The Oligosaccharyltransferase Complex from Saccharomyces cerevisiae

The key step of N-glycosylation of proteins, an essential and highly conserved protein modification, is catalyzed by the hetero-oligomeric protein complex oligosaccharyltransferase (OST). So far, eight genes have been identified in Saccharomyces cerevisiae that are involved in this process. Enzymatically active OST preparations from yeast were shown to be composed of four (Ost1p, Wbp1p, Ost3p, Swp1p) or six subunits (Ost2p and Ost5p in addition to the four listed). Genetic studies have disclosed Stt3p and Ost4p as additional proteins needed for N-glycosylation. In this study we report the identification and functional characterization of a new OST gene, designated OST6, that has homology to OST3 and in particular a strikingly similar membrane topology. Neither gene is essential for growth of yeast. Disruption of OST6 orOST3 causes only a minor defect inN-glycosylation, but an Δost3Δost6double mutant displays a synthetic phenotype, leading to a severe underglycosylation of soluble and membrane-bound glycoproteinsin vivo and to a reduced OST activity in vitro. Moreover, each of the two genes has also a specific function, since agents affecting cell wall biogenesis reveal different growth phenotypes in the respective null mutants. By blue native electrophoresis and immunodetection, a ∼240-kDa complex was identified consisting of Ost1p, Stt3p, Wbp1p, Ost3p, Ost6p, Swp1p, Ost2p, and Ost5p, indicating that probably all so far identified OST proteins are constituents of the OST complex. It is also shown that disruption of OST3 and OST6 leads to a defect in the assembly of the complex. Hence, the function of these genes seems to be essential for recruiting a fully active complex necessary for efficient N-glycosylation.

The key step of N-glycosylation of proteins, an essential and highly conserved protein modification, is catalyzed by the hetero-oligomeric protein complex oligosaccharyltransferase (OST). So far, eight genes have been identified in Saccharomyces cerevisiae that are involved in this process. Enzymatically active OST preparations from yeast were shown to be composed of four (Ost1p, Wbp1p, Ost3p, Swp1p) or six subunits (Ost2p and Ost5p in addition to the four listed). Genetic studies have disclosed Stt3p and Ost4p as additional proteins needed for N-glycosylation. In this study we report the identification and functional characterization of a new OST gene, designated OST6, that has homology to OST3 and in particular a strikingly similar membrane topology. Neither gene is essential for growth of yeast. Disruption of OST6 or OST3 causes only a minor defect in N-glycosylation, but an ⌬ost3⌬ost6 double mutant displays a synthetic phenotype, leading to a severe underglycosylation of soluble and membrane-bound glycoproteins in vivo and to a reduced OST activity in vitro. Moreover, each of the two genes has also a specific function, since agents affecting cell wall biogenesis reveal different growth phenotypes in the respective null mutants. By blue native electrophoresis and immunodetection, a ϳ240-kDa complex was identified consisting of Ost1p, Stt3p, Wbp1p, Ost3p, Ost6p, Swp1p, Ost2p, and Ost5p, indicating that probably all so far identified OST proteins are constituents of the OST complex. It is also shown that disruption of OST3 and OST6 leads to a defect in the assembly of the complex. Hence, the function of these genes seems to be essential for recruiting a fully active complex necessary for efficient N-glycosylation.
Asparagine-linked glycosylation is one of the most common types of eukaryotic protein modifications (1)(2)(3). N-Glycans are essential for cell viability (4 -6) and can have a profound role in the biological function and physicochemical properties of many secreted and integral membrane proteins (7). The key step of this pathway is the en bloc transfer of the core oligosaccharide Glc 3 Man 9 GlcNAc 2 from dolichyl pyrophosphate to selected asparagine residues in an Asn-X-Ser/Thr consensus sequon of nascent polypeptides, where X can be any amino acid except proline (8 -13). The reaction is catalyzed by the ER 1 -resident enzyme N-oligosaccharyltransferase (OST) (for review see Refs. 14 and 15).
