Degradation of Misfolded Endoplasmic Reticulum Glycoproteins in Saccharomyces cerevisiae Is Determined by a Specific Oligosaccharide Structure

Strains used are detailed in Table I. Wild-type denotes a strain with both normal biosynthesis of lipid-linked oligosaccharides and trimming of protein-bound oligosaccharides but harboring the prc1-1 mutation. Oligonucleotides (Microsynth, Balgach, Switzerland) used for gene deletion and screenings are listed in Table II. The integrative plasmid pRS306-prc1-1 containing the mutated CPY gene was provided by Dr. D.H. Wolf (University of Stuttgart, Germany). The antiserum against yeast hexokinase was provided by Dr. S. Schröder (Biozentrum, University of Basel, Switzerland).

Standard protocols were followed for growth of yeast, mating, sporulation, and ascus dissection (Guthrie and Fink, 1991). If not otherwise stated, the cells were grown at 30°C in either YPD medium (2% Bacto-Peptone, 1% Yeast extract [both from Difco Laboratories, Detroit, MI], 2% glucose) or for metabolic labeling experiments in MV medium (0.67% Yeast nitrogen base [Difco Laboratories], 2% glucose and the appropriate supplements).

The MNS1 locus ORF YJR131w (these data are available from GenBank/EMBL/DDBJ under accession number Z49631; Grondin and Herscovics, 1992) was inactivated by replacing a major part of the locus with the KanMX cassette (Wach et al., 1994). The sequence of the kanamycin resistance gene was amplified by PCR by using the template pFA6a-KanMX4 plasmid (Wach et al., 1994) and the primers MNS1forKan and MNS1revKan (Table II). The resulting DNA was transformed into strain SS328 and the cells were selected on G418 plates (200 μg/ml). Transformants were analyzed for correct integration by whole cell PCR (Sathe et al., 1991) using KanMXu and the MNS1-specific MNS1-68u and MNS1 + 431L primers.

The ALG12 locus ORF YNR030w (these data are available from GenBank/EMBL/DDBJ under accession number Z71645; Lussier et al., 1997) was inactivated by the same procedure using the primers ALG12forKan and ALG12revKan for amplifying the KanMX cassette and KanMXu and ALG12for primers for verifying the correct gene deletion (Table II).

The BglII-linearized plasmid pRS306-prc1-1 (Knop et al., 1996) containing the mutated form (G255R) of CPY (Wolf and Fink, 1975) was integrated into the PRC1 locus of various yeast strains, resulting in a duplication of the PRC1 locus. Strains in which an excision of the duplication by homologous recombination had occurred were selected on 5-FOA plates and the resulting colonies screened by PCR for the prc1-1 locus. A fragment of the PRC1 locus was amplified by PCR using the primers CPY462u and CPY855L (Table II) giving raise to a product of 423 bp. Due to the prc1-1 mutation, a BstXI restriction site is destroyed. Therefore, strains containing solely the prc1-1 locus were identified by the resistance of the PCR fragment towards BstXI digestion. Western blot analysis confirmed that they only expressed mutant CPY*.

Stationary grown cells from a YPD overnight culture were inoculated in minimal medium and cultivated to an OD_546nm of 1.0. The cells were harvested by centrifugation, washed in minimal medium containing 0.1% glucose and then incubated in the same medium at 30°C for at least 3 h. For pulse–chase experiments, 2 × 10⁷ cells per time point were labeled by the addition of 50 μCi [³⁵S]methionine (Tran³⁵S-label, 10 mCi/ml; ICN Pharmaceuticals) for 10 min and then chased with a 100-fold excess of nonradioactive methionine. The chase was terminated by the addition of NaN₃ (50 mM final concentration) and immediate freezing in liquid nitrogen. Protein extractions, immunoprecipitation, and SDS-PAGE were performed as described (Franzusoff et al., 1991; te Heesen et al., 1992). The dried gels were exposed and analyzed using a PhosphoImager. The kinetics of CPY* degradation were calculated by setting the counts of time point zero as 100%. For the studies of the transport kinetics of CPY in the strains with and without Mns1p, the cells were labeled for only 5 min at 26°C. The chase, protein extraction, and immunoprecipitation were performed as described above.

