Activation of the Ras-cAMP Signal Transduction Pathway Inhibits the Proteasome-independent Degradation of Misfolded Protein Aggregates in the Endoplasmic Reticulum Lumen*

Many kinds of misfolded secretory proteins are known to be degraded in the endoplasmic reticulum (ER). Dislocation of misfolded proteins from the ER to the cytosol and subsequent degradation by the proteasome have been demonstrated. Using the yeast Saccharomyces cerevisiae , we have been studying the secretion of a heterologous protein, Rhizopus niveus aspartic pro-teinase-I (RNAP-I). Previously, we found that the pro sequence of RNAP-I is important for the folding and secretion, and that (cid:1) pro, a mutated derivative of RNAP-I in which the entire region of the pro sequence is deleted, forms gross aggregates in the yeast ER. In this study, we show that the degradation of (cid:1) pro occurs independently of the proteasome. Its degradation was not inhibited either by a potent proteasome inhibitor or in a proteasome mutant. We also show that neither the export from the ER nor the vacuolar proteinase is required for the degradation of (cid:1) pro. These results raise the possibility that the (cid:1) pro aggregates are degraded in the ER lumen. We HI- Xho I fragment was inserted between the Bam HI and Sal I sites of YIplac211 to generate plasmid pYIP3841. To prepare pYIL3841, the 2.9-kb Bgl II- Bgl II fragment of LEU2 , which was prepared from YEp13, was inserted into the Bam HI site of pYIP3841. pYIL3841 was digested with Apa I and then introduced into R27-7C-1B, KUY20, WCGY4a and WCGY4 11/22a to be integrated at the ura3 locus. To generate pYIT3841, the 1.5-kb Eco RI- Eco RI fragment of TRP1 , which was prepared from YRp7, was inserted into the Sma I site of pYIP3841 after both Eco RI ends were filled in with T4 DNA polymerase. pYIT3841 was digested with Apa I and then introduced into R27-7C-1C and KUY20 to be integrated at the ura3 locus. To generate pYIA3841, the 2.2-kb Bgl II- Bgl II fragment of ADE2 from plasmid pASZ11 (41) was inserted into the Bam HI site of pYIP3841. pYIA3841 was digested with Hpa I and then introduced into ANY21 and MBY10-7A to be integrated at the ADE2 locus. Correct integration was confirmed by Southern hybridiza-tion or polymerase chain reaction. Plasmid pRS416 YEp24 used as a vector control.

proteins to the cytosol (14,(22)(23)(24)(25). Other ER membrane proteins, Der1p, Der3p/Hrd1p, and Hrd3p, are involved in the dislocation process (16, 19, 26 -28), although the detailed functions of these proteins remain to be clarified. This degradation mechanism is called ER-associated protein degradation (ERAD). Studies on ERAD have raised two important questions. First, on the ER luminal side, how are misfolded proteins recognized and recruited to the Sec61 channel? Second, how is the direction of transport across the Sec61 pore regulated?
Despite the above findings, the existence of a proteasomeindependent pathway for ER protein degradation cannot be excluded. It would be of particular interest to determine whether or not proteolytic activity exists in the ER lumen. A recent study suggests that proteasome-independent protein degradation occurs in the ER lumen of yeast (29). Moreover, the existence of ER integral membrane proteases has been demonstrated (30 -32). Another report suggested that the signal peptidase complex might play a role in the degradation of misfolded proteins in addition to its intrinsic role of cleaving signal peptides. This is based on the finding that a mutation in SEC11, which encodes a component of yeast signal peptidase complex, rendered an abnormal membrane protein stable (33).
We have previously found that certain heterologous secretory proteins were retained and degraded in the ER of the yeast Saccharomyces cerevisiae (34). These were derivatives of Rhizopus niveus aspartic proteinase-I (RNAP-I (35)), and we designated one of these "⌬pro." In ⌬pro, the entire region of the pro sequence, which is important for the folding and secretion of RNAP-I (36), is deleted. Upon overproduction in yeast, ⌬pro forms large aggregates in the ER lumen, which induce ER membrane proliferation and elevated synthesis of ER resident chaperones (37). That is, overproduction of ⌬pro leads to biogenesis of a novel, ER-derived compartment that accommodates the ⌬pro aggregates. We have also shown that induction of ER chaperones, especially BiP, by the unfolded protein response (UPR) is essential for cell growth upon overproduction of ⌬pro (38). Overproduction of ⌬pro in the UPR-deficient ⌬ire1 mutant led to the inhibition of protein translocation in the ER, protein folding/transport in the ER, and cell growth (38). From these studies, we presumed that overproduction of ⌬pro in yeast will elicit the mechanism of ER degradation of the misfolded protein aggregates as well as the biogenesis of a novel ER subcompartment. In this study, we provide evidence that the degradation of ⌬pro does not depend on the proteasome. Most likely, unidentified proteolytic activity in the ER lumen participates in the degradation of the ⌬pro aggregates. To investigate how the ⌬pro aggregates are degraded, we have isolated and characterized a mutant defective in the degradation of ⌬pro. We show that the mutated gene is IRA2, which encodes a GTPase-activating protein for Ras. Our results suggest that the proteolytic activity for the ⌬pro aggregates is regulated by the Ras-cAMP signal transduction pathway.
S. cerevisiae strains used in this study are listed in Table I. Strain R27-7C-1C was transformed with plasmid pYPR2841, which carries the ⌬pro gene downstream of the constitutive glyceraldehyde-3-phosphate dehydrogenase (GAPDH) promoter (36). The resultant transformants were mutagenized and tested for the accumulation of ⌬pro. Eight mutants were isolated, and one of them, mutant 22, is described in this report. In other experiments, strains were transformed with plasmid pYPR3841, which carries the ⌬pro gene downstream of the GAL1 promoter (37), and the production of ⌬pro was induced by galactose. Yeast transformation was performed using a lithium acetate procedure (39).
