Degradation of Mutated Bovine Pancreatic Trypsin Inhibitor (BPTI) in the Yeast Vacuole Suggests Post-Endoplasmic Reticulum Protein Quality Control

vacuole. These data suggest a “second line” of protein quality control in the yeast secretory pathway.

To begin to address this question, we expressed wild type and mutant forms of bovine pancreatic trypsin inhibitor (BPTI) in yeast. BPTI is a 58 amino acid polypeptide and the native structure contains three disulfide bonds, formed by C14-C38, C5-C55 and C30-C51, and distinct disulfide-bonded intermediates in the folding pathway are N* (C5-C55, C14-C38), N' (C30-C51, C14-C38) and N SH,SH (C30-C51, C5-C55). Two of the most highly populated intermediates are N* and N', and the conversion of N' to N SH,SH represents the rate-limiting step during BPTI folding in vitro (33,34).
In native BPTI a Tyr at position 35 is buried and immobile (35,36). Although the final conformation of Y35L BPTI is not radically altered, some of the normally buried contacts become solvent-exposed and the protein is unstable relative to wild type BPTI (37)(38)(39)(40), however the rate of disulfide bond formation and overall folding is enhanced in this mutant (40). The conformation of C5A BPTI, in which a disulfide bond in the final and intermediate states cannot form, is also thought to be quite unstable because the C5-C51 bond is buried in the hydrophobic core of the protein. Not surprisingly, significant conformational deviations between wild type and C5A BPTI have been noted (38).
The fates of Y35L and C5A BPTI in the yeast secretory pathway were examined previously, and although neither mutant protein was secreted, Y35L was degraded and C5A accumulated (27,28). Based on these and other data (41,42) the thermodynamic stability of a secreted protein-but not the rate of folding-appears to determine secretion-competence. In this study, we set-out to investigate whether mutant BPTI derivatives were handled by ERAD and/or aggregated. We found that Y35L escaped ERAD and was degraded in the yeast by guest on March 23, 2020 http://www.jbc.org/ Downloaded from

Yeast and molecular methods
The yeast strains used for this study are listed in Table I. Plasmids for the constitutive expression of Wild Type, Y35L and C5A BPTI were described previously (43) and DNA inserts corresponding to these genes and promoter elements in YE112-GPD-BPTI (27) were sub-cloned into PRS426 (44). When the resulting plasmids are introduced into yeast cells, BPTI is expressed from the glyceraldehyde-3-phosphate dehydrogenase promoter. The Nterminus of the protein contains a synthetic pre-pro sequence derived from pre-pro alpha factor but lacks the core oligosaccharyl consensus sites ("G" BPTI). The pro region is cleaved in the Golgi by the Kex2p protease, and an added glutamate-alanine C-terminal to the existing cleavage site improves Kex2p processing. Similar levels of BPTI expression and secretion in yeast are evident if the bona fide pre-pro alpha factor sequence is fused to BPTI or if the synthetic pre-pro sequence contains the glycosylation site (data not shown). The individual BPTI expression vectors were transformed into the desired strains using lithium acetate (45) and transformants were selected for growth on Synthetic Complete (SC) medium lacking uracil but containing glucose at a final concentration of 2%. Cells expressing BPTI were grown to mid-log phase and glycerol stocks were made and stored at -70°C. All experiments were performed using cells freshly broken-out onto solid selective medium.

Cycloheximide chase assay to monitor BPTI maturation and degradation
Yeast were grown to mid-log phase in selective liquid medium at 26°C and were collected and resuspended to a final OD (Optical Density at 600 nm) of 4/ml in the same medium. The cells were incubated at 37°C for 10 min before cycloheximide was added to a final washed three times for 10 min with TBS-T and then incubated at room temperature with secondary antibody (horseradish peroxidase conjugated donkey anti-rabbit IgG; Amersham) diluted 1:5,000 in TBS-T/5% nonfat dry milk for 1 h. The filters were washed three times for 10 min with TBS-T and then treated with the Super Signal West Pico chemiluminescence reagent (Pierce Chemical), exposed to X-ray film and where indicated data were quantified using films exposed in the linear range and the Image Gauge software (Fuji Film).

