Functional evidence for UDP-galactose transporter in Saccharomyces cerevisiae through the in vivo galactosylation and in vitro transport assay.

The oligosaccharide profiles in glycoproteins are determined by a series of processing reactions catalyzed by Golgi glycosyltransferases and glycosidases. Recently in vivo galactose incorporation in Saccharomyces cerevisiae has been demonstrated through the expression of human beta-1,4-galactosyltransferase in an alg1 mutant, suggesting the presence of a UDP-galactose transporter in S. cerevisiae (Schwientek, T., Narimatsu, H., and Ernst, J. F. (1996) J. Biol. Chem. 271, 3398-3405). However, this is quite unexpected, because S. cerevisiae does not have galactose residues in its glycoproteins. To address this question we have constructed S. cerevisiae mnn1 mutant strains expressing Schizosaccharomyces pombe alpha-1,2-galactosyltransferase. The mnn1 mutant of S. cerevisiae provides endogenous acceptors for galactose transfer by the expressed alpha-1,2-galactosyltransferase. We present here three lines of evidences for the existence of UDP-galactose transporter in S. cerevisiae. (i) About 15-20% of the total transformed mnn1 cells grown in a galactose medium were stained with fluorescein isothiocyanate-conjugated alpha-galactose-specific lectin, indicating the presence of alpha-galactose residues on the cell surface. (ii) Galactomannan proteins can be precipitated with agarose-immobilized alpha-galactose-specific lectin from a whole cell lysate prepared from transformed mnn1 cells grown in a galactose medium. (iii) The presence of UDP-galactose transporter was demonstrated by direct transport assay. This transport in S. cerevisiae is dependent on time, temperature, and protein concentration and is inhibited by nucleotide monophosphate and Triton X-100. The overall UDP-galactose transport in S. cerevisiae is comparable with that in S. pombe, indicating a more or less similar reaction velocity, while the rate of GDP-mannose transport is higher in S. pombe than in S. cerevisiae.

N-Linked glycosylation is an essential modification that is highly conserved among all eukaryotic cells (1)(2)(3)(4). The complex N-linked oligosaccharides have a wide variety of structure in animal cells, while they are relatively simpler in lower eukaryotes, such as Saccharomyces cerevisiae. However, in both cases the initial steps are nearly identical and involve the synthesis of Glc 3 Man 9 GlcNAc 2 -PP-dol, transfer of oligosaccharide from lipid to protein, and subsequent trimming to Man 8 GlcNAc 2 in the endoplasmic reticulum. In the latter stage S. cerevisiae elongates it to polymannose-type structure without adding any N-acetylglucosamine, galactose, and sialic acid residues that are characteristics of complex oligosaccharides in mammalian cells. Recently, interest in glycosyltransferases arose by their potential usefulness as tools for the synthesis of oligosaccharides in vitro (5) and for the remodeling of glycan chains of natural or recombinant glycoproteins. Yeast offers an attractive host for the production of heterologous proteins (6), and a number of recombinant glycoproteins were successfully produced in S. cerevisiae, although the sugars attached to proteins were confined to high mannose type. In this regard, the ⌬och1mnn1 double mutant cells (7,8) and ⌬och1⌬mnn1⌬mnn4 triple mutant cells (9) are novel to produce glycoproteins containing N-linked core oligosaccharide Man 8 GlcNAc 2 and to reduce a yeast specific acidic oligosaccharide (mannosylphosphorylated Man 8 GlcNAc 2 ) content, respectively. In addition, the MNN6 gene encoding mannosylphosphate transferase is useful to produce mannosylphosphorylated Man 8 GlcNAc 2 (10), which can be converted to phosphorylated Man 8 GlcNAc 2 that functions as a sorting signal of lysosomal glycoproteins in mammalian cells (1), after releasing a terminal mannose residue. However, we are interested in further attempts to make an in vivo sugar structure remodeling toward the mammalian type structure in S. cerevisiae. To achieve this goal we need to express a number of cDNA encoding mammalian glycosyltransferases and specific sugar nucleotide transporters with an authentic localization having correct topology in the lumen of the Golgi in appropriate yeast mutant cells that provide specific endogenous acceptors for the expressed gene product (glycosyltransferase). As an essential step