Plasmids and Recombinant DNA Methods-Ub-Leu-lacZ, a URA3marked plasmid, was used for GAL promoter-driven expression of a fusion protein between ubiquitin and ␤-galactosidase, whose aminoterminal residue is leucine. This plasmid was kindly provided by Andreas Bachmair.
Plasmids pYPR3841 and pYPR3841U, with which ⌬pro is expressed under the control of the GAL1 promoter and the GAPDH terminator, were described before (37,38). To express the ⌬pro gene in the yeast genome, three integration plasmids, pYIL3841, pYIT3841, and pYIA3841, were constructed. The 1.8-kb BamHI-HindIII fragment, which contained the GAL1 promoter, the ⌬pro coding region, and the GAPDH terminator, was prepared from pYPR3841. This fragment was inserted between the BamHI and HindIII sites of pBluescript II KSϩ, and then re-isolated by digestion with BamHI and XhoI. The 1.8-kb BamHI-XhoI fragment was inserted between the BamHI and SalI sites of YIplac211 to generate plasmid pYIP3841. To prepare pYIL3841, the 2.9-kb BglII-BglII fragment of LEU2, which was prepared from YEp13, was inserted into the BamHI site of pYIP3841. pYIL3841 was digested with ApaI and then introduced into R27-7C-1B, KUY20, WCGY4a and WCGY4 11/22a to be integrated at the ura3 locus. To generate pYIT3841, the 1.5-kb EcoRI-EcoRI fragment of TRP1, which was prepared from YRp7, was inserted into the SmaI site of pYIP3841 after both EcoRI ends were filled in with T4 DNA polymerase. pYIT3841 was digested with ApaI and then introduced into R27-7C-1C and KUY20 to be integrated at the ura3 locus. To generate pYIA3841, the 2.2-kb BglII-BglII fragment of ADE2 from plasmid pASZ11 (41) was inserted into the BamHI site of pYIP3841. pYIA3841 was digested with HpaI and then introduced into ANY21 and MBY10-7A to be integrated at the ADE2 locus. Correct integration was confirmed by Southern hybridization or polymerase chain reaction. Plasmid pRS416 (42) or YEp24 was used as a vector control.
YEpPDE2 and pKF56, a URA3-marked multicopy plasmid carrying PDE2, and a URA3-marked multicopy plasmid carrying IRA2, respectively, were kindly provided by Akio Toh-e. To disrupt IRA2, the 2.8-kb SacI-SacI fragment encoding the amino-terminal region of IRA2 (43) was inserted into the SacI site of pUC18 to form plasmid pUCIRA2. Next, marker fragments were inserted into the BglII site of pUCIRA2, which was located about 200 bp downstream of the start codon of IRA2. As the markers, the 2.9-kb BglII-BglII fragment of LEU2 prepared from YEp13, and the 1.2-kb HindIII-HindIII fragment of URA3 prepared from YEp24 were used to generate pUCira2::LEU2 and pUCira2::URA3, respectively. The terminal HindIII sites of the URA3 fragment were filled in with T4 DNA polymerase before the insertion. pUCira2::LEU2 was cut with SacI and then introduced into R27-7C-1C to generate KUY34. pUCira2::URA3 was also cut with SacI and introduced into R27-7C-1B to generate KUY33. Alternatively, the 2.8-kb SacI-SacI fragment, which was located in the middle region of the IRA2 open reading frame (ORF (43)), was subcloned into the SacI site of pUC18. Then, the 2.9-kb BglII-BglII fragment of LEU2 was inserted into the unique BglII site of the plasmid. The resultant plasmid was cut with SacI and introduced into R27-7C-1B to generate KUY25.
Isolation of Mutants That Accumulate Higher Intracellular Levels of ⌬pro-R27-7C-1C cells harboring pYPR2841 were grown overnight in YNBDCU medium (ϳ2 ϫ 10 8 cells/ml). Cells from 1.25 ml of the culture were washed once with 50 mM potassium phosphate buffer (pH 7.0) and then resuspended in 5 ml of the same buffer. 150 l of ethyl methanesulfonate was added, and the cells were incubated for 30 min at 30°C. Then, ethyl methanesulfonate was inactivated by adding 5 ml of 10% (w/v) sodium thiosulfate solution, and cells were washed three times with water. After being grown in YNBDCU medium for 24 h at 25°C, cells were spread onto YNBD plates supplemented with histidine-HCl (20 g/ml), leucine (30 g/ml), and uracil (50 g/ml) up to 1 ϫ 10 3 cells/plate and incubated for 3 days at 25°C.
For isolation of mutants, colonies on the plates were covered with the nitrocellulose membrane, Hybond-C (Amersham Pharmacia Biotech, Buckinghamshire, England). To lyse the cells, the membranes to which cells were attached were exposed to chloroform vapor in a closed container for 7 min. Then, ⌬pro in each colony was detected using an anti-RNAP-I antibody. The ECL Western blotting system (Amersham Pharmacia Biotech) was used for detection. Colonies that exhibited more intense signals than those of wild-type were pooled and re-examined for the phenotype twice. Only the mutants that reproducibly exhibited intense signals were selected.
Preparation of Whole Protein Extract and Immunoblotting-Whole protein extracts were prepared by lysing cells with alkali (37) or by agitation with glass beads (38). Immunoblotting was performed as described previously (37).