Sucrose density gradient analysis
A total volume of 100 ml of exponentially growing cells (OD 0.5-0.8) were pelleted, washed, and concentrated to 0.5 ml in STED10 buffer (10% wt/vol sucrose, 10 mM Tris-Cl, pH 7.6, 10 mM EDTA) containing protease-inhibitors (1 µg/ml leupeptin, 0.5 µg/ml pepstatin A and ice. Zirconia/silica beads were added to the meniscus and the samples were lysed using a 60 sec burst in a mini Bead-Beater (as above) for 2 min. An additional 500 µl of STED10 containing protease inhibitors was added and the resulting solution was centrifuged at 400g for 10 min at 4°C. The resultant supernatant was transferred to a new tube and 0.3 ml of the extract was layered onto a 30-70% linear sucrose gradient, which was formed by layering 2. were separated on 4-12% Bis-Tris gradient gels as described above.

Indirect Immunofluoresence
The cellular distribution of BPTI and intracellular proteins in yeast were visualized using a protocol supplied by M. Snyder's laboratory (Yale University; personal communication).
Yeast were grown to mid-log phase at 26°C, a 1/10 th the volume of 37% formaldehyde was added and the cells were incubated for 60 min at room temperature with gentle shaking. The solution (70% glycerol/PBS containing 2% n-propyl gallate). The samples were sealed with a cover-slip and clear nail polish, allowed to dry, and immunofluoresence was either assessed immediately or the samples were stored at -20°C in the dark for several months. Cell fluorescence was visualized on an Olympus BX60 microscope fitted with a Hamamamatsu digital camera and images were analyzed using QED Imaging Software (Pittsburgh, PA).

Concanavalin A immunoprecipitation
A total of 300 µl of Concanavalin A Buffer (500 mM NaCl, 1% TritonX-100, 20 mM Tris pH 7.5, 2mM NaN 3 ) and 20 µl of Concanavalin beads (added from a 1:1 stock in Concanavalin A Buffer) were added to 15 µl of cellular lysate, prepared as described in the cycloheximide chase protocol (see above). This mixture was incubated for 16 h at 4°C. The beads and those proteins bound were harvested by centrifugation at 16,000g for 1 min at 4°C.
The beads were washed three times with Concanavalin A Buffer and 1 x LDS sample buffer /0.06M DTT buffer was added to the pellet at a 1:1 bead:sample buffer ratio. The samples were heated to 72°C for 10 min, and proteins were separated on 4-12% Bis-Tris NuPage gels and analyzed by western blotting as described above for the cycloheximide chase protocol.

Y35L BPTI degradation is independent of BiP, calnexin and proteasome function
Previous studies indicated that wild type BPTI is secreted from yeast but the Y35L mutant is degraded intracellularly (28). To determine whether Y35L BPTI was degraded via the ERAD pathway, wild type and Y35L BPTI-expressing plasmids were transformed into cells containing the kar2-1 allele in the gene encoding BiP (KAR2; (46)), into a strain lacking calnexin (cne1∆; (47)), and into the pre1-1pre2-1 mutant in which ~95% of the activity of the proteasome is abrogated (48). We and others have noted that the degradation of soluble ERAD substrates is defective in these strains or in ER-derived microsomes or cytosol derived from these mutants when ERAD is examined in vitro (12,15,16,49,50). Isogenic wild type yeast strains expressing these proteins were examined in parallel. As an additional control, yeast were transformed with a URA-marked 2 micron vector, PRS426 (44), lacking an insert.
Cycloheximide chase experiments were performed as described in the Experimental Procedures and Bis-Tris polyacylamide gels (4-12% acrylamide) were used to resolve the ER-resident, immature BPTI, which contains a pro region, from mature BPTI, which results from Kex2 protease-mediated removal of the pro region in the Golgi. The mature and proregion-containing BPTI species were detected by immunoblot analysis of cellular extracts using anti-BPTI antiserum. A signal was absent when immunoblotted extracts from yeast transformed with the vector control were probed with anti-BPTI antiserum (data not shown).
As shown in Fig. 1., we noted a time-dependent decrease in the amount of the immature ("pro") ER-resident form of the wild type and Y35L proteins, regardless of which strain was examined. In these experiments the level of the mature form either decreased or remained constant, suggesting that the immature form is "chased" into the mature species, which in the case of wild type BPTI is then secreted (27; see below), and/or that a variable population of the mature protein is stable. To differentiate between these possibilities, total yeast protein was radiolabeled with 35 S-methionine/cysteine, cell extracts and extracellular fluid/media were isolated (see Experimental Procedures), and BPTI was detected by immunoprecipitation and radiography as previously described (27,28). As shown in Fig. 2, we observed precursor-product relationships between immature pro-BPTI and the mature intracellular and extracellular BPTI species. It is important to note that only wild type, mature BPTI was secreted, regardless of whether it was expressed in the wild or mutant strains (data not shown, but see Fig. 2 and 6 and 27,28). These results suggest that the Y35L mutant protein is degraded intracellularly, independent of the ERAD pathway, or that it aggregates and cannot be resolved by immunoblot analysis.

