenous acceptors such as mucoric acid or mucoran. Brij-treated membrane fractions showed maximum fucosyl transferase activity at pH 6.5, and at a temperature between 22 and 28°C. The cations Mn²⁺, Mg²⁺, Co²⁺, Zn²⁺, Fe²⁺, and Ca²⁺ activated the enzyme about twofold. The former was slightly more stimulatory at 4 mM. Km for GDP-fucose was 10 μM. Evidence was obtained that mucoric acid serves as acceptor for fucosyl moieties. Acid hydrolysis of the product synthesized from GDP-fuc by Brij-treated membrane fractions revealed fucose as the major radioactive sugar.     

Glycosyltransferases with endogenous acceptor activity in plasma membranes isolated from rat liver

Plasma membrane fractions from rat liver exhibited glycosyltransferase activity with endogenous membrane-associated acceptors and either UDP-galactose, UDPglucose, UDP-N-acetylglucosamine, or GDPmannose donors. Of these, incorporation into non-lipid acceptors was most active with UDP-galactose and only with UDPgalactose and UDPmannose was there incorporation into endogenous lipid acceptors. CMP-N-acetylneuraminic acid was inactive as a donor with the isolated plasma membranes. In order to demonstrate transferase activity, low concentrations of substrate sugar nucleotides and short incubation times were used as well as sulfhydryl protectants and a phosphatase inhibitor (NaF) in the reaction mixtures. The findings support the concept of surface localization of at least a galactosyl transferase in cells of rat liver.     

SYNTHESIS OF GLYCOPROTEINS IN BRAIN: IDENTIFICATION, PURIFICATION AND PROPERTIES OF GLYCOSYL TRANSFERASES FROM PURIFIED SYNAPTOSOMES OF GUINEA PIG CEREBRAL CORTEX

Abstract— Four glycoprotein:glycosyl transferases (a fetuin:N-acetylglucosaminyl transferase; a bovine submaxillary mucin: N-acetylgalactosaminyl transferase; a collagen: glucosyl transferase and an orosomucoid: galactosyl transferase) were purified 34-, 45-, 37- and 47-fold, respectively, from synaptosomes prepared from guinea pig cerebral cortex. Purifications were achieved by centrifugation and by column chromatography on Sephadex G-100 and G-150 of 0 , 1% (w/v) Triton X-100 extractsof the purified cerebral cortical synaptosomes. The enzymes were separated from endogenous acceptors and were highly specific for specific macromolecular acceptors; small molecules were ineffective as acceptors. The fetuin: N-acetylglucosaminyl transferase functioned only with fetuin minus N-acetylneuraminic acid, galactose and N-acetylglucosamine; the bovine submaxillary mucin: N- acetylgalactosaminyl transferase with bovine submaxillary much minus N-acetylneuraminic acid and N-acetylgalactosamine; the collagen: glucosyl transferase with collagen minus glucose; and the orosomucoid: galactosyl transferase with either orosomucoid minus N-acetylneuraminic acid and galactose or fetuin minus N-acetylneuraminic acid and galactose. Each transferase required a specific (XDP)-monosaccharide for transfer. The transferases were entirely dependent on either Mn²⁺ or Mg²⁺ for activation and Fe²⁺ and Hg²⁺ inhibited each of the four enzymes. The optimum pH's for the enzymes were: for fetuin: N-acetylglucosaminyl transferase, 7 , 4–8.0; for bovine submaxillary mucin: N-acetylgalactosaminyl transferase, 7 , 7; for collagen: glucosyl transferase, 7 , 7 and for orosomucoid: galactosyl transferase, 6 , 6. The enzymes were distributed subsynaptosomally primarily in the synaptosomal plasma membrane and in the mitochondria of the synaptosome. The respective values for K_m (μM) and V_mex (pmoles/h/mg of protein) for the transferases were: fetuin: N-acetylglucosaminyl transferase, 12 and 143; for bovine submaxillary mucin: N-acetylgalactosaminyl transferase, 25 and 166; for collagen: glucosyl transferase, 4 and 10 and for orosomucoid:galactosyl transferase, 8 and 111.     