Pulse-chase Experiment and Immunoprecipitation-Pulse-chase experiments were typically carried out as follows. Yeast cells were grown for 3 days in YNBD medium. Appropriate supplements were included in the media. Cells were collected, washed, and transferred to YNBGal medium to induce the expression of ⌬pro. Cells were grown for 4 h and then the cell density was adjusted to 10 A 600 units/ml. Labeling was carried out by adding 0.06 mCi of Tran 35 S-label (ICN Pharmaceuticals, Inc., Costa Mesa, CA) or redivue Pro-mix L-35 S in vitro cell-labeling mix (Amersham Pharmacia Biotech) to 0.6 ml of the suspension. To initiate the chase, cells were collected and washed once, each by centrifugation for 1 min, and then resuspended in 0.6 ml of the chase medium (YNB-Gal medium supplemented with 2% of casamino acids, 75 mM methionine, and 75 mM cysteine). At appropriate times of the chase, 200 l were withdrawn, mixed with 22 l of 100 mM NaN 3 solution, and placed on ice for 5 min. Then, cells were collected by centrifugation and resuspended in 150 l of Tris-buffered saline supplemented with 1% SDS and proteinase inhibitors mixture (5 g/ml of leupeptin, antipain, chymostatin, and pepstatin A; 2.5 g/ml aprotinin; and 1 mM phenylmethylsulfonyl fluoride). Glass beads were added to the suspension. The tube was vortexed for 30 s and chilled on ice for 30 s, repeating four times, and then heated at 100°C for 6 min. After cooling, 600 l of immunoprecipitation dilution buffer (1.25% Triton X-100, 190 mM NaCl, 6 mM EDTA, and 60 mM Tris-HCl (pH 7.4)) supplemented with the proteinase inhibitors mixture was added to the tube. After centrifugation at 10,000 rpm for 2 min, the supernatant was transferred to a fresh tube for immunoprecipitation.
In the case of R2497 strain harboring Ub-Leu-lacZ or pYPR3841U, transcriptional induction from the GAL promoter was initiated by adding 4% galactose to YNBRaf medium. Induction of Ub-Leu-␤-galactosidase and ⌬pro was kept for 4 and 6 h, respectively. The cell density was adjusted to 20 A 600 units/ml, and 0.3 ml was preincubated for 30 min with 200 M MG132 (Peptide Institute, Inc., Osaka, Japan) or an equal volume of Me 2 SO, the solvent. Then, 0.1-0.15 mCi of Tran 35 S-label was added and cells were metabolically labeled for 7 min. Subsequently, chase was carried out in the presence of 200 M MG132 or equal volume of Me 2 SO. YNBRaf medium supplemented with 50 mM methionine, 50 mM cysteine, amino acid mixture (20 g/ml histidine-HCl, 30 g/ml leucine, 20 g/ml arginine-HCl, 30 g/ml tyrosine, 30 g/ml isoleucine, 30 g/ml lysine-HCl, 50 g/ml phenylalanine, 100 g/ml glutamic acid, 100 g/ml aspartic acid, 150 g/ml valine, 200 g/ml threonine, and 400 g/ml serine), and 0.1 g/ml cycloheximide was used as a chase medium.
Immunoprecipitation of ⌬pro and CPY was carried out essentially as described previously (34). To precipitate the antibody-antigen complex, protein A-Sepharose CL-4B (Sigma Chemical Co., St. Louis, MO) was used. Disruption of cells and immunoprecipitation for Leu-␤-gal were carried out as described by Bachmair et al. (44). The immunoprecipitates were loaded onto the SDS-polyacrylamide gel electrophoresis (PAGE) gel. Gels were dried and analyzed by the Fuji FUJIX BAS2500 Bioimaging Analyzer (Fuji Photo Film Co. Ltd., Tokyo, Japan).
Electron Microscopy-Thin sections of yeast cells were prepared and viewed as described previously (37). For immunoelectron microscopy, thin sections were incubated with the anti-RNAP-I antibody and subsequently with 15-nm gold particle-conjugated secondary antibody as described previously (37).

Degradation of ⌬pro Occurs Independently of the Vacuolar
Proteinases-Our previous study indicated that ⌬pro was localized not to the vacuole but to the ER in pep4 prb1 cells, whose vacuolar proteinase activity is significantly reduced, suggesting that ⌬pro is not degraded in the vacuole (34). However, in that study, ⌬pro was produced by the GAPDH promoter and the dilation of the ER accompanying the accumulation of the ⌬pro aggregates was not remarkable, as judged from indirect immunofluorescence (34). Because the overproduction of ⌬pro by the GAL1 promoter resulted in extensive ER proliferation (37), we shifted cells from galactose-to glucose-containing medium to examine whether the dilated ER and its contents, the ⌬pro aggregates, were degraded in the vacuole after the production of ⌬pro was shut off. In wild-type cells, the amount of ⌬pro gradually decreased after the medium change, and most was degraded within 8 h (Fig. 1A). In ⌬pep4 prb1 cells, like wild-type cells, the accumulated ⌬pro was degraded A, BJ5407 (wild-type) and BJ5459 (⌬pep4 prb1) harboring pYPR3841 were grown in yeast minimal medium containing 5% glucose, 2% casamino acids, and uracil for 48 h. Then, cells were shifted to yeast minimal medium containing 5% galactose, 0.2% sucrose, 2% casamino acids, and uracil to induce the expression of ⌬pro. After 15 h, cells were collected and a portion was lysed to prepare whole protein extracts. The remaining was shifted to yeast minimal medium containing 5% glucose, 2% casamino acids, and uracil to stop the expression of ⌬pro. After 4 and 8 h, part of the culture was withdrawn and whole protein extracts were prepared. In each lane, 10 g of proteins was loaded, separated by SDS-PAGE, and transferred to the nitrocellulose membrane. ⌬pro was detected with the anti-RNAP-I antibody. B, wild-type strain R27-7C-1C harboring pYPR3841 was grown in YNBDCU medium for 48 h and cells were shifted to YNBGalCU medium to induce the expression of ⌬pro. After 12 h, cells were shifted to YNBDCU medium to stop the expression of ⌬pro and grown for 3 h. Cells were then fixed by the freeze substitution technique (45) for electron microscopic observation. The arrow indicates the connection between the proliferated membrane and the nuclear