The Highly Conserved Stt3 Protein Is a Subunit of the Yeast Oligosaccharyltransferase and Forms a Subcomplex with Ost3p and Ost4p*

The oligosaccharyltransferase has been purified from Saccharomyces cerevisiae as an hetero-oligomeric complex composed of four or six subunits. Here, the in vivosubunit composition and stoichiometry of the oligosaccharyltransferase were investigated by attaching an epitope coding sequence to a previously characterized subunit gene, OST3. Five (Ost1p, Wbp1p, Swp1p, Ost2p, and Ost5p) of the seven polypeptides that were coimmunoprecipitated with the epitope-tagged Ost3p were identical to those obtained by the conventional purification procedure. Two additional coprecipitating polypeptides with apparent molecular masses of 60 and 3.6 kDa were identified as the 78-kDa Stt3 protein and the 36-residue Ost4 protein, respectively. Stt3p and Ost4p were previously identified in screens for gene products involved inN-linked glycosylation. Quantification of the in vivo radiolabeled subunits and the radioiodinated purified enzyme shows that the yeast oligosaccharyltransferase is composed of equimolar amounts of eight subunits. Exposure of the immunoprecipitated oligosaccharyltransferase to mild protein denaturants yielded a subcomplex comprised of Stt3p, Ost3p, and Ost4p. These experiments, taken together with genetic and biochemical evidence for subunit interactions, suggest that the enzyme is composed of the following three subcomplexes: (a) Stt3p-Ost4p-Ost3p, (b) Swp1p-Wbp1p-Ost2p, and (c) Ost1p-Ost5p.

N-Glycosylation of proteins is an essential, highly conserved protein modification reaction that occurs in all eukaryotic organisms. The oligosaccharyltransferase (OST) 1 catalyzes the transfer of a preassembled high mannose oligosaccharide (Glc 3 Man 9 GlcNAc 2 ) onto Asn-X-Ser/Thr acceptor sites on nascent polypeptides as they are translocated into the lumen of the rough endoplasmic reticulum (for reviews see Refs. 1, and 2). Biochemical, molecular biological, and genetic studies have led to the identification of a surprisingly large number of proteins that are required for the expression of wild-type OST activity.
The yeast OST was initially purified as an oligomeric complex composed of six subunits that are designated as Ost1p (62/64 kDa), Wbp1p (45 kDa), Ost3p (34 kDa), Swp1p (30 kDa), Ost2p (16 kDa), and Ost5p (9 kDa) (3). However, catalytically active tetrameric OST complexes that appear to lack Ost2p and Ost5p have also been described (4,5). In addition, the 34-kDa Ost3 protein appears to be present in reduced amounts relative to the other three subunits in the purified OST tetramer (4), raising the possibility that oligomeric forms of the OST may exist that differ with respect to the presence of regulatory or accessory subunits.
Prior to purification of the yeast enzyme, genetic and biochemical studies had established that Wbp1p and Swp1p were essential for in vivo and in vitro expression of OST activity (6,7). Mutations in genes encoding Ost1p, Ost2p, Ost3p, and Ost5p also cause substantial reductions in both the N-linked glycosylation of proteins in vivo and the transfer of dolichollinked oligosaccharides to acceptor peptide substrates in vitro (8 -11). Defects in the transfer and modification of N-linked oligosaccharides cause morphological, biochemical, and structural alterations in the yeast cell wall (12). Yeast mutants that are defective in the synthesis of the fully assembled dolichol oligosaccharide donor (alg mutants, i.e. asparagine-linked glycosylation) or in the elongation of the N-linked oligosaccharide display an enhanced resistance to sodium vanadate (13) and are hypersensitive to aminoglycoside antibiotics (14). The ost4 mutants, which were isolated based upon enhanced resistance to sodium vanadate, express greatly diminished OST activity in vivo and in vitro (15). The OST4 gene encodes a 36-residue hydrophobic protein that was not detected when the OST was purified.
The glycosylation defect caused by the wbp1-2 mutation is exaggerated when assembly of the optimal oligosaccharide donor (Dol-PP-GlcNAc 2 Man 9 Glc 3 ) for the OST is prevented by a mutation in the ALG5 gene, resulting in a synthetic lethal phenotype (16). Based upon this observation, two genetic screens were devised, one of which was selective for mutations that affect assembly of the oligosaccharide donor (17), and a second that was selective for genes encoding the OST subunits (11). In addition to mutant alleles of genes that encode six of the known OST subunits (Wbp1p, Swp1p, Ost1p, Ost2p, Ost3p, and Ost5p) the latter screen yielded mutants in the STT3 locus (11). Because the 78-kDa Stt3 protein had not been detected as a subunit in the purified OST complex (3,4,5), several alternative roles were proposed for the Stt3p (17).
To determine the composition and subunit stoichiometry of the native yeast OST, we appended an epitope recognized by an antibody raised against the influenza virus hemagglutinin (HA) to either the C terminus of Ost3p or Stt3p and expressed the epitope-tagged proteins in Saccharomyces cerevisiae. Nondenaturing immunoprecipitation of the OST from radiolabeled cultures of a yeast strain expressing the epitope-tagged Ost3 protein showed that the yeast enzyme is composed of eight subunits (Stt3p, Ost1p, Wbp1p, Ost3p, Swp1p, Ost2p, Ost5p, and Ost4p) in approximately equimolar amounts. After exposure to mild denaturants Stt3p, Ost3p, and Ost4p remain bound to the immunoaffinity reagent suggesting that these subunits form a stable subcomplex within the oligosaccharyltransferase. These results are discussed in the context of a model for the structural organization of the OST complex.