for the above goal, it is necessary to clone UDP-Gal transporter gene in S. cerevisiae. However, recently, the in vivo galactose addition was demonstrated by expressing the genes of mammalian ␤-1,4-galactosyltransferase (GalT) 1 in S. cerevisiae (alg1 mutant), which suggests the presence of unexpected UDP-Gal transporter, allowing ␤-1,4-GalT to act inside the lumen of the Golgi to add galactose onto endogenous acceptors under nonpermissive growth conditions (11). According to the hypothesis of a sugar nucleotide transport system in the Golgi apparatus, an antiporter for the corresponding sugar nucleotide must be present in the lumen of the Golgi apparatus, and in the case of UDP-Gal transport, UMP must be present to act as an antiporter (12). A galactosyltransferase inside the lumen is necessary to transfer galac-tose from UDP-Gal to endogenous acceptors to produce UDP, which can be subsequently hydrolyzed to UMP by a UDPase. Since the presence of a UDPase is also demonstrated in S. cerevisiae without any apparent function (13), we are interested in checking the presence of the UDP-Gal transport system in S. cerevisiae by a direct transport assay. In this report we describe in vivo galactose addition to the endogenous mannoproteins under growing conditions in S. cerevisiae (mnn1 mutant) through the expression of the gene encoding S. pombe ␣-1,2-GalT, and provide direct evidence for the presence of a UDP-Gal transport system in S. cerevisiae by in vitro transport assay. It is quite surprising to us that S. cerevisiae has UDP-Gal transport activity without any physiological need. At present it remains unsolved without further evidence that this transport activity is from a specific UDP-Gal transporter or other sugar nucleotide transporter having a broader substrate specificity. However, this finding is promising for the achievement of our goal for the incorporation of galactose residues into sugar chain of the glycoproteins in S. cerevisiae, when the appropriate acceptor glycans are provided.
In Vitro Amplification of S. pombe gma12 ϩ Gene-PCR was carried out essentially as described by Saiki et al. (19). The oligonucleotide primers used for PCR were as follows: 5Ј-cccccctgcaggaattcATGCGGT-TCGCTCCTTATTTA-3Ј and 5Ј-cccccctcgagCTAGGATGATGGTTTCA-AAAGA-3Ј, where capital letters indicate the region of gma12 ϩ open reading frame (20). PCR was carried out using 0.1 M of these primers and 10 ng of genomic DNA of S. pombe strain 972h Ϫ (kindly provided by Dr. Y. Yamamoto, University of Tokyo) as a template in the 20 l reaction mixture. LA Taq polymerase (Takara Shuzo, Kyoto, Japan) was used for reactions. The reactions were performed in 30 cycles as follows. The samples were heated at 98°C for 20 s, cooled at 54°C for 1 min, and heated at 72°C for 2 min.
Construction of Plasmids-Plasmid pSKϩϩgma12 carries a 1.1-kb EcoRI-XhoI fragment containing gma12 ϩ open reading frame amplified by PCR. Plasmid YEpUGAP-gma12 ϩ was constructed as follows. The 0.7-kb EcoRI-XhoI fragment of pSKϩϩgma12 was inserted into plas-mid pKT10 (21) to make a plasmid pKT10ϩgma12C. Subsequently, the 0.4-kb EcoRI fragment of pSKϩϩgma12 was inserted into the plasmid pKT10ϩgma12C to make YEpUGAP-gma12 ϩ . This plasmid carries the entire open reading frame of the gma12 ϩ gene in pKT10, which contains URA3, the origin of the 2-m plasmid and the TDH3 promoter of S. cerevisiae. This construction placed the gma12 ϩ under the control of TDH3 promoter.
Preparation of Solubilized Microsomal Proteins-Ten grams of freshly grown cells were washed with ice-cold water, resuspended in 10 ml of TMS buffer (20 mM Tris-Cl, pH 7.5, 5 mM MgCl 2 , 0.25 M sucrose). Cells were broken with 0.5-mm glass beads in a bead beater (B. Braun Biotech International, Melsungen, Germany). Glass beads, unbroken cells and large cell debris were removed by centrifugation at 10,000 ϫ g at 4°C. The resultant supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C. The membrane pellet was resuspended in TMS buffer to a protein concentration of 25 g/ml and microsomes were disrupted and solubilized by the addition of Triton X-100 to a final concentration of 2% (v/v). After incubating on ice for 30 min, the mixture was centrifuged at 100,000 ϫ g for 60 min in 70.1Ti Beckman ultracentrifuge rotor (Beckman ultracentrifuge, L-80 Optima). The supernatant was used as the source of solubilized enzyme for galactosyltransferase assay.
