|Home | About | Journals | Submit | Contact Us | Français|
Membrane and secretory proteins fold in the endoplasmic reticulum (ER), and misfolded proteins may be retained and targeted for ER-associated protein degradation (ERAD). To elucidate the mechanism by which an integral membrane protein in the ER is degraded, we studied the fate of the cystic fibrosis transmembrane conductance regulator (CFTR) in the yeast Saccharomyces cerevisiae. Our data indicate that CFTR resides in the ER and is stabilized in strains defective for proteasome activity or deleted for the ubiquitin-conjugating enzymes Ubc6p and Ubc7p, thus demonstrating that CFTR is a bona fide ERAD substrate in yeast. We also found that heat shock protein 70 (Hsp70), although not required for the degradation of soluble lumenal ERAD substrates, is required to facilitate CFTR turnover. Conversely, calnexin and binding protein (BiP), which are required for the proteolysis of ER lumenal proteins in both yeast and mammals, are dispensable for the degradation of CFTR, suggesting unique mechanisms for the disposal of at least some soluble and integral membrane ERAD substrates in yeast.
The endoplasmic reticulum (ER) is the site in which membrane and secretory proteins fold, and only properly folded proteins usually exit the ER. Incompletely folded proteins may be retained in the ER, and if folding cannot be achieved, they may form aggregates or be targeted for ER-associated protein degradation (ERAD) (recent reviews by Bonafacino and Weissman, 1998 ; Brodsky and McCracken, 1999 ; Plemper and Wolf, 1999 ; Römisch, 1999 ). From studies with both yeast and mammalian cells, the molecular mechanism of ERAD has recently begun to emerge. The selection of ERAD substrates is highly specific because misfolded proteins have to be distinguished from correctly folded and folding-competent proteins. Molecular chaperones may participate in the selection process because they assist in protein folding and a prolonged association between ER lumenal and cytosolic molecular chaperones and misfolded proteins has been observed for several ERAD substrates (Yang et al., 1993 ; Ping et al., 1994 ; Knittler et al., 1995 ; Schmitz et al., 1995 ; Sawa et al., 1996 ; de Virgilio et al., 1999 ). In addition, mutations in some ER lumenal chaperones prevent ERAD in yeast (McCracken and Brodsky, 1996 ; Plemper et al., 1997 ; Brodsky et al., 1999 ; Gillece et al., 1999 ).
Both in vivo and in vitro data suggest that ERAD substrates are targeted to the proteasome (Biederer et al., 1996 ; Hampton et al., 1996 ; Hiller et al., 1996 ; Qu et al., 1996 ; Werner et al., 1996 ; Wiertz et al., 1996 ), which is a multicatalytic, proteolytic complex in the cytoplasm (reviewed by Voges et al., 1999 ). For ERAD, ubquitination is necessary for most (Jensen et al., 1995 ; Ward et al., 1995 ; Hiller et al., 1996 ; Biederer et al., 1997 ; Hampton and Bhakta, 1997 ; Loayza et al., 1998 ; Zhou et al., 1998 ) but not all substrates (McGee et al., 1996 ; Werner et al., 1996 ; Yu et al., 1997 ). Because the proteasome resides in the cytoplasm, soluble ERAD substrates must be retro-translocated from the ER, and Sec61p, the primary component of the translocon, is required for the degradation of soluble proteins (Pilon et al., 1997 ; Plemper et al., 1997 ). For the proteolysis of membrane proteins, a role for Sec61p has also been proposed (Wiertz et al., 1996 ; Bebök et al., 1998 ; Plemper et al., 1998 ). Membrane proteins may be extracted from the ER membrane before being degraded by the proteasome, or their cytosolic portions may be “shaved” or “clipped” by the proteasome, and the lumenal loops and transmembrane domains may then be handled either by the proteasome or by an undefined protease. Indeed, there is experimental evidence supporting the involvement of multiple ERAD pathways. Specifically, both signal peptidase (Mullins et al., 1995 ), and cysteine and serine proteases have been implicated in the degradation of some ER proteins (Wileman et al., 1991 ; Wikstrom and Lodish, 1992 ; Gardner et al., 1993 ; Moriyama et al., 1998 ; Fayadat et al., 2000 , and references therein).
The importance of the ERAD pathway in cellular physiology is underscored by the fact that several disease-associated molecules are ERAD substrates (reviewed by Brodsky and McCracken, 1999 ). One example is the cystic fibrosis transmembrane conductance regulator (CFTR), the protein in which mutations give rise to cystic fibrosis (Riordan et al., 1989 ). CFTR is a plasma membrane chloride channel and is composed of two membrane spanning domains (MSD), each of which has six transmembrane segments, two nucleotide-binding domains (NBD1 and NBD2), and a central regulatory (“R”) domain. CFTR folding and maturation in the ER is an inefficient, temperature-sensitive process, as indicated by the fact that ~80% of wild-type CFTR is degraded via ERAD (Cheng et al., 1990 ; Denning et al., 1992 ; Lukacs et al., 1994 ; Jensen et al., 1995 ; Ward et al., 1995 ).
