A mutant yeast deficient in Golgi transport of uridine diphosphate N-acetylglucosamine.

Mannan chains of Kluyveromyces lactis mannoproteins are similar to those of Saccharomyces cerevisiae except that they have terminal α12-linked N-acetylglucosamine and lack mannose phosphate. In a previous study, Douglas and Ballou (Douglas, R. K., and Ballou, C. E. (1982) Biochemistry 21, 1561-1570) characterized a mutant, mnn2-2, which lacked terminal N-acetylglucosamine in its mannoproteins. The mutant had normal levels of N-acetylglucosaminyltransferase activity, and the partially purified enzyme from wild-type and mutant cells had the same apparent size, heat stability, affinity for substrates, metal requirement, and subcellular location. No qualitative or quantitative differences were found between mutant and wild-type cells in endogenous mannan acceptors and pools of UDP-GlcNAc. Chitin was synthesized at similar rates in wild-type and mutant cells, and the latter did not have a soluble inhibitor of the N-acetylglucosaminyltransferase or a hexosaminidase that could remove N-acetylglucosamine from mannoproteins. Together, the above observations led Douglas and Ballou ((1982) Biochemistry 21, 1561-1570) to postulate that the mutant might have a defect in compartmentation of substrates involved in the biosynthesis of mannoproteins. We determined whether the above mutant phenotype is the result of defective transport of UDP-GlcNAc into Golgi vesicles from K. lactis. Golgi vesicles which were sealed and of the same membrane topographical orientation as in vivo were isolated from wild-type and mnn2-2 mutant cells and incubated with UDP-GlcNAc in an assay in vitro. The initial rate of transport of UDP-GlcNAc into Golgi vesicles from wild-type cells was temperature dependent, saturable with an apparent K of 5.5 μM and a V of 8.2 pmol/mg of protein/3 min. No transport of UDP-GlcNAc was detected into Golgi vesicles from mutant cells. However, Golgi vesicles from both cells translocated GDP-mannose at comparable velocities, indicating that the above transport defect is specific. In addition to the above defect in mannoproteins, mutant cells were also deficient in the biosynthesis of glucosamine containing lipids.

Mannan chains of Kluyveromyces lactis mannoproteins are similar to those of Saccharomyces cerevisiae except that they have terminal ␣132-linked N-acetylglucosamine units and lack mannose phosphate groups (1). Ballou and co-workers isolated and characterized mutants of K. lactis without terminal N-acetylglucosamine in their mannoproteins (2). One mutant, mnn2-1, lacks the terminal N-acetylglucosaminyltransferase activity while the other mutant, mnn2-2, has wild-type levels of this enzyme. This latter observation prompted Douglas and Ballou (3) to study the biochemical defect underlying this latter phenotype. They found that wild-type and mutant cells had similar pool sizes of UDP-GlcNAc, synthesized chitin at similar rates, and that endogenous mannan acceptors were the same in both cell types. Highly purified preparations of N-acetylglucosaminyltransferase were obtained from both cells and were found to have the same apparent size, heat stability, apparent K m values for substrates, Mn 2ϩ requirements, V max , and subcellular location. No evidence was found for an inactive enzyme precursor in mutant cells nor did these cells contain a soluble inhibitor of the transferase or a hexosaminidase which could remove N-acetylglucosamine from the mannoproteins during their biosynthesis and secretion. These observations led the above authors to postulate, among other possibilities, that the mutant might have a defect in compartmentation of substrates thereby preventing the normal biosynthesis of the mannoproteins.
