|Home | About | Journals | Submit | Contact Us | Français|
Capture of newly synthesized proteins into endoplasmic reticulum (ER)-derived coat protomer type II (COPII) vesicles represents a critical juncture in the quality control of protein biogenesis within the secretory pathway. The yeast ATP-binding cassette transporter Yor1p is a pleiotropic drug pump that shows homology to the human cystic fibrosis transmembrane conductance regulator (CFTR). Deletion of a phenylalanine residue in Yor1p, equivalent to the major disease-causing mutation in CFTR, causes ER retention and degradation via ER-associated degradation. We have examined the relationship between protein folding, ERAD and forward transport during Yor1p biogenesis. Uptake of Yor1p into COPII vesicles is mediated by an N-terminal diacidic signal that likely interacts with the “B-site” cargo-recognition domain on the COPII subunit, Sec24p. Yor1p-ΔF is subjected to complex ER quality control involving multiple cytoplasmic chaperones and degradative pathways. Stabilization of Yor1p-ΔF by inhibiting its degradation does not permit access of Yor1p-ΔF to COPII vesicles. We propose that the ER quality control checkpoint engages misfolded Yor1p even after it has been stabilized by inhibition of the degradative pathway.
Eukaryotic cells maintain a strict policy of protein quality control to prevent the improper deployment of aberrant proteins within the cell, which may impair cellular function and ultimately result in proteotoxicity. This quality control checkpoint is of particular importance with respect to secretory proteins, which enter the endoplasmic reticulum (ER) as linear, unfolded polypeptides and generally must attain the correct quaternary structure before gaining transport competence to be delivered to downstream compartments (Ellgaard and Helenius, 2001 ; Bukau et al., 2006 ). Misfolded and aberrant proteins are disposed of by ER-associated degradation (ERAD), which targets proteins for reverse translocation out of the ER, ubiquitination, and destruction by the cytoplasmic proteasome (Ahner and Brodsky, 2004 ). Thus, the ER plays a fundamental role in coordinating protein synthesis, folding, degradation and transport to ensure accurate and efficient delivery of proteins to the compartments of the secretory pathway.
With recent progress in defining many of the components that participate in these processes, the complexity of these interrelated pathways is beginning to be understood. For example, ERAD is now recognized to be a multifaceted program that uses multiple machineries to degrade misfolded proteins with differently located lesions (Ahner and Brodsky, 2004 ; Ismail and Ng, 2006 ). Substrates within the ER lumen or ER membrane each have distinct topological requirements in terms of both lesion detection and protein extraction before the common endpoint of proteasomal degradation. Thus, rather than being a simple protein detection and degradation service, ERAD is attuned to dispose of proteins from different environments. Despite the advanced state of knowledge with respect to the destructive side of protein quality control, the precise nature of the interplay between the biosynthetic and degradative pathways remains poorly understood. Specifically, the mechanisms by which a single protein is diverted according to its particular folding status are largely unknown.
One critical juncture in the process of protein quality control is the uptake of newly synthesized proteins into ER-derived transport vesicles, known as coat protomer type II (COPII) vesicles for the cytoplasmic coat proteins that sculpt vesicles from the ER membrane and populate them with the appropriate cargo (Lee et al., 2004 ). Membrane proteins are generally captured into COPII vesicles via cytoplasmic ER export signals that bind directly to the COPII subunit Sec24p, which is responsible for efficiently recruiting cargo into nascent vesicles as they form from the ER membrane (Miller et al., 2002 ). Similarly, soluble secretory proteins bind to receptors that in turn likely contain Sec24p-binding sites (Otte and Barlowe, 2004 ). Misfolded proteins seem to be largely excluded from COPII vesicles, but the mechanism by which these proteins fail to enter vesicles is not clear.
Recent experiments using chimeric ERAD substrates demonstrated that degradation of a misfolded protein can be overcome when a strong ER export motif is appended to the protein (Kincaid and Cooper, 2007 ). This suggests that rather than using an active ER retention mechanism, quality control may result from a competitive interaction between ERAD and forward transport. Misfolded proteins may be rapidly destroyed by the ERAD machinery, leaving little opportunity to engage the COPII coat. Thus, disabling the degradative pathway might be expected to stabilize a misfolded protein sufficiently to drive forward transport. Indeed, inhibiting ERAD can allow rescue of forward transport, and, in some cases, the deployment of a functional protein (Kota et al., 2007 ). One mechanism that may play a role in establishing a hierarchy of degradation over forward transport is the presentation of an ER export motif; misfolded proteins may simply fail to present an appropriate signal, resulting in their default degradation. This may be especially relevant to soluble secretory proteins, which contain ER export motifs that mediate interaction with their transmembrane receptors (Otte and Barlowe, 2004 ). Simple failure to engage the receptor would ultimately cause degradation. However, misfolded membrane proteins, which interact directly with Sec24p, present a more difficult quality control problem (Mossessova et al., 2003 ). Where the misfolding lesion is located in a cytosolic domain, it is easy to imagine how structural changes may obscure an ER export motif. This is thought to be true of the human cystic fibrosis transmembrane receptor (CFTR); the most common disease-causing mutation, ΔF508, is located in the same cytosolic domain as the ER export signal, potentially disrupting interaction with Sec24p (Wang et al., 2004 ). Conversely, where the misfolding lesion is in a lumenal or transmembrane region, or in a cytosolic domain distinct from that containing the ER export signal, interaction with Sec24p may be preserved. How these proteins are excluded from COPII vesicles remains unclear, but lumenal or cytosolic chaperones that might bind the misfolded domain may play a role by actively retaining the protein or obscuring an interaction with Sec24p. Finally, the discovery that some misfolded proteins require a round-trip to the Golgi before ERAD suggests that misfolded proteins may not be entirely absent from COPII vesicles (Vashist et al., 2001 ). As for ERAD, the mechanism that excludes a protein from COPII vesicles may be unique for each specific aberrant protein.
We study the close relationship between protein folding and forward transport through the secretory pathway in the budding yeast, Saccharomyces cerevisiae. We describe here the biogenesis of a plasma membrane protein, Yor1p, which is an ATP-binding cassette (ABC) transporter that functions as a pleiotropic drug pump to clear toxic substances from the yeast cytoplasm (Katzmann et al., 1999 ). ABC transporters are a large family of proteins that contain distinct arrangements of membrane spanning domains (MSDs), which form a channel in the lipid bilayer, and nucleotide-binding domains (NBDs) that provide the driving force for substrate translocation across the membrane. The arrangement of MSDs and NBDs in Yor1p places it in the same class of ABC transporters as human CFTR (Riordan et al., 1989 ). Deletion of a phenylalanine residue in NBD1 of Yor1p (F670), equivalent to the ΔF508 mutation in CFTR, causes ER retention and proteasomal degradation of Yor1p (Katzmann et al., 1999 ). In CFTR, this mutation results in a channel that is capable of conducting chloride (Drumm et al., 1991 ; Li et al., 1993 ), but it is recognized by the ER quality control machinery and retained in the ER where it is degraded via ERAD instead of being deployed to the apical membrane (Cheng et al., 1990 ; Denning et al., 1992 ; Ward et al., 1995 ). Extensive work elucidating the mechanisms of destruction of aberrant CFTR have revealed that it is a complex process involving multiple cytoplasmic chaperone complexes (Zhang et al., 2001 ; Youker et al., 2004 ; Younger et al., 2004 , 2006 ), but the relationship between the degradative and forward transport itineraries remains unclear.