Molecular chaperones have been suggested to facilitate CFTR maturation. Cytosolic chaperones heat shock protein 70 (Hsp70) (Yang et al., 1993 ), Hdj-2 (Meacham et al., 1999 ), and Hsp90 (Loo et al., 1998 ), and the ER lumenal chaperone calnexin (Ping et al., 1994 ) transiently associate with CFTR in the ER, and their dissociation coincides with CFTR maturation. Purified Hsp70 suppresses the aggregation of the NBD1 in vitro (Strickland et al., 1997 ) and acts synergistically with Hdj-2 to this end (Meacham et al., 1999 ). Recently, it was demonstrated that perturbing Hsp90-CFTR association through the use of Hsp90-interacting compounds accelerates CFTR degradation (Loo et al., 1998 ), suggesting that if Hsp90 cannot fold CFTR, CFTR is instead targeted for degradation.
Because the secretory pathway and ERAD mechanism are conserved between yeast and mammalian cells, and because yeast present powerful genetic tools, we expressed wild-type CFTR in this organism to begin to dissect the pathway by which an integral membrane ERAD substrate is targeted for proteolysis.
S. cerevisiae strains used were as follows: the ssa1 temperature-sensitive strain JB67 (Matα, his3-11,15, leu2-3112, ura3-52, trp1Δ1, lys2, ssa1-45, ssa2-1, ssa3-1, ssa4-2) and isogenic wild-type JN516 (Matα, his3-11,15, leu2-3112, ura3-52, trp1Δ1, lys2, SSA1, ssa2-1, ssa3-1, ssa4-2) (Becker et al., 1996 ); Δcne1 (Mata, ade2-1, can1-100, ura3-1, leu2-3112, trp1-1, his3-11,15, cne1::LEU2) and isogenic wild-type W301-1a (Mata, ade2-1, can1-100, ura3-1, leu2-3112, trp1-1, his3-11,15) (Parlati et al., 1995 ); proteasome mutant WCG4/2 [Mata, leu2-3112, his3-11,15, ura3Δ5, can(s), pre1-1,pre2-2] and isogenic wild-type strain WCG4 (Mata, leu2-3112, his3-11,15, ura3Δ5) (Heinemeyer et al., 1991 ); ubiquitin-conjugating mutant MHY552 (Matα, his3-Δ200, ura3-52, leu2-3112, lys2-801, trp1-1, ubc6-Δ1::HIS3, ubc7::LEU2) and isogenic wild-type MHY501 (Matα, his3-Δ200, ura3-52, leu2-3112, lys2-801, trp1-1) (Chen et al., 1993 ); BiP mutant strains MS1111 (Mata, ura3-52, leu2-3112, ade2-101, kar2-1), MS193 (Mata, ura3-52, leu2-3112, ade2-102, kar2-133), and isogenic wild-type RSY801 (Mata, ura3-52, leu2-3112, ade2-101) (Brodsky et al., 1999 ); vacuolar protease-deficient strain BJ5461 (Mata, ura3-52, trp1, lys2-801, leu2Δ1, his3Δ200, pep4::HIS3, prb1Δ1.6R, can1) and related wild-type BJ5242 (Mata, ura3-52, trp1, leu2-Δ1, his3-Δ200) (Jones, 1991 ); HRD mutants RHY1952 (Matα, lys2-801, his3Δ200, ura3-52, trp1-1, leu2-3112, hrd1::LEU2) and RHY1903 (Matα, lys2-801, his3Δ200, ura3-52, trp1-1, leu2-3112, hrd3::LEU2) and isogenic wild-type (Matα, lys2-801, his3Δ200, ura3-52, trp1-1, leu2-3112) (Wilhovsky et al., 2000 ).
Yeast transformation was performed by the lithium acetate procedure as described (Ito et al., 1983 ).
CFTR expression in yeast was driven by the constitutive phosphoglycerate kinase promoter in a 2 μ plasmid containing URA3 as the selectable marker. Construction of the plasmids used to express untagged and triple-hemagglutinin (HA)-tagged (at the carboxyl terminus) forms of CFTR in yeast will be described elsewhere (Zhang et al., 2001 ). Yeast containing the 2 μ plasmid pRS426 (Christianson et al., 1992 ) lacking insert were used as a control where indicated.
