Nucleoside diphosphatase and glycosyltransferase activities can localize to different subcellular compartments in Schizosaccharomyces pombe.

Nucleoside diphosphates generated by glycosyltransferases in the fungal, plant, and mammalian cell secretory pathways are converted into monophosphates to relieve inhibition of the transferring enzymes and provide substrates for antiport transport systems by which the entrance of nucleotide sugars from the cytosol into the secretory pathway lumen is coupled to the exit of nucleoside monophosphates. Analysis of the yeast Schizosaccharomyces pombe genome revealed that it encodes two enzymes with potential nucleoside diphosphatase activity, Spgda1p and Spynd1p. Characterization of the overexpressed enzymes showed that Spgda1p is a GDPase/UDPase, whereas Spynd1p is an apyrase because it hydrolyzed both nucleoside tri and diphosphates. Subcellular fractionation showed that both activities localize to the Golgi. Individual disruption of their encoding genes did not affect cell viability, but disruption of both genes was synthetically lethal. Disruption of Spgda1+ did not affect Golgi N- or O-glycosylation, whereas disruption of Spynd1+ affected Golgi N-mannosylation but not O-mannosylation. Although no nucleoside diphosphatase activity was detected in the endoplasmic reticulum (ER), N-glycosylation mediated by the UDP-Glc:glycoprotein glucosyltransferase (GT) was not severely impaired in mutants because first, no ER accumulation of misfolded glycoproteins occurred as revealed by the absence of induction of BiP mRNA, and second, in vivo GT-dependent glucosylation monitored by incorporation of labeled Glc into folding glycoproteins showed a partial (35-50%) decrease in Spgda1 but was not affected in Spynd1 mutants. Results show that, contrary to what has been assumed to date for eukaryotic cells, in S. pombe nucleoside diphosphatase and glycosyltransferase activities can localize to different subcellular compartments. It is tentatively suggested that ER-Golgi vesicle transport might be involved in nucleoside diphosphate hydrolysis.

Almost all nucleotide sugar-dependent glycosyltransferases in the secretory pathway generate nucleoside diphosphates that are converted into monophosphates to relieve inhibition of the transferring enzymes and provide substrates for antiport transport systems by which the entrance of nucleotide sugars from the cytosol into the lumen of the secretory pathway is coupled to the exit of nucleoside monophosphates (1). Several secretory pathway nucleoside diphosphatases have been described already. There are two enzymes displaying such enzymatic activity in Saccharomyces cerevisiae, a GDPase/UDPase and an apyrase (denominated Gda1p and Ynd1p, respectively) (2)(3)(4). Apyrases differ from nucleoside diphosphatases, as the former are able to degrade not only nucleoside diphosphates but also triphosphates. Both enzymatic activities are localized to the Golgi apparatus. Irrespective of their cellular origin, all enzymes able to hydrolyze nucleoside diphosphates (nucleoside diphosphatases proper and apyrases) share four highly similar sequences that have been called apyrase conserved regions. Analysis of the S. cerevisiae genome showed that Gda1p and Ynd1p were the only enzymes displaying those sequences (4). Three nucleoside diphosphatases have been described to date in the mammalian cell secretory pathway, i.e. a Golgi apyrase and two (soluble and membrane bound) endoplasmic reticulum (ER) 1 GDPase/UDPases (5)(6)(7). The ER soluble enzyme is functional with Ca 2ϩ , Mg 2ϩ , or Mn 2ϩ , whereas the insoluble enzyme strictly requires the first cation for activity.
The initial in vitro characterization of nucleoside monophosphates as the metabolites involved in the antiport transport of nucleotide sugars (8) received firm support when it was shown that ablation of the GDA1 or YND1 genes in S. cerevisiae affected both N-and O-protein glycosylation and glycolipid formation (3,9). In the case of GDA1 disruption, it was shown that the effects observed were a consequence of a reduced entrance rate of GDP-Man into the Golgi lumen (10). Although no evidence was provided, it was assumed that mammalian cell ER nucleoside diphosphatases were required for hydrolysis of UDP generated by the UDP-Glc:glycoprotein glucosyltransferase (GT), an enzyme involved in the quality control of glycoprotein folding (11). In addition to this ubiquitous enzyme (S. cerevisiae is the only eukaryotic cell known so far to be devoid of GT; Refs. 12 and 13), the other nucleotide sugar-dependent glycosyltransferases already described in the ER lumen are the glucuronosyltransferases that also generate UDP but only occur in certain high eukaryote tissues (14). UDP-Glc-and UDP-GlcUA-specific transporters have already been described in the rat liver cell ER (15,16). In the case of UDP-Glc, it was determined that the exit of UMP was coupled to entrance of the nucleotide sugar (15).