We aimed to better define the protein folding, degradation, and forward transport pathways accessed by both the wild-type and ΔF forms of Yor1p to gain insight into how the intracellular fate of aberrant ABC transporters is controlled. Specifically, we have defined the ER export motif used by Yor1p in gaining access to ER-derived COPII vesicles and determined the site on Sec24p that likely recognizes this signal. Furthermore, we have used yeast mutants in several chaperones and the ERAD machinery to test how perturbation of these key processes directly impacts the intracellular itineraries of newly synthesized Yor1p. Because many components of the ER folding, degradation and export pathways are remarkably conserved among all eukaryotes, increasing our basic understanding of the regulation of these processes in yeast is likely to be directly applicable to protein folding problems in mammalian systems.
S. cerevisiae strains used in this study are listed in Table 1. Strains bearing mutant forms of SEC24 as the sole copy of this essential gene were created by transforming YTB1 (sec24Δ::LEU2 with wild-type SEC24 borne on a URA-containing plasmid) with either wild-type or mutant forms of SEC24 contained on a HIS-marked plasmid. Cells were cured of the wild-type SEC24::URA plasmid by growth on 5-fluoroorotic acid (0.1% final concentration), leaving the plasmid-borne copy as the sole copy of SEC24. UBC7 was deleted in various strain backgrounds by transformation with a polymerase chain reaction (PCR) product composed of the ubc7Δ::KANMX disruption cassette amplified from the EUROSCARF (Frankfurt, Germany) deletion strain collection. The allelic replacement of UBC7 was confirmed by PCR. LMY314 (sel1Δ::KanMX), LMY285 (hrd1Δ::KanMX), and LMY115 (doa10Δ:: KanMX) were constructed by integration of the respective disruption cassettes into HLJ1/YDJ1. Disruption cassettes were previously generated by PCR amplification from the pUG6 plasmid (Guldener et al., 1996 ). Retrieval of the KanMX gene in LMY285 (hdr1Δ::KanMX) was performed by cre/lox-mediated excision (Guldener et al., 1996 ), and the doa10Δ::KanMX disruption cassette was introduced to create LMY313 (hrd1Δ doa10::KanMX). All allelic replacements were confirmed by PCR. Cultures were grown at 30°C in standard rich media (YPD: 1% yeast extract, 2% peptone, and 2% glucose) or synthetic complete media (SC: 0.67% yeast nitrogen base and 2% glucose, supplemented with amino acids appropriate for auxotrophic growth). For testing sensitivity to oligomycin, strains were grown to saturation, and then they were diluted to an OD600 of 0.5 and fourfold serial dilutions were applied to YPEG plates (1% yeast extract, 2% peptone, 3% ethanol, and 3% glycerol), that were supplemented with oligomycin (Sigma-Aldrich, St. Louis, MO) from a 1 mg/ml stock in ethanol. Drug sensitivities of cells expressing mutant forms of Sec24p were assayed by streaking the strains indicated onto YPEG plates supplemented with 0.1 μg/ml oligomycin, YPD plates supplemented with 400 μg/ml rhodamine B (Sigma-Aldrich) or onto YPD plates supplemented with 10 μg/ml methyl 1-(butylcarbamoyl)-2-benzimidazolecarbamate (benomyl, 95%; Sigma-Aldrich).
Plasmids used in this study are listed in Table 2. pEAE83 bearing YOR1-HA in pRS316 was a gift from Scott Moye-Rowley (University of Iowa). This plasmid was the basis for site-directed mutagenesis by using QuikChange mutagenesis (Stratagene, La Jolla, CA) to obtain various hemagglutinin (HA)-tagged Yor1p mutants (Table 2). pRL001 represents a marker-swapped version of pEAE83, where the URA3 marker was replaced with TRP1 (Cross, 1997 ). This plasmid was used as the template for site-directed mutagenesis to introduce unpaired cysteine substitutions to test whether cross-links form from intra- or intermolecular associations. pEAE93 was a gift from Scott Moye-Rowley and contains an in-frame fusion of green fluorescent protein (GFP) to the C terminus of Yor1p; this plasmid was the basis for site-directed mutagenesis to introduce N- and C-terminal diacidic mutations. pRS315-PDR1-3 bearing a dominant active allele of the transcription factor Pdr3p was a gift from Scott Moye-Rowley, and it was used to overexpress Yor1p–GFP fusions (Epping and Moye-Rowley, 2002 ).
The subcellular localization of Yor1p and various mutants was visualized by fluorescence microscopy of Yor1p–GFP fusion proteins. Strains expressing the appropriate gene fusion were grown in selective medium to mid-log phase and examined with an IX 81 inverted microscope (Olympus, Melville, NY) equipped with a 60× numerical aperture 1.4 PlanApo optics and a Cooke Sensicam QE air-cooled charge-coupled device camera. Images were collected with IPLab 4.0 and analyzed using Adobe Photoshop (Adobe Systems, Mountain View, CA). To visualize Yor1p and the N- and C-terminal diacidic mutants, HLJ1/YDJ1 cells were cotransformed with Yor1p–GFP plasmids and pRS315-PDR1-3 containing the dominant-active transcription factor pdr1–3, which causes up-regulation of several ABC transporters (Carvajal et al., 1997 ).
COPII proteins Sar1p, Sec23p/24p, and Sec13/31p were prepared as described previously (Barlowe et al., 1994 ).
Microsomal membranes were purified from RSY620 cells expressing Yor1p- HA or Yor1pΔF-HA as described previously (Wuestehube and Schekman, 1992 ). In vitro vesicle budding was performed essentially as described previously (Miller et al., 2002 ), except the urea wash was omitted. Briefly, membranes were washed twice with buffer B88 (20 mM HEPES, pH 6.8, 250 mM sorbitol, 160 mM potassium acetate, and 5 mM magnesium acetate), and 125 μg of membranes per reaction were incubated with COPII proteins (10 μg/ml Sar1p, 10 μg/ml Sec23p/24p, and 20 μg/ml Sec13/31p) either in the presence of 0.1 mM GTP with a 10× ATP regeneration system or 0.1 mM GDP. Vesicles were separated from donor membranes by centrifugation at 16,000 rpm for 5 min, and the vesicle fraction was further concentrated by high-speed centrifugation at 55,000 rpm for 20 min. Vesicle pellets were resuspended in SDS sample buffer and heated at 55°C for 5 min before separation by SDS-polyacrylamide gel electrophoresis (PAGE). Proteins were transferred to polyvinylidene difluoride (PVDF) membranes and analyzed by immunoblotting. HA-tagged Yor1p was detected with a monoclonal anti-HA antibody (Covance, Princeton, NJ), and control proteins Erv46p and Sec22p were detected with polyclonal antibodies, gifts from C. Barlowe (Dartmouth Medical School) and R. Schekman (U. C. Berkeley), respectively.