envelope. The arrowhead indicates the electron-dense material localized in the nuclear envelope. The scale bar indicates 1 m. N, nucleus. C, thin sections prepared in B were incubated with the anti-RNAP-I antibody and then with a colloidal gold-conjugated secondary antibody. Note that gold particles were seen on the electron dense regions in the dilated ER but not on the vacuole. The scale bar indicates 0.5 m. N, nucleus; V, vacuole. after its production was shut off, and little was remaining after 8 h (Fig. 1A), indicating that vacuolar proteinases do not participate in the degradation of ⌬pro. We also observed wild-type cells by electron microscopy using the freeze-substituted fixation method (45). At 3 h after the medium shift, the dilated ER was still observed in the cells (Fig. 1, B and C). As we reported previously (37), negatively stained membranes, which seemed to be derived from the nuclear envelope (arrow in Fig. 1B), proliferated in the cytoplasm (Fig. 1, B and C). Electron-dense materials, which represent the ⌬pro aggregates, were present in the lumen of the proliferating membrane structure as well as in the nuclear envelope (arrowhead in Fig. 1B). Unlike methylotrophic yeasts, in which autophagic degradation of peroxi-somes has been observed (46,47), no degradative intermediates were observed through autophagy. In S. cerevisiae, spherical bodies called autophagic bodies, which contain intracellular structures, including rough ER, mitochondria, cytosolic ribosomes, are induced upon nutrient starvation and become visible in the vacuole (48). However, in this case, such spherical bodies containing the dilated ER with the ⌬pro aggregates were not observed in the vacuole (Fig. 1, B and C). Immunoelectron microscopy with the anti-RNAP-I antibody detected ⌬pro on the electron-dense regions in the dilated ER but not on the vacuole (Fig. 1C). These data exclude the possibility that FIG. 2. Vesicular transport from the ER is not required for degradation of ⌬pro. Plasmid pYIA3841 was integrated into ANY21 (wild-type) and MBY10-7A (sec12-4) to express the ⌬pro gene in the yeast genome. Pulse-chase experiments were carried out using these strains as described under "Experimental Procedures" with some modifications. To induce the synthesis of ⌬pro, cells were grown at 23°C for 6 h using 5% galactose and 0.2% sucrose as carbon sources. After preincubation for 20 min at 37°C, 12 A 600 units of cells were labeled with 0.12 mCi of [ 35 S]methionine/cysteine and chased at 37°C. At each chase period, 4 A 600 units of cells were collected, and the cell extracts were divided into halves; one for immunoprecipitation of ⌬pro and the other for CPY. Relative percentages of the remaining ⌬pro, which were normalized to the 0-min chase value, are indicated below the lanes.
FIG . 3. Degradation of ⌬pro is not inhibited by the potent proteasome inhibitor MG132. A, R2497 (⌬erg6) harboring Ub-Leu-lacZ was grown, metabolically labeled and chased as described under "Experimental Procedures." To avoid prolonged inhibition of the proteasome that may induce pleiotropic and secondary effects, the chase periods were shortened by stopping the synthesis of Leu-␤-galactosidase and ⌬pro with 0.1 g/ml cycloheximide in the chase medium. Cells were disrupted, and Leu-␤-galactosidase was immunoprecipitated. The arrowhead probably represents a degradation product of Leu-␤-galactosidase. B, R2497 harboring pYPR3841U was grown, metabolically labeled, and chased as in A. ⌬pro was detected after immunoprecipitation from whole protein extracts.
FIG. 4. Degradation of ⌬pro is not inhibited in the proteasome mutant pre1 pre2. Plasmid pYIL3841 was integrated into WCGY4a (wild-type) and WCGY4 11/22a (pre1-1 pre2-2) to express the ⌬pro gene in the yeast genome. A, to induce the synthesis of ⌬pro, these strains were grown at 23°C for 6 h using 5% galactose and 0.2% sucrose as carbon sources. Then, cells were shifted to 37°C and incubated for 1 h before lysis. 15 g of proteins was loaded in each lane, separated by SDS-PAGE, and transferred to the nitrocellulose membrane. CPY* and ⌬pro were detected with the anti-CPY and the anti-RNAP-I antibodies, respectively. B, pulse-chase experiments were carried out as described under "Experimental Procedures" with some modifications. Synthesis of ⌬pro was induced as described in A. After preincubation for 30 min at 37°C, 9 A 600 units of cells were labeled for 10 min with 0.09 mCi of [ 35 S]methionine/cysteine and chased at 37°C. At each chase period, 3 A 600 units of cells were collected. Relative percentages of the remaining ⌬pro, which were normalized to the 0-min chase value, are indicated below the lanes. C, the set of pulse-chase analysis shown in B was independently carried out three times, and the mean values are diagrammed.
the ⌬pro aggregates are degraded by a vacuolar proteinase-dependent autophagic pathway.
Export from the ER Is Not Required for the Degradation of ⌬pro-We next examined the degradation of ⌬pro when vesicular transport from the ER was blocked. To this end, the sec12-4 mutant defective in vesicle budding from the ER (49) was used, and a pulse-chase experiment was carried out at the restrictive temperature (37°C). In wild-type cells, CPY was transported to the vacuole and processed to the mature form (m CPY, Fig. 2). In the sec12-4 mutant, however, CPY persisted in the ER (p1) form throughout the chase (Fig. 2), indicating that the transport from the ER was completely blocked. Under this condition, degradation of ⌬pro was not inhibited in the sec12-4 mutant (Fig. 2). After 60 min of the chase, the remaining ⌬pro was 63% in wild-type cells and 55% in sec12-4. Thus, ⌬pro was degraded independently of the ER export. Together with our previous finding that ⌬pro is accumulated in the ER (34, 37), we concluded that degradation of ⌬pro occurs in the ER.