⌬ost4 Strains-A two-step PCR-based gene disruption method (25) was used to produce a linear DNA fragment containing a heterologous HIS3 marker flanked by 5Ј (Ϫ215 to 88) and 3Ј (199 to 398) untranslated homology regions from the OST4 gene (15). The pFA6a-HIS3MX6 plasmid, which was used as a template for the PCR reaction, was provided by Dr. Peter Philippsen (University of Basel, Switzerland). RGY330 and RGY340 were transformed to histidine prototrophy using the PCR product to obtain strains designated as RGY333 and RGY344. Integration of the HIS3 marker into the OST4 gene was confirmed by PCR.

Cell Labeling, Extract Preparation, and Immunoprecipitation
Yeast cells were grown for 15-20 h at 25°C in synthetic minimal media (2% glucose) supplemented with the appropriate amino acids until mid log phase (A 600 of 0.8 -1.6). Cells were collected by centrifugation and resuspended at 5 A 600 /ml in minimal medium. Cells were labeled for 20 min or 1 h, as indicated, with 100 Ci of [ 35 S]methionine (Ͼ1000 Ci/mmol, NEN Life Science Products) per A 600 units of cells. Labeling was terminated by the addition of NaN 3 to 10 mM. Hereafter, all the procedures were carried out at 4°C. To prepare lysates for nondenaturing immunoprecipitation, cells were washed once with 20 mM Tris-Cl (pH 7.4), 100 mM NaCl, and resuspended in 5% glycerol, 20 mM Tris-Cl (pH 7.4), 5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, and a protease inhibitor mixture (3). After rapid lysis of cells with glass beads, the homogenates were adjusted to 1.5% digitonin, 0.5 M NaCl, 20 mM Tris-Cl (pH 7.4), 3.5 mM MgCl 2 , and 1 mM MnCl 2 . The extracts were centrifuged for 20 min at 135,000 ϫ g in an airfuge (Beckman Instruments, Palo Alto, CA) using a A-100/30 rotor, and the clarified supernatants were used for immunoprecipitations. For immunoprecipitation with the HA.11 antibody (Babco, Richmond, CA), cell lysates from radiolabeled cells were mixed with unlabeled lysates prepared from an equal quantity of cells that do not express the HA-tagged protein. Extracts that were precleared by prior incubation with protein A-Sepharose beads were incubated overnight with the HA.11 antibody. After a 2-h incubation with the protein A-Sepharose beads, the beads were recovered by centrifugation and washed three times with Nikkol buffer (1 M NaCl, 50 mM Tris-Cl (pH 7.4), 1 mM MgCl 2 , 1 mM MnCl 2 , 1 mM CaCl 2 , 0.02% Nikkol (Nikko Chemical Co., Ltd., Tokyo, Japan)) and once with 50 mM Tris-Cl (pH 7.4), 150 mM NaCl, 5 mM EDTA. Denaturing immunoprecipitations were performed as described previously (28). Endoglycosidase H (Endo H; New England Biolabs, Beverly, MA) digestions were performed according to the manufacturer's recommendations. Immunoprecipitated proteins were incubated for 30 min at 55°C in SDS sample buffer and resolved on either Tris glycine-buffered or Tris-Tricine-buffered SDS-polyacrylamide gels (29).

Protein Immunoblots and Radioiodination of the Purified OST Complex
Microsomal membranes were isolated from wild-type or mutant cells as described (10). Microsomes or samples from nondenaturing immunoprecipitations were resolved by SDS-PAGE and transferred to PVDF membranes. The membranes were probed with antibodies to Ost1p, Wbp1p, the HA epitope, Swp1p, Ost2p, and Ost5p as described (10). Antibody to Ost5p was raised against the N-terminal sequence that precedes the first membrane spanning segment and was provided by Dr. Markus Aebi (ETH, Zurich, Switzerland). Horseradish peroxidaseconjugated secondary antibodies were visualized using enhanced chemiluminescence (ECL Western blotting detection kit, Amersham Corp.).
Radioiodination of the purified yeast OST complex using a chloramine-T oxidation procedure was performed after denaturation in SDS as described previously for the canine OST (30).

RESULTS
Immunoprecipitation of the Yeast OST Complex-A 30-amino acid sequence corresponding to three tandem repeats of the HA epitope was appended to the C terminus of Ost3p to serve as an affinity tag for immunopurification of the yeast OST. A one-step allele replacement method was used to substitute the epitope-tagged OST3 allele for the chromosomal copy of OST3 in a haploid yeast strain. Replacement of the wild-type Ost3 protein with the epitope-tagged Ost3 protein had no detectable effect on cell growth or in vivo OST activity indicating that HA-tagged Ost3p is fully functional (not shown).