Antibodies used in this study were as follows: monoclonal anti-HA mouse (12CA5 clone, 400 μg/ml; Roche Molecular Biochemicals, Indianapolis, IN), monoclonal anti-C mouse (Genzyme, Cambridge, MA), polyclonal anti-Sec61p rabbit (Stirling et al., 1992 ), and polyclonal anti-binding protein (BiP) rabbit (Brodsky and Schekman, 1993 ) antibodies. Primary antibodies were detected with sheep anti-mouse IgG horseradish peroxidase-conjugated or donkey anti-rabbit horseradish peroxidase-conjugated antibody (Amersham Pharmacia Biotech, Piscataway, NJ) and the Amersham ECL Western blotting system was used to detect the signal according to the manufacturer's specifications. For quantitative analysis, 125I-protein A was used in place of secondary antibody, and the signal was visualized using a Fuji PhosphorImager and quantified with MacBas software (version 2.4) (Fujiphotofilm; Koshin Graphic Systems, Stanford, CT).
Yeast microsomes were prepared from a wild-type strain containing the HA-CFTR expression plasmid as described (Brodsky and Schekman, 1993 ). Then, a 10-μl aliquot of microsomes (~100 μg of total protein) was mixed with 1 ml of 100 mM Na2CO4 (pH 11.5) and incubated on ice for 30 min (Fujiki et al., 1982 ). After centrifugation at 230,000 × g at 4°C for 1 h, the pellet was dissolved in 35 μl of 2× SDS-PAGE sample buffer. Proteins in the supernatant were precipitated by incubation on ice for 30 min with trichloroacetic acid (TCA) added to a final concentration of 10%, followed by a 10-min centrifugation at 16,060 × g at 4°C. The pellet was resuspended in 35 μl of 2× SDS-PAGE sample buffer. Proteins from both the pellet and the supernatant were subjected to SDS-PAGE and immunoblot analysis as described above.
To confirm that CFTR was membrane-embedded and could thus float in a sucrose gradient, wild-type and pre1-1pre2-2 yeast transformed with the CFTR-expression vector were grown to midlog phase (OD600 = ~0.5) in selective medium, and the cells were harvested, washed, and then resuspended in membrane storage buffer/EDTA (MSB: 50 mM HEPES pH 7.6, 150 mM NaCl, 5 mM EDTA, 1 mM dithiothreitol supplemented with phenylmethylsulfonyl fluoride, leupeptin, and pepstatin according to the manufacturer's specifications) to a final concentration of ~10 OD600/ml. Glass beads were added to the meniscus and the suspension was agitated on a Vortex mixer four times for 30 s with a 1-min incubation on ice between each agitation. The extract was removed, the beads were washed with an equal volume of MSB, and the combined extracts were centrifuged two times at 300 × g for 2 min to remove unbroken cells. A total of 100 μl of the supernatant was mixed with 300 μl of MSB containing 2.3 M sucrose, and this solution was layered onto 300 μl of MSB containing 2.3 M sucrose in a centrifuge tube. MSB supplemented with 1.5 M sucrose (600 μl) and 0.25 M sucrose (500 μl) were then successively layered onto the gradient and the tube was centrifuged in a Beckman SW55 rotor at 100,000 × g for 5 h at 4°C. Aliquots of 150 μl were removed from the top of the gradient and protein profiles were analyzed by SDS-PAGE and immunoblotting.
Indirect immunofluorescence microscopy was performed essentially as described (Pringle et al., 1989 ). Log phase cells (OD600 = ~0.5–0.7) were fixed for 1 h at room temperature by adding formaldehyde to a final concentration of 3.7%. The yeast were washed twice with sorbitol buffer (50 mM potassium phosphate, pH 7.5, 1.2 M sorbitol), and resuspended to a concentration of 106 cells/10 μl in sorbitol buffer containing 0.1% β-mercaptoethanol and 20 μg/ml zymolase 20T (U.S. Biological, Swampscott, MA). After incubation at 37°C for 1 h, cells were washed with sorbitol buffer twice and applied to poly-l-lysine-coated glass slides (20 μl/well) before being fixed for 5 min with 3.7% formaldehyde in phosphate-buffered saline (PBS) (40 mM K2HPO4, 10 mM KH2PO4, pH 7.5, 0.15 M NaCl). A final concentration of 50 mM NH4Cl was used to quench the formaldehyde. The cells were permeabilized with 0.1% NP-40 in PBS supplemented with 0.1% bovine serum albumin (BSA) (PBS-0.1%BSA) and then incubated with anti-HA (1:250 dilution) or anti-C (1:40 dilution), and anti-BiP (1:500) antibodies in PBS-0.1%BSA overnight at 4°C in a humid chamber. After three washes with PBS/BSA and one wash with PBS/BSA/NP-40, the primary antibodies were detected with anti-rabbit IgG TRITC conjugate (1:250 dilution; Sigma, St. Louis, MO) and anti-mouse IgG fluorescein conjugate (1:250 dilution; Roche Molecular Biochemicals, Indianapolis, IN) in PBS-0.1%BSA.