Previous attempts to purify and characterize this enzyme have met with limited success due to its lability upon solubilization (16 -18) and probably also due to the fact that it is a multimeric membrane protein complex. Meanwhile, however, OST complexes have been purified from different sources, such as dog pancreas (19,20), yeast (21)(22)(23), hen oviduct (24), and human (25) and pig (26) liver. The subunit composition of the various isolated complexes and the protein sequences, so far obtained, reveal in part a high conservation of the structural organization of this enzyme throughout evolution. Independent OST purifications from Saccharomyces cerevisiae have yielded active complexes consisting of four polypeptides (Ost1p (64/62 kDa), Wbp1p (47 kDa), Ost3p (34 kDa), and Swp1p (30 kDa)) (21,23) or of six subunits (Ost1p, Wbp1p, Ost3p, Swp1p, Ost2p (16 kDa), and Ost5p (9.5 kDa)) (22). Cloning and functional analysis of OST1 (23,27), WBP1 (28), SWP1 (29), and OST2 (30) have indicated that these genes are essential for the vegetative growth of the yeast cell and reveal significant homology to components of the canine complex: to ribophorin I (27), OST48 (31), to the C-terminal half of ribophorin II (22), and DAD1 (defender against apoptotic cell death) (20), respectively. In contrast, OST3 (32) and OST5 (33) coding for the 34-and 9.5-kDa subunits, respectively, are not essential, but their deletion yields glycosylation defects and reduces OST activity in vitro. In addition to these six proteins, genetic screens have identified two other loci, OST4 (34) and STT3 (35), that are required for optimal OST function in vivo and in vitro. Recent evidence indicates now that the derived proteins are indeed part of the complex (36,37). OST4 encodes an unusually small, hydrophobic polypeptide of 3.6 kDa; its deletion leads to underglycosylation of N-glycoproteins and a temperature-sensitive growth phenotype. STT3 encodes a 78-kDa transmembrane protein with the highest conservation among the proteins associated with OST function. The essential Stt3p was found to be necessary for stability or assembly of the complex, and its lack affects the substrate specificity for the lipid-linked oligosaccharide donor (35).
So far, the specific function of the various enzyme subunits is obscure and remains to be defined. In addition, the question must be answered which subunit is a bona fide constituent of the complex or serves only an auxiliary function. In this report, we describe the isolation and functional characterization of a new, not essential gene, designated OST6, that has sequence homology and in particular a very similar membrane topology to OST3. Disruption of OST6 causes only a minor defect in N-glycosylation, both in vivo and in vitro. However, a ⌬ost6⌬ost3 double mutant exhibits a synthetic phenotype with a strong underglycosylation of soluble and membrane-bound glycoproteins as well as a defect in complex formation. By blue native electrophoresis, we demonstrate that the OST complex has a molecular mass of about 240 kDa and consists of all hitherto defined OST subunits (except for the small 3.6-kDa Ost4p, for which no antibody is available for detection).
Cloning and Disruption of OST6 -The OST6 coding sequence was originally reported as an ORF of unknown function on chromosome XIII by the yeast genome sequencing project (ORF YML019W). A genomic clone of OST6 was isolated from a yeast gene bank with 4-kilobase pair Sau3A-DNA fragments in the multicopy vector YEp352 (40). As a probe for colony hybridization, a PCR-amplified, digoxygenin-labeled genomic OST6 fragment was used. A genomic clone containing the OST6 open reading frame together with 5Ј-and 3Ј-flanking sequences was isolated from ϳ20.000 Escherichia coli colonies.
For disruption of the OST6 locus by homologous recombination, DNA fragments of 0.9 and 1.2 kilobase pairs from the 5Ј-and 3Ј-untranslated regions of OST6 were amplified from genomic wild-type DNA by PCR and ligated to an URA3 cassette. The 5Ј fragment covered Ϫ840 to ϩ58 base pairs (relative to the start ATG of the OST6 locus), the 3Ј fragment 41 base pairs upstream of the stop codon and 1159 base pairs of the 3Ј-untranslated region. Successful replacement of the OST6 locus was confirmed by PCR using genomic DNA from uracil prototrophic transformants.
Isolation of Microsomal Membranes and Preparation of Solubilized Enzyme Extract-Rough microsomal membranes were isolated as described (21). The membranes were resuspended in membrane storage buffer (30 mM Tris-HCl (pH 7.5), 3 mM MgCl 2 , 1 mM DTT, 35% glycerol, v/v) at a concentration of 12 mg/ml protein. For solubilization, membranes were adjusted to 15 mM Tris-HCl (pH 7.5), 1.5 mM MgCl 2 , 0.5 M KCl, 0.5 mM DTT, and 17% glycerol. After adding 1% Nikkol (final concentration), the reaction was incubated 20 min on ice. The solubilized extract was then separated from insoluble material by centrifugation at 150,000 ϫ g for 40 min. Unless otherwise stated, all steps were carried out at 4°C.