Protein immunoblot and denaturing immunoprecipitation experiments showed that the monoclonal anti-HA antibody recognized a protein of about 36 kDa in total cell extracts prepared from a strain that expresses the HA-tagged Ost3 protein but not from a strain that expresses untagged Ost3p (not shown). To identify proteins that were specifically associ-ated with Ost3p under nondenaturing conditions, total cell homogenates prepared from [ 35 S]methionine-labeled yeast cultures were solubilized with the nonionic detergent digitonin. Six radiolabeled polypeptides ranging in molecular mass from 3.6 to 64 kDa were immunoprecipitated using the monoclonal antibody to the HA epitope from extracts containing the HAtagged Ost3p (Fig. 1A). Coimmunoprecipitation of the radiolabeled polypeptides with the HA-tagged Ost3 protein was specific, as none of the proteins were recovered in immunoprecipitates from the control strain that expresses untagged Ost3p. To facilitate the assignment of the previously characterized OST subunits, we transferred the immunoprecipitated proteins from a polyacrylamide gel onto a PVDF membrane for protein immunoblot analysis (Fig. 1B). Each of the six polypeptides that had been described as subunits of the OST complex (3) were unambiguously identified. The single N-terminal methionine residue in the Ost5p sequence is removed from the mature protein (11); therefore, Ost5p was not detected as a radiolabeled product in the immunoprecipitate from [ 35 S]methionine-labeled cells ( Fig. 1A) but was detected when cultures were labeled with a mixture of [ 35 S]cysteine and [ 35 S]methionine (not shown). Protein immunoblots using an antibody raised against the N terminus of Ost5p revealed an immunoreactive 9-kDa protein. We hypothesized that the polypeptide that migrates slightly slower than the 3.4-kDa molecular weight marker is the 36-residue Ost4 protein. Support for this conclusion is presented in Fig. 4.
To calculate the relative molar amounts of the OST subunits in the immunoprecipitates, polyacrylamide gels ( Fig. 1A) containing [ 35 S]methionine-labeled samples were quantified. The relative intensity of each radiolabeled polypeptide should be directly proportional to the methionine content of each subunit provided that the subunits are present in equimolar amounts. The calculated amounts of the OST subunits relative to the tagged Ost3p protein were as follows: Ost1p (3.2), Wbp1p (2.4), Ost3p (1.0), Swp1p (0.9), Ost2p (1.5), and Ost4p (0.4). The apparent excess of Ost1p and Wbp1p relative to the other subunits prompted further examination of this region of the autoradiogram. Protein immunoblots of the native immunoprecipitate revealed a glycoform doublet for Ost1p (Fig. 1B) rather than the broad band that was visible in the native immunoprecipitate (Fig. 1A). A successive immunoprecipitation experiment was performed to determine whether the excess radiolabel in the Ost1p region of the gel could be explained by a fortuitous comigration of Ost1p with another polypeptide. After the OST was immunoprecipitated under nondenaturing conditions from the HA-tagged Ost3p strain ( Fig. 2A, lane b), the immune complexes were denatured in SDS and subjected to a second immunoprecipitation using an antibody to Ost1p. The Ost1p glycoform doublet that was recovered after denaturing immunoprecipitation ( Fig. 2A, lane c) migrated slightly slower than an intensely labeled diffuse band that was recovered in the unbound fraction with the other OST subunits ( Fig. 2A, lane d). Given the genetic evidence supporting a role for the Stt3 protein in N-linked glycosylation (11), the 78-kDa Stt3 protein (17) was the most likely candidate for the diffusely migrating polypeptide.
Subunit Composition and Stoichiometry of the Yeast OST-To obtain more direct evidence that Stt3p is a subunit of the OST, total cell extracts were prepared from [ 35 S]methionine-labeled cultures of a yeast strain that expresses HAtagged Stt3p (Fig. 2B). At least four of the polypeptides (Wbp1p, Swp1p, Ost2p, and Ost4p) that copurified with HAtagged Ost3p (Fig. 2B, lane b) comigrated precisely with proteins that copurified with HA-tagged Stt3p (Fig. 2B, lane c). The identity of the HA-tagged Stt3 protein was confirmed by immunoprecipitating the tagged protein from SDS-denatured cell extracts (Fig. 2B, lane d). The addition of the 3-kDa affinity tag to the C terminus of Stt3p is responsible for the decreased mobility of HA-tagged Stt3p relative to untagged Stt3p (Fig.  2B, compare lanes b and c). Endoglycosidase H digestion of the HA-tagged Stt3p (Fig. 2B, lane e) increased the mobility of the protein consistent with a previous report that the lumenal domain of Stt3p contains at least one N-linked oligosaccharide (17). Deglycosylated Stt3p migrated as a broad band indicating that heterogeneous glycosylation is not responsible for the diffuse electrophoretic migration of the polypeptide. As expected, the untagged Ost3 protein migrated slightly slower than Swp1p (Fig. 2B, lane c).
Although the preceding experiments demonstrated that the immunopurified OST complex contains Stt3p, a protein of this size had not been detected when the hexameric or tetrameric OST complexes were resolved by SDS-PAGE and stained with either Coomassie Blue or with Silver (3)(4)(5). To resolve this discrepancy, the OST was purified as described previously (3), denatured in SDS, and radiolabeled with 125 I. Resolution of the radioiodinated OST subunits by PAGE in SDS revealed an intensely labeled polypeptide that migrated in the vicinity of Ost1p (Fig. 2B) rather than the well resolved Ost1p glycoform doublet that we detect by staining with Coomassie Blue. Radiolabeled bands corresponding to the other OST subunits (Wbp1p, Ost3p, Swp1p, Ost2p, and Ost5p) were also detected. Radiolabeled aprotinin (ϳ6.5 kDa), a protease inhibitor that is included in the buffer used for the OST purification (Fig. 2B), was not effectively separated from 9.5-kDa Ost5p on this gel but was resolved on a 7.5-17.5% polyacrylamide gel (see Fig. 3). We were not able to unambiguously assign a radiolabeled band to Ost4p due, in part, to the presence of a single tyrosine residue in this protein.