We determined whether the above mutant phenotype of K. lactis is the result of defective transport of UDP-GlcNAc into Golgi vesicles by comparing the initial rates of transport of UDP-GlcNAc into Golgi vesicles from wild-type and mutant mnn2-2 and mnn2-1 cells. Vesicles from wild-type and mutant mnn2-1 cells were able to transport in vitro UDP-GlcNAc in a temperature-dependent and saturable manner while no transport of UDP-GlcNAc into vesicles from mutant mnn2-2 cells was detected. However, vesicles from all these cells translocated GDP-mannose at a comparable velocity; this indicates that the above defect for UDP-GlcNAc transport is specific. In addition to their previously reported lack of N-acetylglucosamine in outer chain of mannoproteins, mnn2-2 mutant cells were also found deficient in their biosynthesis of glucosamine containing lipids. This phenotype is not secondary to the lack of GlcNAc in glycoproteins since mnn2-1, the glycoprotein GlcNAc-transferase mutant, has the same GlcNAc containing lipids as wild-type. Yeast strains were grown at 30°C in shaker liquid culture consisting of 1% yeast extract, 2% bactopeptone, containing 2% glucose (YPD) unless otherwise specified. Cultures were maintained on solid YPD media containing 2% Bacto-agar.

Radioactive Substrates-UDP-N-acetyl-[6-
GDPase Assay-GDPase was determined essentially as described * This work was supported by National Institutes of Health Grants GM45188 (to P. W. R.) and GM30365 (to C. B. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
previously (4). Briefly, incubation mixtures in a final volume of 0.1 ml, contained 10 -50 g of P 3 fraction enzyme protein (4) or 50 -250 g of total homogenate protein, CaCl 2 (1 mol), Triton X-100 (100 g), GDP or ADP (0.2 mol), and imidazole buffer, pH 7.6 (20 mol). Incubations were done for 5 min at 30°C. The reaction was stopped by adding 10 l of 10% SDS. Inorganic phosphate was determined by the Ames method (5). GDPase activity was calculated as the difference between GDP and ADP hydrolysis. ␣1,2-N-Acetyl-L-glucosaminyltransferase Assay-Transferase activity was measured as described previously (2). In a final reaction volume of 0.2 ml enzyme (0.2 to 1.0 mg/protein of a P 3 fraction or 0.5 to 2 mg of total homogenate protein was incubated for 20 min at 30°C with 0.5 mM man␣1,3-Man␣-O-CH 3 mannose (Sigma) as exogenous acceptor and a final concentration of 10 mM MnCl 2 , 10 mM imidazole HCl buffer, pH 6.5, 0.5% Triton X-100, and 1 M UDP-[ 3 H]GlcNAc (100 mCi/mmol). The reaction was stopped by addition of 0.3 ml of 10 mM EDTA and applied to a Dowex AG 1-X2 column (0.5 ϫ 3 cm). The neutral material was collected in a scintillation vial by washing the column with 0.5 ml of water and subjected to liquid scintillation spectrometry.
Nucleotide Sugar Translocation Assay-The theoretical basis for the translocation assay of nucleotide derivatives into vesicles has been described previously (6). Briefly, it consists of (i) determining the total radioactive solutes associated with the vesicle pellet (S t ) after incubation of vesicles with radioactive substrate and (ii) subtracting the radioactive solutes outside the vesicles in the pellet (S o ). This yields the total radioactive solutes within vesicles (S i ). The (V o ) volume outside the vesicles in the pellet was determined using [ 3 H]acetate as a nonpenetrator and was found to be 1.8 l/mg protein for wild-type as well as for mnn2-2 and mnn2-1 mutant-derived vesicle fractions. Incubations were done in 1 ml of buffer A: 0.25 M sucrose, 20 mM Tris-HCl, pH 7.5, 5 mM MnCl 2 . Centrifugation and counting were carried out as described previously (6).