Yeast strains were grown at 30°C in YPD or minimal medium containing the appropriate supplements into stationary phase. 3 × 10⁸ cells were harvested and broken with glass beads in 50 mM Tris-HCl, pH 7.5, 1% SDS, 2 mM PMSF (Franzusoff et al., 1991; te Heesen et al., 1992). Protein extract equivalent to 7 × 10⁶ cells was subjected to reducing SDS-PAGE, transferred to nitrocellulose membranes, and probed with specific antibodies. Binding was visualized by chemiluminescence (SuperSignal ULTRA; Pierce Chemical Co., Rockford, IL). The x-ray films were scanned and the intensity of the protein bands was determined. The antibody conjugates on the nitrocellulose membranes were stripped by treatment in 62.5 mM Tris-HCl, pH 6.7, 2% SDS, 100 mM β-mercaptoethanol at 65°C for 45 min, and the membranes were reprobed with another antibody. As an additional control for equal protein concentrations, protein contents were determined using the method of Sailer and Weissmann (1991).

The analysis of lipid- and protein-linked oligosaccharides has been described (Cacan et al., 1993; Zufferey et al., 1995; Jakob et al., 1998). For pulse–chase labeling of the oligosaccharides, typically 3 × 10⁹ cells of a logarithmically growing culture were pelleted, washed with YP0.1D (2% Yeast extract, 1% Bactopeptone, 0.1% glucose), and resuspended in 450 μl YP0.1D containing 400 μCi 2-[³H]mannose (30 Ci/mmol; ICN Pharmaceuticals). The oligosaccharides were labeled for 1 min at 26°C and the radioactivity was chased by adding nonradioactive (±)D-mannose (111 mM final concentration). At the given time points, 5 × 10⁸ cells were removed, placed in 1 ml of CM 3:2 (chloroform/methanol 3:2 vol/vol) and mixed by vortexing. Extraction, work-up and analysis of lipid-linked oligosaccharides (LLO) and NLO was as described above. For detailed verification of oligosaccharide structure, endo H–released NLO were further digested with α1,2-specific mannosidase from Aspergillus saitoi (15 μU; Oxford Glycosystems, Abingdon, UK) in the supplied buffer. After the digest the NLO were separated by HPLC (see above).

To understand in more detail the role of NLO in glycoprotein degradation, we determined the kinetics of protein-bound oligosaccharide trimming in the ER in vivo. For this, two yeast strains that carry the sec18-50 mutation were used. The sec18-50 mutation results in a temperature-sensitive phenotype and prevents the fusion of ER-derived vesicles with the Golgi apparatus at nonpermissive temperature. No processing of protein-bound oligosaccharides by Golgi glycosyltransferases was observed in sec18 mutant strains at nonpermissive conditions (Novick et al., 1980; Eakle et al., 1988). We performed the experiments at the permissive temperature for the sec18-50 mutation, nevertheless, the export rate of secretory proteins to the Golgi apparatus was slower in sec18-50 cells as compared with wild-type cells and we were able to analyze the trimming of the NLO in the ER. One of the strains carried in addition a deletion in the GLS2 locus inactivating glucosidase II. No growth phenotype was associated with the Δgls2 mutation. Cells were labeled with ³H-mannose for 1 min at 26°C and the radioactivity was chased by adding an excess of nonradioactive (±)D-mannose. At the given time points, the chase was terminated, NLO were released from protein by endoglycosidase H (endo H), and then analyzed by HPLC (see Materials and Methods section).

In the Δgls2 strain, only two trimming events occur in the ER: the removal of the terminal α1,2-linked glucose residue by glucosidase I and the cleavage of an α1,2-linked mannose residue by mannosidase I. As expected, a protein-bound oligosaccharide with the putative structure Glc₂Man₉GlcNAc₂ (G2M9)² was found, which was slowly converted to Glc₂Man₈GlcNAc₂ (G2M8) with a half-life of ∼10 min (Fig. 2, A and B, left). We were unable to detect protein-bound Glc₃Man₉GlcNAc₂.

