INTRODUCTION

Most secretory and membrane proteins of eukaryotic cells are modified by glycosylation. Glycosylation endows protein with a wide variety of characteristics that are important in their cellular functions. Glycosylation occurs in the secretory pathway, and three types of modification are known: (i) attachment of N-linked saccharides to asparagine residues; (ii) attachment of O-linked saccharides to serine or threonine residues; and (iii) attachment of glycosylphosphatidylinositol anchors at the COOH termini. This is also the case with the yeast Saccharomyces cerevisiae, although mannose is the main component of N- and O-linked saccharides, and varieties of sugars are found in those of higher eukaryotes. In yeasts, the fourth type of protein-saccharide linkage was suggested for the covalent linkage between the cell wall proteins and glucans, although details are yet to be studied (1). Several excellent reviews on glycosylation in yeasts have been published (2-5).

N-Linked saccharides are first transferred to proteins as a Glc₃Man₉GlcNAc₂ unit from the Dol¹-PP-intermediate when proteins are translocated into the endoplasmic reticulum. They are then trimmed to form the core N-linked saccharide, Man₈GlcNAc₂, which is common in eukaryotes. After proteins are transported from the endoplasmic reticulum to the Golgi apparatus in S. cerevisiae, core N-linked saccharides receive various quantities of mannose, depending on the protein species, one by one from GDP-Man. O-Linked saccharides that are composed exclusively of mannose in S. cerevisiae are also transferred to proteins in the Golgi apparatus sequentially from GDP-Man except for the first mannose, which is transferred from Man-P-Dol in the endoplasmic reticulum. As mannose is the major component of both N- and O-linked saccharides in S. cerevisiae, formation of GDP-Man, the activated form of mannose, should be vitally important.

In S. cerevisiae, genetic approaches have been utilized in many scientific studies. To study the mechanism and function of glycosylation, mutants defective in glycosylation have been obtained in various ways. The mnn mutants were selected based on changes in cell surface characteristics such as binding of dyes, ion exchange matrix, or antisera (2). [³H]Mannose suicide selection was used to obtain the alg mutants, which are defective in the formation of oligosaccharide-PP-Dol (6), and the och1 mutant (7). Some of the sec mutants show the glycosylation-defective phenotype because glycosylation occurs during the process of secretion. Glycosylation mutants were also obtained as resistant or hypersensitive to chemicals or killer toxins; the vrg mutants were obtained by vanadate resistance (8), the cwh mutants were obtained by Calcofluor White hypersensitivity (9), and the kre2 mutant was obtained by K1 killer toxin resistance (10). The erd1 mutant was obtained by a retention defect of the endoplasmic reticulum resident proteins (11). Studies of these mutants and the cloned wild type genes have been helpful in elucidating the biosynthetic enzymes and intermediate structures of oligosaccharides.

To increase the collection of glycosylation-defective mutants, we screened for vanadate-resistant colonies as described previously but not screened exhaustively by Ballou et al. (8). This selection is excellent because mutants are positively selected, although the biochemical rationale of enrichment is not clear. We isolated glycosylation-defective mutants of nine complementation groups.² One of them, vig9 (vanadate-resistant and immature glycosylation), was found to be a novel mutation defective in the formation of GDP-Man. We report here the cloning and analysis of VIG9, the structural gene for GDP-Man pyrophosphorylase of S. cerevisiae.

MATERIALS AND METHODS

S. cerevisiae TM10 (MATa, leu2 his3 ura3 trp1), W303 (MATa/, ura3/ura3 leu2/leu2 trp1/trp1 his3/his3 ade2/ade2 can1/can1), M9-2D (MAT, mnn9 ade2 lys2 leu2 trp1 his3 ura3), H17-6C (MAT, vig9-1 leu2 ura3 trp1), J12 (MATa, vig9-2 lys1), and H17W (MATa/, vig9-1/vig9-1 ADE2/ade2 HIS3/his3 leu2/leu2 ura3/ura3) were used. Escherichia coli DH5 (F, supE44 lacU169 80lacZM15 hsdR17 recA1 endA1 gyrA96 thi-1 relA1) was used in plasmid propagation, and E. coli BL21/pT-groE (F, ompT hsdS gal dcm, pT-groE) was used in preparation of GST fusion proteins (12). Yeast was usually grown at 30 °C in YEPD (1% Bacto yeast extract (Difco), 2% Bacto-peptone (Difco), and 2% glucose) medium or in SD (0.67% Bacto yeast nitrogen base without amino acids (Difco), 2% glucose, and appropriate supplements) medium. Invertase-inducing medium, YEP2S, contained 2% sucrose instead of glucose in YEPD. Solid media were supplemented with 2% Bacto-agar (Difco).