The subunit stoichiometry of the yeast OST was recalculated FIG. 1. Immunopurification of the yeast OST using epitopetagged Ost3p. Nondenaturing immunoprecipitates from [ 35 S]methionine-labeled cell extracts prepared from yeast strains that express either untagged Ost3p (Ϫ) or HA-tagged Ost3p (ϩ) were resolved in SDS using Tris-Tricine-buffered 10% polyacrylamide gels. The gels were either (A) dried and exposed to an x-ray film or (B) transferred to a PVDF membrane and probed with antibodies to Ost1p, Wbp1p, HA-Ost3p, Swp1p, Ost2p, and Ost5p. The asterisk indicates the IgG heavy chain. A, light and dark exposures of the gel are shown to permit the detection of Ost2p and Ost4p.
in light of the discovery that Stt3p was not resolved from Ost1p in Fig. 1. The value presented in Table I for the molar ratio of Ost1p and Stt3p relative to Ost3p assumes that equal amounts of Ost1p and Stt3p are present. The results of this analysis suggest that the OST subunits are present in equimolar amounts in the immunopurified complex. Wbp1p and Ost2p each contain a single methionine residue; consequently, the values we obtain for these subunits are subject to the greatest error. The subunit stoichiometry was also calculated after quantification of the radioiodinated OST subunits (Fig. 2B and   Fig. 3). If the OST complex contains equimolar amounts of the subunits, the incorporation of 125 I should be proportional to the tyrosine content of the subunits provided that the radioiodination efficiency of individual tyrosine residues is comparable after denaturation with SDS. Quantification of the 125 I-labeled OST subunits also indicated that the seven larger OST subunits are present in roughly equimolar amounts in the purified yeast OST complex (Table I).
The epitope-tagged Ost3p strain provided a means to determine whether Ost3p is present in reduced stoichiometry relative to the other subunits of the OST complex as suggested by Pathak et al. (4). If that were the case, yeast may contain a population of OST complexes that lack Ost3p in vivo. To address this question, the supernatants from nondenaturing immunoprecipitates of the untagged and epitope-tagged Ost3 strain were precipitated with trichloroacetic acid, resolved by PAGE in SDS, and subjected to protein immunoblot analysis using antibodies specific for the HA epitope, Swp1p and Wbp1p  SDS (a and b) or were denatured and reprecipitated using an antibody to Ost1p (c). Denatured subunits that were not precipitated with the antibody to Ost1p were precipitated with trichloroacetic acid prior to electrophoresis (d). Lanes a-d are derived from a single gel; intervening blank lanes were removed. B, nondenaturing (a-c) or denaturing (d and e) immunoprecipitates from radiolabeled cultures of yeast that express either untagged (Ϫ) or HA-tagged Ost3p or Stt3p (ϩ) were resolved by PAGE in SDS using a 10% Tris-Tricine-buffered gel. The sample in lane e was digested with Endo H prior to PAGE. Samples of 125 I-labeled aprotinin and the 125 I-labeled yeast OST were resolved on the same gel. C, the OST was immunodepleted from digitonin extracts prepared from cells that express untagged (Ϫ) or HA-tagged Ost3p (ϩ) using the anti-HA antibody. The supernatants after immunodepletion (a and b) or the total cell extracts (c) were precipitated with trichloroacetic acid and resolved by SDS-PAGE. After transfer to a PVDF membrane, the blots were probed with antibodies to Wbp1p, the HA epitope, or Swp1p. The IgG heavy chain and several cross-reactive bands are designated by an asterisk and triangles.
FIG. 3. Immunoisolation of the OST complex from the ⌬ost3 mutant. A, the OST was immunoprecipitated with the anti-HA antibody under nondenaturing conditions from radiolabeled cultures of the wild-type (WT) and ⌬ost3 mutant strains that express either untagged (Ϫ) or HA-tagged Ost3p or Stt3p (ϩ). The immunoprecipitates and samples of 125 I-labeled yeast OST and 125 I-labeled aprotinin were resolved by PAGE in SDS using a 7.5-17% gradient gel. The radiolabeled proteins corresponding to OST subunits are labeled. A 31-kDa polypeptide that migrates slightly slower than Swp1p in HA-tagged Stt3p samples is indicated by an arrowhead.  1A) or with 125 I (Fig. 2B and Fig. 3), were estimated by scanning polyacrylamide gels using a PhosphorImager. The signal produced by each of the subunits was normalized to the number of methionine or tyrosine residues predicted by the amino acid sequence of the mature protein as indicated.
b As Ost1p and Stt3p were not resolved on the gels, they were treated as a group, and the signal produced together by both of the subunits was normalized to the total number of methionine or tyrosine residues in these two proteins. (Fig. 2C, lanes a and b). Virtually all of the epitope-tagged Ost3 protein that was detectable in total cell extracts (Fig. 2C, lane  c) was immunodepleted from the supernatant fraction (Fig. 2C,  lane b). The majority of the Wbp1p and Swp1p in the total cell extract (Fig. 2C, lane c) was removed by coprecipitation with epitope-tagged Ost3p (Fig. 2C, lane b). The supernatant fraction from an immunoprecipitate from an untagged strain (Fig.  2C, lane a) did not show reduced amounts of Swp1p and Wbp1p. We conclude that the majority of the OST complexes in the rough endoplasmic reticulum contains Ost3p as a subunit. The low amounts of Swp1p and Wbp1p that were recovered in the supernatant fraction (Fig. 2C, lane b) may be derived from a small population of OST complexes that lack Ost3p as a subunit or may indicate that microsomes contain a small pool of incompletely assembled complexes.