The presence of nucleoside diphosphatase and nucleotide sugar-dependent glycosyltransferase activities in the mammalian cell ER and Golgi compartments and the exclusive presence of the former activity in the S. cerevisiae Golgi, that is, in the only secretory pathway compartment in which nucleoside diphosphates are known to occur in this yeast, supported the notion that the presence of glycosyltransferase-generated nucleoside diphosphates in a subcellular compartment necessarily implied the presence of an enzyme able to hydrolyze them in the same compartment. To further study the influence of ER nucleoside diphosphatase on GT-mediated glycoprotein folding, we chose Schizosaccharomyces pombe as model system for the ease by which this microorganism can be genetically manipulated and the well documented presence in it of an ER GT involved the quality control of glycoprotein folding (11). On the characterization of S. pombe nucleoside diphosphatases we made the unexpected observation that this yeast, like S. cerevisiae, only expresses two Golgi-located nucleoside diphosphatase activities, thus demonstrating that, contrary to what has been assumed so far, nucleoside diphosphatase and glycosyltransferase activities can localize to different subcellular compartments.

EXPERIMENTAL PROCEDURES
Materials-[ 14 C]glucose (301 Ci/mol) was from New England Nuclear. Nucleoside tri, di, and monophosphates, protease inhibitors, supplements for culture media, lysing enzyme, 1-deoxynojirimycin, mannan, and endo-␤-N-acetylglucosaminidase H (Endo H) were from Sigma. Restriction enzymes and other enzymes used for DNA procedures were from New England Biolabs. Vector pGEMT-Easy was from Promega. The Pwo polymerase and digoxigenin (DIG DNA labeling and detection kit) were from Roche Applied Science. Zymolyase 100T was from Seikagaku Kogyo Co.
N-Glycan Labeling-For assessing the Golgi extension, cells in the exponential growth phase were harvested, extensively washed with 1% yeast nitrogen base (Difco), resuspended in the same medium, and incubated (0.3 g in 1 ml) for 30 min at 28°C in the presence of 5 mM [ 14 C]glucose (30.1 Ci/mol). After chasing cells with 50 mM Glc (final concentration), cells were incubated for an additional 60 min. For assessing ER N-glycosylation, incubations lasted for only 15 min and were not chased. Where indicated, 5 mM 1-deoxynojirimycin or the same plus 5 mM dithiothreitol were added to incubations; the first reagent for 30 min and the second one for 5 min before the label. For further details on the labeling procedure and preparation of whole cell Endo H-sensitive N-glycans, see Ref. 12.
O-Glycan Labeling-Cells, extensively washed previously with 1% yeast nitrogen base, were resuspended in the same medium (0.16 g in 1 ml) and incubated for 60 min in the presence of 5 mM [ 14 C]Glc (30.1 Ci/mol), chased with 50 mM Glc (final concentration), and incubated for an additional 60 min. Cells were then processed as above for obtaining N-glycans, but the denatured cell pellets obtained after organic solventwater treatments were incubated with 0.4 ml of 0.1 N NaOH overnight at room temperature. Solutions were neutralized with 0.1 N HCl, and 0.3 ml of water was then added. The supernatants obtained on the addition of 2 ml of methanol were dried, resuspended in water, and desalted with a mixed bead (H ϩ /acetate) resin.