We analyzed the structure of the different oligosaccharides in more detail. The endo H–released Glc₂Man₉GlcNAc₂ NLO comigrated with the Glc₂Man₉GlcNAc₂ oligosaccharide obtained from endo H–treated LLO of a Δalg10 strain (Fig. 2,C). This strain accumulates lipid-linked Glc₂Man₉ GlcNAc₂ due to the inactivation of the α1,2 glucosyltransferase (Burda and Aebi, 1998). Digestion of the two major protein-derived oligosaccharides from a Δgls2 sec18-50 strain by exo-α1,2-mannosidase (from Aspergillus saitoi) converted both of them to a single species (Fig. 2,C). Therefore, the two oligosaccharides found on protein in a Δgls2 sec18 strain differed by one α1,2-linked mannose. This observation, the mobility of the exo-α1,2-mannosidase digestion product and the comigration of one oligosaccharide with the Glc₂Man₉GlcNAc₂ marker showed that protein bound Glc₂Man₉GlcNAc₂ and the corresponding mannosidase product Glc₂Man₈GlcNAc₂ were present in the Δgls2 sec18-50 strain. The minor peak observed after exo-α1,2-mannosidase digestion (Fig. 2 C, left) was possibly due to incomplete digestion, since the presence of glucose residues on oligosaccharides reduced the efficiency of the enzyme (Burda, P., unpublished data). Taken together, the results confirmed that release of the α1,2-linked glucose from protein-bound oligosaccharide by glucosidase I was a rapid process in the Δgls2 strain. However, trimming of the oligosaccharide by endogenous α1,2-mannosidase (Mns1p) occurred much slower.

When we analyzed the NLO processing in a strain fully competent for trimming (sec18-50; Fig. 2, right), we were again unable to detect the complete protein-bound Glc₃ Man₉GlcNAc₂ oligosaccharide, even in preparations obtained shortly after the pulse (Fig. 2, A and B, right). The largest oligosaccharide, detected after 2 min of chase (1-min pulse), comigrated with the Glc₂Man₉GlcNAc₂ oligosaccharide but represented a minor fraction (<10%) of the total NLO. In contrast, significant amounts of Glc₁ Man₉GlcNAc₂ were detected at this time point (the structural analysis of this oligosaccharide is described below). However, from this oligosaccharide one or two hexose units were rapidly trimmed. To determine whether this trimming was due to glucosidase II or mannosidase I activity, we analyzed the NLO preparation from the 5-min chase point by digestion with the exo-α1,2-mannosidase (Fig. 2 C). This preparation contained small amounts of oligosaccharides comigrating with the Glc₁Man₉GlcNAc₂ standard and significant levels of oligosaccharides that migrated as expected for Man₉GlcNAc₂ and Man₈GlcNAc₂ oligosaccharide. Indeed, digestion by exo-α1,2-mannosidase revealed that only the Glc₁Man₉GlcNAc₂ oligosaccharides contained a protective glucose residue and was converted to Glc₁Man₇GlcNAc₂, whereas the majority of the oligosaccharides was trimmed to Man₅GlcNAc₂. These results showed that the Glc₁Man₉GlcNAc₂ protein-bound oligosaccharide was converted primarily to Man₉GlcNAc₂ and that the peak representing this oligosaccharide contained no significant amounts of mannosidase I–trimmed, monoglucosylated oligosaccharide.

The analysis of the structure of protein-bound oligosaccharide species as well as their temporal appearance showed that the removal of the terminal α1,2-glucose on protein-bound oligosaccharides by glucosidase I was a rapid process in vivo. Similarly, since we observed only small amounts of diglucosylated oligosaccharides (Fig. 2, A and B, 2-min chase), the hydrolysis of the first α1,3-linked glucose by glucosidase II was a rapid process. We concluded that under our experimental conditions, the monoglucosylated oligosaccharide Glc₁Man₉GlcNAc₂ was converted to the Man₉GlcNAc₂ oligosaccharide with a half-life of ∼2 min and that this occurred before processing by mannosidase I, which was a relatively slow process (half-life 10 min). Evidently, removal of glucose-linked residues was not a prerequisite for mannosidase I action because mannose hydrolysis occurred with approximately the same kinetics in both glucosidase II–proficient or –deficient strains (Fig. 2 B).

Removal of a mannose residue by α1,2-mannosidase concludes the trimming of NLO in the ER of S. cerevisiae (Byrd et al., 1982). Since this cleavage occurred at a slow rate, we speculated that it represents a rate-limiting step and thus is important for efficient glycoprotein transport and maturation. Therefore, we analyzed this aspect in detail by studying the processing of vacuolar proteinase CPY. In the ER, CPY receives four N-linked oligosaccharides (p1CPY, glycosylated proCPY, 67 kD) that are modified in the Golgi apparatus (p2CPY, 69 kD). Upon reaching the vacuole, CPY maturates by proteolytic cleavage of the propeptide (mCPY, 63 kD). In a pulse–chase experiment, we compared the transport rates of CPY, from ER to Golgi and to vacuole in wild-type and Δmns1 strains lacking α1,2-mannosidase activity. We observed that CPY was transported at the same rates (Fig. 3). This demonstrated that trimming of NLO by mannosidase I was not required for export of glycosylated CPY to the Golgi apparatus and the transport to the vacuole.