Subcellular Fractionation-Cells were grown in YPD medium to an OD 600 of 4. The 2-liter culture was chilled, centrifuged, and converted to spheroplasts as described previously (7) using a total of 20 mg of Zymolyase 100T (ICN). This spheroplast suspension was then centrifuged at 450 ϫ g for 10 min. Cells were broken by suspending the pellet in 40 ml of 10 mM triethanolamine, pH 7.2, 0.8 M sorbitol, 1 mM EDTA and drawing the cells rapidly several times into a narrow bore serological pipette. Cell breakage was incomplete but vesicle integrity was well maintained. The suspension was centrifuged successively at 450 ϫ g, 10,000 ϫ g, and 100,000 ϫ g to prepare, respectively, the P 1 , P 2 , and P 3 fractions of Goud et al. (8). The P 3 fraction was enriched in Golgi markers such as GDPase (9, 10) (3.5-fold) and ␣1,2-GlcNAc transferase (4.5-fold) for both wild-type and mnn2-2-derived vesicles and was used for nucleotide sugar transport experiments. The P 3 fraction from mutant mnn2-1 was enriched 3.4-fold in GDPase and had no measurable ␣1,2-GlcNAc transferase activity.
In Vivo Labeling of Radioactive Glucosamine Containing Lipids-For in vivo labeling of glucosamine containing lipids, exponentially growing cultures (OD 600ϭ1 ) in YEP, 2% sucrose, were used. Thirty minutes prior to labeling, 4 OD 600 of cells were harvested and suspended in 2 ml of YEP, 0.5% sucrose. D-[6-3 H]Glucosamine (1 mCi/ml; 60 Ci/mmol; American Radiolabeled Chemicals, Inc., St. Louis, MO) was added for 60 min. Following the incubation, the cell culture was centrifuged, the supernatant removed, and the cells washed four times each with water containing 10 mM sodium azide. Cells, suspended in 0.1 ml of 10 mM aqueous NaN 3 , were broken by vortexing with glass beads, 3 times for 1 min each. 10 l were used to determine protein concentration and total incorporation of radioactive glucosamine. Chloroform/ methanol (1:1) was added to the remaining suspension to achieve a final concentration of chloroform/methanol/aqueous cell suspension of 10: 10:3 (v/v/v). Following centrifugation, the pellet was extracted two more times with 0.6 ml of chloroform/methanol/water (10:10:3) (v/v/v). The pooled lipid extracts were dried under a stream of nitrogen and desalted by butanol extraction (11).
Lipids were analyzed by ascending thin layer chromatography on 0.2-mm Silica Gel 60 plates (EM Science) in a solvent system of chloroform, methanol, 0.2% aqueous KCl (55:45:10). Approximately 40,000 cpm were applied per sample. The developed plates were sprayed with EN 3 HANCE (DuPont NEN) and fluorograms were exposed on Reflection TM film (DuPont NEN) for 3 weeks at Ϫ80°C. To determine sensitivity to mild alkaline hydrolysis, an aliquot of the lipid extract was dried under nitrogen and resuspended in ethanol/water/diethyl ether/ pyridine (15:15:5:1) (v/v/v/v). 100 l of 0.2 N KOH in methanol was added. Following 1 h incubation at room temperature, 20 l of 1 N acetic acid was added to neutralize the reaction. The mixture was evaporated under nitrogen, desalted, and separated on thin layer chromatography as described above.

Differential Translocation of UDP-GlcNAc into Golgi
Vesicles from Wild-type, mnn2-1, and mnn2-2 Mutant K. lactis-Before transport of nucleotide sugars could be measured into Golgi vesicles from the above cells, it was important to determine that these vesicles were sealed, of the same membrane orientation as in vivo, and of comparable purity. A Golgi vesicleenriched fraction was obtained by a procedure similar to that previously described for vesicles from S. cerevisiae. ␣1,2-N-Acetylglucosaminyltransferase activity and guanosine diphosphatase were used to determine their purity, topography, and intactness. As can be seen in Table I, vesicles from wild-type and mutant cells were enriched 4 -5-fold over the homogenate for these two marker enzymes (mutant mnn2-1 had no measurable ␣1,2-GlcNAc transferase activity). In addition, based on the specific activity differences of the GDPase activity with or without Triton X-100, vesicle preparations were found to be 95% latent. Thus, the above vesicles, from wild-type and mutants mnn2-1 and mnn2-2, which were of similar purity, integrity, and membrane topographical orientation as in vivo, could be used for measurements of transport of UDP-GlcNAc and GDP-mannose into a lumenal compartment in an assay in vitro.