Yeast either containing or lacking the CFTR-expression vector were grown to midlog phase and whole cell electron microscopy was performed as published by Kaiser and Schekman (1990) .
Cells expressing CFTR were grown to midlog phase (OD600 = ~0.5) at 26°C before cycloheximide was added to a final concentration of 50 μg/ml, and were incubated either at 26°C or shifted to 40°C with shaking before they were harvested at the indicated time points to prepare cell extracts. For each time point, a total of 2.5 OD600 of cells was harvested, washed with cold water, and resuspended in 1 ml of cold water. An aliquot of 150 μl of freshly prepared 2 N NaOH/1.12 M β-mercaptoethanol was added, and the yeast were resuspended and left on ice for 15 min. Then, 150 μl of 50% TCA was added, and the extract was incubated on ice for an additional 20 min. After centrifugation at 16,060 × g for 5 min at 4°C, the pelleted proteins were resuspended in TCA sample buffer (80 mM Tris-Cl, pH 8.0, 8 mM EDTA, 120 mM dithiothreitol, 3.5% SDS, 0.29% glycerol, 0.08% Tris base, 0.01% bromphenol blue) and incubated at 37°C for 30 min. Total protein was resolved by SDS-PAGE, followed by immunoblot analysis or quantitative immunoblot analysis (see above).
The degradation of the misfolded form of carboxypeptidase Y (CPY*; Hiller et al., 1996 ) was determined using an HA-epitope-tagged version of CPY* (Ng et al., 2000 ). In brief, cells transformed with the CPY* expression vector were grown to logarithmic phase in selective medium containing glucose and a pulse-chase analysis by using [35S]methionine was performed as described (Brodsky et al., 1998 ). CPY* was precipitated from ~5 × 106 cpm of 35S-labeled protein extract with 10 μg of anti-HA antibody (Roche Molecular Biochemicals, Indianapolis, IN) and protein A-Sepharose (Amersham Pharmacia Biotech). Yeast BiP was precipitated using anti-BiP antibody and protein A-Sepharose.
To assess the subcellular localization of CFTR expressed in yeast, indirect immunofluorescence microscopy was performed using cells grown exclusively at 26°C or that had been shifted to 40°C for 1 h. A temperature of 40°C was chosen because CFTR stabilization in temperature-sensitive yeast became maximally pronounced at 40°C (see below). Yeast harboring the CFTR-expressing plasmid (either containing or lacking an HA tag at the C terminus, referred to as HA-CFTR or untagged CFTR below, respectively) were subjected to indirect immunofluorescence microscopy. As shown in Figure Figure1,1, A and B, HA-CFTR in wild-type cells (denoted SSA1 and PRE) grown at either 26°C or shifted to 40°C for 1 h exhibit a strong perinuclear punctate pattern and subplasma membrane residency and colocalize with BiP, an ER lumenal Hsp70. For untagged CFTR, an antibody against the C terminus was used instead of α-HA, and identical results were obtained (our unpublished data). These data indicate that CFTR expressed in yeast resides primarily in the ER, as also suggested by others (Huang et al., 1996 ).
Expression of heterologous proteins in yeast, or overexpression of endogenous proteins may lead to an altered intracellular morphology (Wright et al., 1988 ; Umebayashi et al., 1997 ; Becker et al., 1999 ). To determine whether expression of CFTR similarly affects yeast, logarithmically growing cells containing either a control or the HA-CFTR expression plasmid were prepared for whole cell electron microscopy as described in MATERIALS AND METHODS. As shown in Figure Figure2,2, elongated tubular and enlarged vesicular structures were evident only in cells expressing CFTR. These structures are described in greater detail elsewhere (Kuehn, Nijbroek, and Michaelis, unpublished data), and also arise from the expression of mutant forms of the Ste6, the a factor mating pheromone transporter in yeast. They are distinct from those observed when other ER membrane proteins are overexpressed in yeast (Wright et al., 1988 ; Umebayashi et al., 1997 ; Becker et al., 1999 ), or when the secretory pathway is compromised in a sec mutant strain (Nishikawa et al., 1994 ). In contrast, we failed to observe impaired growth or defects in secretory protein translocation in yeast expressing CFTR (our unpublished results). Because ER membrane proliferation may arise from induction of the unfolded protein response in the ER (Cox et al., 1997 ), we introduced an unfolded protein response (UPR) reporter plasmid into yeast already containing the CFTR-expression plasmid or the plasmid lacking an insert, but found that the UPR was not induced by CFTR expression. Thus, the molecular basis for the membrane proliferation observed in the CFTR-expressing yeast is unknown.