Standard methods were used (13, 14) unless otherwise stated. Enzymes were purchased from either Boehringer Mannheim or Takara Shuzo Co. pRS series plasmids (15) and pYES2.0 (Invitrogen) were used as vectors.

The secretory invertase was analyzed according to the method of Gabriel and Wang (16). Cells were grown to mid-logarithmic phase in 1 ml of YEPD medium, centrifuged, washed with water, resuspended in 1 ml of YEP2S, and incubated further for 2 h. Cells were collected by centrifugation, washed once, and incubated in 100 µl of 10 mM Tris-HCl, pH 8.0, containing 0.9 M sorbitol, 0.1 M EDTA, 10 mM dithiothreitol, and 100 µg/ml Zymolyase 100T (Seikagaku Kogyo) at 37 °C for 30 min. The released periplasmic fraction was separated from spheroplasts by centrifugation.

To analyze the native enzyme, the periplasmic fraction was subjected to 7.5% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. The gel was bathed in 0.1 M sodium acetate, pH 5.1, containing 0.1 M sucrose at 37 °C for 1 h, washed with water, placed in 0.1% 2,3,5-triphenyltetrazolium chloride, 0.5 M NaOH, and boiled to detect red bands.

To analyze the denatured enzyme, the periplasmic proteins were labeled with biotin by adding D-biotinoyl--aminocapronic acid-N-hydroxysuccinimide ester (100 µg/ml, Boehringer Mannheim), and the reaction was stopped by NH₄Cl (50 mM). This solution was incubated at 4 °C for 1 h after adding anti-invertase antiserum and then 8 h after adding protein A-Sepharose (Pharmacia Biotech Inc.). Beads were washed with TBSN (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40) three times. The immunoprecipitated samples were analyzed by SDS-polyacrylamide gel electrophoresis. Proteins were blotted to a polyvinylidene difluoride membrane and detected with Streptoavidin-peroxidase. Endoglycosidase H (0.3 milliunits/µl, Seikagaku Kogyo) treatment was done in 0.1 M sodium acetate, pH 5.2, 0.5 mM 4-amidophenylmethylsulfonyl fluoride at 37 °C for 12 h.

Native chitinase was isolated from the supernatant of the saturated cultures of S. cerevisiae grown in YEPD as described by Kuranda and Robbins (17). The supernatant of the 10-ml culture was mixed with 30 mg of purified chitin (Sigma) to allow binding of chitinase. Chitin was then pelleted by centrifugation and washed three times with TBS (50 mM Tris, pH 7.5, 150 mM NaCl). The washed pellet was suspended in 100 µl of SDS sample buffer, heated to 100 °C for 10 min, and analyzed by SDS-polyacrylamide gel electrophoresis on 6% resolving gel. The gel was stained with Coomassie Brilliant Blue.

H17-6C (vig9-1) was transformed with a yeast genomic library constructed on pRS314 (ARSH4 CEN6 TRP1; a kind gift of A. Yamamoto) by the lithium acetate method (18). Trp⁺ transformants were selected and maintained on SD medium and tested for complementation of the geneticin-sensitive phenotype. Plasmids were isolated from the candidate transformants and propagated in E. coli. All of the unique plasmids were retransformed into H17-6C to verify the ability to complement the glycosylation defect.

Appropriate fragments were subcloned on plasmid pBluescript II SK+ (Stratagene), and deletions were constructed by a double-stranded nested deletion kit (Pharmacia). These clones were sequenced by using the Sequenase modification of the dideoxy chain termination method (19) with dye primers and Applied Biosystems Inc. DNA sequencer model 373A. Sequences were analyzed with computer programs (Genetyx CD, Software Development Co., Tokyo).