Deletion of Ost3p Does Not Disrupt the Oligosaccharyltransferase Complex-Disruption of the OST3 gene is not lethal in a haploid yeast strain but instead causes hypoglycosylation of proteins in vivo and a 2-fold reduction in the in vitro OST activity (10). Protein immunoblot analysis of the ⌬ost3 mutant did not disclose an obvious reduction in the membrane content of Ost1p, Wbp1p, Swp1p, and Ost2p. Although this suggests that Ost3p is not critical for the assembly or stability of the OST complex, we sought more direct evidence concerning the composition of the oligosaccharyltransferase in the ⌬ost3 mutant. The OST complex was immunopurified from radiolabeled cultures of the wild-type or ⌬ost3 mutant strains expressing epitope-tagged OST subunits (Fig. 3). With the exception of Ost3p, comparable amounts of the radiolabeled OST subunits were present in the nondenaturing immunoprecipitates from the two wild-type strains and the ⌬ost3 strain. These results strongly suggest that Ost3p is not required for the assembly or stability of the OST.
A radiolabeled band of 31 kDa (denoted by an arrowhead) was resolved on the 7.5-17% polyacrylamide gradient gel shown in Fig. 3. The 31-kDa protein was present in the immunoprecipitates from strains expressing HA-tagged Stt3p but not in the untagged control strain or the HA-tagged Ost3p strain. A search of the yeast protein sequence data base revealed a hypothetical protein designated YML019W that is 21% identical in sequence to Ost3p. Moreover, Ost3p and YML019W have a similar predicted arrangement of an Nterminal signal sequence and four membrane spanning segments. The predicted molecular weight of the YML019W (34 kDa) is in reasonable agreement with the estimated size of the 31-kDa polypeptide. Further studies will be required to determine whether the 31-kDa polypeptide is YML019W.
Ost3p, Stt3p, and Ost4p Form a Subcomplex within the Oligosaccharyltransferase-We sought additional evidence to support the hypothesis that the 3.6-kDa polypeptide in the immunopurified OST complex is Ost4p. Because expression of Ost4p is not essential for viability of yeast at 25°C, the OST4 gene was disrupted in yeast strains that express the HA-tagged OST subunits. The compositions of the OST complexes that were immunopurified from the wild-type and ⌬ost4 mutant strains were compared (Fig. 4). With the exception of HA-tagged Ost3p, only trace amounts of the OST subunits were recovered in nondenaturing immunoprecipitates from the ⌬ost4 mutant that expresses HA-tagged Ost3p (Fig. 4, lane b). The 3.6-kDa polypeptide was clearly absent when the OST was immunoisolated from a ⌬ost4 mutant that expresses HA-tagged Stt3p (Fig. 4, lane e). Since the absence of the 3.6-kDa polypeptide in the immunopurified OST complex correlates with disruption of the OST4 gene, this experiment supports the identification of the 3.6-kDa protein as Ost4p. Untagged Ost3p migrated slightly slower than Swp1p on this gel (Fig. 4, lane d); a com-parison of lanes d and e reveals that lane e also lacks Ost3p. The absence of untagged Ost3p in complexes isolated from the ⌬ost4 mutant (Fig. 4, lane e) is consistent with the inability to immunoisolate the OST complex from the ⌬ost4 mutant that expresses HA-tagged Ost3p (Fig. 4, lane b).
As in the preceding experiments, the immunopurified complexes were washed several times at 4°C with a nonionic detergent/high salt buffer (Nikkol buffer) to maintain the integrity of the complex and to prevent nonspecific interactions (Fig. 4, lanes a, b, d, and e). Two sequential washes of the immunoisolated complex with a Triton X-100/SDS-mixed micelle buffer followed by the mixed micelle buffer supplemented with 2 M urea eluted Wbp1p, Swp1p, Ost2p, and Ost1p (Fig. 4,  lanes c and f). Elution of Ost1p from the immunoprecipitates was confirmed by protein immunoblotting using the antibody to Ost1p (not shown). Surprisingly, Stt3p, Ost3p, and Ost4p were retained after successive washes with mixed micelle buffer supplemented with 2 M urea (Fig. 4, lanes c and f) or 500 mM NaCl (data not shown). The recovery of the 36-residue Ost4 protein in the denaturant-washed immunoprecipitates from the strain that expresses HA-tagged Stt3p was lower (Fig. 4, lane f), due, in part, to the reduced recovery of both Stt3p and Ost3p in the immunoprecipitate. Whereas only one of the three OST subunits contained the epitope tag in each strain, retention of all three subunits is diagnostic of an Stt3p-Ost4p-Ost3p subcomplex that is resistant to a short (5-10 min) exposure to mild denaturants. As expected, Ost3p was not present in the subcomplex isolated from the ⌬ost4 mutant (Fig. 4, lane g).
