Involvement of Protein N-Glycosyl Chain Glucosylation and Processing in the Biosynthesis of Cell Wall ß-1,6-Glucan of Saccharomyces cerevisiae

Corresponding author: Howard Bussey, Department of Biology, McGill University, 1205 Dr. Penfield Avenue, Montréal, QC, Canada H3A 1B1, hbussey{at}monod.biol.mcgill.ca (E-mail).

Communicating editor: M. JOHNSTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ß-1,6-Glucan plays a key structural role in the yeast cell wall. Of the genes involved in its biosynthesis, the activity of Cwh41p is known, i.e., the glucosidase I enzyme of protein N-chain glucose processing. We therefore examined the effects of N-chain glucosylation and processing mutants on ß-1,6-glucan biosynthesis and show that incomplete N-chain glucose processing results in a loss of ß-1,6-glucan, demonstrating a relationship between N-chain glucosylation/processing and ß-1,6-glucan biosynthesis. To explore the involvement of other N-chain-dependent events with ß-1,6-glucan synthesis, we investigated the Saccharomyces cerevisiae KRE5 and CNE1 genes, which encode homologs of the "quality control" components UDP-Glc:glycoprotein glucosyltransferase and calnexin, respectively. We show that the essential activity of Kre5p is separate from its possible role as a UDP-Glc:glycoprotein glucosyltransferase. We also observe a ~30% decrease in ß-1,6-glucan upon disruption of the CNE1 gene, a phenotype that is additive with other ß-1,6-glucan synthetic mutants. Analysis of the cell wall anchorage of the mannoprotein -agglutinin suggests the existence of two ß-1,6-glucan biosynthetic pathways, one N-chain dependent, the other involving protein glycosylphosphatidylinositol modification.

THE cell wall of the yeast Saccharomyces cerevisiae is a complex structure containing mannoproteins, ß-1,3-glucan, ß-1,6-glucan and chitin, all of which are covalently interconnected by a variety of linkages (reviewed in KLIS 1994 ; CID et al. 1995 ; ORLEAN 1997 ). The ß-1,3-glucan consists of linear polymers with an average length of 1500 residues (MANNERS et al. 1973A ) which is connected via branching to ß-1,6-glucan (KOLLAR et al. 1997 ) and chitin (KOLLAR et al. 1995 ). ß-1,6-Glucan has been shown to be a polymer with an average size ranging from 140 units (MANNERS et al. 1973B ) to 350 units (KOLLAR et al. 1997 ), and is connected to ß-1,3-glucan, chitin and wall mannoproteins (LU et al. 1995 ; KAPTEYN et al. 1996 ; KOLLAR et al. 1997 ). It appears that a key function of ß-1,6-glucan is the covalent anchoring of mannoproteins to the cell wall. Few cell wall proteins have been definitively identified; however, those that are known share a carboxyterminal consensus for the attachment of a glycosyl-phosphatidylinositol (GPI) lipid anchor in the endoplasmic reticulum (ER) (see CARO et al. 1997 , and references therein). This observation implicates GPI modification as a critical step in cell wall biosynthesis, a hypothesis supported by the lethality of GPI biosynthetic mutations in S. cerevisiae (reviewed in ORLEAN 1997 ). The current model entails GPI addition in the ER, after which the lipid-anchored protein translocates to the cell surface via the secretory pathway, and subsequently the GPI inositol phospholipid portion is removed and the remaining protein-attached GPI glycan remnant is somehow coupled to cell wall ß-1,6-glucan (LU et al. 1995 ; KAPTEYN et al. 1996 ; KOLLAR et al. 1997 ). However, non-wall-anchored GPI-proteins, such as the plasma membrane anchored protein Gas1p, exist (NUOFFER et al. 1991 ; VAI et al. 1991 ; RAM et al. 1995 ), suggesting the existence of a mechanism to confer wall "targeting," concerning which some insight may be gained by the genomic analysis of S. cerevisiae putative GPI-proteins (CARO et al. 1997 ). Such wall targeting will likely entail a key ER event(s) followed by cell surface reactions conferring further specificity. Therefore, the elucidation of the ER mutants which affect ß-1,6-glucan synthesis is critical for the understanding of such protein anchorage, and, in turn, overall cell wall biosynthesis.

ß-1,6-Glucan appears to play a critical role in wall biosynthesis and integrity due to its ability to crosslink a variety of cell wall components (KOLLAR et al. 1997 ). Knowledge of ß-1,6-glucan synthesis is derived mostly from the kre mutants, identified based on their resistance to the K1 killer toxin (HUTCHINS and BUSSEY 1983 ). The KRE gene products and their functional homologs have been localized to the endoplasmic reticulum [ER; KRE5 (MEADEN et al. 1990 )], Golgi [KRE6 and SKN1 (ROEMER et al. 1994 )], the cytoplasm [KRE11 (BROWN et al. 1993 )] and the cell surface [KRE9 (BROWN and BUSSEY 1993 ), KNH1 (DIJKGRAAF et al. 1996 ), KRE1 (ROEMER and BUSSEY 1995 )], which suggests that ß-1,6-glucan synthesis occurs via a series of intracellular events along the secretory pathway and is subsequently completed at the cell surface. The exact functions of these gene products, however, have yet to be characterized, and therefore the biochemical steps involved in ß-1,6-glucan biosynthesis remain largely unknown. Recent studies have implicated another ER protein, Cwh41p, in ß-1,6-glucan synthesis, as a ~50% decrease in the cell wall levels of this polymer have been observed upon disruption of the CWH41 gene (JIANG et al. 1996 ). The CWH41 gene was identified in a screen of hypersensitivity to calcofluor white (RAM et al. 1994 ), a phenotype often indicative of cell wall defects (RAM et al. 1994 ; LUSSIER et al. 1997B ). Cwh41p has since been demonstrated to be the enzyme glucosidase I (ROMERO et al. 1997 ), which is responsible for the first step of glucosyl-N-chain processing (see Figure 1). This observation suggested the possibility that the ER events required for ß-1,6-glucan biosynthesis are related to the glucosylation and glucose processing of nascent protein N-chains.

View larger version (26K): [in this window] [in a new window] [Download PPT slide]   Figure 1. —N-chain biosynthesis in the ER lumen of S. cerevisiae (adapted from ORLEAN 1997 ).