Previous studies have shown that a specifically mutated form of vacuolar proteinase CPY (CPY*) is retained in the ER and degraded by the proteasome (Hiller et al., 1996) in an oligosaccharide-dependent manner (Knop et al., 1996). Moreover, deletion of the MNS1 locus affects the degradation of CPY*. These observations suggested that trimming of the oligosaccharides is required for processing of misfolded CPY in the ER. To precisely define the NLO moieties important for proteasome-dependent degradation of CPY*, we generated yeast strains containing CPY* carrying defined NLO structures on glycoproteins (for details, see Fig. 1 and Table III).

In a first step, we investigated the influence of glucose residues of oligosaccharides on CPY* degradation (Fig. 4). For that purpose, we constructed mutant strains that produced the following NLO structures: Glc₂Man₈GlcNAc₂ (G2; Δalg10 Δgls2), Glc₁Man₈GlcNAc₂ (G1; Δalg8 Δgls2), Man₈GlcNAc₂ (G0; wild-type and Δalg6 Δgls2). It is important to note that α1,2-mannosidase can act on glucosylated oligosaccharides in vivo (Fig. 2). When we analyzed the processing of CPY* by pulse–chase experiments, a differential degradation of CPY* was observed depending on the number of glucose residues present on the NLO (Fig. 4). The G2 CPY* (Δalg10 Δgls2) was degraded at the slowest rate, whereas the G0 CPY* was degraded at the same rate as CPY* in a strain with normal oligosaccharide biosynthesis and trimming (Fig. 4, wild-type). By Western blot analysis, another G2 CPY*, Δgls2, behaved similarly as G2 CPY* (Δalg10 Δgls2, see below and Fig. 5,C). For the G1 CPY*, we observed an intermediate degradation rate (Fig. 4). From the initial degradation rates (time points 30 and 60 min), we calculated the half-life of CPY* in the various strains (Table IV). In the wild-type and the G0 cells it was 21 min, similar to published data (Hiller et al., 1996). In comparison, the half-life of CPY* in the G1 cells was 43 min and in the G2 cells 82 min under our experimental conditions. Thus, the larger the number of glucose residues the NLO of CPY* contained, the slower its degradation rate.

Next, we determined whether mannose residues of the NLO influenced degradation of CPY*. The core mannose structure of LLO is synthesized by a set of sequentially acting ER mannosyltransferases (Alg3p, Alg9p, Alg12p; see Fig. 1, Orlean, 1997). Due to the fact that incompletely assembled oligosaccharides can be transferred to protein, albeit with a reduced efficiency, we were able to obtain yeast strains with incomplete, but defined NLO structures, namely Man₅GlcNAc₂ (M5, Δalg3; Aebi et al., 1996), Man₆GlcNAc₂ (M6, Δalg9; Burda et al., 1996), Man₇ GlcNAc₂ (M7, Δalg12, Burda and Aebi, manuscript in preparation), Man₈GlcNAc₂ (M8, wild-type with respect to NLO; M8, Δalg6; Reiss et al., 1996), and Man₉GlcNAc₂ (M9, Δmns1; Puccia et al., 1993; see also Table III).

When the half-lives of mutant CPY* bearing various oligosaccharide structures were analyzed, we found that CPY* containing Man₈GlcNAc₂ NLO was rapidly degraded (Fig. 5,A, lanes 1 and 6 and B). The Man₅GlcNAc₂ CPY* (Fig. 5, A lane 2 and B) was degraded at a reduced rate compared with the Man₈GlcNAc₂ CPY*. However, a greatly reduced rate of CPY* degradation was observed when the NLO were of Man₆GlcNAc₂ (Fig. 5,A, lane 3 and B), Man₇GlcNAc₂ (Fig. 5,A, lane 4 and B), and Man₉ GlcNAc₂ (Fig. 5,A, lane 5 and B) structures. By Western blot analysis, CPY* with diglucosylated oligosaccharides (Δgls2) was also degraded at a slower rate than Man₈ GlcNAc₂ CPY* (Fig. 5,C, compare with Fig. 4). However, the stabilization was not as prominent as for Man₉ GlcNAc₂ CPY* (Δmns1; Fig. 5 C). Taken together, our results showed that efficient degradation of CPY* was dependent on the NLO structure. Man₈GlcNAc₂ CPY* was rapidly degraded, whereas the trimming intermediate Man₉GlcNAc₂ CPY* and the incompletely assembled Man_6-7GlcNAc₂ CPY* were inefficiently degraded.