Y35L mutant BPTI is degraded in the vacuole
Because the time-dependent disappearance of Y35L BPTI occurred in mutants defective for ERAD, and because the vacuole/lysosome has been found to play a distinct role or complementary role to ERAD in the degradation of some aberrant, soluble proteins in yeast (51-54) we investigated whether the vacuole was the site of Y35L BPTI degradation. To this end, plasmids expressing wild type and Y35L BPTI were transformed into wild type and pep4 mutant yeast, in which vacuolar protease activity is negligible (55). Cycloheximide chase reactions were again performed but in this case quantitative data for the levels of mature and pro-BPTI were obtained. In these experiments we found that the mature form of the Y35L BPTI protein, which was not secreted, was markedly stabilized (Fig. 3A).
Moreover, we noted significantly greater quantities of Y35L BPTI in the pep4 mutant strain compared to the wild type strain (compare "short" and "long" exposures; Fig. 3A), suggesting that the protein accumulated in yeast defective for vacuolar proteolytic processing. We also measured the stability of wild type BPTI in these strains. Although the wild type protein can be secreted ( Fig. 2 and (27, 28)), we noted that the intracellular population (see Experimental Procedures) was stabilized in the pep4 mutant (Fig. 3B). These data are consistent with recent results suggesting that the vacuole can degrade proteins that arise via secretory pathway "overflow", particularly when the cell is stressed (54). We surmise that wild type BPTI can escape degradation and be secreted but that the secretory pathway is overwhelmed, leading to vacuolar degradation of the intracellular pool. In contrast, Y35L mutant BPTI is quantitatively retained in the cell and degraded.
The results presented in Fig. 3 suggest that Y35L BPTI, and to some extent wild type BPTI, is targeted to and degraded by the vacuole. To obtain additional support for this hypothesis the residence of BPTI in wild type and pep4 mutant yeast was assessed using anti-BPTI antiserum and indirect immunofluoresence microscopy. A vacuolar-resident protein was identified using a monoclonal antibody against a subunit of the H + -ATPase, Vma2p (56).
mutant was evident (pep4 Y35L "Merge"). Combined with the data presented in Fig. 3A we conclude that the Y35L mutant protein is quantitatively disposed of by vacuolar proteases.
These combined data also indicate that the cell can distinguish between wild type and Y35L BPTI.

Y35L mutant BPTI is not aggregation-prone
Although the data presented above indicate that at least a portion of Y35L BPTI is degraded by vacuolar proteases, another explanation for the complete lack of secretion and disappearance of intracellular Y35L BPTI is that the protein aggregates, an effect that might be exacerbated when vacuolar protease activity is compromised. To test this hypothesis, wild type and pep4 yeast expressing wild type and Y35L BPTI were lysed and the extracts were analyzed by sucrose gradient density centrifugation (see Experimental Procedures). We also assessed the fractionation of the non-secreted C5A mutant form of BPTI that is more stable than Y35L (27). Gradients were fractionated from the top (fraction 1) to the bottom (fraction 13), and pelleted, potentially aggregated material was analyzed as "fraction 14".
The proteins were resolved on 4-12% gradient gels and the migration of BPTI through the gradients was assessed by immunoblot analysis. In each of these extracts the vast majority of BPTI sedimented between fractions 1 and 8 (Fig. 5), which correspond to sucrose densities of 25-52% wt/wt and in identical experiments represent intra-cellular organelles in the secretory pathway (57,58). Notably, low amounts of immature and mature forms of BPTI were observed in fraction 14 in gradients from cells expressing wild type and Y35L BPTI, regardless of whether Pep4p was active. In contrast, we observed a greater population of the C5A mutant form of BPTI in fraction 14, suggesting that this intracellularly retained and stable form of BPTI aggregates to some extent.