Figure 1 illustrates the pathway of N-chain biosynthesis in the ER (reviewed in ORLEAN 1997 ). The dolichol-anchored GlcNAc₂Man₅ structure is synthesized on the cytoplasmic face of the ER membrane and subsequently translocated into the ER lumen. Following this translocation, four more mannose residues are attached, two by the known mannosyltransferases Alg3p (HUFFAKER and ROBBINS 1983 ; AEBI et al. 1996 ) and Alg9p (BURDA et al. 1996 ), and the others via as yet unknown enzyme(s), generating the dolichol pyrophosphate-GlcNAc₂Man₉ structure. Three glucose residues are added sequentially by Alg6p (REISS et al. 1996 ), Alg8p (STAGLJAR et al. 1994 ) and Alg10p (BURDA and AEBI 1998 ), after which the GlcNAc₂Man₉Glc₃ structure is transferred en bloc to suitable protein asparagine residues. Following this transfer, processing of the three glucose residues occurs; the third, outer (-1,2) residue by the glucosidase I enzyme Cwh41p (KILKER JR. et al. 1981; ESMON et al. 1984 ; ROMERO et al. 1997 ), and the remaining two (both -1,3) by the glucosidase II enzyme Rot2p, also known and referred to here as Gls2p (SAUNIER et al. 1982 ; TROMBETTA et al. 1996 ). This transient glucosylation improves the efficiency of oligosaccharide transfer to protein (TRIMBLE et al. 1980 ; SHARMA et al. 1981 ; BREUER and BAUSE 1995 ), although the nonglucosylated GlcNAc₂Man₉ and even the incompletely mannosylated GlcNAc₂Man₅ structures still undergo transfer (HUFFAKER and ROBBINS 1983 ; VEROSTEK et al. 1993 ). Such glucose processing has been shown in vitro to be intrinsic to the ER "quality control" of protein folding in mammalian systems (reviewed in HELENIUS 1994 ; HELENIUS et al. 1997 ), and very recent results suggest the presence of a related system in S. cerevisiae (JAKOB et al. 1998 ). In this process, the N-chains of un-, partially or malfolded proteins are reglucosylated by the enzyme UDP-glucose:glycoprotein glucosyltransferase (UGGT; TROMBETTA and PARODI 1992 ; SOUSA et al. 1992 ; FERNANDEZ et al. 1994 ; PARKER et al. 1995 ; CHOUDHURY et al. 1997 ) following glucose removal, to regenerate a protein-coupled GlcNAc₂Man₉Glc₁ structure to which the ER chaperone calnexin binds to facilitate another round of protein folding (OU et al. 1993 ; HEBERT et al. 1995 ; WARE et al. 1995 ; ZAPUN et al. 1997 ) and is subsequently released via the action of glucosidase II (SAUNIER et al. 1982 ; TROMBETTA et al. 1996 ). Since the putative UGGT and calnexin homologs Kre5p (MEADEN et al. 1990 ; PARKER et al. 1995 ) and Cne1p (PARLATI et al. 1995 ), respectively, have been identified in S. cerevisiae, and since a severe ß-1,6-glucan defect of kre5 cells has been previously established (MEADEN et al. 1990 ), we could address whether Cne1p is also involved in ß-1,6-glucan synthesis, and perhaps establish a relationship between these quality control protein homologs and this process.

This study examines the relationship between glucosylation/processing and ß-1,6-glucan synthesis at the level of the ER to extend the results found in the study of Cwh41p (JIANG et al. 1996 ; ROMERO et al. 1997 ). The recent identification of the ALG gene products involved in N-chain biosynthesis and the processing enzymes Cwh41p and Gls2p has enabled us to take a genetic approach to study several aspects of these processes. From this we suggest that defects in ß-1,6-glucan shown previously in the cwh41 (JIANG et al. 1996 ) mutant and also demonstrated here for the gls2 mutant are a result of the failure of the trimming of N-chain glucosyl residues. These results have enabled us to propose the putative attachment of cell wall ß-1,6-glucan to protein N-chains, which, when combined with data obtained from mutants defective in outer chain synthesis, allows us to make strong predictions on the exact site of such attachment. In addition, we have identified a ß-1,6-glucan defective phenotype associated with the ER protein Cne1p, and present results that suggest that the essential role of Kre5p is independent of protein N-chain glucosylation. Given the cell wall crosslinking function of ß-1,6-glucan, as well as the severe growth defects or lethality seen in mutants lacking this polymer, ß-1,6-glucan appears to serve an important function in S. cerevisiae, and the studies described below contribute to the elucidation of its biosynthesis.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast and bacterial strains and culture conditions:
The S. cerevisiae strains used in this study are listed in Table 1. Standard growth conditions and media for yeast cells were used, as described previously (BUSSEY et al. 1982 ). Genetic crosses, sporulation of diploids, and tetrad dissection were performed using standard methods, as described previously (SHERMAN et al. 1982 ). Yeast transformations were performed using two methods: transformation of plasmids was performed using the one-step method of CHEN et al. 1992 ; transformation of PCR- or restriction digest-derived fragments for the purpose of effecting genetic disruption was performed using the high efficiency method of GIETZ et al. 1995 . Transformants were selected on synthetic minimal medium with auxotrophic supplements, with the exception of selections using the kanMX2 module, where YEPD containing 200 µg/ml geneticin (GIBCO, Burlington, ON) was used. Plasmids were routinely propagated using the Escherichia coli strain DH10B, except for the preparation of uracil containing DNA for mutagenesis, which used the E. coli host strain CJ236 (dut, ung1). Bacterial cells were cultured and transformed using standard media and methods, as described previously (SAMBROOK et al. 1989 ).