Radiolabeled semi-intact cells were prepared essentially as described previously (Kuehn et al., 1996 ). Briefly, cells were grown to mid-log phase in synthetic complete medium, and a total of 5.0 OD600 of cells were harvested and starved of methionine/cysteine for 10 min before addition of Express protein labeling mix (~70μCi/OD600 of cells; MP Biomedicals, Irvine, CA). Cells were labeled for 15 min at 30°C, and then they were metabolically killed and converted to spheroplasts. Cells were gently lysed, washed once with low acetate B88 (20 mM HEPES, pH 6.8, 250 mM sorbitol, 50 mM potassium acetate, and 5 mM magnesium acetate) and twice with B88 before incubation with COPII proteins (10 μg/ml Sar1p, 10 μg/ml Sec23p/24p, and 20 μg/ml Sec13/31p) in a final reaction that contained 2.5 OD of cells either in the presence of 0.1 mM GTP with a 10× ATP regeneration system or 0.1 mM GDP. Vesicles were separated from donor membranes by centrifugation at 16,000 rpm for 5 min, solubilized with 1% SDS (final concentration), and diluted with immunoprecipitation (IP) buffer (50 mM Tris, pH 7.5, 160 mM NaCl, 1% Triton X-100, and 2 mM NaN3). Proteins were immunoprecipitated using monoclonal anti-HA antibodies (Covance) precoupled to protein G-Sepharose beads (GE Healthcare, Little Chalfont, Buckinghamshire, United Kingdom), or polyclonal antibodies against Gas1p, Sec22p or Vph1p (gifts from R. Schekman) coupled to protein A-Sepharose beads (GE Healthcare). Immune complexes were separated by SDS-PAGE and analyzed by PhosphorImage analysis using a Storm PhosphorImager (GE Healthcare). Proteins were quantified using ImageQuant software (GE Healthcare).
In total, 5.0 OD600 of cells expressing wild-type HA-tagged Yor1p or HA-tagged mutant Yor1p were collected during mid-log phase, washed with 250 μl of lysis buffer (25 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, and 1 mM dithiothreitol [DTT]), and lysed by vortexing with glass beads for 10 min at 4°C. The lysate was transferred to a fresh tube, membranes collected by centrifugation at 15,000 rpm for 5 min and proteins solubilized with 100μl of lysis buffer with 1% digitonin. After incubation on ice for 30 min, the soluble fraction was separated from the insoluble material by centrifugation at 5000 rpm for 2 min, and the supernatant was mixed with 10× BNP sample buffer (5% Coomassie G250, 10% 6-aminocaprioic acid, 50 mM Bis-Tris, and 10% glycerol). Samples were run on a 7% acrylamide Bis-Tris gel (50 mM Bis-Tris, 6.55% 6-aminocaproic acid, and 13.9% glycerol) by using a two-buffer system consisting of a cathode buffer (50 mM Tricine, 15 mM Bis-Tris, and 0.02% Coomassie G250) and anode buffer (50 mM Bis-Tris). Proteins were transferred to PVDF overnight at 20 V and analyzed by immunoblotting with a monoclonal anti-HA antibody (Covance).
Cells expressing either wild-type Yor1p-HA or HA-tagged Yor1p mutants were harvested during mid-log phase. In total, 10.0 OD600 of cells were collected, resuspended in 200 μl of B88 buffer, and glass bead lysed at 4°C for 5 min. Cell lysates were subjected to centrifugation at 16,000 rpm for 5 min, and the membrane pellet was resuspended in 100 μl of B88 and divided into four 25-μl reactions (2.5 OD600/reaction). Each reaction was treated with a final concentration of 0, 25, 50, or 100 ng/μl trypsin (Sigma-Aldrich) for 10 min on ice. Digestion was terminated by addition of 0.2 μg/ml (final concentration) soybean trypsin inhibitor (Sigma-Aldrich) to all reactions and incubated on ice for 15 min. Proteins were separated by SDS-PAGE, transferred to PVDF, and the pattern of Yor1p fragments analyzed by immmunoblot by using an anti-HA antibody.
For cells bearing thermosensitive alleles, radiolabeled semi-intact cells were used as the source of membranes. Cells were grown to mid-log phase in complete synthetic medium at 30°C, and then they were harvested, resuspended in fresh medium lacking methionine/cysteine to an OD600 of 5, and incubated for 10 min with gentle shaking at 37°C. Cells were metabolically labeled at 37°C for 10 min by adding 60 μCi of Express protein labeling mix (MP Biomedicals) per OD600 unit of cells. Spheroplasts prepared from these labeled cells were resuspended in 100 μl of B88 and divided into four 25-μl reactions. Each reaction was treated with a final concentration of 0, 100, 200, or 400 ng/μl trypsin for 10 min on ice. Digestion was terminated by addition of soybean trypsin inhibitor to all reactions followed by incubation on ice for 15 min and by two washes with B88. After solubilization with SDS (1% final concentration) and heating to 55°C for 5 min, the resulting protein extracts were diluted with 5 volumes of IP buffer (50 mM Tris, pH 7.5, 160 mM NaCl, 1% Triton X-100, and 2 mM NaN3) and cleared by centrifugation. Yor1p fragments were immunoprecipitated from the cleared supernatant and analyzed by SDS-PAGE and PhosphorImage analysis as described above.
Cells expressing cysteine-substituted forms of Yor1p-HA were grown to mid-log phase, and then they were harvested and converted to spheroplasts. Spheroplasts were washed twice in 20 mM HEPES, pH 7.4, and incubated with increasing concentrations of M8M (prepared as a 100× stock in dimethyl sulfoxide) as indicated. Cells were cross-linked for 15 min at room temperature, and then they were harvested, resuspended in 100 μl of 1% SDS, 50 μl of 3× SDS sample buffer without β-mercaptoethanol or DTT. Cells were disrupted by glass bead lysis (15 min; 4°C), heated to 55°C for 5 min, and proteins were separated by SDS-PAGE followed by immunoblot analysis using anti-HA antibodies.