In Vivo Labeling with [ 35 S]Methionine and Immunoprecipitation-Yeast cells were grown to mid-log phase in yeast complete medium. 1.5 ϫ 10 8 cells were collected at 1,200 ϫ g for 3 min, washed once, and resuspended in 10 ml of minimal medium (0.67% yeast nitrogen base, 2% glucose, and the appropriate amino acids). After a 30-min adaptation at 30°C, or 25°C in case of temperature-sensitive strains, 75 Ci of [ 35 S]methionine was added and cells were labeled for 30 -45 min. In case of invertase, cells were derepressed for 20 min in yeast minimal medium with 0.1% glucose prior to labeling. Incubation was stopped with 10 mM NaN 3 and any 35 S label not incorporated was removed by washing the cells once with 10 mM NaN 3 . Cells were resuspended in 300 l of buffer (50 mM Tris-HCl (pH 7.5), 1 mM EtSH, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine) and broken with glass beads for 10 min on a Vibrax shaker at 4°C. After collecting the supernatant, the glass beads were washed once with 300 l of buffer. The extract was centrifuged for 30 min at 48,000 ϫ g, and the supernatant was incubated with 1% SDS for 5 min at 95°C. SDS was then reduced to Ͻ 0.2% by addition of TNET (50 mM Tris-HCl (pH 7.5), 3 mM EDTA, 150 mM NaCl, 1% Triton X-100), the appropriate antiserum was added, and the mixture incubated at 4°C over night. For precipitation of the antigen/ IgG complexes, 5 mg of Protein A-Sepharose equilibrated in TNET buffer were incubated with the solubilized extract by rolling end over end for 2 h at 4°C. The Sepharose beads were centrifuged for 1 min at 14,000 ϫ g, washed four times with 1 ml of TNET buffer and once with 1 ml of 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, resuspended in 50 l of SDS-PAGE sample buffer, and incubated at 95°C for 5 min. After another centrifugation, the supernatant was analyzed by SDS-PAGE on 8% polyacrylamide gels followed by autoradiography.
Denaturing SDS-polyacrylamide gel electrophoresis was carried out according to Refs. 43 or 44, respectively, and visualized by silver staining (45). For Western blotting, proteins were transferred to nitrocellulose or Immobilon (in case of BN-PAGE) membranes using the semidry blotting technique (46).
Immunodetection was performed according to standard procedures and was visualized by the ECL method (Amersham Pharmacia Biotech). Polyclonal antisera against Wbp1p and Swp1p were raised in rabbits using synthetic peptides corresponding to the 16 C-terminal amino acids coupled to bovine serum albumin with glutaraldehyde (47) as described (21). The polyclonal anti-Ost3p and Ost6p antisera were raised against amino acids 26 -350 and amino acids 25-192, respectively, fused to the maltose-binding protein from E. coli. The fusion protein was isolated by amylose column chromatography using the protein fusion and purification system from New England Biolabs and used for immunization of rabbits.