As can be seen in Table II, while transport of UDP-GlcNAc into Golgi vesicles from wild-type cells was concentration dependent, transport into Golgi vesicles from mnn2-2 mutants was virtually zero and did not change with a 10-fold increase in nucleotide sugar concentration. Vesicles from the mnn2-1 mutant translocate UDP-GlcNAc at rates comparable to vesicles from wild-type cells.
It was important to determine that the above defect in UDP-GlcNAc transport was specific and not a general defect of the Golgi membrane of mutant mnn2-2 cells. We therefore measured the ability of vesicles to transport GDP-mannose. As can be seen in Table II, this nucleotide sugar was translocated into all vesicles with a similar velocity, demonstrating that the a One unit of activity is defined as 1 mol of inorganic phosphate released per min under standard assay conditions described under "Materials and Methods." b One unit of activity catalyzes the formation of 1 pmol of product found per min under standard assay condition described under "Materials and Methods." above defect of UDP-GlcNAc transport was specific. The V max of GDP-mannose transport into K. lactis vesicles was similar to that previously found with vesicles from S. cerevisiae; this was expected as both yeasts have similar mannan chains.
Transport of UDP-GlcNAc into Golgi Vesicles from Wild-type K. lactis Is Saturable and Temperature Dependent-Because transport of UDP-GlcNAc into Golgi vesicles from yeast has never been previously characterized, it was important to determine whether it had the characteristics of protein carrier mediated transport. Transport was temperature dependent; at 0°C it was 30% of that at 30°C. Transport was linear with time up to 4 min and with protein between 0.7 and 1.9 mg. Transport was found to be saturable with an apparent K m of 5.5 M and a V max of 8.2 pmol/mg/3 min (Fig. 1).
Biosynthesis of Glucosamine Containing Lipids in Wild-type K. lactis and Its Impairment in the Golgi UDP-GlcNAc Transport Deficient Mutant-Previous studies with mammalian and yeast cells have shown that translocation of nucleotide sugars into the Golgi lumen was necessary for glycosylation of proteins, proteoglycans, and glycolipids (10 -16). We therefore wanted to determine whether glycolipid biosynthesis was also impaired in the mnn2-2 mutant. Wild-type and mnn2-2 mutants were grown in the presence of radiolabeled glucosamine; glycolipids were processed as described under "Materials and Methods" and separated by thin layer chromatography. As can be seen in Fig. 2, the GlcNAc radiolabeled glycolipid profile of the mnn2-2 mutant and wild-type cells (lanes 1 and 4) was drastically different: glycolipid species A, B, and C were decreased in the mutant cells while species D was increased. Only species B appeared to be mild base resistant. This suggests that this difference in GlcNAc radiolabeled glycolipids may be due to the inability of mutant cells to transport UDP-GlcNAc into the Golgi lumen to use as substrates in the biosynthesis of these lipids. Further evidence for this was obtained by examining the GlcNAc radiolabeled glycolipid profile of mutant mnn2-1, which has been shown to have a similar phenotype of glycoproteins as mutant mnn2-2 but where the defect has been ascribed to lack of ␣1,2-N-acetylglucosaminyltransferase (3). Because this enzyme is highly specific, we predicted that mutant mnn2-1 would not have a defect in its biosynthesis of glucosamine containing glycolipids. As shown in Fig. 3, this was indeed the case, with the GlcNAc radiolabeled glycolipid profile of wild-type and mnn2-1 mutants being essentially the same (lanes 1 and 4); this also applies for the mild alkaline-  1 and 4) and lipids following mild alkaline hydrolysis (lanes 2 and 3) were separated by thin layer chromatography in chloroform, methanol, 0.22% aqueous KCl (55:45:10) and visualized by fluorography after 27 days.