Interaction of Wbp1p with Ost3p-We next asked whether the HA-tagged Stt3 and Ost3 proteins could be used to rapidly isolate the OST from the strains that express mutant OST subunits. The well characterized wbp1-1 and wbp1-2 mutants were selected for this analysis (6, 7). Nondenaturing immuno-  (lanes a, b, d, and e), or Triton X-100/SDS-mixed micelle buffer (50 mM Tris-Cl (pH 7.4), 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, and 0.2% SDS) supplemented with 2 M urea (lanes c, f, and g). The OST subunits were resolved in SDS-PAGE.
precipitates from the epitope-tagged wbp1 mutants were resolved by PAGE in SDS (Fig. 5A). Regardless of whether the cells were radiolabeled for 20 min (not shown) or 1 h at the permissive temperature (Fig. 5A), the yield of all of the OST subunits, including epitope-tagged Ost3p, was greatly reduced when the HA-tagged Ost3p was the affinity ligand for isolation of the complex from the wbp1-1 or wbp1-2 mutants. With the exception of the tagged Ost3 protein, the radiolabeled OST subunits were scarcely more intense than protein contaminants that precipitate from an untagged wild-type strain. When the HA-tagged Stt3 protein was used as the affinity ligand for isolation of the complex from the wbp1 mutants, we observed reduced yields of Wbp1p, Swp1p, Ost3p, and Ost2p (Fig. 5A) and slightly reduced yields of the HA-tagged Stt3p.
Microsomes isolated from the wbp1-2 mutant contain reduced amounts of Wbp1p and Swp1p relative to microsomes isolated from wild-type cells (9). The results presented in Fig.  5A suggest that the Ost3 protein may also be unstable in the wbp1 mutants. To address this possibility, microsomes were isolated from the wild-type, wbp1-1, and wbp1-2 mutant strains that express HA-tagged Ost3p or Stt3p. Protein immunoblots that were probed with the anti-HA antibody showed a substantial reduction in the membrane content of Ost3p in both the wbp1-1 and wbp1-2 mutants (Fig. 5B). The steady state levels of HA-tagged Ost3p were extremely low in the wbp1-2 mutant but were detectable when greater amounts of membrane protein (80 g) were loaded on the gel (not shown). In contrast, the Stt3p content of microsomes isolated from the wbp1 mutants and wild-type cells were quite similar (Fig. 5C).
The OST3 gene was disrupted in the wbp1-2 strain to explore the possible biological interaction between these gene products. The wbp1-2 ⌬ost3 double mutant is viable at 30°C, yet grows significantly slower than the wbp1-2 mutant. Protein immunoblots of microsomes from wild-type yeast and from the wbp1-1, wbp1-2, ⌬ost3, and wbp1-2 ⌬ost3 mutants were probed with an antibody to Ost1p as a simple method to evaluate in vivo OST activity (Fig. 5D). Wild-type yeast express two glycoforms of Ost1p that contain four and three N-linked oligosaccharides, respectively (3). As observed previously, hypoglycosylated forms of Ost1p are synthesized by the wbp1-2 mutant and the ⌬ost3 mutant (9, 10). When the two mutations are combined in a single strain, a more severe glycosylation defect is observed.
Notably, the membrane content of Ost1p is not reduced in the wbp1 mutants, the ⌬ost3 mutant, or the wbp1-2 ⌬ost3 double mutant. High copy expression of Ost3p in the wbp1-1 and the wbp1-2 mutants does not suppress the temperature-sensitive growth phenotype of either strain (not shown). However, overexpression of Ost3p partially corrects the glycosylation defect of the wbp1-1 mutant (Fig. 5E). Overexpression of Ost3p in the wbp1-2 mutant did not noticeably alter the average number of oligosaccharides transferred to Ost1p.

In Vivo Analysis of the Subunit Composition of the OST-
Previous biochemical studies of the yeast oligosaccharyltransferase suggested that the enzyme contains four, five, or six subunits (3)(4)(5)31). Here we have utilized yeast strains that express epitope-tagged OST subunits to define the in vivo subunit composition and stoichiometry of the enzyme. The results we have obtained demonstrate that the native enzyme has a composition that is consistent with the subunit composition that has been suggested by genetic studies (11) and is remarkably similar to the OST complex that was purified by Kelleher and Gilmore (3). In addition to the six previously characterized subunits (Ost1p, Wbp1p, Ost3p, Swp1p, Ost2p, and Ost5p) two additional polypeptides routinely copurified with the epitopetagged Ost3 protein. The radiolabeled 3.6-kDa polypeptide was identified as Ost4p based upon its migration on SDS-polyacrylamide gels and its absence from the immunopurified OST complex in a ⌬ost4 mutant. The OST4 gene was identified by Chi et al. (15) in a genetic screen for mutations that cause defects in the biosynthesis of N-linked oligosaccharides. Although the 3.6-kDa Ost4 protein is essential for viability of yeast at 37°C, loss of this subunit causes a severe defect in the in vivo transfer of oligosaccharides to nascent glycoproteins at 30°C, suggesting that Ost4p is either a structural component of the enzyme or an accessory component that is important for OST assembly (15). Here, we present biochemical evidence that the Ost4 protein is a structural subunit of the mature OST complex. Although the quantitative data suggest that Ost4p is not as abundant as the other subunits, a reduced recovery of proteins of this size is not unexpected when gels are fixed and dried. Notably, open reading frames of this size were not catalogued in the yeast genome project nor are proteins of this size re- solved and detected by commonly used gel electrophoresis methods.