Quantification of Glycosylation-The average number of oligosaccharides for the various glycoproteins was determined by scanning of ECL images using a Bio-Rad MultiImager, or by radioanalytic scanning of fluorographs using a Packard PhosphoImager. The ratio between the average number of oligosaccharides on a given glycoprotein in the different mutants and the average number of glycan chains for that glycoprotein in the wild-type strain was expressed as the percentage of glycosylation relative to wild-type. In

RESULTS
Isolation and Sequence Analysis of the OST6 Gene-In the process of cloning and characterizing the OST3 gene encoding the 34-kDa subunit (our accession number X79596, GeneEMBL data base) a report appeared dealing with the same gene (32). Our nucleotide sequence analysis is in agree-ment with the published data. However, the N terminus of the mature Ost3 protein, isolated by us and determined by gas phase sequencing, was found to start three amino acids further downstream at residue 26 with asparagine (Fig. 1A). We do not know whether this difference is strain-specific or due to proteolytic degradation. A search of protein sequence data bases using the BLASTP protein sequence comparison algorithm (48) revealed a homology between Ost3p and an open reading frame of unknown function on chromosome XIII of S. cerevisiae (YML019W). The sequence identity (20%) and similarity (46%) between OST3 and the yeast ORF, which we named OST6 on account of its function in N-oligosaccharide transfer (see below), are rather modest (Fig. 1A). However, the two proteins have a strikingly similar membrane topology (Fig. 1B) with a typical N-terminal cleavable signal sequence (49) and a Cterminal arrangement of four transmembrane domains using the algorithm of Kyte and Doolittle (50). Searches of protein sequence data bases have revealed that OST3 and OST6, respectively, have homology to three other ORFs: to an ORF of Caenorhabditis elegans on chromosome 3 (identity 21%, similarity 46%) encoding a 37.7-kDa protein, to a candidate human tumor suppressor gene on chromosome band 8p22, named N33 (51), and to the mammalian protein ␣TRAP, a component involved in protein translocation into the ER (identity 23%, similarity 48%; in both cases) (52). Whereas the first two proteins have a similar hydropathy profile, and thus may be structural homologues to Ost3p/Ost6p, ␣TRAP is a type I membrane protein with only one transmembrane domain and probably different function. OST6 encodes a protein of 332 amino acid residues with a calculated molecular mass of 37.9 kDa, but runs on SDS gels as 32 kDa, somewhat faster than Ost3p (34 kDa) and hardly to resolve from Swp1p (see below). The presumed signal sequence cleavage site (Fig. 1A) is in agreement with the predictive method of von Heijne (53). A consensus site for N-glycosylation is located at Asn-175, but not used, since endo H treatment or the N-glycosylation inhibitor tunicamycin do not alter the mobility of Ost6p (data not shown). Thus OST6, like OST3, is not a glycoprotein. A genomic clone of OST6 was obtained by PCR techniques amplifying a fragment that comprises the sequence from 233449 through 234455 from chromosome XIII (YML019W).
OST3 and OST6 Are Required for N-Glycosylation in Vivo, and Their Disruption Leads to a Synthetic Phenotype-To analyze the function of OST6 and OST3 and, in particular, to disclose a possible distinction between the two genes, we examined the glycosylation pattern of well characterized soluble and membrane glycoproteins in single and double mutant strains. Either cell extracts from metabolically labeled cells followed by immunoprecipitation and gel electrophoresis (in case of carboxypeptidase, invertase, dipeptidyl aminopeptidase B), or protein immunoblots (in case of Wbp1p) were analyzed. As depicted in Fig. 2 and Table I, soluble vacuolar carboxypeptidase Y (CPY) shows in single mutants only a slight reduction in glycosylation indicated by the appearance of a minor band of CPY lacking one N-linked chain; the defect in ⌬ost3 is consistently more pronounced than in ⌬ost6. Mature CPY contains four N-linked core type oligosaccharide chains and has a molecular mass of 61 kDa (54). In striking contrast, the double mutant ⌬ost3⌬ost6 severely underglycosylates CPY and lacks one, two, and, to some extent, also three and four chains. Such a reduction in glycosylation can also be observed in OST mutants defective in the essential genes WBP1 (28) or OST1 (27). All underglycosylated bands were endo H-sensitive and shifted to a 53-kDa band, the predicted size of unglycosylated mature CPY, indicating that the bands differ in oligosaccharide content rather than in polypeptide mass (data not shown). A similar underglycosylation pattern in the order ⌬ost6 Ͻ ⌬ost3 Ͻ ⌬ost3⌬ost6 was demonstrated also for the soluble, extracellular glycoprotein invertase (only the core glycosylated ER form has been considered, since the high molecular weight cell wall form does not allow a reliable quantification of the glycosylation defect), for the vacuolar membrane protein dipeptidyl aminopeptidase B (DPAP B) and for the OST subunit Wbp1p (Fig. 2; Table I). The glycoproteins vary to some extent in the degree of underglycosylation. Our results, however, do not allow to postulate a bias for underglycosylation of membrane glycoproteins, as has been previously discussed for the case of ost3 (32).
Since OST3 and OST6 show a synthetic phenotype, we asked whether an overexpression of OST6 in a multicopy vector can functionally replace OST3 and thus cure underglycosylation of CPY in an ⌬ost3 single and an ⌬ost3⌬ost6 double mutant, respectively. Indeed, this is the case as shown in Fig. 3, lanes 1  and 4. No suppression of CPY underglycosylation was observed, when OST1, WBP1, or SWP1 were overexpressed (data not shown). We did not examine overexpression of OST3 in an ⌬ost6 mutant, since the underglycosylation defect in this strain is rather mild.