To demonstrate that the differences in degradation rate of CPY* were due to the altered oligosaccharide structure, we transformed the yeast cells with plasmids complementing the deleted gene loci. The Δalg3, Δalg9, and Δalg12 cells were transformed with plasmids containing the ALG3 (pALG3), ALG9 (pALG9) and ALG12 (pALG12) locus. Further, the CPY* wild-type strain was transformed with the vector plasmid (YEp352). The NLO structure-dependent CPY* degradation phenotype could be reverted by complementing the deleted alg gene loci with the appropriate plasmids (Fig. 5,D, lanes 3–8), but the degradation in wild-type cells was not influenced by the transformation with the empty plasmid (Fig. 5 D, lanes 1 and 2).

Incomplete assembly of the LLO leads to a reduced oligosaccharide transfer to protein by the oligosaccharyltransferase (Sharma et al., 1981; Silberstein and Gilmore, 1996), which is then apparent by underglycosylation of glycoproteins (Huffaker and Robbins, 1981, 1983; Stagljar et al., 1994; te Heesen et al., 1994; Burda et al., 1996; Burda and Aebi, 1998). The equal degradation of Man₈GlcNAc₂ CPY* in glycosylation wild-type cells and in Δalg6 cells (Fig. 5 A, lanes 1 and 6) indicated that the degradation was not due to impaired synthesis of LLO, but rather to the altered oligosaccharide structure.

Also, in contrast to wild-type CPY, CPY* was efficiently glycosylated in both Δalg3 (Man₅GlcNAc₂) and Δalg9 (Man₆GlcNAc₂) cells under the conditions used. We observed a single CPY* glycoform in Δalg3 prc1-1 and Δalg9 prc1-1 cells (Fig. 6, lanes 1 and 3), whereas correctly folded CPY was incompletely glycosylated in Δalg3 and Δalg9 cells (Fig. 6, lanes 7 and 9). This hypoglycosylation of CPY was visualized by the distinct bands upon Western blot analysis representing different glycoforms of this protein (Stagljar et al., 1994). In addition, this experiment showed that such forms of CPY* did not reach the vacuole because endo H treatment resulted in the deglycosylated pro-form of CPY* in Δalg9 cells (Fig. 6, lanes 3 and 4, compare with lanes 9 and 10). As expected, the oligosaccharides in Δalg3 cells were resistant to endo H digestion, whereas in Δalg9 cells they were endo H sensitive (Fig. 6, lanes 8 and 10).

In Δalg3 or Δalg9 cells, a much higher steady state level of CPY was observed compared with CPY* in various mutant cells (Fig. 6). This indicated that degradation of CPY* was not completely blocked by the altered oligosaccharide structure. However, our results demonstrated that oligosaccharide structures specifically affected the degradation of misfolded CPY accumulating in the ER, whereas processing and secretion of wild-type CPY was not altered.