The disruption plasmid pSV924 was constructed by cloning the HpaI/SalI fragment containing the LEU2 gene in the HpaI/SalI site of the VIG9 gene. A diploid yeast, W303, was transformed with the vig9::LEU2 fragment excised from pSV924 by HindIII and SacI. Leu⁺ transformants were selected, and disruption of one copy of VIG9 was confirmed by Southern blotting. Spores were developed and dissected to examine essentiality of the VIG9 gene.

Yeast cells were collected, washed in cold washing solution (1.4 M sorbitol, 10 mM sodium azide), and converted to spheroplasts. The spheroplasts were pelleted and resuspended in ice-cold lysis buffer (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 0.001 volume of protease inhibitor mixture (1 mg/ml each of leupeptin, chymostatin, pepstatin, aprotinin, and antipain)) and lysed osmotically. The lysate was immediately centrifuged at 750 × g for 3 min to remove debris and further centrifuged at 100,000 × g for 1 h to produce a supernatant. The protein concentration was determined according to Bradford (20).

An appropriate amount of the above cell extract was incubated in a 50-µl reaction mixture containing 40 mM HEPES, pH 7.6, 8 mM MgCl₂, 0.1 mM mannose 1-phosphate, 0.2 mM GTP, 50 mM GDP, 50 mM GMP, and 0.1 mCi/ml [-³²P]GTP (specific activity 3,000 Ci/mmol, ICN) at 25 °C. Samples were withdrawn, mixed with equal amounts of stop solution (100 mM GTP, 100 mM EDTA), and frozen in dry ice-ethanol to stop the reaction. Amounts of [³²P]GDP-Man formed in samples were determined with an Image Analyzer (Fuji, BAS2000) after separating the GDP-Man by chromatography on a polyethyleneimine cellulose thin layer (Merck). The amounts of radioactive materials are expressed as photostimulated luminescence, calculated by the Image Analyzer.

The DNA encoding Vig9 protein was prepared by polymerase chain reaction using primer oligonucleotides 5-CGGGATCCATGAAAGGTTTAATTTTAGTCGG and 5-CGCTCGAGGGCGCAGAACAGATCATCA and inserted in pGEX-4T-3 (Pharmacia) after cleavage with BamHI and XhoI to construct pSV930, which has an inducible gene encoding GST-Vig9 fusion protein. E. coli BL21/pT-groE (12) was transformed with pSV930 or pGEX-4T-3, and the transformants were grown in 100 ml of Luria broth (1% Bacto-trypton (Difco), 0.5% Bacto yeast extract (Difco), and 0.5% NaCl) until the A_600 nm reached 0.5. Isopropyl-1-thio--D-galactopyranoside (0.05 mM) was added, and incubation was continued at 26.5 °C for 2 h and 4 °C for 1 h before harvest. Cell lysates were prepared by sonication, and GST-Vig9 protein or GST was adsorbed to 10 µl of glutathione-Sepharose 4B beads (Pharmacia) at 4 °C for 1 h. After five washes with 1 ml of TBS containing 1% Triton X-100, the adsorbed proteins were eluted with 0.5 ml of 50 mM Tris-HCl, pH 8.0, containing 10 mM glutathione.

GDP-Man pyrophosphorylase assay was done in 50 µl of reaction mix containing 0.05 µg of GST-Vig9 protein or GST at 25 °C for 10 min.

Yeast cells that were grown to A_600 nm = 1.0 in YEPD medium were transferred to YEP4G (containing 4% galactose instead of glucose in YEPD) medium containing 1 µg/ml [¹⁴C]mannose and incubated for 2 h (21). Cells were then converted to spheroplasts and lyzed by suspending in lysis buffer (0.2 M sorbitol, 0.1 M NaCl, 5 mM MgCl₂, 20 mM HEPES, pH 7.4, 100 µg/ml phenylmethylsulfonyl fluoride) and incubated for 30 min on ice. After centrifugation at 100,000 × g for 1 h, the supernatant was analyzed by silica thin layer chromatography (n-butyl alcohol:acetic acid:water, 4:1:2). The amount of intracellular GDP-Man was determined as the radioactivity of the spot of GDP-Man by Image Analyzer.