Assay of Galactosyltransferase-The standard transferase assays were carried out according to Chappell and Warren (22). The reaction mixture in a total volume of 50 l contained 100 mM Hepes-NaOH (pH 7.0), 1 mM MnCl 2 with 25 nmol of UDP-[ 3 H]Gal (specific activity, 40 nCi/nmol, after dilution of commercially available UDP-[ 3 H]Gal with unlabeled UDP-Gal) and 5 mol of given sugar acceptor. Reaction mixture contained 0.1-0.2% (v/v) Triton X-100, depending on the detergent concentration in the enzyme fraction. After incubation at 30°C for 30 min, the reaction was terminated by adding 200 l of ice-cold water and loaded onto a 1.0 ml of Dowex-1 (Cl Ϫ form, 200 -400 dry mesh) anion exchange column packed in a 3-ml syringe. The column was washed twice with 1.0 ml of water and the combined eluents were mixed with 2.5 volumes of scintillation mixture. The Dowex columns were regenerated by washing 2.5 ml of 5 M NaCl, 2.5 ml of 0.1 M HCl, and then 8.0 ml of water. One unit of enzyme activity is defined as the amount of enzyme that catalyzes the transfer of 1.0 nmol of galactose from UDP-Gal/min under the standard assay conditions. GDPase Assay-The assay was done essentially as described by Abeijon et al. (23). Briefly, incubation mixture in a final volume of 0.1 ml contained enzyme (20 g of Triton X-100 solubilized P 3 fraction) (see "Subcellular Fractionations"), CaCl 2 (1 mol), Triton X-100 (100 g), GDP (0.2 mol), and immidazole buffer, pH 7.6 (20 mol). Incubation was for 5 min at 30°C. The reaction was stopped by adding 10 l of 10% (w/v) SDS. Released inorganic phosphate was determined by the Ames (24) method. The absorbance was measured at 820 nm and the amount of inorganic phosphate released was calculated from a calibration curve prepared by using KH 2 PO 4 as a standard. One unit of activity was defined as 1 mol of inorganic phosphate released per min under standard assay conditions. Specific activity is calculated as units/mg of protein. Latency of GDPase was calculated according to Abeijon et al. (23).
Protein Quantitation-The protein concentration was measured by BCA reagent (Pierce) using bovine serum albumin as a standard.
Preparation of G. simplicifolia I-B4-agarose-GS I-B4 was coupled with CNBr activated Sepharose 4B according to Roy et al. (28) with some modifications. Five hundred milligrams of CNBr-Sepharose 4B were swollen for 1 h and washed on a sintered glass (G3) with 1 mM HCl. Five hundred milligrams of pure lectin (GS I-B4) were used in NaHCO 3 buffer (10 mM, pH 8.3) containing NaCl (0.5 M), mixed with the gel, and kept overnight at 4°C. For blocking the remaining active groups, the gel suspension was transferred to blocking agent, 0.2 M glycine (pH 8.0). To wash away excess unbound adsorbed proteins, the gel was alternatively washed three or four times with high pH coupling buffer (pH 8.3) and then low pH Na-acetate buffer (pH 4.2), each containing 0.5 M NaCl. This ensured that no free ligand remained ionically bound to the immobilized ligand.
Subcellular Fractionations-The Golgi-rich fractions were isolated as described by Walworth and Novick (29) with some modifications. Briefly, cells were grown at 30°C in an appropriate medium to an A 600 3.0 -3.5. The cells were washed with 10 mM NaN 3 and resuspended in the spheroplast buffer (1.4 M Sorbitol, 50 mM potassium phosphate, pH 7.5, 10 mM NaN 3 , 40 mM 2-mercaptoethanol, and zymolyase 100T, 1 2 Y. Shimma, unpublished results. mg/g of wet cells). The cell suspension in spheroplast buffer was incubated at 37°C for 45 min with occasional gentle mixing. The spheroplasts were pelleted in a table top centrifuge (Beckman, GS 6KR), and washed twice with 1.0 M ice-cold sorbitol to remove zymolyase. The cells were then suspended in ice-cold lysis buffer (0.8 M Sorbitol in 10 mM triethanolamine, pH 7.2, and anti-protease mixture). The spheroplasts were transferred to a 50-ml Dounce homogenizer (Wheaton Scientific, Millville, NJ) on ice and lysed with 20 strokes of a Teflon pastel. Low speed centrifugation at 1,000 ϫ g for 10 min yielded a large pellet containing unlysed cells and cell wall debris. The post 1,000 ϫ g supernatant (S 1 ) was carefully collected and then centrifuged at 10,000 ϫ g for 15 min at 4°C (Hitachi, Himac SCR20B, Japan), which yielded a pellet (P 2 ) of nuclear and mitochondrial fraction. The post mitochondrial supernatant (S 2 ) was then centrifuged at 100,000 ϫ g (Beckman ultracentrifuge L-80 Optima, in 70.1Ti rotor) yielded a pellet (P 3 ) of Golgi-enriched fraction and the cytosolic supernatant (S 3 ).