Cells were grown to mid-log phase in complete synthetic medium, and then they were harvested, resuspended in fresh medium lacking methionine/cysteine, and incubated for 15 min with gentle shaking at either 30°C (for cells lacking a temperature-sensitive allele) or at 37°C (for cells bearing a temperature-sensitive allele). Cells were metabolically labeled for 5 min by adding 30 μCi of Express protein labeling mix (MP Biomedicals) per OD600 unit of cells. A 10× chase solution (10 mM l-cysteine, 50 mM l-methionine, 4% yeast extract, and 2% glucose) was added and 2 OD aliquots of cells harvested at different times. At each time point, cells were transferred to chilled tubes, and sodium azide was added to a final concentration of 20 mM. Cells were washed once with 20 mM sodium azide and resuspended in 100 μl of 1% SDS. Glass beads were added, and cells were lysed by vortexing for 15 min at 4°C. Cell lysates were heated at 55°C for 5 min, diluted with 5 volumes of IP buffer (50 mM Tris, pH 7.5, 160 mM NaCl, 1% Triton X-100, and 2 mM NaN3), and cleared by centrifugation. Yor1p and Gas1p were immunoprecipitated from the cleared lysate and analyzed by SDS-PAGE and PhosphorImage analysis as described above.
The yeast ABC transporter Yor1p is a polytopic membrane protein that functions at the plasma membrane as a pleiotropic drug pump to clear toxins from the cytosol. A mutant form of Yor1p, Yor1p-ΔF, which carries a deletion of a phenylalanine residue at position 670, is trapped in the ER, destabilized in vivo, and degraded by the cytoplasmic ubiquitin/proteasome system (Katzmann et al., 1999 ). To independently confirm that ER-export of Yor1p-ΔF is impaired, we used an in vitro vesicle budding assay that recapitulates COPII vesicle formation from purified ER membranes. Microsomal membranes were isolated from cells expressing an HA-tagged copy of either wild-type Yor1p or Yor1p-ΔF. These purified membranes were incubated with the COPII components Sar1p, Sec23p/Sec24p, and Sec13p/Sec31p in the presence of either GTP or GDP. This incubation is sufficient to generate transport vesicles, which can be separated from donor membranes by differential centrifugation and specific capture of cargo proteins into vesicles analyzed by immunoblotting (Barlowe et al., 1994 ). Whereas wild-type Yor1p was efficiently packaged into COPII vesicles, very little Yor1p-ΔF was detected in the vesicle fraction (Figure 1B). Two proteins that cycle between the ER and Golgi, Sec22p and Erv46p served as positive controls to demonstrate efficient vesicle biogenesis in this assay. The specific absence of Yor1p-ΔF in COPII vesicles generated in vitro is consistent with its in vivo ER localization and suggests that the mutant protein has engaged the ER quality control checkpoint that prevents deployment of aberrant cellular proteins.
We next developed assays to probe the folding, assembly state, or both of Yor1p. We first examined the migration of Yor1p by using BNGE, which allows the preservation of native structures by solubilizing proteins with mild detergents (Schagger and von Jagow, 1991 ). Solubilized proteins are coated with Coomassie Blue, which imparts a net negative charge, allowing proteins to migrate uniformly in one-dimensional nondenaturing gel electrophoresis. Microsomal membranes containing HA-tagged forms of Yor1p were solubilized with digitonin, and protein complexes analyzed by BNGE followed by immunoblotting. Wild-type Yor1p was detected as a single band of ~350 kDa, whereas Yor1p-ΔF formed a high-molecular-weight smear of ~670 kDa (Figure 1C). An additional sample of Yor1p was solubilized with SDS and reducing agents, which also yielded a single band of ~350 kDa, suggesting that this species corresponds to monomeric Yor1p that, under these electrophoretic conditions, migrates as a larger protein than its expected ~150 kDa. Similar aberrant protein migration profiles are characteristic of other yeast multispanning membrane proteins, Gap1p (Kota and Ljungdahl, 2005 ) and Pma1p (Lee et al., 2002 ). The precise nature of the high-molecular weight smear that corresponds to Yor1p-ΔF is not clear; the mutant protein may form large aggregates or may be in complex with other proteins, including cytoplasmic or ER chaperones.
As a separate assay to probe the folding and assembly of Yor1p, we subjected membranes containing either wild-type or mutant Yor1p to limited proteolysis. Microsomal membranes were treated with increasing concentrations of trypsin and analyzed by immunoblotting. Yor1p presented a persistent ~160-kDa band and a single major cleavage product of ~60 kDa with increasing amounts of trypsin (Figure 1D, left). The pattern of Yor1p-ΔF cleavage products was markedly different, with both the 160- and 60-kDa bands further digested into smaller products with increasing trypsin (Figure 1D, right). The increased susceptibility of Yor1p-ΔF to trypsinolysis supports the hypothesis that this mutant protein is improperly assembled, thereby exposing additional trypsin cleavage sites that are obscured in the correctly folded protein.
Finally, we developed a cross-linking assay to probe the assembly of the transmembrane domains (TMDs) of Yor1p. Previous studies on the related ABC transporters P-glycoprotein and CFTR had used cysteine residues, substituted within specific transmembrane domains, to probe the ability to form cross-linked disulfide bonds (Chen et al., 2004 ). Misfolding mutations, including the ΔF508 lesion, resulted in an inability to form specific cross-links, suggesting that these transmembrane domains fail to assemble correctly. We introduced cysteine residues at specific positions in the 6th and 12th TM domains of Yor1p, equivalent to the sites used for CFTR. Membranes expressing wild-type Yor1p that contained cysteine substitutions at F481 (TM6) and L1162 (TM12) were exposed to a methanethiosulfonate cross-linker with a 13-Å spacer arm and the mobility of the protein monitored by nonreducing SDS-PAGE and anti-HA immunoblotting (Figure 1E). On exposure to cross-linker, a species with reduced mobility was detected, similar to that observed for CFTR. This cross-linked protein was also detected with a L479C/L1162C substitution pair, analogous to a second cross-linking pair that was used to probe transmembrane domain assembly in CFTR (Pagant, unpublished data). The cross-linked species likely represents an intramolecular modification, because when the individual F481C and L1162C substitutions were introduced separately on two different plasmids and cotransformed into cells, no cross-linked proteins were detected (Figure 1E, left). When the paired F481C/L1162C substitutions were introduced into Yor1p-ΔF, no cross-linked species were detected; instead, the unmodified protein disappeared and the majority of the protein presented as a very-high-molecular weight aggregate that largely failed to enter the resolving gel. Similar higher order aggregation was also detected for CFTR-ΔF, and it is thought to represent nonspecific cross-linking of exposed cysteine residues (Chen et al., 2004 ). Thus, like CFTR-ΔF, Yor1p-ΔF seems to contain improperly assembled transmembrane domains.