To verify that CFTR expressed in yeast inserts into the ER membrane, ER-derived microsomes (Brodsky and Schekman, 1993 ) were prepared from HA-CFTR–expressing cells and treated with sodium carbonate (Fujiki et al., 1982 ). As presented in Figure Figure3A,3A, HA-CFTR and Sec61p, the ER translocation channel, were found exclusively in the membrane pellet, whereas BiP, an ER lumenal protein, is primarily (~66%) located in the supernatant. Incomplete extraction of BiP might have been observed from its interaction with components of the translocation machinery (Brodsky and Schekman, 1993 ).
To confirm that CFTR integrates into the membrane, cell extracts from CFTR-expressing cells were prepared and mixed with a dense sucrose solution, overlayed with sucrose-containing buffers of lower densities, and then subjected to ultracentrifugation. As shown in Figure Figure3B,3B, we observed that CFTR floated in the sucrose gradient coincident with BiP. We did note that some of the BiP failed to float, suggesting that a portion of it became liberated from the vesicles, most likely during the preparation of the cell extracts. These combined immunofluorescence and biochemical data indicate that CFTR is integrated into the ER membrane in yeast.
We next addressed whether wild-type CFTR is degraded via the ubiquitin-proteasome pathway in yeast, as in mammalian cells (Jensen et al., 1995 ; Ward et al., 1995 ). The degradation of CFTR was assayed in growing cells after the addition of cycloheximide. The amount of CFTR remaining at various time points was quantified by immunoblot analysis by using 125I-protein A. We found that HA-CFTR was stabilized in the pre1-1pre2-2 mutant (Figure (Figure4),4), a strain with mutations that abrogate ~95% of the activity of the proteasome (Heinemeyer et al., 1993 ). HA-CFTR was also stabilized in the ubc6,7 strain, which had been deleted for the ubiquitin conjugation enzymes Ubc6p and Ubc7p (Figure (Figure4).4). We note in this and other experiments (see below) that the extent of CFTR degradation in unique wild-type yeast strains differs, indicating the necessity of using isogenic strains to measure ERAD in vivo. Regardless, our results indicate that the proteasome and ubiquitin conjugation facilitate the degradation of CFTR in yeast.
The intracellular degradation of many proteins in yeast occurs in the vacuole, and misfolded secretory proteins can be targeted to this organelle (Hong et al., 1996 , and references therein). To confirm that the degradation of CFTR was independent of vacuolar proteases, we examined the fate of HA-CFTR in a Δpep4 mutant in which most vacuolar proteases are inactivated (Jones, 1991 ), but found that CFTR degradation was unaffected (Zhang et al., 2001 ).
The ER lumenal chaperones calnexin and BiP are required for the efficient degradation of soluble ERAD substrates (Le et al., 1994 ; Knittler et al., 1995 ; Schmitz et al., 1995 ; McCracken and Brodsky, 1996 ; Qu et al., 1996 ; Plemper et al., 1997 ; Brodsky et al., 1999 ), and both ER lumenal (e.g., calnexin) and cytosolic chaperones (e.g., Hsp70 and Hsp90) associate with CFTR in mammalian cells (Yang et al., 1993 ; Ping et al., 1994 ; Loo et al., 1998 ; Meacham et al., 1999 ). S. cerevisiae has two classes of cytosolic Hsp70s, encoded by the SSA and SSB genes (Boorstein et al., 1994 ). Although the Ssb proteins are involved in protein translation (Pfund et al., 1998 ), the Ssa proteins are more similar to mammalian Hsp70 and are required for a variety of chaperone-dependent activities (Miao et al., 1997 ). There are four Ssa proteins, Ssa1-4p, of which the expression of at least one is essential for viability (Werner-Washburne et al., 1987 ). Therefore, to explore whether Hsp70 is required for CFTR degradation, we used a strain containing an SSA1 temperature-sensitive allele, ssa1-45, and in which ssa2-4 had been inactivated (Becker et al., 1996 ). The isogenic “wild-type” strain contains SSA1 and similarly lacks functional ssa2-4. We found that CFTR degradation is robust at both 26 and 40°C in wild-type cells (Figure (Figure5A);5A); ~60% of CFTR is degraded after 90 min in wild-type cells, regardless of the temperature at which the experiment was performed. In contrast, in the ssa1-45 mutant strain, CFTR degradation is proficient at 26°C, but the protein is significantly stabilized at 40°C. When the degradation of untagged CFTR was examined using an antibody against the C terminus of the protein, CFTR was also stabilized in the ssa1-45 strain at 40°C (our unpublished results). We conclude that Hsp70 is required to facilitate CFTR degradation in yeast, although it is dispensable for the ERAD of two soluble proteins both in vivo and in vitro (Brodsky et al., 1999 ).
Because the Ydj1p cochaperone stimulates the ATPase activity of Ssa1p (Cyr et al., 1992 ) and YDJ1 and SSA1 interact genetically (Becker et al., 1996 ), Ydj1p may also facilitate CFTR degradation. In addition, the ydj1-151 mutant strain is defective for the ubiquitin-dependent degradation of some cytoplasmic substrates (Lee et al., 1996 ). We found, however, that CFTR was degraded efficiently in ydj1-151 yeast (our unpublished results).