Glycosyl transferases of baby-hamster-kidney (BHK) cells and ricin-resistant mutants. N-glycan biosynthesis

Five cell lines of ricin-resistant BHK cells have been assayed for gross carbohydrate analysis of cellular glycoproteins, for the activities of several glycosidases and of specific glycosyl transferases active in assembly of N-glycans of glycoproteins. The latter enzymes include sialyl transferase using asialofetuin as glycosyl acceptor, fucosyl transferases using asialofetuin and asialoagalactofetuin acceptors, galactosyl transferases using ovalbumin, ovomucoid and N-acetylglucosamine as acceptors and N-acetylglucosaminyl transferases using ovalbumin and glycopeptides as acceptors. Cell line RicR14, binding less ricin than normal BHK cells, contains reduced amounts of sialic acid, galactose and N-acetylglucosamine in cellular glycoproteins and lacks almost completely N-acetylglucosamine transferase I, an essential enzyme in assembly of ricin-binding carbohydrate sequences of N-glycans. These cells also contain reduced levels of N-acetylglucosamine transferase II active on a product of N-acetylglucosamine transferase I action. Sialyl transferase activity is severely depressed while fucose-(alpha 1 leads to 6)-N-acetylglucosamine fucosyl transferase activity is increased. Cell lines RicR15, 17, 19 and 21 showed partial deficiencies in galactosyl and N-acetylglucosaminyl transferases. A hypothesis is put forward to account for the different carbohydrate compositions and ricin binding properties of glycoproteins synthesised by these cells in terms of the determined enzyme defects, the normal level of sialyl transferases detected in RicR15 and RicR21 cells and the elevated levels of sialyl and fucosyl transferases detected in RicR17 and 19 cells. None of the above changes in glycosyl transfer reactions in the RicR cell lines are due to enhanced glycosidase or sugar nucleotidase activities in the mutant cells.     

Biosynthesis of yeast mannan. Characterization of mannan-synthesizing enzyme systems from mutants defective in mannan structure

The yeastSaccharomyces cerevisiae X2180-1A (wild) and its mutants X2180-1A-4 (mnn 1) and X2180-1A-5 (mnn 2) defective in mannan biosynthesis were used as enzyme sources to catalyzein vitro mannosyl transfer from GDP-[¹⁴C-U]-mannose to endogenous glycoproteins as well as to exogenous, low-molecular weight acceptors. While the enzyme preparation from the wild strain exhibited all mannosyl transferase activities involved in mannan biosynthesis by catalyzing the synthesis of characteristic mannoprotein, the enzyme frommnn 1 mutant failed to catalyze the synthesis of α(1→3) mannoside linkages both with endogenous as well as with exogenous acceptors. The enzyme preparation from themnn 2 mutant catalyzed the formation of mannoprotein very similar to that obtained with the enzyme from the wild strain. The most important difference was the formation of a higher number of unsubstituted mannosyl units in the α(1→ 6) linked mannan backbone. The observed results support the hypothesis that in themnn 1 the mutation has altered the structural gene involved in biosynthesis of an α(1→3) mannosyl transferase catalyzing the addition of α(1→3) linked mannosyl units to α(1→2) linked mannotrioses in the polysaccharide side chains and in the oligosaccharides attached to serine and/or threonine in the protein part of mannan molecule. Themnn 2 mutant represents most probably a kind of regulatory mutation where the activity of an α(1→2) mannosyl transferase adding the mannosyl units directly to α(1→6) linked backbone in the outer region of polysacoharide part of yeast mannan is repressedin vivo but becomes significantin vitro.     

Ectogalactosyltransferase. Presence of enzyme and acceptors on the rat lymphocyte cell surface.

Experiments are described to demonstrate the existence of ectogalactosyltransferase activity on the lymphocyte surface. The procedures described enable us to exclude the possibility of misleading results due to precursor hydrolysis and intracellular utilization of the free galactose. This depicted transferase is able to catalyse the transfer of a galactosyl residue from UDP-galactose to a nonphagocytosable exogenous acceptor and to endogenous membrane acceptors. The cells galactosylated in this way acquired new agglutinating properties with soybean agglutinin, which proves the external position of the galactosyl residues incorporated on the cell surface.     