TABLE II Translocation of UDP-GlcNAc and GDP-mannose into a Golgi-enriched fraction; concentration of radioactive solutes within vesicles after a 3-min incubation UDP-[ 3 H]GlcNAc (600 dpm/pmol) or GDP-[ 3 H
]mannose (3000 dpm/pmol) were incubated for 3 min with a P 3 vesicle fraction (1.3 mg of protein) in 1 ml of buffer A. Translocation was determined as described previously in detail (6). S t , S o , and S i are, respectively, total solutes in the pellet, solutes outside, and solutes inside the vesicles pellet (6). Results are an average Ϯ S.E. of triplicate determinations.  2 and 3). This strongly supports the above hypothesis, namely that the defect in GlcNAc incorporation into glycolipids of the mnn2-2 mutants is only the consequence of deficient UDP-GlcNAc transport into the Golgi lumen and is not secondary to a defect in glycoprotein biosynthesis. DISCUSSION This study demonstrates that the N-acetylglucosamine-deficient phenotype of the mnn2-2 mutant of K. lactis is the result of impaired transport of UDP-GlcNAc into Golgi vesicles. This is the first mutant yeast cell line characterized as deficient in Golgi transport of any solute and differs from known nucleotide sugar transport mutants of mammalian cells since the latter show 5% of Golgi in vitro transport activity compared to their corresponding wild-type vesicles and also have greatly reduced, but not zero, levels of the corresponding sugar in proteins, proteoglycans, and glycolipids. Because both wild-type and mnn2-2-derived Golgi vesicles from K. lactis transport GDPmannose at comparable rates, this study strongly suggests that the defect is specific. The apparent K m for UDP-GlcNAc transport into Golgi vesicles from wild-type K. lactis is comparable to that of other nucleotide sugars such as GDP-mannose into Golgi vesicles from S. cerevisiae (9). Although S. cerevisiae does not have N-acetylglucosamine in its outer chains and therefore would not be expected to require a Golgi UDP-GlcNAc transporter for glycoproteins, mammalian cells have been shown in previous studies in vitro to possess such transport activity both in the Golgi apparatus (17) and the endoplasmic reticulum (17,18). The latter organelle presumably uses UDP-GlcNAc for the biosynthesis O-GlcNAc containing glycoproteins. Other yeast strains, such as Schizosaccharomyces pombe, contain galactose in their outer chains which presumably is transported into the Golgi lumen via UDP-galactose.
Although biochemical and genetic studies with mammalian cells strongly suggest that the different uridine containing nucleotide sugars are transported into the lumen of the Golgi via individual specific transporters (12)(13)(14), it will be important to determine whether there are structural similarities between the UDP-GlcNAc transporter of K. lactis, the presumed UDPgalactose transporter of S. pombe, mammalian UDP-GlcNAc transporters of the Golgi and endoplasmic reticulum membrane, as well as other uridine sugar nucleotide transporters (see below). Mammalian studies suggest that UMP is the antiporter for UDP-GlcNAc and UDP-galactose and that this nucleoside monophosphate is a competitive inhibitor of both transporters suggesting that they both may contain common structural features (19,20).
The antiporter for yeast Golgi transport of GDP-mannose has been characterized as being GMP; this molecule arises by the action of a Golgi lumenal GDPase. Biochemical, cell biological, and genetic studies have shown that the GDPase plays a pivotal role in the entry of GDP-mannose into the Golgi and subsequent mannosylation of proteins and lipids in vivo (10,16,21). Previous studies in vitro (4) have also shown that this GDPase has significant UDPase activity; thus, it is possible that the same GDPase protein may give rise to UMP, the putative antiporter for the UDP-GlcNAc transporter in K. lactis. At this time, it is not clear whether this yeast has a UDPase protein separate from the GDPase or whether the latter protein also mediates the in vivo degradation of UDP to UMP.
The availability of the yeast mutant described here makes this strain an attractive candidate for the cloning of the corresponding transporter. This in turn should allow determination of whether the transporter of UDP-GlcNAc for K. lactis is homologous to other yeast and mammalian transporters.  4) and those following mild alkaline hydrolysis (lanes 2 and 3) were separated as described in the legend of Fig. 2. Visualization by fluorography was after 10 days.