RESULTS

We obtained two independent vig9 mutants among 331 spontaneous vanadate-resistant isolates. H17 (vig9-1) and J12 (vig9-2) were able to grow on YEPD medium containing up to 5 mM vanadate and on SD medium containing up to 4 mM vanadate. Both mutants grow well at either 25 or 37 °C as the parent yeast. Sensitivity of the mutants to 5-fluorouracil, hygromycin B, and cycloheximide was not altered from the parent strain. However, vig9 mutants showed hypersensitivity to geneticin (G418). They could not grow on YEPD plates containing geneticin at 50 µg/ml, but the parent yeast grew well in the presence of geneticin at 100 µg/ml.

The degree of glycosylation was determined by analyzing with secretory invertase for N-glycosylation and extracellular chitinase for O-glycosylation. As shown in Fig. 1A, invertase of the vig9-1 mutant migrated faster than that of the wild type in native gel electrophoresis. The difference of mobility is due to the degree of glycosylation because the invertase polypeptides showed the same mobility in SDS-polyacrylamide gel electrophoresis after removal of saccharides by endoglycosidase H treatment (Fig. 1B). N-Glycosylation of the vig9-1 mutant was moderately defective as the rate of migration of invertase was between those of the wild type and the mnn9 mutant. Each N-linked saccharide in the mnn9 mutant is only 5 mannose residues larger than the core oligosaccharide (22). O-Glycosylation was also affected in the vig9 mutants (Fig. 1B). Vanadate-resistant, geneticin-hypersensitive, and glycosylation-defective phenotypes were all recessive and cosegregated during three cycles of back-crossing.

To clone the wild type VIG9 gene, we used geneticin hypersensitivity of the vig9 mutant as a selective marker. We transformed H17-6C with a genomic library of the wild type S. cerevisiae constructed on pRS314 and screened for Trp⁺ transformants, which can grow on a YEPD plate containing geneticin at 100 µg/ml. Twenty-seven candidate transformants were tested further for vanadate sensitivity and N-glycosylation of invertase. Plasmid DNAs from four transformants that showed the wild type level of vanadate sensitivity and glycosylation of invertase were recovered in E. coli, and the restriction maps were constructed. These plasmids contained three kinds of overlapping inserts. Deletion and subcloning analysis of the fragment in pSV903 indicated that a 2.5-kilobase HindIII-Sau3A1/BamHI fragment was responsible for complementation.

The 2.5-kilobase minimal fragment was cloned in a integration vector (pRS305) to make pSV921. This plasmid was integrated in the chromosome of a vig9-1/vig9-1 diploid yeast (H17W) by homologous recombination. Tetrad analysis of 26 asci of the transformant indicated that pSV921 was integrated at the vig9-1 locus. Thus, the insert contains the authentic VIG9 gene.

We determined the nucleotide sequence of the 2498-base pair insert of pSV914 and found an open reading frame (ORF) of 1083 nucleotides which encodes a polypeptide of 361 amino acids. After submission of our manuscript, the complete nucleotide sequence of the S. cerevisiae genome was determined, and this ORF had a systematic name of YDL055C on the chromosome IV. The predicted molecular mass is 39,565 Da, and the pI is 5.93. There are three potential TATA sequences of the promoter elements at nucleotide 258 to 251 (TATATATA), 70 to 67 (TATA), and 14 to 8 (TATAAA) in the 5 upstream of the ORF. If the last candidate is selected, the putative initiation methionine (nucleotides 1-3) is too close to it, and an alternate methionine at nucleotides 97-99 will be used in this case. However, this possibility is unlikely because the deduced NH₂-terminal amino acid sequence is highly conserved with the NH₂-terminal sequences of bacterial proteins of similar function as described below. Consensus poly(A)-polymerase recognition sequences (AATAAA) are found at nucleotides 1238-1243, 1256-1261, and 1308-1313 in the 3 downstream of the ORF. No consensus splicing motif (5-GTAPyGT ... TACTAAC ... PyAG-3) was found in the insert of pSV914. This ORF should represent the VIG9 gene.

To examine whether VIG9 is essential for yeast cell growth, we constructed a diploid yeast in which one copy of VIG9 was disrupted. Disruption was achieved by replacing a HpaI-SalI fragment of one copy of VIG9 with LEU2 in W303 (Fig. 2A). Disruption was confirmed by Southern blotting (Fig. 2B). After sporulation, 31 asci were dissected on YEPD at 30 °C. Only two spores among each tetrad outgrew, and all viable colonies were auxotrophic for leucine (Fig. 2C). Another gene disruption with TRP1 also gave similar results. Thus, the VIG9 gene is essential for yeast cell growth.