To determine whether BiP and calnexin play a role in the degradation of CFTR, we examined CFTR turnover in kar2-1, kar2-133, and Δcne1 strains in which the degradation of the soluble ERAD substrate protein proalpha factor was debilitated in vitro (McCracken and Brodsky, 1996 ; Brodsky et al., 1999 ). Strains containing the kar2-1 or kar2-133 mutations are also defective for the degradation of the soluble substrate A1PiZ in vivo (Brodsky et al., 1999 ). As shown in Figure Figure5,5, B and C, we observed that the rate of CFTR proteolysis was identical in the kar2 and cne1 mutant strains and their corresponding isogenic wild-type strains. Together, these results indicate that different sets of chaperones are required to degrade several membrane and soluble proteins, and suggest that the respective ERAD pathways are distinct.
To confirm that the degradation of soluble ERAD substrates is compromised in the kar2-1 and kar2-133 strains, we examined the fate of CPY* in the wild-type and kar2 mutants by chase analysis after the addition of cycloheximide to cells. The ERAD of CPY* was established by Wolf and colleagues (Hiller et al., 1996 ), and using a distinct kar2 mutant, Plemper et al. (1997) found that the degradation and retro-translocation of CPY* from the ER required BiP. When we measured the proteolysis of CPY* in wild-type and the kar2 mutants used in this study, stabilization of CPY* was observed in the kar2 strains (Figure (Figure6).6). Even after a 45-min chase at a semipermissive temperature of 30°C, the amounts of CPY* remaining as a percentage of the initial levels were 12, 49, and 84% in the wild-type strain and the kar2-1 and kar2-133 mutants, respectively.
Hampton and colleagues (1996) have isolated a number of mutants that are defective for the regulated degradation of HMG-CoA reductase in yeast. Two of the proteins encoded by the corresponding genes, Hrd1/Der3p and Hrd3p, form a stoichiometic complex and cooperate during ERAD (Gardner et al., 2000 ). Although the hrd1 and hrd3 mutants display a mild defect in the degradation of two integral membrane ERAD substrates (Wilhovsky et al., 2000 ), other ERAD substrates are significantly stabilized in the hrd1/der3 mutant (Bordallo et al., 1998 ; Plemper et al., 1998 ). To determine whether CFTR degradation is affected by the hrd1 and hrd3 mutations, we introduced the CFTR expression plasmid into these mutants and an isogenic wild-type strain and measured the levels of CFTR over time, as described above. As shown in Figure Figure7,7, we observed no significant difference in the overall rate of CFTR degradation in wild-type, hrd1, and hrd3 yeast. Thus, CFTR, like UP* (Wilhovsky et al., 2000 ) is UBC6/7-dependent but HRD1-independent.
To gain insight into how CFTR is selected and targeted to the proteasome, we examined CFTR localization under conditions in which CFTR degradation is impeded. Indirect immunofluorescence microscopy was performed with the pre1-1pre2-2 and ssa1-45 mutant strains and isogenic wild-type cells expressing CFTR. As presented above, in wild-type cells, CFTR appears in a strong punctate pattern and colocalizes with BiP (Figure (Figure1A).1A). In ssa1-45 cells, at both 26 and 40°C (Figure (Figure1A),1A), similar results were observed. However, when proteasome function is attenuated and the cells are shifted to 40°C, CFTR can concentrate to one or two “dots” in some cells; a representative section of the cells is shown in Figure Figure1B.1B. We do not believe that these spots represent “aggresomes, ” extracted CFTR that accumulates in a perinuclear site in mammalian cells when CFTR is overexpressed or cells are grown in the presence of proteasome inhibitors (Johnston et al., 1998 ), for the following reasons. First, CFTR and the lumenal chaperone BiP colocalize in these images. Second, CFTR floats in sucrose gradients regardless of whether it derived from wild-type or the proteasome-mutant strain (Figure (Figure3B)3B) in which degradation is attenuated (Figure (Figure4A).4A). The fact that not all pre cells expressing CFTR display one or two dots (Figure (Figure1B)1B) may represent heterogenous levels of CFTR expression. In fact, we have noted that this pattern arises in cells expressing the greatest amount of CFTR (our unpublished results), suggesting that lower levels of CFTR may be “handled” by residual ERAD activity or other proteases. Nevertheless, these results indicate that the residency of CFTR is differentially affected when its proteolysis is abrogated by defects in Hsp70 or the proteasome.