Cloning and Expression of Spgda1p-The corresponding gene of Spgda1p (GenBank TM protein accession number NP_593447) (1671 bp) was synthesized using genomic DNA as template, Pwo polymerase, and primers 5Ј-ACTGGAAGATCTATGACTCCAACTATGAAATC-3Ј and 5Ј-ACGATCCGCTCGAGAATTTCTTCTTTCACGTTAC-3Ј. The gene was then digested with BglII and XhoI and cloned into the BglII/XhoI sites of expression vector p416-GPD, which confers uracil prototrophy. The plasmid was amplified in E. coli, and colonies were analyzed with Spgda1 ϩ internal primers 5Ј-AAATTTCATATGATAGAGCCTGGCT-TATC-3Ј (primer UDPaseM1) and 5Ј-CCAAACACCATCCAAGTTCC-3Ј (primer 824a1). Plasmids were purified and analyzed by SpeI and XhoI digestion. Plasmids were then transfected into S. cerevisiae by electroporation (2.5 kV, 25 F, 200 ohms). Cells were grown for 1 h in rich medium and plated in selective medium. Colonies were then analyzed by PCR using the above mentioned internal primers.
Characterization of Spgda1 Mutants-Mutant genomic DNA was digested with XmnI and SphI, run on 1.2% agarose gels, and submitted to successive Southern blotting analysis using first a 740-bp SacII/ HindIII Spgda1 ϩ fragment derived from the pGEMT-Easy vector containing the 1196-bp fragment and then the entire ura4 ϩ gene as probes. Both probes were labeled with digoxigenin. The expected fragments of 1274-and 3038-bp for wild type and mutant cells were obtained.
Characterization of Spynd1 Mutants-Colony PCR reactions were performed on water-resuspended cells using a primer homologous to a portion of Spynd1 ϩ outside the fragment used for gene disruption (primer c11E10a-EcoRI) and a primer specific for ura4 ϩ (5Ј-TGCTC-CTACAACATTACC-3Ј). Positive colonies displaying the expected 838-bp band were submitted to Southern blotting analysis. For this purpose, genomic DNA was digested with MspI and run on a 1.2% agarose gel. Probes successively used were, first, a 424-bp fragment obtained by NcoI and HindII digestion of plasmid pGEMT-Easy containing the 1309-bp fragment and then the entire ura4 ϩ gene. The probes were labeled with [ 32 P]dCTP by the Random Primer (Promega) method. The expected and detected fragments contained 514 and 2377 bp in the wild type and the Spynd1 mutant cell genomes, respectively.
Enzymatic Assays-S. pombe and S. cerevisiae microsomes were prepared as described previously (12). Nucleoside phosphatase activities were essentially assayed as described with slight modifications (21). Briefly, between 10 and 50 g of membrane proteins were incubated in a total volume of 100 l in 0.2 M imidazole buffer, pH 7.2, 0.1% digitonin, 10 mM CaCl 2 , 2 mM the corresponding nucleoside phosphate, and incubated for 5-10 min at 30°C. Reactions were stopped on the addition of 100 l of 10% SDS and 100 l of water. Liberated phosphate was assayed as described (22), employing 15-min incubations at 45°C. Phosphates present either in microsomes and reagents or liberated during incubations at 45°C were estimated for each tube with blanks in which the SDS was added before the membrane fractions. Glucosidase II was assayed using [glucose-14 C]Glc 1 Man 9 GlcNAc 1 (23) or p-nitrophenyglucoside (24) as substrates, depending on whether microsomes or sucrose gradient fractions were the enzyme source. Galactosyltransferase was assayed in a total volume of 50 l containing 0.1 M HEPES buffer, pH 7.2, 1.2 mM MnCl 2 , 0.2% Triton X-100, UDP-[ 14 C]Gal (300 Ci/mol), 2 mg of S. cerevisiae mannan, and 10 l of gradient fraction. After 60 min at 30°C, a label soluble in 10% trichloroacetic acid but insoluble in 66% methanol containing 0.1 M LiCl was quantified. Values obtained on incubations in which mannan was added after stopping the reactions were subtracted. GT was assayed with denatured thyroglobulin as the acceptor substrate as described previously (25).