Inhibition of the Proteasome Does Not Prevent the Degradation of ⌬pro-Because a variety of ER proteins are known to be dislocated to the cytosol and subsequently degraded by the proteasome (reviewed in Ref. 50), we addressed the question of whether ⌬pro is degraded by the proteasome. First, the effect of a proteasome inhibitor on the degradation of ⌬pro was tested. The potent proteasome inhibitor MG132 (reviewed in Ref. 51) has been shown to permeate the membrane of a yeast erg6 mutant and reversibly inhibit the degradation of short-lived proteins by 80% (52). In addition, MG132 has been shown to inhibit the ERAD of a mutant form of the plasma membrane protein Ste6p (53) and of CPY* (54). Here we performed pulsechase experiments to test whether MG132 prevents the degradation of ⌬pro in ⌬erg6 cells. As a control, the proteolysis of Leu-␤-galactosidase (44), which is known to be degraded by the ubiquitin-proteasome pathway according to the N-end rule, was compared with that of ⌬pro. In the presence of 200 M MG132, cells were preincubated for 30 min before being pulselabeled and chased for the indicated periods. Under this condition, the degradation of Leu-␤-galactosidase was completely inhibited throughout the 60-min chase period (Fig. 3A). The lower band indicated by the arrowhead was observed only in the absence of MG132, suggesting that this band represents a degradation product of Leu-␤-galactosidase. In contrast, MG132 did not inhibit the degradation of ⌬pro under the same condition (Fig. 3B). Regardless of the presence of this drug, most ⌬pro (Ͼ85%) disappeared during the 40-min chase period. Next, we monitored the degradation of ⌬pro in a proteasome mutant (pre1-1 pre2-2) defective in two subunits of the proteolytic core (55). In immunoblotting, this mutant accumulated higher amount of CPY* than the wild-type strain did (Fig. 4A), consistent with the degradation defect reported before (19). In contrast, ⌬pro accumulated in the pre1 pre2 mutant to the same degree as in wild-type (Fig. 4A). In a pulse-chase analysis, the rate of ⌬pro degradation in pre1 pre2 was almost identical to that in wild-type (Fig. 4B). Quantitation from re- peated experiments indicated that the degradation rate in pre1 pre2 was almost indistinguishable from that in wild-type (Fig.  4C). The finding that neither the proteasome inhibitor nor the proteasome mutant impaired the degradation of ⌬pro indicates that ⌬pro is not degraded by the proteasome via the ER-tocytosol dislocation process.
Isolation of Mutants That Accumulate Higher Amounts of ⌬pro-The data presented above show that ⌬pro is degraded independently of two known degradative pathways, the vacuolar proteinase and the cytosolic proteasome pathways. To investigate how ⌬pro is degraded, we attempted to isolate mutants defective in the degradation of ⌬pro.
Previously, we reported that intracellular accumulation of ⌬pro was not obvious when ⌬pro was produced with a multicopy plasmid pYPR2841, where the ⌬pro gene is under the control of the GAPDH promoter (34). In this expression system, we screened for mutants that accumulate higher intracellular levels of ⌬pro. Some of these mutants would be defective in the degradation of ⌬pro. The screening was performed using a colony-blotting assay (see "Experimental Procedures"). To confirm the validity of this method, we investigated the accumulation of wild-type and mutated RNAP-Is that were expressed by the GAPDH promoter in wild-type cells (Fig. 5A). An intense signal was detected from the cells producing wild-type RNAP-I. This is because secreted RNAP-I was directly adsorbed to the nitrocellulose membrane. In contrast, only moderate and faint signals were detected from the cells producing mutated RNAP-Is, M1 and ⌬pro (34), respectively. These results are consistent with our previous observation that both M1 and ⌬pro were degraded in the ER without being secreted, and that M1 was degraded more slowly than ⌬pro (34). Among 80,000 mutagenized colonies tested, eight were found to accumulate ⌬pro reproducibly to the same extent as M1 in wild-type cells, as judged from the intensity of the signals (Fig. 5B). Of these, three mutants, Nos. 2, 22, and 52, exhibited relatively intense signals in the repeated experiments. Genetic analysis revealed that these three mutants fell into different complementation groups and that each mutant contained a single recessive mutation. One of these mutants, mutant 22, was further characterized.
Cell extracts were prepared to detect the accumulation of ⌬pro in the mutant 22. Expression of ⌬pro was induced by the GAL1 promoter, and the time course of intracellular accumulation of ⌬pro was analyzed. Overproduction of ⌬pro had no  Fig. 6A. At 9 and 20 h after the expression of ⌬pro had been started, total RNA was extracted for Northern hybridization using a probe specific for the KAR2 gene. In the ϩ lanes, ⌬pro was overproduced with pYPR3841 whereas in the Ϫ lanes it was not produced with pYPR3831X (37). inhibitory effect on the growth of the mutant 22. 2 In the mutant cells, accumulation of ⌬pro was detected from 6 h after its induction of expression (Fig. 6A). Under the same experimental condition, accumulation of ⌬pro in wild-type cells was detected only after 9 h of induction (Ref. 37 and Fig. 6B). The difference was obvious at 6 and 9 h, however, the mutant 22 cells accumulated ⌬pro to the same extent as wild-type cells after 12 h (Fig. 6B). The mRNA levels for the ⌬pro gene in wild-type and the mutant 22 cells were comparable, 2 suggesting that the marked accumulation of ⌬pro in the early time points by the mutant 22 cells was due to the defect in the degradation of ⌬pro (see Figs. 10 and 11 below).