Fig. 3 shows the hydropathy profile of the predicted Vig9 protein. Vig9 is a hydrophilic protein without any hydrophobic stretch long enough to function as a secretory signal sequence or transmembrane domain. This suggests that Vig9 protein is a soluble cytoplasmic protein.

Comparison of the predicted Vig9 protein sequence with a data base using the FASTA program (23) indicated the sequence was a novel one. However, significant homology between Vig9 protein and several bacterial proteins was found (Fig. 4). Those bacterial proteins have a common characteristic. RfbF protein of Salmonella typhimurium, which is 32.7% identical in 251 amino acids to Vig9 protein, catalyzes the synthesis of CDP-glucose from glucose 1-phosphate and CTP in the biosynthetic pathway of O-antigen (24). RfbF protein of Yersinia pseudotuberculosis is 31.1% identical in 257 amino acids to Vig9 protein (25). RfbA protein of Yersinia enterocolitica is 27.6% identical in 254 amino acids and catalyzes the synthesis of dTDP-L-rhamnose.³ StrD protein of Streptomyces griseus, which is 23.3% identical in 339 amino acids, catalyzes the synthesis of dTDP-streptose (26). So, all of these enzymes catalyze the synthesis of (deoxy)NDP-sugar from sugar phosphate and (deoxy)NTP. Homology with these enzymes having this common characteristic suggests that Vig9 protein may function as GDP-Man pyrophosphorylase, which catalyzes the formation of GDP-Man from mannose 1-phosphate and GTP.

Cell extracts were prepared from yeast cells and assayed for GDP-Man pyrophosphorylase activity. Fig. 5 shows that the amount of radioactive GDP-Man formed from [-³²P]GTP and mannose 1-phosphate was linearly dependent on the amount of protein in the reaction mixture. The amount of the labeled GDP-Man increased in a time-dependent manner (Fig. 6). Fig. 7 shows the effect of adding a 10-fold molar excess (2 mM) of cold NTPs on the formation of labeled GDP-Man. The authentic substrate, GTP, competitively reduced the amount of radioactive GDP-Man to 18% of that of the control reaction. The others also affected the formation of GDP-Man: ATP to 51%, UTP to 67%, and CTP to 87%. Although ATP showed relatively severe interference, we have not examined whether ADP-Man was formed or not. If the cell extract was omitted, the radioactivity in the region of the thin layer film corresponding to the spot of GDP-Man was 2.8% of the control reaction (the background count). When mannose 1-phosphate was omitted, the amount of labeled GDP-Man formed was 7.1% of the control reaction, which may indicate that only a small amount of mannose 1-phosphate is present in the cell extract. Stimulation of the reaction by the addition of pyrophosphatase to remove pyrophosphate, the other reaction product, was not significant in our assay condition (data not shown).

16

16

The extract of the vig9-1 mutant cells had a practically negligible activity to form GDP-Man (Fig. 6, H17-6C). This result suggests that the vig9 mutation affects the GDP-Man pyrophosphorylase activity. A mixture of equal amounts of extracts from wild type and vig9-1 cells supported the formation of the same amount of GDP-Man as did the wild type extract (Fig. 6, TM10 + H17-6C). This result indicates that the vig9 mutation did not function to produce inhibitory products or catalyze a reaction that reduces the amount of GDP-Man; this is in agreement with the recessive nature of vig9 mutations. Upon introduction of a single-copy VIG9 plasmid, the vig9 mutant cells recovered the GDP-Man pyrophosphorylase activity to the wild type level (Fig. 6, H17-6C/pSV914). Furthermore, the extract of wild type cells carrying a multicopy VIG9 plasmid showed a 3-4-fold increased activity to form GDP-Man (Fig. 6, TM10/pSV923). These results further support that VIG9 is the structural gene for GDP-Man pyrophosphorylase in accordance with the prediction from the homology between VIG9 protein and the bacterial enzymes.