To gain a better understanding of how Yor1p engages the ER export pathway, a clear definition of the ER export signals used by Yor1p is required. Two independent diacidic motifs have been identified in Yor1p; when either the N-terminal D71XE73 or C-terminal D1472XE1474, were mutated, Yor1p was trapped in the ER, causing cells to become oligomycin sensitive (Epping and Moye-Rowley, 2002 ). Similarly, ER exit of CFTR is dependent upon the presence of a diacidic motif, although the putative signal in CFTR is contained within NBD1 (Wang et al., 2004 ). We investigated the potential roles of several diacidic motifs in mediating ER exit of Yor1p, focusing on the two terminal DXE motifs and two potential diacidic motifs in NBD1 of Yor1p: D691XD693 and D729XY731. The latter signal is spatially homologous to the aspartic acid residue of the DXE CFTR export signal, but it only contains a single acidic residue (Figure 1A). We generated HA-tagged mutants where each potential motif was replaced with alanine, and we examined in vivo phenotypes and in vitro capture into COPII vesicles of each of the mutant proteins (Figure 2).
Yor1p is required for growth in the presence of low concentrations of the mitochondrial poison, oligomycin, and impaired trafficking of Yor1p to the plasma membrane results in oligomycin sensitivity (Epping and Moye-Rowley, 2002 ). We replicated the oligomycin-sensitive phenotypes of the D71AXE73A and D1472AXE1474A mutants observed by Epping and Moye-Rowley (2002) ; the N-terminal mutant was more sensitive to oligomycin than the C-terminal mutant. Although the D691AXD693A mutant was extremely sensitive to oligomycin, the D729A mutant was relatively resistant to the drug, similar to wild-type Yor1p (Figure 2A). These data suggest that D729 residue does not comprise the ER export motif for Yor1p, but the oligomycin sensitivity phenotype alone did not allow us to distinguish between the three remaining motifs as bona fide export signals.
To better define the signal responsible for ER export of Yor1p, we examined uptake of each of the mutants into COPII vesicles generated in vitro from radiolabeled permeabilized cells. Wild-type Yor1p was efficiently packaged into COPII vesicles, and of all the diacidic mutants that we tested, only the N-terminal D71AXE73A mutant was impaired in its capture into COPII vesicles (Figure 2B). A control cargo protein, Vph1p, was equivalently packaged in each of these experiments, indicative of general budding efficiency. We further tested the localization of Yor1p mutants by microscopy by using Yor1p–GFP fusions (Figure 2C). As reported previously, wild-type Yor1p fused to GFP was found at the plasma membrane, whereas the N-terminal (D71AXE73A) diacidic mutant showed a perinuclear localization and cortical ER localization, consistent with its retention in the ER (Epping and Moye-Rowley, 2002 ). Conversely, the localization of the C-terminal (D1472AXE1474A) diacidic mutant more closely resembled that of the wild-type protein, with strong plasma membrane fluorescence and no detectable internal perinuclear staining. This is consistent with our in vitro budding data, but it conflicts with a previous report that used sucrose gradients to show that Yor1p-D1472AXE1474A was ER retained (Epping and Moye-Rowley, 2002 ). We note that the oligomycin sensitivity phenotype of this C-terminal mutant is intermediate between wild-type Yor1p and the N-terminal mutant, suggesting some in vivo impairment of protein delivery or function. It is possible that the C-terminal mutant protein is turned over more rapidly at the plasma membrane, rendering cells slightly more sensitive to oligomycin. Such destabilization may create a steady-state distribution such that the mutant protein is located both at the plasma membrane and in the ER; the ER pool may be undetectable by GFP fluorescence but more easily resolved by cell fractionation. Consistent with this model, destabilization of CFTR by truncation of the C terminus causes the protein to be rapidly internalized from the plasma membrane and cycled through an endosomal compartment before lysosomal destruction (Sharma et al., 2004 ).
To confirm that the impaired trafficking of the D71AXE73A mutant results from an abrogated exit signal and not a folding defect, we examined the folding status of the mutant protein (Figure 2D). Yor1p-D71AXE73A showed a similar cleavage profile to that of wild-type Yor1p with a persistent band at ~160 kDa and another at ~60 kDa. Similar experiments with BNGE also indicated that each of the mutants was properly folded, with migration patterns similar to wild-type Yor1p (Lee, unpublished data). Finally, we probed the arrangement of transmembrane domains of the di-acidic mutant by using cysteine cross-linking, which revealed a similar cross-linking pattern to that of the wild-type protein (Figure 2E). Together, these data confirm that the N-terminal diacidic motif, D71XE73, is the ER export signal for Yor1p, and that the ER export defects observed when this motif is altered are not caused by misfolding of Yor1p. The other diacidic mutants were also correctly folded, as determined by trypsin sensitivity and BNGE, suggesting that the oligomycin sensitivity of these proteins may stem from defects in protein function rather than impaired protein trafficking or gross protein misfolding.
ER export signals function by interacting with Sec24p, which contains three known cargo-binding sites: the A-, B- and C-sites that each recognize distinct motifs. The observation that Yor1p uses a diacidic export motif (DxE) suggests the B-site of Sec24p as the likely binding site (Mossessova et al., 2003 ). However, because the superficially similar diacidic motif of Gap1 (DID) does not use the B-site, we also explored a potential role for either the A-site or C-site in Yor1p uptake (Miller et al., 2003 ). We tested seven defined mutants of Sec24p by examining various drug sensitivities of strains expressing each of these mutants as the sole copy of Sec24p, anticipating that disrupting the Sec24p–Yor1p interaction would result in increased oligomycin sensitivity. Indeed, B-site mutants of Sec24p that displayed normal growth on YPEG (Figure 3A) were unable to grow in the presence of low concentrations (0.1 μg/μl) of oligomycin (Figure 3B), but viability could be rescued by introducing an episomal copy of Yor1p (Figure 3C). Similar sensitivity was seen in the presence of rhodamine (400 μg/ml; Figure 3D), which is cleared from the cell predominantly by another ABC transporter, the pleiotropic drug pump Pdr5p, but can also use Yor1p (Rogers et al., 2001 ). Neither the Sec24p A-site nor C-site mutants were sensitive to oligomycin or rhodamine. All of the Sec24p mutants were able to grow in the presence of benomyl at concentrations that result in lethality of mutants with impaired microtubule function (10 μg/ml; Figure 3E), suggesting that the sensitivity of the B-site mutants to oligomycin and rhodamine is not a general drug intolerance but results from impaired trafficking of specific transporters. These observations suggest that an intact B-site on Sec24p is critical for the trafficking of Yor1p out of the ER and is most likely the site that binds the N-terminal D71XE73 motif of Yor1p.