Subcellular Fractionation-Subcellular fractionation was performed as described (26,27) with some modifications. Mutant cells (ϳ4 g) grown in rich medium to an OD 620 nm of 2 were harvested, washed with 10 mM sodium azide, and resuspended in 50 mM Tris-HCl buffer, pH 7.5, 1.4 M sorbitol, 10 mM sodium azide, and 40 mM 2-mercaptoethanol. Lysing enzyme and zymolyase 100T were then added to a final concentration of 0.5 mg/ml each. Conversion into spheroplasts was performed at 37°C and was monitored by the decrease in absorbance at 600 nm after a one-hundredth dilution in water. The reaction was stopped by the addition of 1 mM EDTA after ϳ75% of cells had been converted into spheroplasts. The latter were purified by a 12-min centrifugation through a 1.8 M sorbitol cushion. The spheroplasts were resuspended in a hypoosmotic buffer (15 mM triethanolamine-acetate buffer, pH 7.2, 0.3 M sorbitol, and 1 mM EDTA containing 10 M leupeptin, 1 M pepstatin, 1 mM phenylmethylsulfonyl fluoride, 1 mM tosylphenylalanyl chloromethyl ketone, 1 M E-64, and 0.5 mM tosyl-lysine chloromethyl ketone), lysed by several passages through a serological pipette, and homogenized with 15 strokes in a Wheaton B (Wheaton Scientific) homogenizer. Non-lysed cells were removed by centrifugation (450 ϫ g for 3 min). Each of two 1.5-ml aliquots of the post-nuclear supernatant (S 450 ) was poured onto a gradient having 1 ml of 14,18,22,26,30,34,38,40,42,46, and 50% sucrose concentrations (for the Spgda1 mutant the 42% sucrose fraction was omitted) in 15 mM triethanolamine-acetate buffer, pH 7.2, 1 mM EDTA, and 3 mM MgCl 2 . Tubes were centrifuged for 2.5 h at 185,000 ϫ g in a swinging bucket rotor. Fractions were collected from the top (the first fraction had 1.5 ml, and successive ones had 1 ml). Materials in the first fractions and pellets were discarded. Sucrose concentrations were determined with a refractometer, and fractions having similar concentrations in both gradients were pooled, 5-fold diluted with 20 mM imidazole buffer, pH 7.5, 0.25 M sucrose, and 1 mM EDTA, and centrifuged for 1 h at 100,000 ϫ g. The pellets were resuspended in 0.2 ml of the last buffer containing the above indicated protease inhibitor concentrations. Supernatants (S 450 ) were also centrifuged for 1 h at 100,000 ϫ g and resuspended as the gradient fraction pellets to determine total enzymatic activities. The latter as well as the protein concentrations were determined in all fractions.

RESULTS
Analysis of S. pombe Genome-BLAST analysis of proteins encoded in the recently completed S. pombe genome, using as probes sequences of S. cerevisiae Gda1p or Ynd1p, mammalian soluble ER GDPase/UDPase, mammalian membrane-bound ER Ca 2ϩ -dependent GDPase/UDPase, human Golgi apyrase (respective GenBank TM accession numbers NP_010872, AAF17573, CAB45533, CAC85467 and AAC17217), or the protein portions encompassing the four so-called apyrase conserved regions in the three first enzymes, showed that the fission yeast genome codes for the two proteins with putative nucleoside diphosphatase activity (GenBank TM accession numbers NP_593447 and NP_588201) were the same as that of S. cerevisiae. The first sequence corresponds to a 556-amino acid protein with a molecular mass of 61.6 kDa. The protein displays a signal peptide, one potential N-glycosylation consensus sequence, the four canonical apyrase conserved regions and, as with S. cerevisiae Gda1p, no other potential transmembrane region besides the signal peptide. The second sequence corresponds to a 572-amino acid protein with a molecular mass of 64.6 kDa. This protein does not display a signal peptide but does have a potential transmembrane region close to the C terminus, thus suggesting a type I membrane protein. These same features are shared by S. cerevisiae Ynd1p. The second putative S. pombe nucleoside diphosphatase displays the four apyrase conserved regions and three potential N-glycosylation consensus sequences. Proteins coded by sequences of NP_593447 and NP_588201 will be referred to as Spgda1p and Spynd1p, respectively, because of their respective similarity to the S. cerevisiae enzymes. Percent similarities of the above mentioned proteins are shown in Table I.