To demonstrate that the Vig9 protein catalyzes the formation of GDP-Man, we constructed a plasmid that produces a GST-Vig9 fusion protein as described under "Materials and Methods." The affinity-purified GST-Vig9 protein had activity to synthesize [³²P]GDP-Man from [-³²P]GTP and mannose 1-phosphate, whereas the fusion partner GST did not (Fig. 8).

Formation of [³²P]GDP-Man from [-³²P]GTP and mannose 1-phosphate by GST-Vig9 fusion protein.

We further examined the effect of the reduced activity of GDP-Man pyrophosphorylase in the vig9 mutant. Cells of the wild type and vig9-1 mutant were grown in the presence of [¹⁴C]mannose, and the amount of GDP-Man in the cell extracts was determined. Five independent experiments demonstrated that the amount of GDP-Man in vig9-1 mutant cells was 70.6% of that of the wild type cells. This result indicates that the substrate of mannosylation is a limiting factor of glycosylation and the direct cause of the glycosylation defect in the vig9-1 mutants.

DISCUSSION

We have isolated a new glycosylation-defective mutation, vig9, among vanadate-resistant mutants. The biochemical rationale of this enrichment is not clear (8). Mutants of a wide variety of genes including MNN8, MNN9, MNN10, ALG4/SEC53, OCH1, ANP1, VAN1, and VRG4/VAN2 showed vanadate resistance (8, 27-29).² The ALG4/SEC53 encodes phosphomannomutase (30), and OCH1 encodes a glycosyltransferase (7). The common characteristics of these genes seem to be only that the mutations affect glycosylation. So, vanadate resistance may be an indirect effect of glycosylation defect. The protein(s) that concerns the efficiency of vanadate transport into the cell and/or access of vanadate to the main target protein(s) may be glycoprotein(s), and immature glycosylation may affect their activity and/or stability and thus may be responsible for vanadate resistance.

The vig9 mutants showed hypersensitivity to geneticin, and we cloned the wild type gene by restoration from this phenotype. Various glycosylation-defective mutants also shows geneticin hypersensitivity, and glycosylation is suggested to be necessary for the activity of a protein that prevents uptake or mediates removal of aminoglucosides (31).

Mannose is a crucially important sugar for protein glycosylation in S. cerevisiae. In each N-saccharide chain, only two unit sugars are N-acetylglucosamine, and all the rest are composed of mannose. O-Saccharides are composed solely of mannose in S. cerevisiae. Mannose is transferred to proteins from its activated form, GDP-Man, either directly or via dolichol intermediates, Dol-PP-GlcNAc₂Man₉Glc₃ or Dol-P-Man. GDP-Man is synthesized from a glycolytic intermediate, fructose 6-phosphate, by the action of three enzymes. Fructose 6-phosphate is first converted to mannose 6-phosphate by phosphomannose isomerase, then converted to mannose 1-phosphate by phosphomannomutase, and finally GDP-Man is synthesized from mannose 1-phosphate and GTP by GDP-Man pyrophosphorylase. Mutants defective in phosphomannose isomerase and phosphomannomutase were isolated previously, and the structural genes, PMI40 and ALG4/SEC53, respectively, were cloned and sequenced (30, 32, 33). These mutants were isolated among temperature-sensitive lethal mutants that also showed defects in glycosylation or protein secretion (6, 34-36). Mutants of GDP-Man pyrophosphorylase have not been isolated in similar mutant screening, and this is the first report on mutations in this enzyme.

Two independently isolated vig9 mutants showed glycosylation defects in both N- and O-saccharides (Fig. 1). This phenotype is similar to pmi40-1 and sec53/alg4 mutants in good accordance with the prediction that these gene products participate in the same pathway to form GDP-Man from fructose 6-phosphate. The in vitro assay of GDP-Man pyrophosphorylase in yeast lysates suggested that the VIG9 gene encodes the enzyme (Fig. 5). Affinity-purified GST-Vig9 fusion protein catalyzed the formation of GDP-Man, whereas GST protein itself did not (Fig. 8). This result clearly indicates that VIG9 encodes GDP-Man pyrophosphorylase. However, in contrast to pmi40 and sec53/alg4 mutants, vig9 mutants did not show a temperature-sensitive growth defect on the various media we tested. As the VIG9 gene was demonstrated to be essential for cell proliferation, the possibility of a redundant gene is excluded. The intracellular concentration of GDP-Man of the vig9-1 mutant was approximately 70% of that of the wild type, whereas the GDP-Man pyrophosphorylase activity in the cell lysate was less than 5% of that of the wild type. There are several possibilities to explain this apparent discrepancy: (i) GDP-Man pyrophosphorylase may be highly active, and the reduced activity is still enough to support synthesis of 70% of the product of the wild type enzyme; (ii) the metabolic flow tends to accumulate GDP-Man faster than its consumption in vivo; and/or (iii) the mutant enzyme is considerably more unstable in vitro than in vivo.