Plasmids:
The plasmid pAG1 (WOJCIECHOWICZ et al. 1993 ), containing a 6.1-kb genomic HindIII-HindIII fragment containing the AG1 gene (also known as SAG1) encoding -agglutinin, was kindly provided by PETER LIPKE (Hunter College of the City University of New York, NY). The plasmid pVTAG1 was generated by the insertion of the 2.6-kb HindIII-SspI fragment of pAG1 (containing the AG1 gene) into the HindIII and PvuII sites of pVT102U, a 2µ-based expression vector with a URA3 selectable marker and an ADH1 promoter (VERNET et al. 1987 ). A 3.7-kb XbaI-HindIII fragment of pAG1 was inserted into the XbaI and HindIII sites of pRS316 (SIKORSKI and HIETER 1986), to generate pRSAG1, a centromeric vector which contains the AG1 gene under the control of its endogenous promoter. YEp352 (HILL et al. 1993 ) was used to express the SUC2 gene, which encodes invertase, as described previously (LUSSIER et al. 1997A ). The plasmid pFA6-kanMX2 (WACH et al. 1994 ) was kindly provided by PETER PHILIPPSEN (Biozentrum, University of Basel, Switzerland). The plasmid pBGIIW containing the gls2::TRP1 disruption was prepared by WEI-JIA OU (Biotechnology Research Institute, Montréal, QC, Canada) as follows: A 3.9-kb genomic fragment containing the GLS2 gene flanked by a 5' SalI and a 3' SmaI site was obtained by PCR amplification using the 5' primer GTATGTCGACAGTGCTAAACACGACAGTAACGC and the 3' primer TATGCCCGGGTCTGAAGAGCAAAATCTGAAGCC (underlined and bold sequences represent the SalI and SmaI sites, respectively). This fragment was inserted into the SalI and SmaI sites of pBKS^- (Stratagene, La Jolla, CA) to generate the plasmid pBGII. Subsequently the plasmid pBGIIW was prepared via the insertion of the SphI-EcoRV fragment of pJJ248 (JONES and PRAKASH 1990 ) containing the TRP1 gene into the SphI-EcoRV sites of pBGII. The plasmid pFP10.13 containing the cne1::LEU2 disruption was prepared previously in the study of PARLATI et al. 1995 . The plasmid pALG5(SpeBgl)HIS containing the alg5::HIS3 disruption (TE HEESEN et al. 1994 ) was kindly provided by STEPHAN TE HEESEN (Eidgenössische Technische Hochschule, Zurich, Switzerland). The plasmid pSM491 is a PBSK^- (Stratagene)-based vector containing three copies of the influenza virus hemagglutinin (HA) epitope flanked by NotI sites (KOLODZIEJ and YOUNG 1991 ).

DNA purification and recombinant DNA techniques:
Yeast DNA was isolated by the method of HOFFMAN and WINSTON 1987 . Plasmid minipreps were prepared from E. coli via the boiling method described by SAMBROOK et al. 1989 . Restriction endonucleases, shrimp alkaline phosphatase, T4 DNA ligase, Taq DNA polymerase, sequenase, T4 polynucleotide kinase, and the Gene 32 protein were purchased from Bethesda Research Laboratories Inc. (Gaithersburg, MD), Pharmacia LKB Biotechnology (Piscataway, NJ), New England Biolabs (Beverly, MA), Boehringer Mannheim Biochemicals (Indianapolis, IN) or US Biochemicals (Cleveland, OH) and were used according to the manufacturers' instructions.

Gene disruption:
All gene disruptions were performed on the SEY6210 derived autodiploid, followed by sporulation, tetrad dissection, and mating type determination using the tester strains MC75 and MC76 to obtain the disrupted haploid progeny of known mating type. The gls2::TRP1 disrupted strains HAB866 and HAB867 were prepared via the transformation of the 3.9-kb NheI-Eco47III fragment from the plasmid pBGIIW. The cne1::LEU2 disrupted strains HAB868 and HAB869 were prepared via the transformation of the 3.9-kb StuI-SstI fragment from the plasmid pFP10.13. The alg5::HIS3 disrupted strain HAB872 was prepared via the transformation of the 2.9-kb HpaI-BamHI fragment from the plasmid pALG5(SpeBgl)HIS. Strains HAB870-871 (alg3), HAB873-874 (alg6), HAB875-876 (alg8), HAB877-878 (alg9), HAB879 (agl), HAB880 (mnn9) and HAB881 (och1) were prepared by PCR-based disruption using the kanMX2 module as described by WACH et al. 1994 , using the PCR oligonucleotide primers shown in Table 2, and the pFA6-kanMX2 plasmid as a template. The alg5::kanMX2 disruptions in strains HAB884 (alg5 cwh41) and HAB885 (alg5 gls2) were similarly prepared using the PCR-based method of WACH et al. 1994 , except that they were performed in the HAB855 (cwh41) and HAB866 (gls2) strain backgrounds, respectively. Integrants in HAB868, 869 and 872 were verified by Southern analysis; integrants in HAB866, 867,870, 871, 873-881, 884 and 885 were verified by a PCR analysis (not shown). The multiple disruption strains HAB882, 883, 886-895 were prepared by crosses of the appropriate haploid strains, followed by sporulation, tetrad dissection and spore progeny analysis to obtain the desired haploid progeny. The kre5 heterozygous, alg8 gls2 homozygous diploid strain HAB896 was prepared by the transformation of a 5.4 kb ClaI-SalI fragment containing the kre5::HIS3 disruption (see MEADEN et al. 1990 ) into a SEY6210 alg8 gls2 homozygous diploid (prepared by crossing HAB875 and HAB867). kre5::HIS3 transformants were verified by Southern analysis (not shown). Sporulation, tetrad dissection, and histidine selection were subsequently performed to determine spore viability and whether viable spores harbor the kre5::HIS3 disruption.

Epitope tagging of -agglutinin:
Single-stranded, uracil containing DNA of pVTAG1 was prepared using the M13 helper phage R408 (Stratagene) and the E. coli strain CJ236 transformed with pVTAG1 as described previously (SAMBROOK et al. 1989 ). Mutagenesis was performed via the method of KUNKEL et al. 1987 using the primer AGCAGCAAACAGCCAGGCGGCCGCTCCAGTCCCTCATCT, which resulted in the introduction of 9 bases (underlined) containing a NotI site at position 1548 of the AG1 open reading frame. The resulting construct, named pVTAG1-516NotI was confirmed by DNA sequencing. A 110-bp NotI fragment of pSM491, containing three copies of the HA epitope, was subsequently inserted into the NotI site of pVTAG1-516NotI to generate pVTAG1-516HA, which encodes the triply HA-tagged (between Pro516 and Ser517) version of the -agglutinin protein, -agglutinin-516HA. The 1.8-kb EcoRI-HindIII fragment of pVTAG1-516HA was inserted into the EcoRI and HindIII sites of pRSAG1, to generate pRSAG1-516HA, in which -agglutinin-516HA is expressed under the control of the endogenous promoter. ag1 MAT cells (HAB879) transformed with pRSAG1-516HA were able to mate with tester MATa cells (MC76) in liquid culture with wild-type efficiency (not shown), demonstrating wild-type functionality of -agglutinin-516HA.

Cell wall analyses:
The levels of cell wall alkali insoluble ß-1,3- and ß-1,6-glucan were determined as described previously (DIJKGRAAF et al. 1996 ), except that zymolyase digestion was extended to 20 hr, and there was one additional change of water during dialysis of the digested material. Values were calculated as micrograms of alkali insoluble glucan per milligram dry weight cell wall and are reported as percentages (WT = 100%) to facilitate comparison (absolute values for wild-type SEY6210 cells were in the order of 90 µg ß-1,6-glucan and 150 µg ß-1,3-glucan/mg dry weight cell wall). Susceptibility of yeast strains to the K1 killer toxin was determined as described previously (BROWN et al. 1993 ). Cell wall fractions of strains transformed with pVTAG1-516HA were isolated, SDS-extracted and -digested with the recombinant ß-1,3-endoglucanase Quantazyme (Quantum Biotechnologies, Laval, QC, Canada) as described previously (KAPTEYN et al. 1996 ). Quantazyme released wall material was digested with a recombinant fusion protein of endoglycosidase H (EndoH) and maltose binding protein (New England Biolabs) according to the manufacturer's protocol. This treatment was performed to improve the electrophoretic resolution of the extremely large (>300-kD) wall form of -agglutinin, since it has been shown that its treatment with EndoH results in the removal of ~100 kD of mass (LU et al. 1994 ).