Characterization of Putative S. pombe Nucleoside Diphosphatase Activities-Genes coding for Spgda1p and Spynd1p were transfected and overexpressed in S. cerevisiae ⌬gda1 mutants. Figs. 1, A and B and 2, A and B show the nucleoside diphosphate specificity and cation requirements of Spgda1p and Spynd1p. The first enzyme is essentially a GDPase/ UDPase, whereas the second one is an apyrase as it hydrolyzes not only nucleoside diphosphates but also triphosphates. In addition, no marked specificity toward the base was observed in Spynd1p. Both enzymes required a bivalent cation for activity (Ca 2ϩ , Mg 2ϩ , or Mn 2ϩ ). Results show that Spgda1p and Spynd1p closely resemble S. cerevisiae Gda1p and Ynd1p, respectively.
Disruption of Spgda1p-and Spynd1p-encoding Genes-Genes were disrupted as indicated under "Experimental Pro- cedures." Single mutants did not show any morphological or growth phenotype. Fig. 3A shows nucleoside diphosphatase activities found in microsomes isolated from wild type, Spgda1, and Spynd1 mutant cells. From results shown in Figs. 1, A and B and 3A, it may be concluded that Spynd1p is the sole enzyme able to hydrolyze ADP and that the activity of Spgda1p toward UDP and GDP in wild type cells is 6 -7 fold higher than that of Spynd1p. In vitro control assays performed on a Golgi (galactosyltransferase; Ref. 28) and two ER (glucosidase II and GT; Ref. 11) enzymes showed that their activities were not affected in Spgda1 and Spynd1 mutant cells (Fig. 3B).
Disruption of Spgda1 ϩ and Spynd1 ϩ Is Synthetically Lethal-Colonies yielded by 24 tetrads obtained on sporulation of Spgda1⁄Spynd1 diploids were analyzed by colony PCR using primers that yield different length fragments for Spgda1 ϩ , Spgda1::ura4 ϩ , Spynd1 ϩ , and Spynd1::ura4 ϩ , as well as by the ability of individual spores to grow in the absence of uracil (the original strains were ura4 Ϫ , and the ura4 ϩ gene was used to disrupt both Spgda1 ϩ and Spynd1 ϩ ). Seven tetrads were parental ditype, five were non-parental ditype, and twelve were tetratype. Only 74 spores generated colonies, and the missing 22 corresponded precisely to all of the double mutants that were expected to be found. Observation under the microscope showed that, in all cases, non-growing spores had germinated and duplicated 1-4 times. As attempts to grow those cells in rich liquid medium failed, it may be concluded that disruption of Spgda1 ϩ and Spynd1 ϩ is synthetically lethal for vegetative growth. This result is at variance with what was observed in S. cerevisiae (3). ⌬gda1/⌬ynd1 double mutants were viable, although they showed severe growth and cell wall defects.
Subcellular Localization of Spgda1p and Spynd1p-Sucrose gradient centrifugation of post-nuclear supernatants isolated from Spgda1 and Spynd1 mutants showed that, like their homologs in S. cerevisiae, both enzymes localize to the Golgi (Figs. 4, A and B). It is worth remarking that, even in mutants lacking the main nucleoside diphosphatase activity (Spgda1p), no UDPase peak was detected in the ER (Fig. 4B). The markers used were enzymes related to glycoprotein processing (galactosyltransferase for Golgi and glucosidase II for the ER; localization of the latter agreed with that of NADPH-cytochrome c reductase; Ref. 27).
Effect of Spgda1 ϩ and Spynd1 ϩ Disruption on Golgi N-and O-Glycosylation-Golgi N-glycosylation in S. pombe involves the addition of Man and Gal residues to the Man 9 GlcNAc 2 core transferred to proteins in the ER (29). To study the effect of the absence of either one of the nucleoside diphosphatase activities on Golgi N-glycosylation, wild type, Spgda1, and Spynd1 mutant cells were incubated with 5 mM [ 14 C]Glc for 30 min. This was followed by further incubation with 50 mM Glc for 60 min.