There are other genes that encode NDP-sugar pyrophosphorylases in S. cerevisiae. The well known GAL7 gene encodes UDP-galactose pyrophosphorylase (37), and more than a dozen of proteins in eukaryotes are homologous to the Gal7 protein in the data base. UGP1 encodes UDP-glucose pyrophosphorylase (38), and there are also more than a dozen of its homologs in the data base. However, no significant homology was found among Vig9, Gal7, and Ugp1 proteins. Although the complete nucleotide sequence of S. cerevisiae genome has been determined, no protein in the genome shares significant homology with Vig9. A putative protein YHL012W has 20.8% homology with Ugp1 protein and may possibly encode an NDP-sugar pyrophosphorylase. The ORF for UDP-GlcNAc pyrophosphorylase is yet to be discovered, and YHL012W is a possible candidate.

Bacteria also have enzymes to synthesize (deoxy)NDP-sugars which are necessary for the synthesis of capsular polysaccharides, outer membrane lipopolysaccharides (O-antigen), or aminoglycoside antibiotics. Yeast phosphomannose isomerase has homology with the bacterial phosphomannose isomerases from E. coli and S. typhimurium at 35.2 and 34.9% identity, respectively (33). The sequence of bacterial phosphomannomutase has not been reported yet. Yeast GDP-Man pyrophosphorylase (Vig9 protein) has sequence identity between 22.9 and 32.7% with various enzymes that catalyze the formation of (deoxy)NDP-sugars. So, these genes were probably derived from an ancestral gene in the course of evolution. There are several highly conserved sequences such as the NH₂-terminal 34 amino acids, and many residues are identical among the sequences of five proteins (Fig. 4). The conservation in these primary sequences may indicate their importance in the catalytic function and/or folding of these proteins.

GDP-Man is synthesized from mannose 1-phosphate and GTP in the cytosol. For protein mannosylation in the Golgi apparatus, GDP-Man should be translocated in the lumen of this organelle. In mammalian cells, a specific membrane carrier mediates the entry of GDP-Man into the Golgi and the exit of GMP, which is converted from GDP, the other reaction product of glycosylation, by the Golgi GDPase (39). This is also the case in S. cerevisiae, where the yeast Golgi GDPase gene (GDA1) has been cloned, and the mutant shows a partial block in O- and N-glycosylation (40). On the other hand, the cloning or mutation of the GDP-Man transporter gene has not be reported. S. cerevisiae has many excellent techniques for a genetic approach, such as the selection of multicopy suppressor genes or synthetically lethal mutations, to find related gene products in the same biochemical pathway. The GMP/GDP-Man transporter gene may function as a multicopy suppressor of the vig9 mutation, or its mutation may be synthetically lethal with it. Such approaches should be helpful in understanding the whole process of protein glycosylation.

After our first manuscript was submitted, three sequences were registered in the EMBL data base. The nucleotide sequence of accession number [GenBank] was a part of the genome project and completely coincided with our nucleotide sequence. The sequence of accession number [GenBank] (41) was the same except the bases 807-808 were GG, whereas they were CC in our data. Correspondingly, the 16th amino acid of the ORF is glycine, but it is proline in our data. The sequence of accession number [GenBank] was the same except the base 149 was C while it was T in our data. Correspondingly, the 50th amino acid of the ORF is alanine while it is valine in our data.

Benton et al. (41) reported PSA1 gene as a suppressor of temperature-sensitive growth of cln1 cln2 erc14-1. As described above, PSA1 is identical to VIG9, and the authors also found homology between the protein and NDP-hexose pyrophosphorylases. They constructed temperature-sensitive alleles of the gene but failed to show defect in glycosylation with carboxypeptidase Y. The reason for this apparent discrepancy with our data is not clear.