We observed a significant effect of the structure of N-linked oligosaccharides on degradation of misfolded CPY*, a model protein for degradation of glycoproteins retained in the ER (Wolf and Fink, 1975; Finger et al., 1993; Hiller et al., 1996; Knop et al., 1996). When expressed in Δgls2 or Δmns1 cells, we found a reduced degradation of CPY* (Fig. 5,C). The effect of the mannosidase I inactivation was significantly stronger than that of the glucosidase II deletion. A similar stabilization of CPY* as in Δmns1 cells was observed in Δalg9 and Δalg12 cells, where incompletely mannosylated oligosaccharides (Man₆GlcNAc₂ and Man₇GlcNAc₂, respectively) were transferred to protein. In Δalg3 cells, which contain protein-bound Man₅Glc NAc₂, degradation of CPY* was reduced, however, not to the same extent as in Δalg9 or Δalg12 cells. Interestingly, in Δalg6 cells, where nonglucosylated oligosaccharide Man₉GlcNAc₂ was transferred to protein and trimmed to Man₈GlcNAc₂, no effect on the stabilization of CPY* was observed. We concluded, that the protein-bound Man₈GlcNAc₂ structure was an important recognition element in the degradation pathway of CPY*. Our results were best explained by a model (Fig. 7) where the Man₈GlcNAc₂ oligosaccharide on a misfolded glycoprotein acted as a positive signal for degradation. This oligosaccharide structure might be recognized by a lectin, since any changes in its structure reduced the efficiency of degradation. Such a carbohydrate-binding protein has also been postulated in the degradation of glycoproteins in mammalian cells (Yang et al., 1998) and might represent “the additional signal to direct them (soluble misfolded proteins) to the dislocation and the ubiquitation machinery” (Kopito, 1997). Our results showed that glucosylated oligosaccharides also reduced the degradation rate of CPY*, albeit to a lesser extent than the Man₉GlcNAc₂, Man₇GlcNAc₂ and Man₆GlcNAc₂ structures. We concluded that the α1,2-α1,2-dimannose branch of the Man₈ GlcNAc₂ oligosaccharide was a less important structural element for oligosaccharide recognition than both the α1,6- and α1,3-branch affected by the alg3, alg9 and alg12 mutations (see Fig. 1). The observation that the Man₅ GlcNAc₂-producing Δalg3 mutation had a less severe effect on degradation was explained by the hypothesis that this oligosaccharide structure represents an intermediate in the degradation of glycoprotein as shown in higher eukaryotic cells (Villers et al., 1994; Ermonval et al., 1997).

Whether the postulated lectin additionally recognizes unfolded protein domains, as does UDP-glucose/glycoprotein glucosyltransferase, involved in the ER quality control pathway of higher eukaryotes (Sousa and Parodi, 1995), is not known. It is possible that the binding of both, the chaperones and the postulated lectin constitute a signal which targets the glycoprotein to the degradation pathway. Indeed Kar2p, the yeast homologue of BiP, transiently binds to wild-type CPY (te Heesen and Aebi, 1994; Simons et al., 1995). Furthermore, this Hsp70 protein was shown to be involved in the degradation of misfolded protein by the proteasome pathway (Plemper et al., 1997).

It has been proposed that the trimming of the protein-bound oligosaccharide in the endoplasmic reticulum represents a biological timer for the protein maturation in the ER of higher eukaryotes (Helenius et al., 1997). This timer function might be required to prevent permanent residence of misfolded glycoproteins in the ER due to the binding to calnexin and calreticulin in higher eukaryotes. Our results are fully compatible with this timer model: as in higher eukaryotic cells (Hubbard and Robbins, 1979), the protein bound oligosaccharide underwent a step-wise trimming process. Removal of the terminal α1,2-glucose by glucosidase I and the first α1,3-glucose by glucosidase II was a very rapid process, whereas the second α1,3-glucose was removed more slowly. The same difference in glucose hydrolysis was observed for glucosidase II of higher eukaryotic cells (Hubbard and Robbins, 1979) and it has been proposed that two different substrate binding sites of glucosidase II are responsible for this difference: the high affinity site would be responsible for the hydrolysis of the first glucose, the low affinity site for the hydrolysis of the second glucose (Alonso et al., 1993). Similar to the findings in higher eukaryotic cells, the mannose trimming was a slow process as compared with hydrolysis of the glucose residues of the protein-bound oligosaccharide. The removal of one specific α1,2-mannose residue by ER-α1,2-mannosidase might represent the time-point after which a misfolded protein is routed to the degradation pathway. We speculated that the processing by mannosidase I determined the time-scale in which a protein had to be correctly folded. If this was not achieved, the glycoprotein was degraded. Importantly, maturation and transport of correctly folded CPY is oligosaccharide-independent (Schwaiger et al., 1982; Winther et al., 1991) and was also not influenced by trimming or oligosaccharide structure (Figs. 3 and 6). It was also postulated that there is a selective export of proteins out of the ER in yeast (Kuehn and Schekman, 1997). Selective export of only correctly folded proteins in combination with degradation of misfolded proteins, timed by oligosaccharide trimming, might therefore represent an effective quality control system for glycoprotein folding in the ER of S. cerevisiae, where the calnexin/calreticulin cycle (reglucosylation of misfolded proteins) has not been found (Fernandez et al., 1994, Jakob et al., 1998).