Interaction of the components of the lactose synthetase system

The modifier action of α-lactalbumin upon galactosyl transferase is pH-dependent. When either N-acetyl glucosamine or glucose is the acceptor, the pH profile of activity is altered in the presence of α-lactalbumin. The effect of α-lactalbumin upon the kinetic constants at a series of pH's has been interpreted in terms of a molecular mechanism of modifier activity.Fluorescence polarization studies indicated that a definite molecular complex between α-lactalbumin and galactosyl transferase is formed in the presence of substrate. Estimates of the equilibrium constants have been made.     

Galactose transfer to endogenous acceptors within Golgi fractions of rat liver

The distribution of galactosyl transferase was studied using trans and cis Golgi fractions isolated by a modification of the Ehrenreich et al. procedure (1973. J. Cell Biol. 59:45-72) as well as an intact Golgi fraction isolated by a new one-step procedure. Two methods of assay were used. The first method analyzed the ability of Golgi fractions to transfer galactose (from uridine diphosphogalactose [UDP-gal] substrate) to the defined exogenous acceptor ovomucoid. The second method assessed the transfer of galactose from UDP-gal substrate to endogenous acceptors (endogenous glycosylation). The trans Golgi fraction (Golgi light) was highly active by the first method but revealed only low activity by the second method. Golgi fractions enriched in central and cis elements (the Golgi intermediate, heavy and especially the intact Golgi fraction) were highly active in both methods of assay. The endogenous glycosylation approach was validated by gel fluorography of the endogenous acceptors. For all Golgi fractions, transfer of galactose was revealed to secretory glycopeptides. It is concluded that galactosyl transferase activity in vivo occurs primarily in central and cis Golgi elements but not trans Golgi vesicles.     

Galactosyl transferase of a Golgi fraction from cultured neoplastic mast cells

Suspension cultures of neoplastic mouse mast cells were used to obtain large quantities of a homogeneous cell population as starting material for cell fractionation. A Golgi fraction was prepared by slight modification of established techniques and identified by electron microscopy. Assay of galactosyl transferase activity using ovalbumin, desialylated degalactosylated orosomucoid, and N-acetylglucosamine as galactose acceptors showed that the Golgi fraction was enriched in specific activity over the homogenate. The Golgi galactosyl transferase was examined in detail. Acceptor concentrations for optimal galactose incorporation were determined, and substrate inhibition effects were shown with higher concentrations of all three acceptors. Manganese was shown to be necessary for galactose incorporation. A higher concentration of manganese afforded some protection from substrate inhibition by acceptors, but at the same time was itself inhibitory. All three acceptors competed with one another for galactose incorporation, indicating that a single enzyme catalyzed the transfer of galactose for all acceptors.     

Effectivity of dolichyl phosphates with different chain lengths as acceptors of nucleotide activated sugars

Chemical synthesis of different S-forms of dolichyl-P was performed in order to investigate the use of these polyprenes in mannosyl, glucosyl and glucosaminyl transferase reactions. Determination of the Vmax values for a series of dolichyl-P demonstrated that the velocities of transferase reactions with all those dolichyl-P derivatives present in animal tissues are largely the same. The apparentK _m values for the various dolichyl-P in the transferase system studied differed, but this property does not appear to have physiological importance.     

Glucosyl transferase activity of bovine galactosyl transferase

Bovine galactosyl transferase was found to utilize UDPglucose as a substrate and elicit disaccharide biosynthesis with glucose and N-acetylglucosamine as acceptors. The relative rate of glycosyl transferase with N-acetylglucosamine as acceptor was 0.3%, the rate for N-acetyllactosamine biosynthesis. This activity was also evidenced indirectly from NMR water proton relaxation experiments, and from Mn(II) ESR experiments. In direct experiments with radioactive UDPglucose, paper chromatography showed a product which migrated with cellobiose when glucose was the acceptor and a new, glucose-containing product which resulted when GlcNAc was the acceptor.Despite this marginally expanded specificity of the donor site, spin-label experiments with a covalently bound UDPgalactose analog reaffirmed the restrictive nature of the donor site against this non-glycosyl-like analog.     