Isolation of Total Cellular Proteins-Yeast cells were grown in an appropriate medium and harvested at 3,000 rpm and washed with 10 mM NaN 3 and 10 mM dithiothreitol, suspended in LIP buffer (50 mM Tris-Cl, pH 7.4, 1% Triton X-100, 150 mM NaCl, 5 mM EDTA). Cells were disrupted by glass beads by vortexing (at full power 4 ϫ 1 min) and centrifuged at 5,000 rpm for 15 min at 4°C, and the clear supernatant were isolated. The total cellular proteins were precipitated by adding an equal volume of 7% (v/v) trichloroacetic acid to the supernatant and kept on ice for at least 1 h. The precipitated proteins were isolated by centrifugation at 10,000 ϫ g for 45 min at 4°C, washed twice with cold acetone, and reisolated by centrifugation. Isolated proteins were dissolved in iodination buffer (100 mM Na-phosphate buffer, pH 7.0).
Radioiodination of Whole Cell Proteins-Proteins were iodinated by IODO-BEADs according to the manufacturer's instructions. The iodination reaction was started by adding Na 125 I (200 Ci) and two IODO-BEADs to 500 g of protein, dialyzed, and dissolved in 0.1 M Naphosphate buffer, pH 7.0. The reaction mixture was incubated at room temperature for 10 min. The iodination reaction was terminated by removal of IODO-BEADs, and the iodinated proteins were dialyzed extensively against water and then PBS to remove free iodine.
Lectin Precipitation-Lectin precipitation was performed according to Schwientek et al. (11). In short, 25 l of 30% (w/v) concanavalin A-agarose 4B and 35 l of 50% (w/v) of GS I-B4-agarose was added to 100 g of iodinated whole cell proteins in PBS to precipitate mannoproteins and galactomannoproteins, respectively. The precipitated pellet was washed twice with PBS, solubilized with Laemmili (31) solubilization buffer, boiled for 5 min, and subjected to 10% SDS-PAGE. The gels were fixed, dried, and exposed to Kodak X-Omat film at Ϫ70°C for 12 h.
Lectin Staining of Cells-Yeast cells were stained with FITC-conjugated GS I-B4 as follows. Freshly grown cells (2 ϫ 10 6 ) of various strains were washed in PBS and taken in 50 l of PBS. GS I-B4-FITC (2.5 l, 1.0 mg/ml) was added and incubated on ice for 30 min in dark. After incubation the cells were washed three times with 200 l of PBS, resuspended in 10 -15 l of PBS, and observed under fluorescence microscope (Olympus BX50, Japan).
FACS Analysis-The samples were prepared same as described above using same lectin. The cells were sorted in a fluorescent cell sorter (Becton Dickinson FACSCalibur). The positive cells were isolated and concentrated in a "built-in" cell concentrater.
Isolation of Yeast Mannans-Mannans were isolated according to Peat et al. (32) with modifications. Freshly grown yeast cells were isolated by centrifugation, washed with 1% (w/v) KCl, and resuspended in 20 mM citrate buffer (pH 7.0). The cell suspension was then heated in an autoclave at 120°C for 1 h. Supernatant was isolated by centrifugation, and 2 volumes of ethanol were added. The crude mannan was precipitated by centrifugation (10,000 ϫ g for 20 min). The precipitated mannan was then dissolved and dialyzed against distilled water with two changes. Protein concentrations were estimated and used as acceptor substrates for galactosyltransferase assay.