The work described here establishes CFTR as an ERAD substrate in the yeast S. cerevisiae. We found that CFTR expressed in yeast is an integral membrane protein retained in the ER, and that proteasome activity and ubiquitin conjugation systems are necessary for maximal degradation. Furthermore, we found that the cytoplasmic Hsp70 Ssa1p, but neither the lumenal Hsp70 BiP nor lumenal chaperone calnexin is required for CFTR proteolysis. This is opposite to the chaperone requirements for the degradation of soluble ERAD substrates: the ER lumenal proteins calnexin and BiP are required, but the cytoplasmic chaperone Ssa1p is not (McCracken and Brodsky, 1996 ; Plemper et al., 1997 ; Brodsky et al., 1999 ; Figure Figure6).6). Together, these results suggest unique mechanisms for the quality control of at least some integral membrane and soluble ER proteins.
Our results are consistent with work from other laboratories investigating the chaperone requirements for the degradation of integral membrane proteins in the yeast ER. First, Plemper et al. (1998) observed that BiP function is dispensable for the degradation of Pdr5p*, an integral membrane ERAD substrate. Second, while this manuscript was in preparation, Hill and Cooper (2000) reported that Ssa1p was necessary for the degradation of Vph1p, an integral membrane subunit of the vacuolar ATPase that becomes an ERAD substrate when the VMA22 gene product is absent, whereas mutations in KAR2 had no effect on the degradation of Vph1p. Third, we found that Ssa1p is required, and BiP is dispensable for the destruction of an ER-retained, mutated form of Ste6p (“Ste6p*”), the integral membrane a-factor transporter in yeast that is homologous to CFTR (Harper, Brodsky, and Michaelis, unpublished data).
Because cytosol prepared from the ssa1-45 mutant strain shifted to the nonpermissive temperature supports the degradation of soluble ERAD substrates (Brodsky et al., 1999 ), the defect in CFTR degradation that we observed in the ssa1-45 mutant cannot arise from proteasome inactivation. Instead, we favor a model in which Ssa1p retains cytoplasmic domains of ER membrane proteins in a protease-accessible conformation, an activity that would be essential if the proteasome or other protease initially shaves or clips this domain. We previously found that approximately equal amounts of Ssa1p are associated with ER-derived microsomes and “free” in yeast cytosol (Brodsky et al., 1999 ), suggesting that much of this chaperone may be positioned to associate with ERAD substrates. Several other observations support this model. First, each of the integral membrane ERAD substrates (i.e., Vph1p, CFTR, Ste6p*) that require Ssa1p activity for degradation contain large, cytoplasmic domains. If this model is correct, one might expect that Ssa1p prevents the formation of protein aggregates and promotes protein folding. In fact, we found previously that cytoplasm prepared from the ssa1-45 mutant is unable to refold heat-denatured firefly luciferase in vitro, whereas cytosol prepared from the isogenic wild-type strain refolds luciferase (Brodsky et al., 1999 ). Second, others have observed that mammalian Hsp70 suppresses the aggregation of the NBD1 of CFTR in vitro (Strickland et al., 1997 ; Meacham et al., 1999 ). Third, if Ssa1p “holds” CFTR in a conformation that is accessible to the proteasome or unidentified protease, then cleavage may initiate within the large, cytosolically disposed NBD and/or R domains. Indeed, upon overexposure of gels in which degradation was assayed, we noted CFTR degradation intermediates of molecular weights ~80–120,000 (our unpublished results), a size consistent with cleavage within the NBD and R domains. CFTR degradation intermediates in this molecular weight range have also been observed when CFTR biogenesis was examined in mammalian cells (Lukacs et al., 1994 ; van Oene et al., 2000 ). And fourth, Ssa1p may not be essential for the degradation of integral membrane ERAD substrates presenting less prominent cytoplasmic domains. Indeed, only a minor effect on the proteolysis of Sec61-2p was observed when its stability was assessed in the ssa1-45 strain (S. Nishikawa, S. Fewell, Y. Kato, J. Brodsky, T. Endo, unpublished data).
To explain the requirement for BiP in the degradation of soluble but not integral membrane ERAD substrates, we suggest that lumenal domains must be preserved in an aggregation-free state, an activity that BiP is known to exhibit (reviewed by Gething, 1997 ). In contrast, the lumenal and transmembrane domains of integral membrane proteins may be removed independent of BiP, perhaps through direct extraction by the proteasome (Mayer et al., 1998 ; Xiong et al., 1999 ). Finally, it is possible that BiP may be required to “unlock” the translocation channel to permit ERAD substrates to retro-translocate after their complete import into the ER (Plemper et al., 1999 ). This model arises from the demonstration by Johnson and coworkers that BiP might gate the translocation pore in the mammalian ER (Hamman et al., 1998 ).