Dilation of the ER and Unfolded Protein Response in the Mutant Defective in the Degradation of ⌬pro-Our previous study using wild-type cells demonstrated that overproduced ⌬pro formed gross aggregates in the ER, which resulted in enlargement of the ER (37). In the present study, electron microscopy revealed similar morphological changes in the mutant 22 cells. Proliferated ER membranes were observed in the cytoplasm (Fig. 7, A and B). They seemed to be connected with the nuclear envelope (Fig. 7B, arrow). Electron-dense materials were observed in the luminal side of the proliferated membranous structures as well as in the nuclear envelope (Fig. 7B,  arrowhead). They were shown to contain ⌬pro by immunoelec- FIG. 10. Mutation in IRA2, which encodes a GTPase-activating protein for Ras, causes the degradation defect of ⌬pro. A, introduction of IRA2 reduced the accumulation of ⌬pro in the mutant defective in the degradation of ⌬pro. To express the ⌬pro gene in the yeast genome, plasmid pYIL3841 was integrated into strains R27-7C-1C (wild-type) and KUY20 (mutant). The resultant strains were transformed with the indicated plasmids and grown as described in Fig. 6A. At 4.5 h after the expression of ⌬pro had been started, protein extracts were prepared. In each lane, proteins prepared from 1 A 600 unit of cells were loaded. ⌬pro was detected with the anti-RNAP-I antibody. B, pulse-chase analysis demonstrated that the ⌬pro degradation defect was complemented by IRA2. Using strains KUY90 (wild-type) and KUY91 (mutant) harboring the indicated plasmids, pulse-chase analysis was carried out (see "Experimental Procedures"). Relative percentages of the remaining ⌬pro, which were normalized to the 0-min chase value, are indicated below the lanes. C, the mutated locus is identical to IRA2. R27-7C-1C (wild-type), KUY20 (mutant), KUY34 (⌬ira2), KUY35 (wild-type ϫ ⌬ira2), and KUY36 (mutant ϫ ⌬ira2) strains harboring pYPR3841 were grown as described in Fig. 6A. At 6 h after the expression of ⌬pro had been started, protein extracts were prepared and 10 g of proteins was loaded in each lane. ⌬pro was detected with the anti-RNAP-I antibody.
FIG. 11. Suppression of the ⌬pro degradation defect of ira2 by overexpression of PDE2, which encodes the high affinity cAMP phosphodiesterase. A, strains KUY90 (wild-type) and KUY91 (ira2) harboring the indicated plasmids were grown as described in Fig. 6A. At 6 h after the expression of ⌬pro had been started, protein extracts were prepared. In each lane, 20 g of proteins was loaded. ⌬pro was detected with the anti-RNAP-I antibody. B, using strains KUY90 (wild-type) and KUY91 (ira2) harboring the indicated plasmids, pulse-chase analysis was carried out (see "Experimental Procedures"). Relative percentages of the remaining ⌬pro, which were normalized to the 0-min chase value, are indicated below the lanes. C, the set of pulse-chase analysis shown in B was independently carried out three times, and the mean values are diagrammed. tron microscopical analysis with the anti-RNAP-I antibody. 3 These results indicate that, in the mutant 22 cells as in wildtype cells, ⌬pro forms gross aggregates in the ER and ER membrane proliferation occurs in response to the ⌬pro aggregates.
Overproduction of ⌬pro has been shown to trigger the UPR (56,57) in wild-type cells. That is, mRNA of KAR2, which encodes yeast BiP, was induced upon overproduction of ⌬pro (37,38). In the mutant cells, Northern analysis (Fig. 8) indicated that overproduction of ⌬pro also increased the level of KAR2 mRNA about 2.3-fold (compare lanes 1 and 2) and 1.5fold (compare lanes 3 and 4) at 9 and 20 h after the expression of ⌬pro was induced, respectively. Thus, the mutation does not impair the ER-to-nucleus UPR pathway.
IRA2, which Negatively Regulates Yeast Ras, Is Required for the Degradation of ⌬pro-The mutant 22 showed the following phenotypes: a defect in fermentation of galactose (Fig. 9A), reduced accumulation of glycogen, and sensitivity to nutrient starvation. 4 These phenotypes are similar to those of mutants in which the Ras-cAMP pathway is hyperactivated (58 -61). In S. cerevisiae, activation of Ras leads to an elevated level of cAMP, which stimulates the activity of protein kinase A (PKA) to ensure cell proliferation (reviewed in Ref. 62). During the course of genetic analysis, we found that the above phenotypes were caused by a single and recessive mutation that was tightly linked with the mutated locus responsible for the ⌬pro degradation defect. 4 Therefore, by complementation of the galactose-nonfermentable phenotype, which was dependent on high temperature (37°C), we tried to isolate the gene that is located very close to, or is identical to, the gene involved in the degradation of ⌬pro. Using a plasmid-borne bank, however, identification of the authentic genomic locus was unsuccessful. Instead, the PDE2 gene, which encodes the high affinity cAMP phosphodiesterase in yeast (63), was identified as a high copy suppressor of the galactose-nonfermentable phenotype (Fig.  9A). Because the PDE2 gene product negatively regulates the Ras-cAMP pathway by decreasing the amount of cAMP, this result strongly suggests that the Ras-cAMP pathway is hyperactivated in the mutant 22. Hence, we reasoned that a gene, the product of which acts upstream of Pde2p to down-regulate the Ras-cAMP pathway, is mutated in the mutant 22. Components of the yeast Ras-cAMP pathway have been extensively identified by genetic approaches. Therefore, the gene in question was likely to have already been characterized. The failure in the cloning by means of a plasmid-borne bank may indicate that the gene is very large. Taking these possibilities into consideration, we suspected that either IRA1 or IRA2 was mutated in the mutant 22. Both of these genes encode the yeast GTPase-activating proteins (GAPs) for Ras, which convert a GTP-bound active form of Ras to a GDP-bound inactive form (64 -66). In addition, each has an unusually large ORF of about 9 kb in length. Because the length of the genomic inserts in the plasmid-borne library was 10 ϳ 20 kb, these inserts have rarely encompassed the entire region of the IRA1 or IRA2 ORF. In fact, several lines of evidence indicated that IRA2 was mutated in the mutant 22. First, IRA2 complemented the galactosenonfermentable phenotype of the mutant 22 (Fig. 9A). Second, when the mutant No.22 was crossed to a ⌬ira2 strain, the resultant diploid showed a galactose-nonfermentable phenotype (Fig. 9B). Third, when this diploid strain was sporulated and the four-spored asci were dissected, the haploid segregants from the 13 asci were all galactose-nonfermentable. 4 Having established that the genetic locus involved in the degradation of ⌬pro was very close to or identical to IRA2, we next attempted to determine which genomic fragment around IRA2 complemented the ⌬pro degradation defect. Our results showed that the marked accumulation of ⌬pro in the mutant was rectified by introduction of IRA2 (Fig. 10A). A pulse-chase analysis (Fig. 10B) shows that the rate of ⌬pro degradation in the mutant was significantly slower than that in the wild-type strain. This defect was complemented by the introduction of IRA2. To confirm that IRA2 is not an extragenic suppressor, the mutant was mated with a ⌬ira2 strain. Accumulation of ⌬pro in the resultant diploid (Fig. 10C, lane 5) was comparable to that in a haploid mutant strain (Fig. 10C, lane 2) and was higher than that in a control strain that was made by mating the wild-type strain to a ⌬ira2 strain (Fig. 10C, lane 4). From these results, we concluded that mutation in IRA2, which encodes a negative regulator of the Ras-cAMP pathway, caused the degradation defect of ⌬pro.