Transport Assay-Sugar nucleotide transport was assayed according to Ma et al. (33) with some modifications. Golgi-enriched vesicles (500 g of protein) were incubated in a reaction mixture (20 mM Tris-Cl, 0.25 M sucrose, 5 mM MgCl 2 , 1 mM MnCl 2 , 10 mM 2-mercaptoethanol) with radiolabeled sugar nucleotides (as indicated in the figure legends). After incubation at 30°C for 5 min, the samples were diluted with 3.0 ml of ice-cold stop buffer (20 mM Tris-Cl, 0.25 M sucrose, 150 mM KCl, 1 mM MgCl 2 ) and placed on ice. Diluted samples were then applied onto filtration apparatus containing HA filters (24-mm diameter, 0.45-m pore size). The filters were washed with another 10 ml of wash buffer. The filters were air-dried and placed in 15-ml counting vials, and 2 ml of ethylene glycol methyl ether were added. The vials were allowed to stand at room temperature with occasional shaking until the filters were dissolved (about 30 min). Ten milliliters of scintillation mixture (Clear-sol) were added, and the samples were counted in a liquid scintillation counter (Beckman LS1701). The amount of radioactivity that was bound nonspecifically to the outside of the vesicles was determined by zero time assay.

RESULTS
Expression of S. pombe ␣-1,2-Galactosyltransferase in S. cerevisiae-gma12 ϩ gene encoding ␣-1,2-GalT of S. pombe was expressed in S. cerevisiae cells. Plasmid YEpUGAP-gma12 ϩ carries gma12 ϩ gene under the control of the S. cerevisiae TDH3 promoter. YEpUGAP-gma12 ϩ and control plasmid pKT10 were introduced into S. cerevisiae R16B mnn1 mutant cells (the choice of this mutant is further described under "Discussion"). GalT activity in solubilized microsomal fractions from these transformant cells was assayed using ␣-methyl D-mannoside as an acceptor substrate (Table I). The enzyme preparation from transformed cells expressing gma12 ϩ gene showed a higher specific activity toward galactose transfer than the control counterpart cells. This GalT activity in transformed cells is acceptor dependent (Table I).
Since the full-length gma12 ϩ gene includes sequences coding for both transmembrane and cytoplasmic domains (20), the expressed enzyme is expected to be localized in the Golgi vesicles with the same topography as in S. pombe. As shown in Table IIA, a considerable amount of GalT activity is associated with the low speed pellet (P 2 ) and about 34% of total enzyme activity is associated with the 100,000 ϫ g pellet (P 3 ), where as about 52% of GDPase activity (Golgi marker), 16% of ␣-mannosidase (vacuolar marker), 18% of NADPH-dependent cytochrome c reductase (endoplasmic reticulum marker) and 24% of cytochrome c oxidase (mitochondrial marker) are associated with vesicle fraction (P 3 ). However, the relative distribution of above proteins on the basis of specific activity indicated that the enzyme (␣-1,2-GalT) was highly abundant in the vesicle fraction (P 3 ) with an overall increase in specific activity of 4.9-fold (Table IIB, 5.43 in P 3 versus 1.1 in S 1 ). In this study the specific activity of GDPase has increased about 5.3-fold, whereas the specific activities of marker enzymes of other organelles such as NADPH-dependent cytochrome c reductase, cytochrome c oxidase, and ␣-mannosidase has reduced by 2.4-, 9.8-, and 5.5-fold, respectively. Regarding the topology of high speed pellet (P 3 ) associated enzymes about 70% of the GalT has a luminal orientation, while the marker enzyme GDPase showed 95% luminal orientation (Table IIC). The endogenous in vivo acceptor substrate for S. pombe ␣-1,2-GalT is not known yet. Therefore, we have tried a number of acceptors including mannoproteins prepared from different mutant strains of S. cerevisiae. Among the sugar acceptors so far used, ␣-methyl D-mannoside is the best acceptor (Table IIIA). Mannoproteins prepared from mnn1 mutant displayed the higher activity compared with wild type and other mutant cells (Table IIIB).