Factors affecting glucosyl and mannosyl transfer to dolichyl monophosphate by liver cell-free preparations

GDP-mannose and UDP-mannose (each at less than 1 micrometer) markedly inhibit glucosyl transfer from UDP-glucose (1.6 micrometer( to dolichyl phosphate in liver microsomal preparations. The biphasic response suggests the presence of two glucosyl transferases only one of which is inhibited. The inhibition appears to be a property of the intact nucleotide phosphate sugars and not due to competition for a limited pool of dolichyl phosphate. UDP-galactose and UDP-xylose cause a less marked inhibition of the same enzyme. The failure of UDP-glucose to inhibit mannosyl transfer suggests that the pool of dolichol monophosphate used by mannosyl transferase is not available to the glucosyl transferase. The relationship between the degree to which an exogenous prenol phosphate acts as an acceptor of mannose and the degree to which it inhibits mannosylation of endogenous dolichyl monophosphate varies among different prenyl phosphates. Mannosyl transferase exhibits two pH optima.     

Partial purification and characterization of a mannosyl transferase involved in O-linked mannosylation of glycoproteins in Candida albicans

Incubation of a mixed membrane fraction of C. albicans with the nonionic detergents Nonidet P-40 or Lubrol solubilized a fraction that catalyzed the transfer of mannose either from endogenously generated or exogenously added dolichol-P-[14C]Man onto endogenous protein acceptors. The protein mannosyl transferase solubilized with Nonidet P-40 was partially purified by a single step of preparative nondenaturing electrophoresis and some of its properties were investigated. Although transfer activity occurred in the absence of exogenous mannose acceptors and thus depended on acceptor proteins isolated along with the enzyme, addition of the protein fraction obtained after chemical de-mannosylation of glycoproteins synthesized in vitro stimulated mannoprotein labeling in a concentration-dependent manner. Other de-mannosylated glycoproteins, such as yeast invertase or glycoproteins extracted from C. albicans, failed to increase the amount of labeled mannoproteins. Mannosyl transfer activity was not influenced by common metal ions such as Mg(2+), Mn(2+) and Ca(2+), but it was stimulated up to 3-fold by EDTA. Common phosphoglycerides such as phosphatidylglycerol and, to a lower extent, phosphatidylinositol and phosphatidylcholine enhanced transfer activity. Interestingly, coupled transfer activity between dolichol phosphate mannose synthase, i.e., the enzyme responsible for Dol-P-Man synthesis, and protein mannosyl transferase could be reconstituted in vitro from the partially purified transferases, indicating that this process can occur in the absence of cell membranes.     

Membrane fusion and glycosylation in the rat hepatic Golgi apparatus

When purified Golgi fractions were incubated with UDP-[3H]galactose in the absence of Triton-X-100, radioactivity was incorporated into an endogenous lipid and several peptide acceptors. Electron microscope analysis of Golgi fractions incubated in the endogenous galactosyl transferase assay medium revealed extensive fusion of Golgi saccules. Systematic removal of constituents in the galactosyl transferase assay medium showed enhanced (minus beta-mercaptoethanol) or reduced (minus ATP, minus sodium cacodylate buffer or minus MnCl2) fusion of Golgi membranes compared to the complete medium, Stereologic analysis revealed a correlation between membrane fusion and galactosyl transferase activity (r = 0.99, P less than 0.001). Electron microscope radioautography was carried out after incubation of Golgi fractions with UDP-[3H]galactose. Silver grains were not observed over trans elements of Golgi but were revealed mainly over large fused saccules with the number of silver grains being proportionate to membrane fusion (r = 0.92, P less than 0.001). Bilayer destabilization at points of Golgi membrane fusion may act to translocate galactose across the Golgi membrane and thereby provide a fusion regulated substrate for terminal glycosylation.     