Western blotting analyses:
Quantazyme-released, EndoH-treated wall protein samples were analyzed by SDS-PAGE (LAEMMLI 1970 ) using 6% polyacrylamide gels and transferred to Hybond-C nitrocellulose membranes (Amersham, Oakville, ON, Canada). To ensure that the loaded material represented the equivalent amount of cells, wall protein samples were standardized prior to loading to total cellular protein in each case, measured via the dye-binding assay of BRADFORD 1976 (BioRad, Mississauga, ON, Canada). Immunoblotting was performed using a 1000-fold dilution of the HA.11 anti-HA monoclonal primary antibody (Babco, Richmond, CA) and a 2000-fold dilution of horseradish peroxidase-conjugated goat anti-mouse secondary antibody (Amersham), and was subsequently detected using the enhanced chemiluminescence procedure (Amersham). Isolation and Western analysis of growth medium samples of strains expressing invertase (transformed with YEp352-SUC2) was performed as described previously (LUSSIER et al. 1997A ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

N-chain glucosylation affects cell wall glucan levels:
To determine the effects of N-chain glucosylation on wall glucan levels, we have used strains disrupted in the genes which encode some of the enzymes involved in this process. We have constructed and studied strains in which only the Glc₀ (WT), Glc₁ (alg8 gls2), Glc₂ (gls2) or Glc₃ (cwh41) glucosylated forms of the N-chain exist (see Figure 1), with the results shown in Table 3. An initial assay to detect defects in cell wall ß-1,6-glucan is to determine the degree of resistance (if any) of such mutant strains to the K1 killer toxin. Since this pore-forming protein has been shown to bind a cell wall ß-1,6-glucan-containing component and subsequently kill cells (reviewed in BUSSEY 1991 ), increased resistance to this toxin (as seen by smaller "kill zones" on seeded plates) is an indirect indicator of a decrease in the amount of cell wall ß-1,6-glucan present. As seen in the last column of Table 3, both cwh41 and gls2 cells, which contain the Glc₃ and Glc₂ forms, respectively, exhibit a noticeable resistance to killer toxin, which is also observed to a lesser extent in alg8 gls2 cells (containing the Glc₁ form). Such results suggest that failure to remove these N-chain glucosyl residues results in a decrease in the amount of cell wall ß-1,6-glucan. It should also be noted that the killer toxin resistance of alg8 cells (in which one glucose residue is added and removed) is essentially identical to that of wild-type cells (in which three glucose residues are added and removed), indicating that in the absence of N-chain glucose residues no ß-1,6-glucan defect exists. To confirm the above results, direct determination of cell wall alkali insoluble glucan was performed in parallel experiments. As seen in Table 3, cwh41 (Glc₃) and gls2 (Glc₂) cells both possess a ~45% decrease in the level of ß-1,6-glucan (a result shown previously for cwh41 cells [ JIANG et al. 1996 ]) and the Glc₁-containing alg8 gls2 cells show an intermediate decrease of 27%, in agreement with the killer toxin results above. alg8 (Glc₀) cells behaved as wild-type (Glc₀) cells in this respect with no defect in this polymer, also consistent with the killer toxin results. It should also be noted that ß-1,6-glucan loss is typically accompanied by an increase in ß-1,3-glucan (see Table 3 Table 4 Table 5), which may potentially represent a form of compensation for the ß-1,6-glucan loss. The results in Table 3 therefore demonstrate a decrease in the amount of cell wall ß-1,6-glucan, and a correlation of this decrease with an increase in the number of N-chain glucosyl residues.

View this table: [in this window] [in a new window]   Table 5. Effect of the cne1 mutation on cell surface ß-1,6-glucan levels, and its interaction with other ß-1,6-glucan biosynthetic mutants

We then sought to determine whether the reduced level of cell wall ß-1,6-glucan was directly attributed to the increased glucosylation of N-chains, and not an indirect consequence of the mutations examined. This was addressed using mutations in the ALG3 (HUFFAKER and ROBBINS 1983 ; AEBI et al. 1996 ), ALG6 (REISS et al. 1996 ), and ALG9 (BURDA et al. 1996 ) genes. ALG3 and ALG9 encode mannosyltransferases required for the synthesis of the nonglucosylated branches of the N-chain; ALG6 encodes the glucosyltransferase that attaches the first -1,3-glucose to the third mannosyl branch (see Figure 1). Because Alg3p generates the substrate required by Alg9p, which generates a substrate later modified by Alg6p (REISS et al. 1996 ), the N-chains from alg3, alg6 and alg9 cells differ in the number of mannosyl residues, but all completely lack glucosylation, and as a result may be used to better analyze the effects of improper glucosylation. As seen in Table 4, alg3, alg6 and alg9 cells all contained wild-type (or in some cases higher) levels of cell wall ß-1,6-glucan and all exhibited a killer toxin sensitivity identical to wild-type cells. The higher than wild-type levels of ß-1,6-glucan in strains containing alg3 or alg9 disruptions is likely due to undermannosylation in these mutants, which inflates the measured ß-1,6-glucan value since all values are standardized to the total wall dry weight. These results indicate that gross differences in the mannosyl branches do not reduce ß-1,6-glucan levels, and support the model that the defects discussed above are primarily due to the presence of the glucosyl residues. Another method of eliminating N-chain glucosylation is via mutation of the ALG5 gene, since Alg5p synthesizes the dolichol phosphate glucose (Dol-P-Glc) glucosyl donor required by the Alg6p, Alg8p and Alg10p glucosyltransferases (TE HEESEN et al. 1994 ). We find that alg5 cells possess wild-type levels of cell wall ß-1,6-glucan (not shown), enforcing the conclusion that the observed defects in cell wall ß-1,6-glucan are directly caused by unprocessed N-chain glucosyl residues. This observation also suggests that any initiation of ß-1,6-glucan synthesis in the ER is not Dol-P-Glc dependent, and is therefore likely to utilize UDP-glucose as a glucosyl donor.