The 78-kDa Stt3p protein has an N-terminal hydrophobic domain with 12 predicted membrane-spanning segments and a C-terminal hydrophilic domain that is located in the endoplasmic reticulum lumen (17,24). The apparent absence of Stt3p in both the tetrameric and hexameric OST preparations (3)(4)(5) led to the hypothesis that Stt3p might be a substoichiometric component of the OST that is important for assembly of the enzyme (17). Here, we have presented several lines of evidence conclusively demonstrating that the Stt3 protein is a subunit of the yeast OST. By using a yeast strain that expresses HA-tagged Ost3p, we were able to determine the subunit stoichiometry of the OST by rapidly isolating the enzyme from radiolabeled yeast cultures. Sequential immunoprecipitation experiments indicated that the bulk of the radiolabeled protein in the Ost1p-Stt3p region of the gel was Stt3p. Quantification of the [ 35 S]methionine-labeled OST subunits strongly supports the conclusion that Stt3p is present in equimolar amounts relative to the six previously characterized subunits. Given that there was no indication of a labile association between Stt3p and the other OST subunits, how can we account for the apparent lack of Stt3p in the purified OST preparations? As presented in this study, the 78-kDa Stt3 protein has an anomalous gel mobility, migrating instead as a diffuse 60-kDa band that is not resolved from Ost1p under most electrophoresis conditions. Radioiodination of the purified OST preparation revealed an intensely labeled diffuse polypeptide that comigrates with Ost1p. Apparently, the Stt3 protein stains poorly with Coomassie Blue and with silver, hence the diffusely migrating Stt3 protein was simply not detected in previous studies. The extreme hydrophobicity of the N-terminal region of Stt3p is presumably responsible for both the aberrant gel mobility and the poor staining characteristics.
The epitope-tagged strains were useful for evaluating the stability and the subunit composition of the OST complexes in mutants that express reduced OST activity. Notably, the composition and integrity of the OST was not detectably altered in the ⌬ost3 mutant, indicating that Ost3p is not required for the assembly or stability of the enzyme. We speculate that Ost3p is a peripherally located subunit within the OST complex. A possible role for Ost3p as an accessory subunit was suggested by the observation that loss of Ost3p causes substantial reductions in oligosaccharide transfer to some but not all acceptors (10). Here, we observed that most of the cellular pool of Wbp1p, Swp1p, or Ost1p copurified with the epitope-tagged Ost3 protein, indicating that the majority of the OST complexes in wild-type yeast contain Ost3p as a subunit.
In the absence of Ost4p, the association between Ost3p and the other OST subunits was labile, suggesting that the interaction between Stt3p and Ost3p is dependent upon Ost4p. In contrast to what was observed in the wbp1 mutants, the recovery of the HA-tagged Ost3p was comparable in the wild-type and the ⌬ost4 mutant. One interpretation of this difference is that the association between Ost3p and the other OST subunits in the ⌬ost4 mutant is destabilized by the detergents used for immunoisolation of the complex.
Evidence for Discrete Subcomplexes within the OST-A physical interaction between the Wbp1p, Swp1p, and Ost2p subunits (Fig. 6) is supported by complementary genetic and biochemical observations. The SWP1 gene and the OST2 gene are allele-specific high copy suppressors of the wbp1-2 mutant (6,9). Microsomes isolated from the wbp1-2 mutant contain reduced amounts of Wbp1p and Swp1p at the permissive temperature (9). Suppression of the wbp1-2 growth defect by overexpression of Swp1p or Ost2p is explained by an enhanced stability of Wbp1p, which results in increased OST activity (6,9). In vivo depletion of either Wbp1p or Swp1p leads to a reciprocal reduction in the membrane content of the other subunit (5). Wbp1p can be cross-linked to Swp1p in yeast microsomes suggesting a direct interaction between these two subunits (6).
Recent studies reveal that the purified canine OST contains DAD1 in addition to ribophorin I, ribophorin II, and OST48 (30). The four mammalian subunits are homologous to Ost2p, Ost1p, Swp1p, and Wbp1p, respectively (2). Cross-linking experiments have provided evidence for a direct physical interaction between the subunits of the canine enzyme (30). Readily identified cross-linked products included DAD1-OST48 heterodimers, ribophorin II-OST48 heterodimers, and ribophorin II-OST48-DAD1 heterotrimers (30). These biochemical findings strongly suggest that the Swp1p-Wbp1p-Ost2p subcomplex is a structural unit in the OST that is conserved between vertebrates and fungi. Several groups have obtained evidence suggesting that the active site of the OST is associated with Wbp1p or its vertebrate homologue OST48 (4,34). Most recently, an epoxyethylglycine derivative of the tripeptide acceptor was shown to react with both OST48 and ribophorin I, suggesting that the active site may reside at an interface between OST48 and ribophorin I (35).