The patterns of glycans released from whole cell glycoproteins by Endo H are shown in Figs. 5, A-C. Whereas the ablation of Spgda1 ϩ did not affect Golgi extension, that of Spynd1 ϩ diminished formation of the larger compounds. Percentages of label in specific ER N-glycans (Man 9 GlcNAc 2 and Man 8 GlcNAc 2 ) were 34, 33, and 53% in wild type, Spgda1, and Spynd1 cells, respectively, the rest corresponding to Golgi-elongated glycans. Evidently, more Man 9 GlcNAc 2 and Man 8 GlcNAc 2 species traversed the Golgi without being elongated in Spynd1 cells than in the other two strains. Because different aliquot volumes of samples were run on Fig. 5, A-C, total amounts of N-glycans synthesized in different strains may not be estimated from total label values in runs shown in those figures, although for each sample a valid calculation of percentages of ER-and Golgi specific N-glycans can be performed. The fact that ablation of the main Golgi GDPase/UDPase activity-encoding gene did not affect Golgi elongation, whereas absence of the minor activity had a limited but detectable effect, is puzzling but not surprising. For instance, disruption of GDA1 in S. cerevisiae hardly affected invertase Golgi N-glycosylation, but it had profound effects in the elongation of carboxypeptidase Y N-glycans and almost totally inhibited formation of mannosylated ceramides (9). The differential effects observed in S. pombe might be related to possible different locations of Spgda1p and Spynd1p within Golgi cisternae.
Golgi O-glycosylation in this yeast also involves the addition of Man and Gal residues, but to the Man unit added to proteins in the ER (30). Wild type and mutant cells were incubated with labeled Glc and chased with the unlabeled monosaccharide, and O-linked glycans were released from whole cell glycoproteins by a mild alkaline treatment. The patterns of oligosaccharides obtained are depicted in Fig. 6, A-C. Ablation of either one of both nucleoside diphosphatase genes did not affect Golgi O-extension of glycans. The null or very limited effects of disruption of only one of both nucleoside diphosphatase genes on Golgi N-and O-glycosylation agree with their similar subcellular localizations.
Effect of Spgda1 ϩ and Spynd1 ϩ Disruption on ER Nucleotide Sugar-dependent N-Glycosylation-As mentioned above, GTmediated N-glycosylation is the only nucleotide sugar-dependent reaction described to date in the S. pombe ER lumen. This reaction is not essential for cell viability under normal growth conditions (31). Moreover, complete or severe reduction in monoglucosylated glycan formation caused by disruption of glucosidase II subunit ␣or ␤or GT-encoding genes did not affect cell growth or morphology. Nevertheless, hindering the formation of monoglucosylated glycans led to ER accumulation of misfolded glycoproteins, as revealed by the induction of BiP-encoding mRNA. Such induction is particularly easy to assess in S. pombe because, whereas under normal conditions only one messenger is formed, the accumulation of misfolded proteins leads to the synthesis of two BiP mRNAs (32). As shown by Northern blotting analysis (Fig. 7, lanes 1-3), only one BiP mRNA was synthesized in wild type, Spgda1, and Spynd1 mutant cells, thus indicating that the formation of monoglucosylated glycoproteins was not severely impaired upon disruption of either one of the nucleoside diphosphataseencoding genes. On the contrary, a gpt1 mutant cell (GT null) yielded the expected two-band pattern (Fig. 7, lane 4).
To confirm that GT-dependent glucosylation was indeed not severely impaired in Spgda1 and Spynd1 cells, double mutants Spgda1⁄alg6 and Spynd1⁄alg6 were constructed. Alg6 ϩ codes for the dolichol-P-Glc-dependent glucosyltransferase that is responsible for the addition of the first Glc unit to Man 9 GlcNAc 2 -P-Pdolichol. Alg6 mutants, therefore, transfer Man 9 GlcNAc 2 to protein, and the formation of monoglucosylated glycans in them is FIG. 7. Northern blotting analysis of BiP mRNA in wild type and mutants cells. Cells were grown in rich medium at 28°C, and RNAs were submitted to Northern blotting analysis. Lanes 1, 2, 3, and 4 correspond to wild type and Spgda1, Spynd1, and gpt1 mutant cells, respectively. The latter are GT null mutants. exclusively mediated by GT activity. The above mentioned double as well as single alg6 mutant cells were incubated for 15 min with 5 mM [ 14 C]Glc in the presence of deoxynojirmycin, a glucosidase II inhibitor. Under these short incubation and low Glc concentration conditions, there is minimal Golgi enlargement of N-glycans. Compounds migrating as Man 8 GlcNAc, Man 9 GlcNAc, Glc 1 Man 9 GlcNAc, and Man 10 GlcNAc appeared in the chromatogram, but patterns of N-oligosaccharides obtained from whole cell glycoproteins were very similar for the three cell types (Fig.  8, A-C). Strong acid hydrolysis of the compounds formed yielded labeled Man, Glc, and Gal units, but whereas the Glc/Man ratio was identical in compounds formed in alg6 and Spynd1⁄/alg6 cells, glycans synthesized in Spgda1⁄alg6 mutants yielded a lower ratio, thus indicating a partially impaired GT-mediated glucosylation in those cells (Fig. 8, D-F). As almost all of the Man units in the compounds formed had been transferred to proteins from a dolichol-P-P derivative, i.e. through a pathway not involving lumenal nucleotide sugar-dependent glycosyltransferases, nucleotide sugar transporters, and nucleoside diphosphatases, the Glc/Man ratio provides a reliable indication of GT-mediated glucosylation levels in intact simple (alg6) and double (Spgda1⁄alg6 and Spynd1⁄alg6) mutant cells.