To further demonstrate that the defects in cell wall ß-1,6-glucan observed above are directly due to the presence of N-chain glucosyl residues, we asked if Cwh41p and Gls2p functions are limited to N-chain glucose processing. This was tested by determining whether alg3, alg6 and alg9 mutations can suppress the phenotype seen in a cwh41 or gls2 background. Since alg3, alg6 and alg9 cells completely lack any N-chain glucosylation, substrates for Cwh41p or Gls2p do not exist in these strains, and therefore the loss of Cwh41p or Gls2p should not affect the wild-type ß-1,6-glucan levels observed in alg3, alg6 and alg9 backgrounds. This is the case, as both ß-1,6-glucan measurements and killer toxin results of the double disrupted strains in Table 4 indicate that alg3, alg6 and alg9 mutations are all epistatic to cwh41 or gls2, and thus suppress the cwh41 and gls2 phenotypes simply by eliminating any initial presence of N-chain glucosyl residues. Similarly, elimination of the Dol-P-Glc glucosyl donor via an alg5 disruption also restores the decreased levels of this polymer seen in cwh41 or gls2 cells to wild-type levels (not shown). Therefore, at least with regard to cell wall ß-1,6-glucan biosynthesis, the functions of Cwh41p and Gls2p appear limited to deglucosylation, since the phenotypes associated with their loss are eliminated in various Glc₀ mutant backgrounds. Taken together, the results suggest that the observed defects in ß-1,6-glucan biosynthesis are solely due to the presence of N-chain glucosyl residues and show some correlation between the severity of the ß-1,6-glucan phenotype and the number of such glucose residues present. A possible explanation to account for this phenotype is that these glucose residues may actually sterically hinder direct attachment of ß-1,6-glucan to protein N-chains. Such attachment could occur at any of the N-chain residues present in an alg3 background (see Figure 1), since this strain retains wild-type ß-1,6-glucan levels. However, attachment to the outer chain, which requires Och1p (NAKAYAMA et al. 1992 ) and Mnn9p (RASCHKE et al. 1973 ; YIP et al. 1994 ) for initiation and elaboration, respectively, remains a possibility. To rule out outer chain involvement in such putative attachment, we have determined that neither och1 nor mnn9 cells have any defect in ß-1,6-glucan, as assayed both directly and using killer toxin (not shown). These results further define the site of putative ß-1,6-glucan attachment to protein N-chains, discussed further below (see DISCUSSION).

The roles of Cne1p and Kre5p in cell wall ß-1,6-glucan biosynthesis:
Since the above results indicate an involvement of N-chain glucosylation with ß-1,6-glucan biosynthesis, we wished to analyze subsequent ER N-chain-dependent events, and their possible effects on the cell wall ß-1,6 glucan phenotype. In vitro studies of ER protein folding in mammalian systems (see HELENIUS 1994 ; HELENIUS et al. 1997 for reviews) have demonstrated that the N-chains of misfolded proteins are deglucosylated by glucosidases I and II (HEBERT et al. 1995 ; ORA et al. 1995), but are then reglucosylated with a single -1,3-glucose at the same branch by the enzyme UDP-glucose:glycoprotein glucosyltransferase (UGGT; SOUSA et al. 1992 ; FERNANDEZ et al. 1994 ; PARKER et al. 1995 ; CHOUDHURY et al. 1997 ), which enables the specific binding of the chaperone protein calnexin to aid folding (OU et al. 1993 ; HEBERT et al. 1995 ; WARE et al. 1995 ; ZAPUN et al. 1997 ). Such quality control mechanisms, however, have yet to be identified in S. cerevisiae, despite the presence of glucosidase I (Cwh41p), glucosidase II (Gls2p), as well as potential homologs of UGGT (Kre5p; MEADEN et al. 1990 ) and calnexin (Cne1p; PARLATI et al. 1995 ). These components are of interest to us, since cwh41 and gls2 cells both possess a ~45% reduction in ß-1,6-glucan as shown above, and it has previously been shown in our laboratory that kre5 cells have the most severe reduction in this polymer observed to date (MEADEN et al. 1990 ), implying an involvement of these activities with cell wall ß-1,6-glucan synthesis.

On the basis of the above similarities, we investigated whether the S. cerevisiae calnexin homolog Cne1p is also involved in ß-1,6-glucan biosynthesis. As seen in Table 5, cne1 cells have a ~30% reduction in this polymer, and an increased resistance to killer toxin. Furthermore, the multiple mutants cne1 cwh41, cne1 gls2 and cne1 cwh41 gls2 all exhibit a ~70% reduction in this polymer, and thus show a more severe phenotype than cne1, cwh41 or gls2 alone. Such genetic interactions involving Cne1p leading to a more severe reduction in ß-1,6-glucan could be caused by extending the partial effects caused by cwh41 or gls2 defects on N-chains, or through some other role of Cne1p. We also found that a cne1 deletion exacerbated the partial ß-1,6-glucan defect in a kre6 strain (see Table 5). This again argues that the cne1 defect can be additive with respect to downstream ß-1,6-glucan defects such as that caused in the Golgi by a kre6 deletion. Efforts to rescue the S. cerevisiae cne1 defect with the heterologously expressed calnexin gene from dog or Schizosaccharomyces pombe failed to correct the ß-1,6-glucan defect in S. cerevisiae (not shown). This failure neither supports nor opposes a quality control function for Cne1p, but it does suggest that Cne1p may have an additional function related to ß-1,6-glucan synthesis in S. cerevisiae, which is lacking in its counterparts from other species. This is consistent with the previous observation that the majority of ER chaperone functions in S. cerevisiae are fulfilled by Kar2p/BiP (SIMONS et al. 1995 ), which may also support a unique role for Cne1p in this organism.

Although the severe ß-1,6-glucan defect of kre5 cells is well established, the nature of this defect remains unknown, and the only hypothesis for Kre5p function is based on its significant but limited sequence similarity to the UGGT enzyme (PARKER et al. 1995 ). To ask if KRE5 encodes a putative UGGT activity, we determined if the Glc₀ and Glc₁ forms of the N-chain represent the substrate and product, respectively, for Kre5p in S. cerevisiae, as they do for the UGGT enzyme in other systems. Since the function of the UGGT is to regenerate the Glc₁ form of the N-chain following glucosidase II activity, our strategy was to generate the Glc₁ form by a genetic approach, and determine whether the essential requirement of Kre5p was dispensed with under these conditions. This was accomplished by analyzing the effects of a kre5 disruption in an alg8 gls2 background, in which the Glc₁ form of the N-chain exists constitutively. Since kre5 disruptions are lethal in a SEY6210 background, if Kre5p is simply a UGGT, kre5 alg8 gls2 cells should be viable, since UGGT activity is no longer required. Diploid transformants harboring a heterozygous kre5 disruption in a homozygous alg8 gls2 disrupted background were obtained as described in MATERIALS AND METHODS. Eighteen of 20 tetrads from two such transformants showed a 2:2 segregation of viable:nonviable spores; the other two contained one and zero viable spores, respectively (data not shown). None of the viable spores contained the kre5::HIS3 disruption. These data, therefore, demonstrate that kre5 alg8 gls2 cells are not viable, strongly suggesting that the essential function of Kre5p is unrelated to N-chain reglucosylation, in agreement with the results of FERNANDEZ et al. 1994 and JAKOB et al. 1998 who were unable to detect any UGGT activity in S. cerevisiae. However, this genetic experiment does not address if Kre5p also has an N-chain UGGT function in addition to its essential activity.