Y35L mutant BPTI is stabilized in pep3 mutant yeast
Proteins can be targeted to the vacuole through multiple routes, including the endosomal, non-endosomal (direct), and cytoplasm-to-vacuole (CVT) pathway (59,60). Vacuolar degradation of some mis-folded soluble proteins through the endosomal route requires the To begin to define how the mutant BPTI is targeted to the vacuole, we first expressed wild type and Y35L BPTI in yeast lacking VPS10/PEP1, but no impact on either the secretion or degradation of these proteins was noted; in contrast, Y35L BPTI was stabilized in yeast deleted for PEP3 (data not shown), which is required for the endosomal, nonendosomal, and CVT pathways to the vacuole (65). These data further support vacuolemediated degradation of Y35L BPTI, and suggest that the degradation of this protein is Pep1/Vps10p-independent.

Kex2p-processing is not required for Y35L BPTI degradation, but is necessary for wild type BPTI secretion
Mature wild type and Y35L BPTI are formed by the Kex2p-mediated cleavage of the synthetic pro sequence fused at the N-termini of these proteins, and previous work using Kex2p-over-expressing strains suggested that the Y35L BPTI secretion defect was not due to by guest on March 23, 2020 http://www.jbc.org/ Downloaded from a limiting amount of cellular Kex2p (43). However, Arvan and colleagues reported that a single-chain insulin derivative was diverted from vacuole-mediated degradation to the cell surface in kex2 mutant yeast (63). We therefore wished to investigate the behavior of the wild type and Y35L protein in a kex2 genetic background in order to determine whether Y35L could be secreted in the absence of Kex2p processing.
Wild type and a kex2 yeast strain were transformed with the wild type BPTI or Y35L expression constructs and cell lysates and extracellular fluid (medium) were collected during a cycloheximide chase assay. As shown in Fig. 6A, Y35L was not secreted from the kex2 mutant strain, and the extent of pro-Y35L formation was significantly reduced. Moreover, the level of the immature Y35L BPTI protein decreased over time, suggesting that the protein was degraded. We also found that the secretion of wild type BPTI was blocked in kex2 yeast (Fig. 6A), consistent with a requirement for this post-translational processing event during protein secretion.
Interestingly, little difference in the extent of processing of the wild type pro region was evident when the wild type and kex2 mutant were compared. This phenomenon could arise from spurious cleavage of the pro region via other endosomal proteases to which the Y35L protein is inaccessible. It is also important to note that in a cycloheximide chase a population of stable, slowly processed material can build-up. Therefore, to establish whether the rapid cleavage of the pro region requires Kex2p, wild type and kex2 yeast expressing BPTI were metabolically labeled (27,28), cell extracts were prepared, and BPTI was immunoprecipitated using anti-BPTI antiserum and Protein A-Sepharose. As shown in Fig.   6B cleavage of the pro region was only evident in the wild type strain.

ER-retained Y35L BPTI is stable and glycosylated
The results presented thus far suggest that Y35L BPTI is not an ERAD substrate, even though the protein is unstable (see Introduction). One scenario to explain these results is that Interestingly, the presence of new immunoreactive wild type and Y35L BPTI species that migrated slower than immature (pro) BPTI during SDS-PAGE was observed in sec12 cells (Fig. 7A). The slower migrating species most likely derived from ER-resident BPTI as aggregation-prone substrate in the ER helps maintain its solubility and promotes secretion.
To determine whether the higher molecular weight BPTI species observed in Fig. 7A represent a novel, glycosylated form of BPTI, we incubated extracts from wild type and sec12 yeast with Concanavalin A-Sepharose and examined for the presence of BPTI in the load ("-") and bound ("+") fractions using anti-BPTI antiserum (Fig. 7B). Although the three forms of BPTI were again evident in sec12-derived extracts, the highest molecular weight species bound preferentially to the lectin ("*" in Fig. 7B). As controls for this assay, we found that Gas1p-a glycosylated secreted protein-also bound to Concanavalin A-Sepharose, but that the cytoplasmic molecular chaperone, Ssa1p-which lacks glycans-did not (Fig. 7B). Because the form of BPTI used for these studies lacks a consensus site for Nlinked glycosylation but contains 4 serine and 5 threonine residues, we suggest that BPTI similarly acquires O-linked mannose side chains when trapped in the ER. Also in accordance with data obtained from other laboratories (see above), we suggest that this modification stabilizes BPTI in the ER.