During the course of these experiments we obtained evidence for a direct interaction between Ost3p, Stt3p, and Ost4p. The Stt3p-Ost4p-Ost3p subcomplex was not disrupted by an exposure to Triton X-100/SDS-mixed micelle buffer that was supplemented with 2 M urea or 0.5 M NaCl. In contrast, at least four (Ost1p, Swp1p, Wbp1p, and Ost2p) of the five other OST subunits dissociated from the Stt3p-Ost4p-Ost3p subcomplex upon exposure to the mixed micelle buffer. As the mixed micelle buffer lacked reducing agents, we cannot rule out the possible stabilization of the Stt3p-Ost4p-Ost3p complex by disulfide bonds. Genetic support for an interaction between Ost3p and Stt3p was provided by the observation that immunopurified OST complexes lack Ost3p when expression of Stt3p was reduced in vivo by placing the STT3 gene under control of the Three subcomplexes designated as the Stt3p-Ost4p-Ost3p subcomplex, the Swp1p-Wbp1p-Ost2p subcomplex, and the Ost5p-Ost1p subcomplex are proposed to be assembled into the octameric oligosaccharyltransferase. An arrowhead designates the N terminus of each subunit. Protein topology predictions for Ost1p, Wbp1p, Swp1p, and Ost3p were reviewed previously (2). Predicted topologies for Stt3p (17), Ost4p (15), and Ost5p (11) are based upon the cited references. glucose-repressible GAL promoter (17). A systematic search for high copy suppressors of mutant alleles of each gene that encodes an OST subunit has revealed that OST3 and OST4 are allele-specific high copy suppressors of stt3 mutants (41). In addition, overexpression of Ost3p suppresses the restrictive phenotype of the ost4 mutant. Allele-specific high copy suppression can provide evidence for a direct physical interaction between proteins (36).
The purified vertebrate OST complex appears to lack proteins that are homologous to the subunits of the Stt3p-Ost4p-Ost3p subcomplex (34,37,38). The canine enzyme is 5-fold less active than the purified yeast enzyme (3); hence it has been speculated that the purified canine enzyme may be the catalytic core of a larger complex that contains homologues of Stt3p, Ost3p, and perhaps Ost4p (30). Indeed, searches of the protein sequence data base have revealed that the yeast Stt3 protein is 60% identical in sequence to Caenorhabditis elegans, murine, and human proteins of unknown function (17). Northern blots have shown that the mRNA encoding the human homologue of Stt3p is expressed in all human tissues that were tested (39). A putative tumor suppressor gene encoding a human homologue of the Ost3p, which is called N33, was localized to chromosome band 8p22 (40). Although the sequence identity between Ost3p and N33 is only 20%, the two proteins share an identical arrangement of four predicted membrane-spanning segments. The expressed sequence tag data base contains mouse and human cDNAs that encode 37-residue proteins that are 36% identical in sequence to Ost4p. Given that yeast Stt3p and Ost4p were not detected after electrophoresis using conventional protein staining methods, the conclusion that the purified vertebrate enzyme lacks the Stt3p-Ost4p-Ost3p subcomplex must be reevaluated.
Overexpression of Ost5p rescues an ost1-5 alg5 double mutant suggesting that Ost1p is in contact with Ost5p (11). Biochemical evidence supporting an Ost1p-Ost5p subcomplex has not been reported. Ribophorin I, the homologue of Ost1p, is a component of the OST that has been purified from canine (37), avian (38), and porcine (34) tissues. A homologue of the Ost5 protein was not detected in the purified vertebrate OST preparation.
The three proposed subcomplexes in the OST are depicted in Fig. 6. It should be noted that we have not obtained evidence supporting an in vivo pool of autonomous subcomplexes. Instead, we propose that the subcomplexes depicted here may represent assembly intermediates that interact to form the fully functional OST complex. Possible sites of interaction between the subcomplexes were suggested by the analysis of strains that contain a mutant allele of an OST subunit. Here, we observed that the membrane content, and hence the stability of Ost3p, is drastically reduced in the wbp1-1 and wbp1-2 mutants suggesting that assembly of Ost3p into the complex is impaired. Although overexpression of OST3 did not rescue the temperature-sensitive phenotype of the wbp1-1 or wbp1-2 mutants, the in vivo glycosylation defect of the wbp1-1 mutant yeast was partially suppressed. Together, these results suggest that Ost3p is physically associated with Wbp1p. However, Ost3p is clearly not uniquely responsible for the interaction between the Sttt3p-Ost4p-Ost3p subcomplex and the Swp1p-Wbp1p-Ost2p subcomplex, as deletion of Ost3p does not destabilize the interaction between Stt3p and the other OST subunits. Cross-linking of ribophorin I to OST48 suggests that these two proteins may be adjacent in the mammalian OST (30); hence, the interaction between the Swp1p-Wbp1p-Ost2p subcomplex and the Ost1p-Ost5p subcomplex may involve contact between Ost1p and Wbp1p. Now that the composition of the yeast OST has been defined, the epitope-tagged strains can be used for the rapid and efficient isolation of the enzyme from wild-type and mutant strains for enzymatic studies. The results we have obtained concerning the Stt3p-Ost4p-Ost3p subcomplex may lead to the isolation of the homologous mammalian proteins once immunological probes for the putative homologues have been produced. Finally, by defining interactions between several OST subunits, we have obtained information about the structural organization of this large integral membrane protein prior to the elucidation of its three-dimensional structure.