Effect of glucose processing mutations on the wall incorporation of -agglutinin:
Several recent studies have demonstrated that ß-1,6-glucan plays an important role in the anchoring of proteins to the cell wall, (KAPTEYN et al. 1996 ; KOLLAR et al. 1997 ), with one of the better studied proteins in this respect being the yeast cell wall protein -agglutinin (LU et al. 1994 , LU et al. 1995 ). Previous results have shown increased secretion of -agglutinin into the medium when expressed in the ß-1,6-glucan-deficient strains kre1, kre5, kre6, kre9 and kre11 (LU et al. 1995 ), implying a defect in cell wall anchorage. We examined the cell wall retention of -agglutinin in the mutant strains described in this study to determine if these novel ß-1,6-glucan mutants exhibit a similar defect in the crosslinking of proteins to the cell wall as observed in the kre mutants. -Agglutinin was chosen as a model cell wall protein since it is known to be N-glycosylated (CHEN et al. 1995 ) and is therefore conducive to the study of the effects of N-chain glucosylation, and has an easily assayable function, which has enabled the preparation of a fully functional epitope-tagged version, -agglutinin-516HA (see MATERIALS AND METHODS). The samples seen in Figure 2A represent the wall-crosslinked fraction of -agglutinin, which was obtained via their release from SDS-extracted cell walls using the recombinant ß-1,3-endoglucanase Quantazyme, and subsequently treated with EndoH to improve electrophoretic resolution. Samples were normalized to the total cell protein in each case, allowing a semi-quantitative comparison of the relative amounts of wall anchored -agglutinin present in each mutant background. As seen in Figure 2A, the ability to anchor this protein in the cell wall is not affected by mutations in Cne1p (cne1; lane 2), glucosidase I (cwh41; lane 3), glucosidase II (gls2; lane 4), or combinations thereof (lanes 5-8). In the case of cwh41 cells, this result is in agreement with the study of JIANG et al. 1996 , where the anchorage of the cell wall protein Cwp1p was also unaffected in this mutant background. However, significantly less wall-anchored -agglutinin is present in kre6 cells (lane 9), consistent with the impaired wall anchorage previously observed in this mutant background (LU et al. 1995 ). kre6 cne1 cells (lane 10) show a defect in -agglutinin wall anchorage similar to kre6 cells, suggesting that this defect is primarily due to the kre6 disruption. We examined in parallel the biosynthesis of an N-glycosylated nonwall protein, the secreted protein invertase (ESMON et al. 1987 ), as seen in Figure 2B. This protein is essentially unaltered in the above mutant backgrounds, indicating that the disruption of these genes does not result in bulk defects in protein secretion or glycosylation and that the decreased wall incorporation of -agglutinin seen in kre6 and kre6 cne1 cells is not due to such gross defects.

View larger version (60K): [in this window] [in a new window] [Download PPT slide]   Figure 2. —Western analysis of cell wall -agglutinin and secreted invertase in the various mutant backgrounds of this study. (A) Cell wall fractions of the indicated strains expressing the epitope-tagged protein -agglutinin-516HA were prepared and immunoblotted with an anti-HA monoclonal antibody as described in MATERIALS AND METHODS. The last two lanes contain a cell wall sample from wild-type cells expressing untagged -agglutinin and a mock reaction containing only quantazyme and EndoH, respectively, confirming the absence of antibody cross-reactivity with endogenous cell wall proteins or the enzymes used in sample preparation. (B) Medium samples of the indicated strains expressing the secreted protein invertase were prepared and immunoblotted with an anti-invertase polyclonal antiserum as described in MATERIALS AND METHODS. The last lane contains a medium sample prepared in parallel from wild-type cells transformed with YEp352, confirming the absence of antibody cross-reactivity with endogenous medium proteins.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

N-chain glucosylation and ß-1,6-glucan synthesis:
Results presented here identify an involvement of nascent protein N-chains in the ER with ß-1,6-glucan synthesis, since failure to remove N-chain glucose residues following the transfer of the dolichol-linked precursor to proteins results in a significant decrease in the amount of cell wall ß-1,6-glucan. The exact mechanisms involved in generating the ß-1,6-glucan phenotype are unclear, but we consider the following three possibilities: (1) Steric hindrance (by the three glucose residues) of direct attachment of ß-1,6-glucan to protein N-chains; (2) the loss of other putative unknown ß-1,6-glucan synthetic functions of the glucose trimming enzymes glucosidases I (Cwh41p) and II (Gls2p); and (3) indirect effects caused by gross defects in glycosylation and/or secretion. Each of these are discussed independently below.