An alpha-D-galactosyltransferase activity in Ehrlich ascites tumor cells. Biosynthesis and characterization of a trisaccharide (alpha-D-galactose-(1 goes to 3)-N-acetyllactosamine)

An alpha-D-galactosyltransferase activity has been detected in membranous fractions (42,000 x g) of Ehrlich ascites cells which transfers galactosyl groups from UDP-galactose to endogenous and exogenous acceptors. The products of the reaction contain alpha-D-galactopyranosyl groups at the nonreducing termini. A solid state assay was developed to follow alpha-D-galactosyltransferase activity in the presence of beta-D-galactosyltransferase. Examination of a variety of insolubilized exogenous acceptors indicated that the most active acceptors for the alpha-D-galactosyltransferase had the structure beta-D-Gal-(1 goes to 4)-beta-D-GlcNAc(1 goes to at their nonreducing termini. Incubation of UDP-[14C]galactose and beta-D-gal-(1 goes to 4)-D-GlcNAc (N-acetyllactosamine) or of UDP-galactose and beta-D-[14C]Gal-(1 goes to 4)-D-GlcNAc in the presence of the alpha-D-galactosyltransferase resulted in the enzymic synthesis of a 14C-labeled trisaccharide. Chemical and enzymic methods of analysis revealed the structure of the trisaccharide to be alpha-D-Gal-(1 goes to 3)-beta-D-Gal-(1 goes to 4)-D-GlcNAc. These data indicate that the alpha-D-galactosyltransferase in Ehrlich ascites cells transfers galactosyl groups to suitable acceptors to form an alpha-(1 goes to 3)-D-galactosidic linkage.     

A dolichol-linked trisaccharide from central nervous tissue: structure and biosynthesis.

Calf brain membranes have previously been shown to enzymatically transfer N-acetyl[¹⁴C]glucosamine from UDP-N-acetyl[¹⁴C]glucosamine into N-acetyl[¹⁴C]glucosami-nylpyrophosphoryldolichol, N,N′-diacetyl[¹⁴C]chitobiosylpyrophosphoryldolichol and a minor labeled product with the chemical and chromatographic properties of a [¹⁴C]trisaccharide lipid (Waechter, C. J., and Harford, J. B. (1977) Arch. Biochem. Biophys.181, 185–198). This paper demonstrates that incubating calf brain membranes containing endogenous, prelabeled N-acetyl[¹⁴C]glucosaminyl lipids with unlabeled GDP-mannose enhances the formation of the [¹⁴C]trisaccharide lipid. The intact [¹⁴C]trisaccharide lipid behaves like a dolichol-bound trisaccharide, in which the glycosyl group is linked via a pyrophosphate bridge, when chromatographed on SG-81 paper or DEAE-cellulose. Mild acid treatment releases a water-soluble product that comigrates with authentic β-Man-(1→4)-β-GlcNAc(1→4)-GlcNAc. The free [¹⁴C]trisaccharide is converted to N,N′-diacetyl[¹⁴C]chitobiose by incubation with a highly purified β-mannosidase. These findings indicate that the trisaccharide lipid formed by calf brain membranes is β-mannosyl-N,N′-diacetylchito-biosylpyrophosphoryldolichol. The two glycosyltransferases responsible for the enzymatic conversion of the N-acetylglucosaminyl lipid to the trisaccharide lipid have been studied using exogenous, purified [¹⁴C]glycolipid substrates. Calf brain membranes enzymatically transfer N-acetylglucosamine from UDP-N-acetylglucosamine to exogenous N-acetyl[¹⁴C] glucosaminylpyrophosphoryldolichol to form [¹⁴C]disaccharide lipid. The biosynthesis of [¹⁴C]disaccharide lipid is stimulated by unlabeled UDP-N-acetylglucosamine under conditions that inhibit N-acetylglucosaminylpyrophosphoryldolichol synthesis. Unlike the formation of N-acetylglucosaminylpyrophosphoryldolichol the enzymatic addition of the second N-acetylglucosamine residue is not inhibited by tunicamycin. Exogenous purified [¹⁴C] disaccharide lipid is enzymatically mannosylated by calf brain membranes to form the [¹⁴C] trisaccharide lipid. The formation of the [¹⁴C]trisaccharide lipid from exogenous [¹⁴C] disaccharide lipid is stimulated by unlabeled GDP-mannose and Mg²⁺, and inhibited by EDTA. Exogenous dolichyl monophosphate is also inhibitory. These results strongly suggest that the calf brain mannosyltransferase involved in the synthesis of the trisaccharide lipid requires a divalent cation and utilizes GDP-mannose, not mannosylphosphoryldolichol, as the direct mannosyl donor.