The identification of IRA2 raised the possibility that hyperactivation of the Ras-cAMP pathway interferes with the ER degradation of ⌬pro. However, it has previously been shown that yeast Ras has a function distinct from the regulation of the cAMP level (67). It is therefore possible that, in addition to the elevation of the cAMP level, Ira2p has another biological function that is related to the ER degradation of ⌬pro. To investigate this possibility, we tested whether the defect of ⌬pro degradation in the ira2 mutant could be suppressed by the overexpression of PDE2 and thereby the decrease of the cAMP level. As shown in Fig. 11A, accumulation of ⌬pro in ira2 cells was reduced by the overexpression of PDE2 (compare lanes 2 and 3). In a pulse-chase analysis, the rate of ⌬pro degradation in the ira2 mutant was consistently accelerated by the overexpression of PDE2 (Fig. 11, B and C). Thus, in both immunoblotting and pulse-chase analyses, overexpression of PDE2 significantly suppressed the ira2 defect. These results strongly suggest that elevation of cAMP level in the ira2 mutant leads to the impaired degradation of ⌬pro.

DISCUSSION
In this article, we investigated whether the degradation of ⌬pro, which forms large aggregates in the yeast ER lumen, depends on vacuolar proteinases-and cytosolic proteasome-dependent pathways. The results presented here demonstrate that the degradation of ⌬pro was not affected when either pathway was blocked, indicating that the ⌬pro aggregates are mostly degraded in the ER lumen. To elucidate the degradative pathway(s) of ⌬pro, we isolated eight mutants that accumulate more ⌬pro than wild-type cells. One of these, the mutant 22, was shown to be defective in the degradation of ⌬pro. We have demonstrated that the mutated gene is IRA2, which encodes a negative regulator of the yeast Ras-cAMP pathway. Therefore, we propose that the ⌬pro aggregates are degraded by an as yet unidentified proteolytic activity in the ER lumen, which is negatively regulated by the Ras-cAMP signaling pathway.
Where Are the ⌬pro Aggregates Degraded?-Several reports have indicated that proliferated organelles are degraded in the vacuole via autophagy. In the methylotrophic yeasts Hansenula polymorpha and Pichia pastoris, peroxisomes that developed during the growth on methanol are degraded in the vacuole in the course of adaptation to nonmethylotrophic medium (46,47). In hepatocytes, smooth ER membranes that proliferated by phenobarbital treatment are removed by autophagic lysosomes (68). In the electron microscopic analysis of wild-type cells, however, we did not observe the sequestration of the proliferated ER, which contained the ⌬pro aggregates, into the vacuole. This result by itself does not rule out the involvement of autophagic processes, because, in a previous study on S. cerevisiae, accumulation of autophagic bodies was not observed in the vacuole of wild-type cells due to rapid degradation in a proteinase B-dependent manner (48). However, our finding that ⌬pro was degraded in ⌬pep4 prb1 cells at a rate similar to that in wild-type cells (Fig. 1A) argues against the possibility that the ER containing the ⌬pro aggregates is degraded by autophagy. We have also shown that the vesicular transport from the ER is not required for degradation of ⌬pro. It was reported that an ER misfolded protein is exported from the ER to the trans-Golgi then captured by Vps10 receptor and delivered to the vacuole for degradation (69). However, it is unlikely that ⌬pro is degraded by following this route because ⌬pro was normally degraded in the sec12-4 mutant, which blocks the exit from the ER. From these results, we concluded that the site of ⌬pro degradation is the ER. Degradation of proliferated membrane structures is often accompanied by the appearance of morphologically distinctive degradation intermediates. Several examples have been reported in yeast cells. One example, as described above, is the sequestration of the peroxisomes into the vacuole. Another is a structure called a "whorl," which has been observed during the degradation of karmellae, an HMG-CoA reductase-induced stacked membrane structure surrounding the nucleus (70). Subsequent analysis has revealed the presence of acidic compartments and lipid particles in the interior of whorls, suggesting that acidification in whorls activates the degradation of membranes to form lipid particles (71). Thus, identification of the degradation intermediates is expected to provide insights into the degradation process. As shown in Fig. 1, we examined the proliferated ER containing the ⌬pro aggregates by electron microscopy, but no other particular morphological changes were observed during the degradation of the ⌬pro aggregates. Thus, from electron microscopy, we have no clues at present to elucidate the mechanism through which ⌬pro aggregates are degraded in the proliferated ER.