In support of our model, we found that the degradation of a mutant form of the oligosaccharyltransferase component Stt3p, a glycoprotein with multiple transmembrane domains (Zufferey et al., 1995) was also controlled by oligosaccharide trimming (Bodmer, D., U. Spirig, and M. Aebi, manuscript in preparation), suggesting that resident ER glycoproteins are subject to the same degradation system as are glycoproteins that are exported from the ER.

Is there a similar role of the Man₈GlcNAc₂ oligosaccharide in the quality control process of glycoproteins in the ER of higher eukaryotic cells? In the trimming process of protein-bound oligosaccharides, removal of the glucose residues precedes the hydrolysis mannose trimming (Hubbard and Robbins, 1979). There is an α1,2-mannosidase activity that leads to the same Man₈GlcNAc₂ oligosaccharide as the yeast MNS1 enzyme in the ER of higher eukaryotic cells (Bischoff and Kornfeld, 1983). However, there are additional ER mannosidase activities in the ER of higher eukaryotic cells and a different Man₈GlcNAc₂ oligosaccharide isomer, where the α1,2-mannose linked to the α1,6-mannose is removed, can be produced (Weng and Spiro, 1993, 1996; Moremen et al., 1994). The trimming process in higher eukaryotes is therefore more complex than in yeast. Nevertheless, our results obtained in yeast are compatible with reports on inhibition of α-mannosidase trimming by deoxymannojirimycin that stabilizes specific misfolded glycoproteins in the ER (Su et al., 1993; Daniel et al., 1994; Liu et al., 1997; Yang et al., 1998). On the other hand, degradation of some glycoproteins in the ER was not affected by mannosidase inhibition (Yang et al., 1998).

Previous work has shown that CPY* remains in the ER, is ubiquitinated and then degraded in a proteasome-dependent pathway (Hiller et al., 1996). When we compared the level of wild-type CPY and mutant CPY* in both the Δalg9 and Δmns1 cells (Figs. 5,A and 6), it was apparent that more mature (vacuolar) CPY was present in wild-type cells than CPY* in the prc1-1 cells. A major portion of CPY* was apparently degraded in these cells. Our results showed that alterations of the oligosaccharide structure did not completely block degradation of CPY*. Moreover, mutations reported to affect CPY* degradation do not completely block CPY* degradation either (Hiller et al., 1996; Knop et al., 1996; Plemper et al., 1997). Taken together, these results suggest an alternative, glycosylation-independent degradation pathway for misfolded glycoproteins in the ER of S. cerevisiae.

N-linked oligosaccharides are added co- and posttranslocationally in yeast. For CPY, competition between glycosylation and folding has been reported (Holst et al., 1996). In cells containing alg mutations, incompletely assembled oligosaccharide is transferred to nascent protein (Stagljar et al., 1994), albeit at a reduced rate (Sharma et al., 1981). This is reflected by incomplete use of potential N-glycosylation sites. However, we noticed that in both Δalg3 and Δalg9 cells, only CPY* with all four potential N-glycosylation sites occupied accumulated, whereas in the Δalg3 and Δalg9 cells, wild-type CPY lacking one or two oligosaccharides were found (Fig. 6). Therefore, we postulated that the prolonged exposure of misfolded CPY* to the oligosaccharyltransferase compensated for the reduced affinity of the oligosaccharyltransferase towards incompletely assembled oligosaccharide and resulted in fully glycosylated CPY* in both Δalg3 and Δalg9 cells.

The assembly of the lipid-linked oligosaccharide and its transfer to selected asparagine residues of polypeptides in the ER is a highly conserved process. The selective processing of the protein-bound oligosaccharide supports the idea that conservation of the transferred oligosaccharide structure is due to the function of specific trimming intermediates in glycoprotein maturation. Genetic tailoring of NLO structures will provide a useful tool to identify additional roles of the oligosaccharide in glycoprotein processing.

We thank Dr. D.H. Wolf (University of Stuttgart, Germany) for providing the prc1-1 integrative plasmid. The hexokinase antiserum was kindly provided by Dr. S. Schröder (Biozentrum, University of Basel, Switzerland). We would also like to thank Drs. A. Helenius and S. te Heesen for helpful discussions and comments on the manuscript and anonymous reviewers for invaluable remarks.

This work was supported by the Swiss National Science Foundation grants 3100-040350 (to M. Aebi) and 31-50835.97 (to J. Roth).

CPY

carboxypeptidase Y

CPY*

mutated, misfolded CPY

endo H

endoglycosidase H

LLO

lipid-linked oligosaccharides

NLO

N-linked oligosaccharides