Ost3p and Ost6p Influence OST Activity in Vitro, but Not Formation of Lipid-linked Oligosaccharide-The phenotype of underglycosylation of proteins could be attributed to a defect in the assembly of lipid-linked oligosaccharide (alg mutants) (55)(56)(57) or to a reduced OST activity (27,28,30). To clarify this, we analyzed the dolichol-linked oligosaccharides from isogenic wild-type, ⌬ost3, ⌬ost6, and ⌬ost3⌬ost6 cells labeled with [ 3 H]mannose. In all strains synthesis of fully assembled core oligosaccharide was observed (Fig. 4). The oligosaccharide pattern in the double mutant was slightly shifted toward Glc 3 Man 9 GlcNAc 2 , which is in accord with the observed underglycosylation in this strain. We conclude that the glycosylation defect is not due to a specific defect in the biosynthesis of the lipid-oligosaccharide precursor, as is the case in different alg mutants.
Next, we tested whether OST activity was affected in vitro in these strains. Since we had previously observed that the stt3-3 mutation results in a sensitivity of OST toward only a suboptimal, non-full-length substrate (35), two different glycosyl donor substrates (lipid-linked chitobiose and lipid-linked Glc 3 Man 9 GlcNAc 2 ) were used; the synthetic hexapeptide Tyr-Asn-Leu-Thr-Ser-Val served as the oligosaccharide acceptor (16). As shown in Table II, the measured OST activity reflects the underglycosylation observed in vivo. Whereas the single mutations display only a small decrease in activity, the ⌬ost3⌬ost6 double mutant is severely affected, in a similar fashion as, e.g., a wbp1 mutant (28). OST activity in ⌬ost6 using the chitobiose donor was consistently somewhat lower than in ⌬ost3, opposite to the situation in vivo; however, glycosyl transfer using the full-length substrate is in accord with the in vivo data. The data also indicate that OST6 may influence the glycosyl donor recognition, but compared with stt3-3 only to a very minor extent.
We have also determined various enzyme activities involved in biosynthesis of lipid-linked oligosaccharide such as DolP-Man, DolP-Glc, and DolPP-GlcNAc 2 , respectively, and found no differences between the mutants and wild-type (data not shown). Similarly, biosynthesis of O-linked glycans, which is initiated in yeast in the ER (3, 58), is not affected. Chitinase, an exclusively O-mannosylated glycoprotein, is not altered in its mobility when analyzed by Western technique and, furthermore, no defect in O-mannosylation activity is detectable in vitro (data not shown).
Ost3p and Ost6p Are Not Involved in Peptide Substrate Recognition but Are Needed for Optimal Oligosaccharyltransfer-In search of a specific function for OST3 and OST6, we considered among other possibilities that these genes could be involved in peptide substrate recognition. To address this point, we expressed simultaneously with wild-type CPY a recombinant form (in single copy as is the case for the wild-type form) containing only one of the four carbohydrate chains, either A, B, C, or D (Fig. 5A). The corresponding carboxypeptidase species are called CPY-A, CPY-B, CPY-C, and CPY-D, respectively. As shown in Fig. 5B, both in the ⌬ost3 single and in the ⌬ost3⌬ost6 double null mutant, all four recombinant CPY forms are glycosylated. This indicates that Ost3p and   Ost6p, respectively, are not involved in the recognition of a particular glycosylation site of the nascent polypeptide chain. The slight increase in mobility of CPY-C mutant (Fig. 5B, lane  3) is an inherent feature of the recombinant form and not due to altered glycosylation (38); it is also observed when expressed in a wild-type strain (Fig. 5B, upper panel).