To confirm that the oligomycin sensitivity of the Sec24p B-site mutants resulted from failure of Yor1p to leave the ER, we examined the localization of Yor1p-GFP in these mutants (Figure 4). In wild-type cells and Sec24p A- and C-site mutants, Yor1p-GFP was found exclusively at the plasma membrane (Figure 4), consistent with full oligomycin resistance of these mutant strains. However, when Yor1p-GFP was expressed in any of the B-site Sec24p mutants, a distinct perinuclear ER fluorescence pattern was observed (Figure 4). This ER retention of Yor1p-GFP is consistent with an inability to be efficiently recognized by the mutant form of Sec24p, causing accumulation in the ER and leading to oligomycin sensitivity. Intracellular localization of Gap1p-GFP was unaffected by the Sec24p B-site mutation (Kung, unpublished data), consistent with in vitro vesicle budding data that suggest this site is not used by Gap1p (Miller et al., 2003 ), and suggesting that the Yor1p trafficking defect is specific. Finally, we examined in vitro vesicle budding of Yor1p from urea-washed membranes by using either wild-type Sec24p or a B-site mutant. Although the vesicle budding efficiency was very low, likely the result of urea treatment, Yor1p was packaged into a vesicle fraction in the presence of wild-type Sec24p, but it was absent from vesicles made with the B-site mutant protein (Supplemental Figure 1). Together, these data suggest that the N-terminal diacidic motif of Yor1p acts as an ER export signal that interacts with the B-site on Sec24p to mediate efficient ER egress.
Given that the diacidic ER export signal for Yor1p resides in a separate domain from the ΔF lesion, it seems unlikely that this particular misfolding mutation directly impacts the interaction between Yor1p and Sec24p in the way that has been proposed for CFTR (Wang et al., 2004 ). We therefore aimed to understand how various cellular folding and degradative pathways might influence ER exit of Yor1p. Because the majority of Yor1p faces the cytoplasm, we hypothesized that protein folding may be assisted by cytoplasmic chaperones, which may in turn influence the recruitment of the COPII machinery. Newly synthesized CFTR in mammalian cells uses heat-shock protein of 90 kDa (HSP90) as a positive folding factor; treatment with geldanamycin (GA), an Hsp90 inhibitor, disrupts the interaction of nascent CFTR with cytosolic HSP90 and enhances its proteasomal degradation (Loo et al., 1998 ). Yeast possess two cytoplasmic HSP90s, one constitutive (Hsc82p) and one heat-inducible (Hsp82p), which are 97% identical at the amino acid level. Due to their functional redundancy, the expression of only one homologue is required to maintain viability (Borkovich et al., 1989 ). We analyzed the stability of wild-type Yor1p in a double mutant, hsc82Δ hsp82Δ, that expresses a temperature-sensitive mutant protein, hsp82pts (G313N), which is extremely unstable and rapidly degraded when cells are shifted to the nonpermissive temperature of 37°C (Bohen and Yamamoto, 1993 ; Fliss et al., 2000 ). The corresponding wild-type strain for this experiment contains a plasmid-borne copy of wild-type HSP82. Cells were grown at permissive temperature and Yor1p degradation was monitored by pulse-chase analysis after shift to 37°C. The rate of wild-type Yor1p degradation was significantly higher in the mutant strain, suggesting that cytoplasmic HSP90 is important in the biogenesis of Yor1p (Figure 5A). Maturation of the cell wall protein, Gas1p, served as a control to demonstrate that loss of HSP90 function did not lead to general defects in protein secretion. Like Yor1p, Yor1p-ΔF was destabilized in the hsp82ts strain, suggesting that HSP90 also plays a role in the attempted folding and biogenesis of this aberrant protein (Figure 5B).
We asked whether these destabilized forms of Yor1p and Yor1p-ΔF could be rescued by perturbation of the ERAD pathway. ERAD of Yor1p-ΔF relies on ubiquitination by the ER-localized E2 ligase, Ubc7p, deletion of which partially stabilizes Yor1p-ΔF (Katzmann et al., 1999 ). Yor1p-ΔF was slightly restabilized in a hsp82ts ubc7Δ double mutant, showing a similar stability to that seen in wild-type cells (Figure 5B, right). Conversely, mutation of UBC7 in the hsp82ts background did not result in the stabilization of wild-type Yor1p, suggesting that perturbation of HSP90 function did not render Yor1p a substrate for the same ERAD accessed by Yor1p-ΔF, but instead it induced degradation via a different mechanism. We examined the folding state of newly synthesized Yor1p by trypsin digestion of membranes isolated from cells that had been radiolabeled after shift to nonpermissive temperature. Yor1p presented a similar cleavage profile in both the wild-type and mutant strains (Figure 5C). These results suggest that Yor1p is able to properly fold in the absence of HSP90 but that it is ultimately destabilized despite this apparent native conformation. We next examined the in vitro packaging of Yor1p into COPII vesicles by using membranes from either wild-type or hsp82ts cells that had been preshifted to 37°C and pulse-labeled at the restrictive temperature. Yor1p was packaged equally well into vesicles generated from wild-type and mutant membranes, and budding of a control protein, Vph1p, demonstrated equivalent levels of COPII vesicle formation (Figure 5D). These data suggest that the destabilizing effect of HSP90 disruption does not directly influence the ER exit of Yor1p, which is not degraded by ERAD in the hsp82ts strain but likely transits the secretory pathway for degradation in the vacuole.
Having probed some of the mechanisms of protein folding and ER export of wild-type Yor1p, we determined which ERAD pathway is used in the degradation of the aberrant Yor1p-ΔF. We expressed Yor1p-ΔF in the ERAD mutants, ubc7Δ and sel1Δ/ubx2Δ, both of which are common to all known ERAD pathways; Ubc7p is the E2 ligase responsible for ubiquitin modification, and Sel1p/Ubx2p is the membrane anchor for the protein extraction apparatus Cdc48p. We examined the stability of Yor1p by pulse-chase analysis in each of these mutants (Figure 6). As reported by Katzmann et al. (1999) , mutation of Ubc7p resulted in a significant stabilization of Yor1p-ΔF, restoring the protein to essentially wild-type Yor1p levels (Figure 6A). Similarly, deletion of Sel1p caused Yor1p-ΔF to be degraded more slowly than in wild-type cells (Figure 6A). We note that the half-life that we observe for Yor1p-ΔF is significantly longer than that described by Katzmann et al. (1999) . Because our experimental approaches are largely the same, we suspect that these differences result from the distinct strain backgrounds used in the two studies. In each case, disruption of Ubc7p resulted in the stabilization of Yor1p-ΔF to essentially wild-type Yor1p levels.