The first model would involve the attachment of ß-1,6-linked glucose residues to protein N-chains, either directly via a glucose-mannose linkage, or indirectly via a spacer containing different sugar residues or linkages. The one, two, or three glucose residues of the N-chain could potentially sterically hinder this attachment to varying extents, accounting for the results observed above. This model is supported by the work of TKACZ 1984 , who demonstrated that ß-1,6-glucan synthesis is partially inhibited following treatment with the N-glycosylation inhibitor tunicamycin, and that material containing ß-1,6-glucan may be removed from wall proteins by EndoH digestion, which specifically removes protein N-chains. KOLLAR et al. 1997 also detected a fraction of wall glucan which was removed from proteins by EndoH digestion, however, the reducing end was in the opposite orientation expected for an N-glycan, and would therefore require a unique linkage in this structure to account for N-chain attachment. If such N-chain attachment were to exist, our results suggest that it would occur at the same mannosyl branch which is glucosylated, since it is unaffected in alg3 or alg9 strains in which the first and second mannosyl branches are truncated (see Figure 3). Attachment at the third, glucosylated mannosyl branch would also be consistent with a steric hindrance mechanism since the site of attachment would be relatively close to the site of glucosylation. The absence of ß-1,6-glucan defects in och1 and mnn9 strains also rules out any attachment of this polymer to the outer chain. Recent structural analysis of the Glc_1-3Man₉GlcNAc₂ structures isolated from mammalian cells (PETRESCU et al. 1997 ) have shown that the three glucosyl residues form a relatively rigid, tight turn extending from the third mannosyl branch, which points increasingly outward as the chain is elongated from one to three glucose residues. It is therefore possible that the Glc₃ and Glc₂ forms may result in a degree of steric hindrance and thus a ß-1,6-glucan phenotype, which would occur to a lesser extent in the Glc₁ form, in agreement with the results obtained here (see Table 3). Formally, attachment could potentially occur at any residue present in the GlcNAc₂Man₅ structure present in alg3 och1 cells (Figure 3, shaded residues); however, the above results and the possible steric hindrance provided by the glucosyl residues strongly suggest that such putative attachment occurs at either the penultimate or the terminal (glucosylated) mannose residues of this branch (see Figure 3). It should be noted that all of the results described above indicate that such putative N-glycan attachment would represent only one of at least two points of attachment of ß-1,6-glucan to proteins, since the glycosyl phosphatidylinositol (GPI) anchor has been shown to be the key site of attachment of proteins to cell wall glucan and is the minimum requirement for wall targeting (SCHREUDER et al. 1993 ; VAN BERKEL et al. 1994 ; KAPTEYN et al. 1996 ; KOLLAR et al. 1997 ). Such multiple attachment may suggest an alternative interpretation of these results, in which the experimental generation of constitutive N-chain Glc_1-3 forms (see Table 3) actually creates nonphysiological acceptors for ß-1,6-glucan attachment, which compete with GPI glycans for the ER ß-1,6-glucan synthetic machinery, thus resulting in a ß-1,6-glucan phenotype. However, mutations in glucosidases I or II did not affect the wall attachment of -agglutinin (Figure 2), suggesting that the essential GPI-dependent wall attachment is not impaired under these conditions. Therefore, all our results are consistent with ß-1,6-glucan attachment to both the N-glycans and GPI remnant of cell wall proteins; providing evidence for N-chain attachment, strongly supporting the first model proposed above.

View larger version (24K): [in this window] [in a new window] [Download PPT slide]   Figure 3. —Schematic representation of potential sites of ß-1,6-glucan attachment to protein N-chains. Shaded residues indicate those present in an alg3 och1 background. Residues are shown using the same symbols as Figure 1.

The second model raises the possibility that the glucose trimming enzymes glucosidases I and II possess other unknown activities that are involved in ß-1,6-glucan synthesis, which are lacking in a cwh41 or gls2 disruptant. Our results appear to eliminate this possibility, since the Glc₀ forms present in alg3, alg5, alg6 and alg9 backgrounds can restore wild type levels of cell wall ß-1,6-glucan to cwh41 and gls2 cells, indicating that the sole activity of these processing enzymes is directed at the N-chain glucosyl residues.

Bulk defects in N-glycosylation and secretion, as proposed in the third model above, are also unlikely, since the N-glycosylated nonwall protein invertase appears to be synthesized, modified and secreted properly in the mutant backgrounds examined. There remains the possibility that the observed phenotype may be indirect, a result of the impaired function of a certain subset of proteins involved in ß-1,6-glucan synthesis. For example, impaired glucosylation and processing may exclusively affect a ß-1,6-glucan synthetic protein, possibly via defects in its folding due to impaired quality control mechanisms, in which case the study of the biosynthesis of other proteins such as invertase would not be indicative of such selective activity. Although possible, however, this hypothesis remains unlikely due to the apparent absence of a UGGT/calnexin-dependent quality control system in S. cerevisiae (see below), a principle further enforced by the inability of the monoglucosylated alg8 gls2 mutant to suppress kre5 lethality. Given the current data, we favour the first model, which proposes the attachment of ß-1,6-glucan to the N-chains of wall mannoproteins. Direct proof of such a mechanism, however, requires further experimentation involving biochemical analysis of the N-glycans of mature wall proteins to detect the presence and nature of any glucose polymer(s) attached at this site.

The roles of Cne1p and Kre5p in ß-1,6-glucan biosynthesis:
Little is known regarding Cne1p and Kre5p activity, despite the existence of putative homologous proteins of known function in other systems. Current knowledge suggests that S. cerevisiae lacks the analogous ER quality control components calnexin and UGGT which ensure the proper folding of proteins in other systems, although this conclusion is inferred primarily by a lack of positive results using techniques which may perhaps not be optimal for S. cerevisiae. The data presented here demonstrate the previously unsuspected involvement of Cne1p in ß-1,6-glucan synthesis, suggesting that Cne1p has a different function or at least an additional function compared to its counterparts in other organisms. This conjecture is supported by the inability of heterologously expressed dog or S. pombe calnexin proteins to suppress the ß-1,6-glucan defect of cne1 cells. Also, Cne1p differs from mammalian calnexin in that it is unable to bind calcium (PARLATI et al. 1995 ), an essential component of mammalian calnexin activity (WADA et al. 1991 ; OU et al. 1995 ), consistent with a unique function. The exact nature of Cne1p activity in ß-1,6-glucan synthesis, however, is more difficult to elucidate. Disruption of the CNE1 gene in cwh41, gls2 or kre6 cells results in an additional ß-1,6-glucan defect compared to the single mutants (Table 5), a result which may be interpreted in the following two ways: In the first model Cne1p may be involved in the same pathway of ß-1,6-glucan synthesis as Cwh41p, Gls2p, and/or Kre6p, the absence of the latter proteins resulting in a partial defect in the ß-1,6-glucan synthetic activity of this pathway. The additional absence of Cne1p may simply cause a further decrease in such activity, resulting in a greater loss of ß-1,6-glucan production by this pathway. Alternatively, Cne1p may act in a ß-1,6-glucan biosynthetic pathway separate from that utilizing Cwh41p, Gls2p, and/or Kre6p, in which case the double disruptants would possess partial defects in both pathways, resulting in an overall additive ß-1,6-glucan phenotype. Both models are supported to some extent by studies of the cell wall incorporation of -agglutinin presented here (Figure 2). Since no significant defect in -agglutinin anchorage is seen in cne1, cwh41, or gls2 cells, the cells therefore share a similar phenotype in this respect, and thus these gene products may perhaps act within the same pathway or step in ß-1,6-glucan synthesis, a hypothesis supported by their ER co-localization. Since mammalian calnexin activity works in conjunction with glucose processing, the simplest assumption is that Cne1p acts within the same pathway of ß-1,6-glucan synthesis as those proteins involved in glucose processing, i.e., Cwh41p and Gls2p, as proposed in the first model. The considerable defect in -agglutinin anchorage in kre6 cells that is not observed in cne1 cells may imply that Kre6p and Cne1p are components of different pathways. Distinction between these two models will likely require the elucidation of such putative pathway(s) to determine the exact role(s) of Cne1p in ß-1,6-glucan biosynthesis.