Inherent in these models, and because of the different CFTR immunofluorescence staining patterns observed in the ssa1-45 and pre1-1pre2-2 strains (Figure (Figure1),1), we suggest that CFTR degradation in yeast is a multistep process. Only when proteasome function was attenuated at 40°C did we observed CFTR localization to one or two sites in the ER. Similar structures have been detected in yeast expressing Ste6p* and are termed ER-associated bodies (ERABs). As shown here for CFTR, Ste6p*-induced ERAB formation does not involve induction of the UPR, and a detailed immunofluorescence and electron microscopic analysis of the morphology of ERABs has been undertaken and will be described elsewhere (Kuehn, Nijbroek, and Michaelis, unpublished data). Unique localization patterns in the ssa1 and pre mutants may arise if Ssa1p acts upstream of the proteasome in the CFTR degradation pathway: The spots we observe in the proteasome mutant strain could represent the final staging points before CFTR proteolysis, whereas CFTR delivery to these sites is halted when Hsp70 is inactivated. Because the CFTR that concentrates to these spots is membrane-associated, as determined by the floatation analysis (Figure (Figure3B),3B), our results also suggest that the catalytic activity of the proteasome is required to degrade or extract CFTR at or from the yeast ER membrane. Such a role for the proteasome was also suggested by Mayer et al. (1998) when the degradation of a hybrid ER membrane protein was examined in strains lacking a functional proteasome.
Because CFTR remains membrane-associated when proteolysis is compromised in yeast, we believe that CFTR does not reside in aggresomes. In contrast, aggresomes form in CFTR-expressing mammalian cells when overexpressed or when proteasome activity is blocked (Johnston et al., 1998 ; Wigley et al., 1999 ). Thus, there may be unique mechanisms to handle accumulated and undegraded CFTR in yeast and mammals. In fact, it has been suggested that aggresome formation may be cell-type specific, because aggresomes have not been observed in every CFTR-expressing mammalian cell line when proteasome function is attenuated (Chen et al., 2000 ). Finally, it is possible that aggresome formation requires high levels of CFTR than cannot be produced in yeast.
Although ~20% of wild-type CFTR in mammalian cells escapes degradation and transits to the plasma membrane, CFTR in yeast appears to reside primarily in the ER. However, we cannot exclude the possibility that a fraction of CFTR in yeast escapes ERAD and migrates through the secretory pathway. Relevant to this hypothesis, we often find that a fraction of CFTR resists degradation (see for example, Figure Figure5C),5C), and that a minor fraction of CFTR (~10%) comigrates with Pma1p, a plasma membrane protein, upon velocity sucrose gradient analysis (our unpublished results).
Finally, the data presented in this article may be pertinent to the study of membrane protein degradation in the mammalian ER. In agreement with our results, Fisher et al. (1997) reported that the degradation of ApoB100 in HepG2 cells is enhanced approximately twofold when cytoplasmic Hsp70 is overexpressed, suggesting that the chaperone facilitates ERAD. A role for both the cytoplasmic Hsp70 and Hsc70 molecular chaperones during CFTR biogenesis in mammalian cells has also been uncovered by the use of modulators of chaperone activity. First, Rubenstein and Zeitlin (2000) showed that sodium phenylbutyrate decreases the amount of Hsc70-ΔF508 CFTR complexes in mammalian cells and that there is a concomitant rescue of the ΔF508 mutant phenotype; however, Hsp70 levels increase under these conditions (P. Zeitlin, personal communication). Second, Jiang et al. (1998) discovered that ΔF508 CFTR-expressing cells treated with the Hsp70/Hsc70-interacting drug deoxyspergualin exhibited a partial restoration in cAMP-stimulated chloride channel activity. These investigators hypothesized that altering the interaction of the chaperone with CFTR may promote maturation, although it is possible that DSG stimulates Hsp70, thus enhancing CFTR folding and increasing the yield of “active” ΔF508-CFTR. Interpreting the conclusions from these diverse studies is further complicated by the fact that mammalian Hsp70 has been shown to facilitate ubiquitin conjugation onto several proteins substrates in vitro (Bercovich et al., 1997 ). Clearly, further work will be directed to better understand the roles of Hsp70/Hsc70 in eucaryotic protein turnover.
We thank Drs. Raymond Frizzell and Robert Bridges for reagents, their invaluable support, and for many helpful discussions, and Dr. Gergely Lukacs for critical reading of the manuscript. We also thank Drs. Elizabeth Craig, Dieter Wolf, Mark Hochstrasser, Caroline Slayman, Elizabeth Jones, Peter Walter, Randy Hampton, and Davis Ng for generously supplying reagents, and Tom Harper for help with imaging technology. This work was supported by a Pilot-Feasibility grant from the Cystic Fibrosis Foundation's Research Development Program at the University of Pittsburgh (to J.L.B.), by Grant MCB-9722889 from the National Science Foundation (to J.L.B. and A.A.M.), and by Grants GM-51508 and DK-58029 from the National Institutes of Health (to S.M.).