Hrd1p/Der3p and Doa10p are components of two different ERAD pathways that are responsible for degrading ERAD substrates with lumenal (ERAD-L) and cytosolic (ERAD-C) lesions, respectively. Surprisingly, Yor1p-ΔF was degraded equivalently in each of the hrd1Δ and doa10Δ mutants, similar to its degradation in wild-type cells (Figure 6B). However, in a hrd1Δdoa10Δ double mutant, Yor1p-ΔF was slightly stabilized, equivalent to the level of stabilization seen in the sel1Δ strain (Figure 6B). These data suggest that Yor1p-ΔF accesses both the ERAD-L and ERAD-C pathways, unlike another misfolded ABC transporter, Ste6p* (Huyer et al., 2004 ) but similar to the unrelated amino acid transporter Gap1p (Kota et al., 2007 ).
We next investigated a possible mechanism by which Yor1p- ΔF might be prevented from entering into COPII vesicles: misfolded or aberrant proteins may be disposed of by ERAD so rapidly that they do not have time to engage the ER export machinery. Moreover, either ubiquitination itself, or simple engagement of the ERAD machinery may mask ER export motifs that would otherwise drive forward transport. We tested this hypothesis by using the ubc7Δ mutant to block the ER ubiquitination of Yor1p-ΔF and by asking whether this stabilization could rescue uptake into COPII vesicles (Figure 7). Stabilization of Yor1p-ΔF did not result in improved folding of Yor1p-ΔF, because newly synthesized Yor1p-ΔF remained highly susceptible to trypsin cleavage in both wild-type and ubc7Δ cells (Figure 7A). Similarly, the ability of Yor1p-ΔF to form transmembrane domain cross-links was not restored in a ubc7Δ strain (Figure 7B). We next tested capture of this stabilized pool of Yor1p-ΔF into COPII vesicles. Loss of UBC7 did not result in uptake of Yor1p-ΔF into COPII vesicles; the vacuolar protein Vph1p served as a positive control to demonstrate efficient vesicle formation (Figure 7C). These data suggest that neither the ubiquitination of Yor1p-ΔF nor the engagement of ERAD components downstream of Ubc7p is responsible for the ER retention of Yor1p-ΔF.
It is possible that the in vitro vesicle budding assay we use to directly monitor egress from the ER is not sensitive enough to detect a very small increase in the budding efficiency of stabilized Yor1p-ΔF. We therefore used a phenotypic growth assay to probe the function of Yor1p at the plasma membrane under conditions that stabilized Yor1p-ΔF in vivo. Consistent with the accumulation of Yor1p-ΔF in the ER, cells that express Yor1p-ΔF as the sole copy of Yor1p are oligomycin sensitive (Katzmann et al., 1999 ). Strains disrupted for either YOR1 alone or YOR1 plus UBC7 were transformed with plasmids containing wild-type Yor1p or Yor1p-ΔF, with an empty vector serving as a negative control. Strains bearing the empty vector were completely sensitive to oligomycin, whereas the presence of plasmid-borne Yor1p conferred oligomycin resistance (Figure 7D). As expected, Yor1p-ΔF expressed in yor1Δ cells was unable to confer oligomycin resistance, and this phenotype was also observed in the yor1Δ ubc7Δ double mutant, suggesting that stabilization of Yor1p-ΔF cannot rescue in vivo function. We observed a similar phenotype, i.e., no rescue of Yor1p-ΔF function, in a yor1Δsel1Δ double mutant (Pagant, unpublished data).
To investigate whether blocking the ERAD pathway upstream of Ubc7p could lead to export of Yor1p-ΔF from the ER, we analyzed the intracellular fate of Yor1p-ΔF in cells mutated for the ER localized Hsp40s, a class of chaperones known to play roles in protein folding and degradation. Ydj1p is the yeast homologue of Hdj-2, which has been shown to be involved in the ER quality control of mammalian CFTR by assisting in polyubiquitination before degradation (Younger et al., 2004 ). However, a second yeast ER-localized Hsp40, Hlj1p, is functionally redundant with Ydj1p in the degradation of human CFTR expressed in yeast cells (Youker et al., 2004 ). We therefore analyzed the biogenesis of Yor1p in the temperature-sensitive double mutant hlj1Δ ydj1-151. Pulse-chase experiments at restrictive temperature showed that Yor1p-ΔF was stabilized in the absence of Hsp40 function in the ER membrane (Figure 8A), consistent with the model that ER-localized Hsp40s divert misfolded proteins for proteosomal degradation. Maturation of the cell wall protein, Gas1p, was identical in wild-type and hlj1Δ ydj1-151 cells, demonstrating that loss of chaperone function does not lead to general defects in secretory protein biogenesis. Interestingly, the trypsin sensitivity of Yor1p-ΔF synthesized at restrictive temperature in the hlj1Δ ydj1-151 strain more closely resembled that of the wild-type protein, suggesting that cytoplasmic Hsp40s participate in the unfolding of Yor1p-ΔF before degradation (Figure 8B). However, despite the stabilization and at least partial refolding of Yor1p-ΔF in the hlj1Δ ydj1-151 mutant, this mutant protein was not packaged into COPII vesicles in an in vitro vesicle budding assay (Figure 8C). We used cysteine cross-linking to probe the assembly of the membrane domains of Yor1p-ΔF in the Hsp40 mutants, which revealed that despite the conversion of Yor1p-ΔF to a trypsin-resistant form, the TMDs did not show a wild-type configuration (Figure 8D). Together, these data suggest that Hsp40-containing chaperone systems actively unfold the cytoplasmic domains of Yor1p before degradation without affecting the arrangement of TMDs, and that even after stabilization of aberrant Yor1p, the ER quality control checkpoint is still enforced.
Quality control of protein biogenesis within the secretory pathway is a fundamentally important process; the impaired folding and trafficking of proteins is the proximal cause of a number of human diseases. We sought to better understand the interplay between protein folding and forward transport through the secretory pathway by examining how the stability of a misfolded protein within the ER influences uptake into ER-derived COPII vesicles. We explored these pathways by investigating the biogenesis of the yeast ABC transporter Yor1p, a plasma membrane-localized pleiotropic drug pump that is homologous to CFTR with respect to domain arrangement and topology. An aberrant form of Yor1p with an analogous phenylalanine deletion within NBD1, Yor1p-ΔF670, is retained in the ER and degraded by ERAD (Katzmann et al., 1999 ). We used several in vitro assays to demonstrate that Yor1p-ΔF fails to enter into COPII vesicles and that this aberrant protein is improperly assembled.