The function of Kre5p is more difficult to study using a genetic approach, since it is lethal in certain strain backgrounds, makes no detectable ß-1,6-glucan and is epistatic to all known mutants that exhibit this phenotype (MEADEN et al. 1990 ). We have shown via genetic interactions that the essential function of Kre5p does not appear to be the generation of a monoglucosylated N-chain as occurs with a classic UGGT, since the constitutive presence of this structure in an alg8 gls2 mutant was unable to rescue the lethality of a kre5 disruption in the SEY6210 background. Such results, however, do not address whether Kre5p also has a non-essential UGGT-like activity, which may only be determined biochemically using assays developed for this purpose. Indeed, the results presented here when combined with previous cell wall studies suggest that Kre5p may play dual roles in ß-1,6-glucan biosynthesis, since kre5 cells have no detectable cell wall ß-1,6-glucan (MEADEN et al. 1990 ) and this mutation is epistatic to all known ß-1,6-glucan mutants. In addition, kre5 cells have a severe defect in the attachment of cell wall proteins, since the cell wall protein Cwp1p is virtually undetectable in the cell wall fractions of these cells (JIANG et al. 1996 ), and growth medium levels of both -agglutinin (LU et al. 1995 ) and Cwp1 (JIANG et al. 1996 ) have been shown to be greatly elevated, due to the virtually complete lack of wall anchorage. Our results suggest that at the level of the ER, two potential ß-1,6-glucan biosynthetic pathways may exist, one involving protein N-chains, the other being dependent on the GPI modification of wall mannoproteins. Current knowledge indicates that the latter pathway is essential for wall protein attachment, wall integrity, and cell viability, and therefore the essential function of Kre5p may be involved with such a GPI-dependent pathway. The complete absence of ß-1,6-glucan in kre5 mutants suggests that Kre5p possesses an additional non-essential function involved in a putative N-chain-dependent pathway. The nature of this additional function is unknown, a possible explanation being that Kre5p may be a glucosyltransferase with dual substrate specificity, having the ability to modify both N-chains as well as wall protein GPI glycans in the ER. The former activity would represent its non-essential function, while the latter would be essential for cell wall biosynthesis. However, the role of Kre5p in ß-1,6-glucan synthesis remains unknown, and evidence of such dual substrate specificity and pathways (discussed further below) may be obtained only via the determination of the biochemical activit(ies) of Kre5p by other means, and relating such activit(ies) to ß-1,6-glucan biosynthesis.

ß-1,6-Glucan biosynthesis in Saccharomyces cerevisiae:
The results provide new insight into the synthesis of cell wall ß-1,6-glucan, implicating the involvement of protein N-chains as well as the ER protein Cne1p in this process. It is becoming increasingly apparent that perhaps two groups of ß-1,6-glucan synthetic mutants exist; those that have defects in cell growth and in the anchoring of wall proteins such as the kre mutants, and those that are able to incorporate proteins in the wall and appear to grow normally despite lower levels of this polymer, such as the cwh41, gls2 and cne1 mutants of this study. Such groups may correspond to two potential parallel biosynthetic pathways for ß-1,6-glucan synthesis, as suggested above. Such a model would propose that while certain polymers, linkages and correctly modified wall proteins are absolutely essential for wall integrity, other components are not. It is therefore feasible for wall proteins to be anchored via their N-chains as our data suggest, but it is also possible that this represents a secondary anchorage point, without which the cell can still sufficiently anchor proteins in the wall via carboxyterminal GPI glycans, maintaining wall integrity and growth. The next phase in this area will entail biochemical analyses which test the predictions implicit in the current genetic data.

	FOOTNOTES

¹ These authors contributed equally to this work.

	ACKNOWLEDGMENTS

We thank Dr. PETER LIPKE (Hunter College of the City University of New York, NY) for providing the plasmid pAG1, Prof. PETER PHILIPPSEN (Biozentrum, University of Basel, Switzerland) for providing the plasmid pFA6-kanMX2, and Dr. STEPHAN TE HEESEN (ETH Zurich, Switzerland) for providing the plasmid pALG5(SpeBgl)HIS. SS was supported by a Medical Research Council/Pfizer Canada Inc. Pharmaceutical Manufacturers' Association of Canada Grant to HB. GJPD was supported by a Max Stern Recruitment Fellowship. This work was supported by a Natural Sciences and Engineering Research Council of Canada Operating Grant to HB, a Canadian Pacific Professor of Biotechnology.

Manuscript received October 17, 1997; Accepted for publication February 25, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AEBI, M., J. GASSENHUBER, H. DOMDEY, and S. TE HEESEN, 1996  Cloning and characterization of the ALG3 gene of Saccharomyces cerevisiae.. Glycobiology 6:439-444[Abstract/Free Full Text].

BRADFORD, M. M., 1976  A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72:248-254[Medline].

BREUER, W. and E. BAUSE, 1995  Oligosaccharyl transferase is a constitutive component of an oligomeric protein complex from pig liver endoplasmic reticulum. Eur. J. Biochem. 228:689-696[Medline].

BROWN, J. L. and H. BUSSEY, 1993  The yeast KRE9 gene encodes an O-glycoprotein involved in cell surface ß-glucan assembly. Mol. Cell. Biol. 13:6346-6356[Abstract/Free Full Text].

BROWN, J. L., Z. KOSSACZKA, B. JIANG, and H. BUSSEY, 1993  A mutational analysis of killer toxin resistance in Saccharomyces cerevisiae identifies new genes involved in cell wall (16)-ß-glucan synthesis. Genetics 133:837-849[Abstract].

BURDA, P. and M. AEBI, 1998  The ALG10 locus of Saccharomyces cerevisiae encodes the -1,2 glucosyltransferase of the endoplasmic reticulum: the terminal glucose of the lipid-linked oligosaccharide is required for efficient N-linked glycosylation. Glycobiology 8:455-462[Abstract/Free Full Text].

BURDA, P., S. TE HEESEN, A. BRACHAT, A. WACH, and A. DÜSTERHÖFT et al., 1996  Stepwise assembly of the lipid-linked oligosaccharide in the endoplasmic reticulum of Saccharomyces cerevisiae: identification of the ALG9 gene encoding a putative mannosyltransferase. Proc. Natl. Acad. Sci. USA 93:7160-7165