In Vivo Galactose Addition-We have hypothesized that, if the recombinant ␣-1,2-GalT has a correct Golgi luminal orientation and the cell wall mannoproteins of mnn1 mutant can serve as an acceptor for galactose transfer, then the in vivo galactose addition to glycoproteins may be possible in S. cerevisiae. To address this possibility, the transformed mnn1 cells were grown on galactose as a sole carbon source to provide UDP-Gal in the cytosol and were stained with FITC conjugated ␣-galactose-specific lectin (GS I-B4-FITC). Under our experimental conditions, 15-20% of the cells were stained with GS I-B4-FITC, while no staining was observed when the mnn1 cells containing control plasmids were grown on galactose or when transformed mnn1 cells were grown on glucose (Fig. 1). It was unsuccessful for enrichment of the stained cells by FACS sorting, and regrowing the positive cells in the galactose medium (SCGal-Ura) does not increase the relative ratio of stained cells, still maintaining only the 15-20% of cell staining. These partial cell staining properties are reproducible and independent of growth phase or galactose concentration in the growth medium (2% and 5%, w/v) (data not shown).
Lectin Precipitation-To further confirm in vivo galactose incorporation, we have tried to precipitate proteins containing specific sugars by using sugar-and linkage-specific lectins from whole cell lysates. The total cellular proteins were isolated from yeast cells and iodinated as described under "Experimental Procedures." A number of bands were detected when con-canavalin A-agarose was used to precipitate mannoproteins from whole cell proteins prepared from S. pombe and different transformants of S. cerevisiae either carrying gma12 ϩ expressing or control plasmids ( Fig. 2A, lanes 1-5). In contrast, when GS I-B4-agarose was used to precipitate galactomannoproteins, only cellular proteins from S. pombe and S. cerevisiae a Values were expressed as the percent of the activity obtained with ␣-methyl D-mannoside as an acceptor which was taken as 100%. Enzyme was assayed as described under "Experimental Procedures." One hundred percent activity corresponds to 10.5 units/mg protein.
b Values obtained with the crude mannoprotein isolated from RA1-1B strain was take as 100%. Enzyme was assayed as described under "Experimental Procedures" except that mannans were used as acceptors. In each assay 0.5 mg (0.142 mM) mannan was used 100% activity corresponds to 1.8 units/mg protein. The molar concentration was calculated on the basis of the average molecular weight (70 kDa) of mannan.  b Enzyme activities were determined under standard assay conditions as described under "Experimental Procedures" using P 3 fraction as enzyme source with or without Triton X-100. 100% activity corresponds to the sum of activities in presence and absence of Triton X-100. transformants carrying the expression plasmid YEpUGAP-gma12 ϩ grown in a galactose medium showed a number of bands (Fig. 2B, lanes 1 and 2). Proteins could not be precipitated when transformant cells containing expression plasmids were grown in a glucose medium or when transformant cells containing control plasmid were grown in a galactose medium (Fig. 2B, lanes 3-5).
Integrity and Topology of Vesicles-The Golgi-rich sealed vesicle fractions were prepared from S. cerevisiae and S. pombe as described under "Experimental Procedures." The integrity and topology of the microsomal vesicles were determined by latency of GDPase, a luminal Golgi activity (34), in the absence of Triton X-100. The results from the GDPase latency assay indicated that at least 95% of the vesicles were sealed and had the identical orientation and topology as in vivo (data not shown for each sample, see Table IIC). Thus, vesicles from both yeast species were sealed with same orientation as in vivo.
Transport of Sugar Nucleotides into Vesicles-To standardize the conditions for sugar nucleotide (GDP-Man and UDP-Gal) transport, it is necessary to characterize the properties of transport system on the basis of dependence for (i) incubation time, (ii) protein concentration, (iii) substrate concentration, and (iv) incubation temperature. The rate of UDP-Gal transport in S. pombe is linear at least up to 8 min and 0.8 mg of protein (Fig. 3). The overall profile of time and temperature dependence of GDP-Man transport in S. pombe is similar to those of UDP-Gal transport (data not shown). In case of S. cerevisiae the rate of UDP-Gal transport is also linear up to 8 min and 0.8 mg of protein (Fig. 4). The profile of incubation time and protein concentration dependence of GDP-Man transport in S. cerevisiae is similar to those of UDP-Gal transport (data not shown).
GDP-Man and UDP-Gal Transport in S. pombe-GDP-Man and UDP-Gal transport is saturable in vesicles derived from S. pombe, as shown in Figs. 5A, and 6A. Although in both cases the transport is saturable, the rate of GDP-Man transport is much higher than that of UDP-Gal under the same substrate concentration. The V max of GDP-Man and UDP-Gal are 420 pmol/mg/5 min and 16 pmol/mg/5 min, respectively. The apparent K m of GDP-Man and UDP-Gal transport are 3.5 M and 6.7 M, respectively. The rate of UDP-Gal transport is also dependent on the incubation temperature (the rate of transport at 4°C is about 42% compared with that obtained at 30°C) and inhibited in the presence of Triton X-100, specific sugar nucleotides and nucleotide di/monophosphate (Table IV).