Although proteasome-dependent degradation of ER misfolded proteins has been reported, we have shown that the proteolytic activities of the proteasome are not required for the ER degradation of ⌬pro by using the proteasome inhibitor MG132 and the proteasome mutant pre1 pre2. Two possibilities can be considered for the degradation of ⌬pro. First, ⌬pro is dislocated from the ER to the cytosol, and then degraded by the protease(s) distinct from the proteasome. Second, ⌬pro is degraded in the ER lumen without being dislocated to the cytosol. We favor the latter possibility, because ⌬pro forms gross aggregates in the ER lumen, which are unlikely to be completely unfolded and become competent for retrograde transport across the Sec61 channel. We cannot rule out the possibility that a minor population of ⌬pro, which has not yet been incorporated into the aggregates in the ER lumen, can be dislocated to the cytosol and degraded by the proteasome. For direct examination whether ⌬pro is retranslocated to the cytosol, we tried to use the dislocation-deficient sec61-R4 mutant (25). We found, however, that the translocation of ⌬pro into the ER lumen was impaired in this mutant 4 and did not pursue this experiment further. To discriminate whether or not ⌬pro is dislocated, it will be necessary to monitor the degradation of ⌬pro in vitro by isolating the proliferated ER that is marked by the presence of the ⌬pro aggregates. Characterization of this compartment will be helpful to identify the components involved in the degradation of ⌬pro.
Although the ERAD pathway, which means dislocation to the cytosol and subsequent degradation by the proteasome, has been extensively characterized, the possibility of other pathway(s) in ER protein degradation is still an open question. Some experimental data are suggestive of an alternative degradation pathway(s). For example, ERAD mutants such as ubc7 and hrd1 do not show sensitivity to tunicamycin, which causes accumulation of misfolded proteins in the ER (72). These ER misfolded proteins may well be eliminated by alternative degradation machinery in these ERAD mutants.
Involvement of the Ras-cAMP Pathway in the Regulation of the Proteasome-independent Degradation of the ⌬pro Aggregates-We have shown that mutation in IRA2, which encodes a negative regulator of Ras, is responsible for the degradation defect of the ⌬pro aggregates. We also showed that downregulation of the Ras-cAMP pathway by overexpression of PDE2 suppresses the defect of the ira2 mutant in the degradation of ⌬pro. It is unlikely that Ira2p itself is responsible for the degradation of ⌬pro in the ER lumen. Rather, the ira2 mutant seems to be defective in the signal transduction pathway upstream of the ER degradation. Recently, it has been revealed that one of the yeast MAP kinase pathways is required for correct localization of ␣-1,3-mannosyltransferase to early Golgi compartments (73). This study and our present one indicate an unanticipated linkage between signal transduction pathways and organelle functions. We propose that the hyperactivation of the Ras-cAMP pathway inhibits the ER degradation of the ⌬pro aggregates through the constitutive activation of PKA. In mouse, tissue-specific localization of PKA to the ER through the interaction with protein kinase A-anchoring protein is suggested (74). Likewise, in yeast, PKA may be at least partially localized to the ER, where it negatively regulates the ER degradation by phosphorylating the components involved in this process. There are three isoforms of yeast PKA catalytic subunit (Tpk1, Tpk2, and Tpk3), and each has recently been shown to have a specific function (75,76). Therefore, it is worth investigating which isoform is engaged in the ER degradation as well as the localization of the isoform.
The UPR is a well-known signal transduction pathway, which regulates the ER quality control involving the degradation of misfolded proteins. We have shown here that transcriptional induction of KAR2 in response to the accumulation of the ⌬pro aggregates in the ER is normal in the ira2 mutant, indicating that the hyperactivation of the Ras-cAMP pathway does not inhibit the UPR. The transient phenotype of ⌬pro accumulation in ira2 cells could be related to the function of the UPR pathway as follows. Soon after the start of ⌬pro expression from the GAL1 promoter, marked accumulation of ⌬pro in the ER of ira2 cells is observed because of the degradation defect. This induces the UPR against the ⌬pro aggregates to prevent further accumulation of ⌬pro. Because one of the functions of molecular chaperones is to facilitate protein degradation (reviewed in Ref. 77), induction of ER resident chaperones by the UPR pathway may promote the degradation of ⌬pro to compensate for the defect of ira2 cells. Therefore, marked accumulation of ⌬pro is restricted to the early stage after the expression of ⌬pro is induced. Similar phenomena have been reported in the ER-associated degradation of ␣-1-protease inhibitor and unglycosylated pro-␣ factor (78). Also, it has been recently shown that the UPR is required for degradation of ER misfolded proteins (72,79,80).
In view of the ER degradation of ⌬pro, what is inhibited by the hyperactivation of the Ras-cAMP pathway? It is known that yeast PKA negatively regulates cellular responses to a variety of stress by inhibiting STRE (stress response element)dependent transcription of a subset of genes (81). It is possible that some components, which act on the ER degradation of ⌬pro, are transcriptionally regulated through the STRE. Recently, the yeast genome has been comprehensively searched for genes with STREs (82). Such genes may include those involved in the ER degradation of ⌬pro. The chaperone Hsp104, for example, may be such a candidate. Hsp104 is not only transcriptionally regulated by STRE (83) but has recently been shown to be required for conformational repair of heat-denatured proteins in the ER (84). In our continued study of this subject, identification of downstream components of the Ras-cAMP pathway and characterization of other mutants isolated in this study should provide additional insights into the ER degradation of the ⌬pro aggregates.