In order to estimate the degree of glycosylation of the wildtype and recombinant CPY forms in the ⌬ost3⌬ost6 double mutant, the following calculation was made. Since the recombinant glycosylated and unglycosylated CPY forms comigrate with the Ϫ3 and Ϫ4 wild-type forms, respectively, the amount of radioactivity of the Ϫ3/Ϫ4 bands of the control (lane 5) was subtracted from the corresponding bands in lanes 1-4 to give the amount of radioactivity in the recombinant forms (assuming that underglycosylation of wild-type CPY is the same, when both species are expressed). We calculate that in ⌬ost3⌬ost6 cells wild-type CPY contains on average 1.8 instead of four N-linked saccharide chains, corresponding to 45% glycosylation compared with wild-type cells (Table I). In the recombinant forms, having only one glycosylation site, 0.58 chains of CPY-A, 0.75 of CPY-B, 0.71 of CPY-C, and 0.65 of CPY-D are glycosylated, corresponding on average to 67% glycosylation. Thus, a nascent polypeptide with multiple glycosylation sites is somewhat more affected, indicating that OST activity is limiting in the double mutant.
In order to exclude a defect in the protein translocation machinery that could also account for the underglycosylation, we investigated by Western analysis whether the amount of components of the translocon, such as Sec61p, Sec62p, and Sec63p, was altered in the ⌬ost3⌬ost6 double mutant. However, no difference was observed as compared with wild-type cells (data not shown).
Growth Phenotypes of OST3 and OST6 Disruptants-Both OST3 and OST6 are not essential for yeast growth, and also the double knock-out mutant does not exhibit an altered growth rate or morphology as compared with wild-type when tested up to 37°C in complete medium (data not shown). Mutations in other OST subunits, like WBP1, OST1, or OST2, cause temperature-sensitive growth defects (27,28,30). However, we noticed that ost3 and ost6 null mutants display at 30°C differential phenotypes, when stressed by agents that interfere with cell wall biogenesis, such as Calcofluor White, caffeine, or SDS. Cells with weakened cell walls are sensitive against Calcofluor White (CFW) (59) at concentrations that do not impair growth of wild-type cells. As shown in Fig. 6A, growth of the ⌬ost6 disruptant is completely abolished in the presence of 5 g/ml CFW, whereas in ⌬ost3 no inhibition occurs. A similar effect is also exhibited by the phosphodiesterase inhibitor caffeine (Fig. 6B). Evidence has been obtained (60 -62) that a protein kinase C is a central element in the regulation of cell integrity via regulating cell wall formation. Likewise, ⌬ost6 but not ⌬ost3 cells are sensitive to caffeine and SDS. In the case of ⌬ost3, it was found that the mutant is even somewhat more resistant against SDS, compared with the isogenic wild-type strain (Fig. 6C). Various mutants with a defective cell wall biosynthesis or N-glycosylation were reported to reveal an SDS sensitivity (63,64). The different behavior of ⌬ost6 and ⌬ost3 null mutants toward these compounds could mean that both genes may in part be involved in different functions important for cell wall biogenesis and/or stability.
Subunit Composition of the OST Complex by Blue Native and Denaturing Electrophoresis-Our affinity purification procedure using an anti-Wbp1p affinity column (21) led to the isolation of a tetrameric, enzymatically active OST complex consisting of Ost1p, Wbp1p, Ost3p, and Swp1p, whereas the complex purified by Kelleher and Gilmore (22) contained in addition Ost2p and Ost5p. In none of the two procedures the essential and genetically identified subunits Stt3p and Ost4p, or the newly discovered Ost6p were present in detectable amounts by silver or Coomassie staining. Using the more sensitive method of Western analysis, we show now that the tetrameric OST preparation from wild-type cells contains also Stt3p, Ost2p, and Ost5p, as well as Ost6p (no antibodies are available for Ost4p; Fig. 7B). We find that Stt3p, which has a predicted molecular mass of 78 kDa, migrates on SDS gel as a diffuse band with an anomalous mobility of ϳ60 kDa that is not resolved from Ost1p (lanes 1 and 3). Proof for the correct identification of the Stt3p signal comes from the observation that it disappears, when Stt3p depleted membranes were analyzed (data not shown). Ost6p turns out to have a mobility almost identical to that of Swp1p and, therefore, may have escaped detection so far.
Affinity purification and silver staining (Fig. 7A) of OST from ⌬ost3⌬ost6 cells reveals a reduction of Ost1p and, as expected, a lack of Ost3p (and Ost6p by Western blot, see Fig. 7B). Ost1p and Wbp1p, respectively, the two glycosylated subunits of the complex are partly underglycosylated (Fig. 7, A, lane 2; B, lanes  4 and 6). By Western analysis, the presence of Stt3p, Ost2p, and Ost5p could also be demonstrated. Furthermore, a comparison of the composition of the complexes from wild-type strain and ⌬ost3⌬ost6 double mutant (Fig. 7B) indicates that the amount of Stt3p, Ost1p, and Ost5p is reduced. Thus, the structural organization of OST is severely affected and leads, as shown above (Table II), to a defect in enzyme activity, observed so far for cells lacking an essential OST protein.