GDP-Man and UDP-Gal Transport in S. cerevisiae-Both GDP-Man and UDP-Gal transports are also saturable in vesicles derived from S. cerevisiae (Figs. 5B and 6B) and likewise, the rate of GDP-Man transport is much higher than that of UDP-Gal transport. The V max of GDP-Man and UDP-Gal are 123 pmol/mg/5 min and 13 pmol/mg/5 min, respectively. The apparent K m values of GDP-Man and UDP-Gal transport are 2.5 and 5.0 M, respectively. UDP-Gal transport in S. cerevisiae is time and protein concentration dependent (Fig. 4, A and B). The transport rate is temperature dependent (the rate of transport at 4°C is about 40% compared with that obtained at 30°C) and inhibited in the presence of Triton X-100. Transport is inhibited in the presence of specific sugar nucleotide and nucleotide di/monophosphates (Table IV). DISCUSSION Glycosyltransferases are responsible for glycoprotein biosynthesis, by transferring sugar residues from a nucleotide sugar to a growing sugar chain, and are resident membrane proteins of the endoplasmic reticulum and Golgi apparatus. The specificity of these transferases for their donor and acceptor substrates constitutes and determines the primary structures of the sugar chains produced by the cell. The glycosylation pathways in eukaryotes from yeast to mammals have been extensively studied and reviewed (1)(2)(3)(4). It has been reported that mamalian glycosyltransferases can be produced in S. cerevisiae in an active form (35,36), even though they do not have any influence on the host sugar structure. However, recently, the in vivo galactose addition in S. cerevisiae has been demonstrated by expressing human ␤-1,4-GalT in the alg1 mutant (11). In their report some evidences were presented that suggest the presence of UDP-Gal transporter in S. cerevisiae.
These indirect but strong evidences stimulated us as an attractive proposition to check the presence of a UDP-Gal transport system in S. cerevisiae by direct transport assay, because this point is critical when we consider a sugar structure remodeling toward mammalian complex type oligosaccharides by using yeast cells. The alg1 mutant combined with the expression of ␤-1,4-GalT is not suitable as far as to study a stable in vivo galactose addition in S. cerevisiae, because the alg1 mutant offers acceptors for galactose transfer only at the nonpermissive temperature. We have chosen mnn1 mutant of S. cerevisiae combined with the expression of S. pombe ␣-1,2-GalT, which is capable of transferring galactose to a variety of mannan acceptors under the growing conditions (22), because galactose addition sites so far studied (37,38) include terminal Gal-␣-1,2-Man residue, that corresponds to terminal Man-␣-1,3-Man residues in S. cerevisiae. The recombinant ␣-1,2-GalT is enzymatically active and displays similar characteristics for acceptor substrate specificity as reported in S. pombe (Table I) tor than those from wild type cells or mnn2 mutant cells (Table  II), which may be reasonable when considering the structural similarity between proposed galactomannan structures in S. pombe (37,38) and mannan structures in mnn1 mutant (14), both of which contain a common ␣-1,2-linked mannobiose or mannotriose structure that can serve as an acceptor.
The results of this study bear out validity of the hypothesis. We were able to stain 15-20% of mnn1 cells expressing S. pombe ␣-1,2-GalT with FITC-conjugated ␣-galactose-specific lectin GS I-B4, only when the cells were grown in a synthetic medium using galactose as a sole carbon source (Fig. 1). We were unsuccessful to concentrate those stained cells by FACS sorting to increase the percentage of stained cells. Although the sorted positive cells were regrown in a galactose medium followed by lectin staining, the percentage of the stained cells remained unchanged. This partial staining may be partly due to the limitation of acceptor recognition by recombinant ␣-1,2-GalT, because more than one Gal-Ts are present in S. pombe (20), and also partly due to the limitation of accessibility (low affinity) of endogenous acceptor toward ␣-1,2-GalT. However, the latter case seems unlikely because the 15-20% staining is reproducible in any transformants in a growth phase independent manner, suggesting that the putative changes of acceptor mannans during the cell growth may not affect the affinity toward ␣-1,2-GalT in S. cerevisiae. In support of this, the precipitation of galactomannan by GS I-B4-agarose (Fig. 2) demonstrates the in vivo galactose addition not only to the cell surface but also to the total cellular glycoproteins. Similar phenomena of partial cell surface staining (30 -50% of total cells) with FITC-conjugated specific lectins were also observed in Chinese hamster ovary cells transfected with ␣-1,2-fucosyltransferase cDNA (39). At this point it is difficult to speculate the actual limiting factor for the staining without further study.