DISCUSSION
Unstable secreted proteins, such as Y35L BPTI, exist in rapid equilibrium between folded and unfolded states, the latter of which transiently exposes normally buried amino acids that might be recognized by ER molecular chaperones. Therefore, we anticipated that Y35L BPTI would be targeted for ERAD, and hoped to begin to correlate the relative instability of a secreted protein with its propensity to become an ERAD substrate. Instead, we found that Y35L underwent Kex2p-dependent maturation in the Golgi and was diverted to the vacuole for degradation. Support for this conclusion comes from the observed stabilization of mature Y35L BPTI in pep4 yeast, from its increased residence in the vacuole when Pep4p is disabled, and from its stabilization in pep3 mutant cells. Although ERAD is probably involved in the degradation of most aberrant secreted proteins, other soluble proteins transit beyond the ER and are similarly degraded by the vacuole. In addition to Y35L BPTI, additional members of this class are mutated forms of a yeast invertase-lambda repressor hybrid (51), a fusion protein consisting of Hsp150 and β-lactamase (52) a CPYproteinase A (PrA) hybrid protein (53) and a chimera containing the yeast pre-pro α factor leader sequence and insulin (63). There is no sequence similarity between these diverse substrates that are subjected to post-ER protein quality control, and at present it is not clear how they are targeted for degradation. Mis-folded human proteins associated with specific diseases are degraded in the vacuole equivalent-the lysosome-and here too it is unclear how they are selected (72).
Our study indicates that post-ER protein quality control exhibits substrate specificity because the Y35L mutant form of BPTI is quantitatively retained in the cell and degraded, whereas ~10-20% of wild type BPTI is secreted based on the data in Fig. 2. We suggest that the relative instability of the mutant compared to the wild type protein provides the foundation for substrate selectivity. Specifically, the Y35L mutant has been estimated to be >3 kcal/mol less stable than wild type BPTI, and although it folds faster than the wild type protein (40), we suspect that during its transit through the secretory pathway it partitions on average into an unfolded or partially folded species more frequently than the wild type protein. The non-native state might be sufficient "tag" the protein for vacuole targeting and degradation.
The recognition mechanism for post-ER protein quality control in yeast and mammals is unclear (72), but in three cases Pep1/Vps10p has been suggested to play a role in targeting soluble substrates to the vacuole in yeast (51)(52)(53). Pep1/Vps10p also recognizes vacuolar hydrolases, such as CPY (61,62). To reconcile these data, Winther and colleagues have suggested that Pep1/Vps10p contains two polypeptide-binding sites, one that is sequencespecific (73) and one that may recognize mis-folded protein structure (53  . Second, 5-fold over-expression of BiP has no effect on wild type BPTI secretion (27).
Third, BPTI lacks BiP binding sites, as determined using an algorithm first developed by   protein in a wild type yeast strain expressing BPTI was radiolabeled and cell extracts and media were resolved. BPTI was detected in the extract by immunoprecipitation using a BPTI-specific polyclonal antibody and in the media BPTI was evident as the only radiolabled protein at this molecular weight, as previously described (28). Duplicate sets of data were quantified and the relative amount of each species was calculated as a fraction of the maximal amount of that species over time.      Wild type (SEC12) and sec12-4 mutant yeast expressing wild type or Y35L BPTI were examined by cycloheximide chase at 37°C to induce the sec12 mutant phenotype. In addition to immature BPTI and a reduced amount of mature BPTI, a higher molecular weight form of BPTI was observed in extracts prepared from the sec12 mutants (*). B. Extracts from the 60 minute time point in part "A" were incubated with Concanavalin A-Sepharose and a portion of the load (-) and the bound material (+) was assayed for the presence of BPTI