In order to estimate the size of the native complex and identify possible subcomplexes (see below), we applied the method of blue native-polyacrylamide electrophoresis that was reported to be a powerful tool for the separation and size determination of native membrane bound complexes (41,42,65). By this technique, in combination with Western analysis, we can identify in wild-type membranes a complex of ϳ240 kDa (Fig. 8), in which all OST subunits are assembled. In the case of Wbp1p, we also observe in low amounts an additional band (larger than the expected mass of Wbp1p) that is recognized by the antibody; its origin remains obscure, however. The differences in signal intensities do not reflect different amounts or the stoichiometry of the subunits in the complex, but are rather due to differences in the antibody titer. DISCUSSION We have described the identification and functional characterization of OST6, a novel gene encoding a subunit of the OST from yeast. Ost6p is a 32-kDa transmembrane protein not disclosed in any of the previously purified OST complexes. It is homolog to Ost3p, the 34-kDa ␥-subunit of yeast OST (32). The identity between both proteins with 21% is rather low, but they display a strikingly similar membrane topology containing four predicted C-terminal membrane-spanning domains. Such a hydropathy profile and the same degree of identity (23%) is also found for a Caenorhabditis elegans ORF of unknown function and a human candidate tumor suppressor. The relationship of OST6 and OST3 to a neoplastic phenotype is obscure, but altered glycosylation of cell surface proteins is a well known feature of tumor cell lines (66,67).
Disruption of OST3 was reported (32), and is confirmed in the present study, to cause an only moderate underglycosylation of proteins in vivo and a small decrease in OST activity measured in vitro. In the case of disruption of OST6 the underglycosylation defect is even less. However, an ⌬ost3⌬ost6 double mutation leads to a synthetic phenotype with a severe underglycosylation, both in vivo and in vitro ( Fig. 2; Table I) (21) were analyzed on a 12.5% SDS gel (43). B, Western analysis. 15 g of protein were loaded on a 10% SDS-polyacrylamide gel (lanes 1-8) or a 10% Tris-Tricine gel (lanes 9 -16), respectively, and a Western blot analysis was performed with antibodies to Stt3p, Ost1p, Wbp1p, Ost3p, Ost6p, Swp1p, Ost2p, and Ost5p as indicated and visualized by ECL detection. In the left lanes within each antibody column, OST from wild-type is loaded; right lanes contain OST from ⌬ost3⌬ost6 . is the case for ts mutations of essential OST genes (27,28,30). Surprisingly, however, growth is not impaired at temperatures up to 37°C. Overexpression of Ost6p not only complements the ⌬ost6 knockout, but also rescues the ⌬ost3 underglycosylation effect. Although we have not performed the reverse experiment (due to the very mild ⌬ost6 defect), it seems that both genes have in part a redundant function and are able to partially replace each other. On the other hand, they also reveal specific effects (cf. Fig. 6). We observed that upon stressing the cells by compounds interfering with cell wall biogenesis, like Calcofluor White, caffeine, or SDS, an ⌬ost6 mutant behaves differently compared with an ⌬ost3. Even though no distinct target reaction can be given for these findings, they may eventually help to unravel the complexity of formation and function of N-linked saccharide chains.