To confirm that GT-mediated glucosylation was indeed diminished in Spgda1 cells, the cells were incubated as above, but in the presence of dithiothreitol. This reagent prevents proper folding of most glycoproteins by hindering disulfide bond formation and, therefore, their ER exit to the Golgi also (33). As depicted in Fig. 9, A and B, a lower proportion of Glc 1 Man 9 GlcNAc was formed in Spgda1⁄alg6 than in alg6 cells. Strong acid hydrolysis of compounds formed showed, as expected, a lower Glc/Man ratio in glycans synthesized in Spgda1⁄alg6 cells (Fig. 9, C and D). It may be concluded, there- fore, that whereas absence of Spynd1p did not affect GT-mediated glucosylation, deficiency of Spgda1p partially (35-50%) reduced such reaction in vivo. DISCUSSION Like S. cerevisiae, S. pombe expresses two enzymes capable of nucleoside diphosphate hydrolysis, Spgda1p and Spynd1p, the former being a nucleoside diphosphatase proper and the latter an apyrase. As mentioned above, a comprehensive search in the fission yeast genome did not reveal sequences coding for additional enzymes catalyzing similar reactions. Like S. cerevisiae Gda1p and Ynd1p, both S. pombe enzymes localized to the Golgi, as shown by sucrose gradient centrifugation of postnuclear supernatants. Furthermore, this technique confirmed the absence of a third ER nucleoside diphosphatase. The Golgi localization of both Spgda1p and Spynd1p agreed with the occurrence of Golgi N-and O-glycosylation reactions in cells in which either one of the encoding genes had been disrupted. Results obtained showed that one enzyme took the place of the other in the hydrolysis of nucleoside diphosphates required for relieving glycosyltransferase inhibition and providing nucleoside monophosphates for the antiport mechanism responsible for nucleotide sugar entrance into the Golgi lumen. Only a decrease in the formation of the larger Golgi N-glycans was observed in Spynd1 mutants. The redundant function of S. cerevisiae Gda1p and Ynda1p in the hydrolysis of nucleoside diphosphates generated by Golgi N-and O-glycosylation reactions has been observed previously (3).
The most puzzling result was that, contrary to what happens in mammalian cells, there was no nucleoside diphosphatase activity in S. pombe ER. The fission yeast, the same as all other eukaryotic cells with the (to date) only known exception of S. cerevisiae (12,13), expresses an ER enzyme (GT) that is involved in the quality control of glycoprotein folding and generates UDP. As mentioned above, this enzyme together with liver glucuronosyltransferase are the only fully described nucleotide sugar-dependent glycosyltransferases occurring in the ER lumen. GT is a soluble protein that localizes to the ER and the ER-Golgi intermediate compartment. Like the other components of the quality control mechanism (glucosidase II, calnexin, and calreticulin), it displays an ER retrieval sequence at its C terminus (11). It may be assumed, therefore, that it is continuously transported between the ER and cis-Golgi cisternae by COPII and COPI vesicles. In fact, not only GT but also glucosidase II and calreticulin have been detected in the mammalian cell ER-Golgi intermediate compartment by quantitative immunogold electron microscopy (34).