S. pombe has a physiological demand for a transport of GDP-Man and UDP-Gal in the Golgi vesicles, while the presence of UDP-Gal transporter is quite unexpected for S. cerevisiae because S. cerevisiae does not add any galactose to its proteins. Our present results on UDP-Gal transport in S. cerevisiae, along with the previous report (11), strongly demonstrate the presence of a UDP-Gal transport system in S. cerevisiae. However, recently, inconsistent results have been presented reporting no detectable UDP-Gal transporter activity in S. cerevisiae (40). We believe that their failure to detect any UDP-Gal transporter activity is due to the absence of UMP generating system in their transport assay system. According to the theory (34,41), sugar nucleotide transport is dependent on the antiporter concentration in the lumen of the Golgi. In their case, even though cells were grown in a galactose medium to drive the gene expression under the control of GAL1 promoter that will also provide UDP-Gal in the cytosol (11), no GalTs were expressed to transfer galactose from UDP-Gal to endogenous acceptors and to produce UDP, which should be in turn hydrolyzed to UMP by a UDPase that is known to be present in the lumen of the Golgi of S. cerevisiae (13). In addition, no appropriate S. cerevisiae mutants, which accumulate intermediate glycan chains that may serve as an endogenous acceptor for galactose transfer (11), were used as host cells. Therefore, the lack of UMP may inhibit the total transport activity even in the presence of a functional transporter.
In this report we have provided direct evidence for the presence of a UDP-Gal transport system in S. cerevisiae. The rate of UDP-Gal transport in S. cerevisiae is comparable with that in S. pombe, while the rate of GDP-Man transport in S. pombe is much higher than that in S. cerevisiae. Several lines of evidence have excluded the possibility that the activity of a UDP-Gal transporter in S. cerevisiae may be derived from a nonspecific binding of UDP-[ 3 H]Gal to the surface of Golgi vesicles. If such counts are originated from nonspecific binding, one might expect that there may not be any changes in counts in both incubation time and incubation temperature independent manners, which is absolutely inconsistent with our results (Figs. 3  and 4). The calculated apparent K m for GDP-Man transport in S. cerevisiae is 2.5 M, which is comparable with the published results, the apparent K m was reported 2.0 M and 3.0 M (23, 34). The apparent K m for UDP-Gal transport for both S. pombe and S. cerevisiae are with in the expected values for sugar nucleotide transport (12). The overall profile of UDP-Gal transport is similar in S. pombe and S. cerevisiae (Fig. 6) with little difference in apparent K m , which are 6.7 and 5.0 M, respectively. The inhibition study of UDP-Gal transport by nucleotide di/monophosphates and sugar nucleotides, together with the loss of counts with the treatment of Triton X-100, provides the reliability of our results, indicating that the inhibition by UMP and UDP (36.8 -48.3%) is much higher than that by GDP-Man (17.7%) (Table IV).
Although the rate of UDP-Gal transport in both yeast species are comparable, it is still difficult to speculate the physiological role of UDP-Gal transporter in S. cerevisiae. It remains to be elucidated that this activity may be derived from other sugar nucleotide transporters with a broader substrate specificity, for instance, UDP-glucose transporter, which is necessary for  a One hundred percent activity corresponds to 16 pmol/mg/5 min and 13 pmol/mg/5 min, respectively, for S. pombe and S. cerevisiae. b This is not due to an inhibition, but due to a titration by an equal amount of hot and cold UDP-Gal.
␤-1,6-glucan synthesis along the secretory pathway (42). We made an attempt to throw some light by inhibition study of UDP-Gal transport by cold UDP-Glc, which showed about 21 and 27% inhibition for S. pombe and S. cerevisiae, respectively (Table IV). However, from these data it is difficult to predict something, because a wide variation of inhibition of UDP-GlcNAc transport by different UDP-sugar nucleotide has been reported (43), suggesting all UDP-sugar nucleotides has different affinity for the active site of transporter.