Previous biochemical investigations of the yeast OST suggested that the enzyme consists of four (Ost1p, Wbp1p, Ost3p, Swp1p) (21,23), five (68), or six subunits (Ost1p, Wbp1p, Ost3p, Swp1p, Ost2p, Ost5p) (22). Genetic screens have identified in addition STT3 (35) and OST4 (34). In particular the lack of the essential and highly conserved Stt3p in all OST preparations was puzzling, and it was hypothesized that Stt3p could be a substoichiometric assembly factor (35). We have shown now by more sensitive probing with antibodies that the "tetrameric," enzymatically active complex isolated by us (21) also contains Stt3p, Ost2p, Ost5p, and the newly discovered Ost6p. The present study, and also recent experiments employing affinity purification of tagged Stt3 protein, or analysis of co-immunoprecipitates of in vivo radiolabeled subunits clearly identify Stt3p (36,37) and also Ost4p (37) as components of the complex. Moreover, an estimation of the amount of radioactivity incorporated into the subunits of the co-immunoprecipitates (37) is compatible with the notion that the eight subunits identified therein (Stt3p, Ost1p, Wbp1p, Ost3p, Swp1p, Ost2p, Ost5p, and Ost4p) may be present in equimolar amounts. Thus, the underrepresentation of some subunits in the various complex preparations seems to be due to their depletion during isolation rather than to a real substoichiometric participation in the in vivo complex. This interpretation is also in agreement with our analysis of the native complex composition using the method of blue native electrophoresis. A complex in the range of ϳ240 kDa was found, which agrees with a calculated molecular mass assuming all 9 subunits are present in equimolar amounts. Nevertheless, the isolation of an active "tetrameric" complex may indicate that for the actual catalysis of the glyco-sylation reaction less subunits are sufficient, and the additional components may be essential or important only for in vivo function.
OST3 and OST6 are not essential for OST activity, but their simultaneous lack drastically decreases glycosylation in vitro and in vivo. Therefore, one could envisage that the products of these genes are needed for optimal OST activity either directly by interacting with and regulating the catalytic subunit, or indirectly by affecting the assembly of an optimal complex, or being involved in proper positioning the OST to the polypeptide at the translocation site or to the site of formation of the lipid-linked oligosaccharide precursor. Analysis of the subunit composition of the affinity-purified complex in the ⌬ost3⌬ost6 double mutant by Western analysis clearly indicates a severe defect in the structural organization of the complex (Fig. 7).
Studies demonstrating genetic interactions among different OST genes, either by using a high copy number suppression approach or by constructing double mutants with a synthetic phenotype, have led to the suggestion that OST subunits can be sorted into three groups: I, Ost1p-Ost5p; II, Wbp1p-Swp1p-Ost2p; and III, Stt3p-Ost4p-Ost3p (15,36,37). The structural similarity of Ost3p and Ost6p, their synthetic interaction, as well as the suppression of ⌬ost3 by OST6 overexpression suggests a grouping of Ost6p to (III). Complementary biochemical evidence indicates a direct physical interaction between Wbp1p and Swp1p (29) and Ost2p, 2 as well as between Stt3p, Ost3p, and Ost4p (37), supporting the idea that these groups may represent discrete subcomplexes. In this context, it is interesting to note that upon simultaneous disruption of the OST3 and OST6 genes, Ost5p and Ost1p (subcomplex I) and Stt3p (subcomplex III) are decreased, whereas Wbp1p, Swp1p, and Ost2p (subcomplex II) are not affected in their amount (Fig. 7). Since glycosylation still occurs in the ⌬ost3⌬ost6 strain, albeit reduced, one could speculate that Wbp1p, Swp1p, and Ost2p represent the catalytic core unit of the complex.
At present it is not clear, however, whether the proposed subcomplexes comprise intermediates in the assembly of a fully functional complex, or whether they are autonomous in vivo pools that may assemble into OST complexes with slightly different composition and function. Recent findings indicate the occurrence of multiple pathways of protein translocation into the ER (69 -72), one dependent on the signal-recognition particle and the other independent. The identified translocation complexes share in part subunit components. The fact that no distinct OST subcomplexes can be detected in wild-type cells by blue native polyacrylamide gel electrophoresis makes it less likely that OSTs of different composition are associated with the respective translocation machineries. In agreement with this idea is also our observation that in the ⌬ost3⌬ost6 mutant proteins are underglycosylated irrespective whether they use the SRP-dependent (e.g. invertase, DPAP B) or the SRP-independent (e.g. CPY) sorting pathway. Nevertheless, this view does not exclude specific interactions of particular OST subunits and/or not yet identified auxiliary proteins with the different targeting pathways.
The combination of biochemistry and powerful yeast genetic methods has advanced tremendously our knowledge of the N-glycosylation of proteins. Now that the composition of one of the most complex enzymes in nature as well as some of the interactions between its nine subunits has been defined, further work needs to concentrate on a number of intriguing issues. These include the specific functions of the various subunits of the complex, the regulation of the enzyme and the coupling of OST to protein translocation and protein folding. Finally, the results obtained in yeast may also lead to the isolation of the homologous mammalian proteins, not yet identified in respective complexes.