How is GT-generated UDP hydrolyzed in S. pombe? Ablation of either Spgda1 ϩ or Spynd1 ϩ did not elicit induction of BiP mRNA, as happened in mutants in which the formation of monoglucosylated N-glycans was severely or totally impaired (gls2␣, gls2␤, and gpt1 mutants, respectively, devoid of glucosidase II ␣ or ␤ subunits or GT; Ref. 19). Moreover, an in vivo assay of GT-mediated glycoprotein glucosylation revealed that the reaction was only partially affected in Spgda1 and not affected at all in Spynd1 mutants. A possibility could be that, in the particular case of the S. pombe ER UDP-Glc transporter, UDP and not UMP is the metabolite involved in the nucleotide sugar transport and, therefore, no UDP hydrolysis would be required. This possibility seems highly unlikely, however, because no such antiport mechanism has ever been described for nucleotide sugar entry into the secretory pathway in any of the cells (fungal, plant, or mammalian) studied to date, and it would not account for the effect of Spgda1 ϩ disruption on GT-mediated glucosylation. Concerning the exit of nucleoside diphosphates from the ER, our results show that probably ADP and not AMP is the still unidentified antiporter in the entrance of ATP into the ER lumen (1,35), because ablation of the only gene coding for an ADPase activity in the S. pombe secretory pathway (Spynd1p) not only did not affect cell viability but also did not elicit induction of BiP mRNA. It would have been expected that, if AMP was the antiporter, a lower rate of ATP entrance into the ER lumen caused by nucleoside diphosphatase deficiency would certainly affect molecular chaperoneassisted protein folding.
A more attractive possibility is that vesicular transport between the ER and cis Golgi cisternae might carry not only the macromolecular components of the quality control mechanism (S. pombe lacks calreticulin but expresses calnexin; Ref. 36) but also metabolites like UDP in the anterograde movement to be hydrolyzed by Spgda1p and/or Spynd1p and UMP in the retrograde transport. This possibility is supported by the partial decrease in the glucosylation of folding intermediates observed in Spgda1 mutants. The involvement of vesicular ER-Golgi traffic in UDP disposal could be tested if the conditional mutants affected in such traffic were available. Unfortunately, although those mutants have been extensively studied in S. cerevisiae, no similar S. pombe mutants have been characterized to date. Future work will be directed to their production and characterization to be used as tools for testing the postulated role of ER-Golgi transport in GT-generated UDP hydrolysis. An additional experiment to be performed would be to assay in vivo GT-mediated glucosylation in Spgda1 or Spynd1 mutant cells expressing the other nucleoside diphosphatase under the control of a repressible promoter. If as proposed, Spgda1p and Spynd1p were indeed involved in GT-generated UDP hydrolysis, then no GT activity would be detected under conditions of total repression. However, the best of such promoters available for S. pombe (nmt, for no message in thiamine) is apparently leaky, as we have been unable to completely shut off GT expression when encoded in plasmids containing that promoter transfected into GT null mutants. As the amount of nucleoside diphosphatase allowing ER glucosylation of folding intermediates is presently unknown, ambiguous results would be obtained.
Finally, a third possibility to account for the observed lack of UDPase activity in S. pombe ER lumen could be the entrance of UDP-Glc into the Golgi by a "classical" transporter followed by vesicular retrograde transport of the nucleotide sugar to the ER lumen and anterograde transport of GT-generated UDP to the Golgi to be hydrolyzed by Spgda1p and/or Spynd1p. This last possibility seems highly unlikely, as we have detected substantial labeling of endogenous glycoproteins on the incubation of isolated intact microsomal vesicles with UDP-[ 14 C]Glc, i.e. under conditions in which ER-Golgi transport is most probably not functional due to the absence of energy sources (ATP, GTP) and cytosolic proteins. 2 Results presented show, therefore, that contrary to what has been assumed to date, nucleoside diphosphatase and glycosyltransferase activities do not necessarily localize to the same subcellular compartments in all eukaryotic cells. It is tentatively suggested that vesicle ER-Golgi transport might be involved in the hydrolysis of ER-generated nucleoside diphosphates.