We used limited proteolysis to demonstrate that Yor1p-ΔF showed enhanced susceptibility to proteolytic attack, consistent with a partially folded or conformationally destabilized structure similar to that seen for the ΔF form of CFTR (Du et al., 2005 ). Biochemical and structural analyses of CFTR NBD1 suggest that the ΔF508 mutation does not dramatically alter the fold of the isolated domain but that it may instead disrupt surface topology. This in turn may perturb an interaction between NBD1 and MSD1, indirectly affecting the interactions between transmembrane helices (Lewis et al., 2005 ; Thibodeau et al., 2005 ). Furthermore, F508 in CFTR also contributes to the posttranslational folding of NBD2, likely through interdomain interactions (Du et al., 2005 ). Because the epitope tag with which we detected Yor1p is located at the C terminus of the protein, trypsinolysis specifically probes the conformational state of the second half of the protein. Thus, the stable ~65-kDa proteolytic fragment observed for wild-type Yor1p likely corresponds to a cleavage event between TM8 and TM9, with the ΔF670 mutation likely destabilizing NBD2, similar to the situation observed for CFTR. We also used cysteine substitution and cross-linking analysis to probe the arrangement of transmembrane domains in Yor1p. The inability of Yor1p-ΔF to form cross-linkable interactions between membrane spanning domains suggests that, like CFTR, multiple interdomain interactions are impaired by this misfolding lesion (Cui et al., 2007 ).
Together, these in vitro assays suggest that Yor1p-ΔF presents multiple assembly defects: improperly arranged transmembrane segments as well as a destabilized cytoplasmic domain (NBD2). The presence of multiple misfolding lesions is consistent with its redundant use of at least two disposal pathways. Both the Hrd1p-dependent pathway, which likely disposes of proteins with membrane-localized lesions in addition to ERAD-L clients (Carvalho et al., 2006 ), and the Doa10p-dependent pathway, which degrades ERAD-C substrates, can divert Yor1p-ΔF for degradation; only when both pathways are disabled is Yor1p-ΔF stabilized. Because this stabilization is only partial, the possibility remains that additional machinery also contributes in vivo to the efficient destruction of aberrant Yor1p. The specific misfolding lesions detected by each of these degradation systems remain to be defined, but they likely result from the exposure of multiple domain interfaces that are usually buried in the native protein. Our evidence that both NBD2 and the transmembrane domains of Yor1p-ΔF are improperly assembled suggests that both cytosolic and intramembrane lesions are potential signals for degradation by the ERAD-C and ERAD-L/M pathways, respectively. The dual disposal mechanism we see for Yor1p-ΔF is similar to that used by the amino acid permease, Gap1p, which is rapidly degraded in the absence of its chaperone, Shr3p (Kota et al., 2007 ). Again, the precise nature of the conformational abnormality presented by Gap1p in the absence of Shr3p is not known, but based on our data for Yor1p-ΔF, it seems likely that Gap1p must present multiple folding defects to access both the ERAD-C and ERAD-L/M pathways.
Although misfolded Gap1p and Yor1p-ΔF share the same redundant degradation pathways, these two proteins differ in their ability to be rescued by inhibiting degradation. Blocking ERAD of Yor1p-ΔF did not permit capture into COPII vesicles, nor did it restore oligomycin resistance. However, when misfolded Gap1p was stabilized by introducing ERAD mutations into an shr3Δ strain, partial recovery of amino acid transport function was observed (Kota et al., 2007 ). This difference in protein recovery may stem from the nature of the specific folding defects associated with each individual protein. Aberrant Gap1p produced in the absence of its chaperone may represent a mixed population of folding intermediates, some of which can attain a folding state compatible with COPII engagement if the degradation pathways are impaired. Conversely, the specific genetic lesion that causes misfolding of Yor1p-ΔF may render it incapable of proper assembly even given prolonged residence time in the ER. Thus, the ability of an aberrant protein to be rescued by impeding its destruction seems likely to be dependent on the specific nature of the misfolding defect, which will in turn influence the folding, degradative and trafficking pathways that each specific protein can access. Indeed, different misfolding variants of a single protein may behave distinctly; a variety of transthyretin mutants showed distinct secretion phenotypes that were not entirely correlated with the degree of global protein stability, even in the absence of stabilization through ERAD inhibition (Sekijima et al., 2005 ). Furthermore, even when considering a single protein, different mechanisms of stabilization may result in different outcomes. Attempts to rescue CFTR-ΔF by inhibiting its destruction have met with mixed results; early experiments using proteasome inhibitors did not result in rescue of CFTR-ΔF (Kopito, 1999 ) but more recent experiments in lung epithelial cells suggest that inhibition of either the proteasome or the protein extraction apparatus, p97/VCP/Cdc48p, can rescue delivery of CFTR-ΔF to the plasma membrane (Vij et al., 2006 ). Indeed, the ability for a misfolded protein to be rescued by simple stabilization may also depend on the cell type, with the specific complement of cellular chaperones associated with a given tissue likely to play an important role. This seems particularly true for polytopic membrane proteins such as Yor1p and CFTR, which are subject to complex quality control processes involving multiple chaperone machineries (Younger et al., 2006 ).
One additional mechanism that may contribute to the differing abilities of misfolded proteins to be rescued by inhibition of ERAD is the presentation of an appropriate ER export signal. Such signals tend to be small peptides with limited structural features that may be obscured or improperly presented in the context of a misfolded protein. The failure to present an appropriate export motif could account for the distinct intracellular fates of misfolded variants of transthyretin; the different mutants may display or obscure their ER export signals independently of the thermodynamics of the particular protein fold (Sekijima et al., 2005 ). In membrane proteins, which interact directly with Sec24p, the presentation of an appropriate export motif in the context of a misfolded protein is likely to be a more complicated affair. For CFTR, the diacidic ER export signal is located in the same domain as the misfolding lesion, suggesting that the structural destabilization induced by the ΔF508 lesion may directly impair the presentation of the signal to Sec24p (Wang et al., 2004 ). In Yor1p, the diacidic export motif is located in a different cytosolic domain from the misfolding lesion, suggesting the misfolding event has a more indirect effect on protein traffic. One key question that remains unanswered from our study is whether the ΔF lesion impacts the interaction between Sec24p and Yor1p directly. We attempted to address this issue by isolating “prebudding complexes” that contain cargo proteins bound to Sar1p/Sec23p/Sec24p and represent an early cargo–capture phase of vesicle biogenesis (Kuehn et al., 1998 ). However, we were unable to reproducibly detect wild-type Yor1p in these complexes, perhaps due to the difficulty in solubilizing a protein with so many transmembrane domains (Kung and Miller, unpublished data). Alternative approaches such as reconstitution of Yor1p into detergent micelles or synthetic liposomes may allow us to more directly assess the binding between Yor1p and Sec24p and ask how the ΔF lesion impacts this interaction. In addition to directly impeding interaction with Sec24p, another mechanism by which a misfolding lesion may impact forward traffic is by indirectly preventing capture into vesicles through the recruitment of cytoplasmic chaperones at early stages of protein synthesis. We have demonstrated that cytoplasmic HSP90s play a role in productive biogenesis of wild-type Yor1p. One can imagine a model whereby correctly folded Yor1p no longer interacts with HSP90, release of which in turn exposes the Sec24p interaction motif. Thus, action of Sec24p would be sterically inhibited by HSP90, and a molecular handoff would occur between the folding chaperones and the forward transport machinery. Detailed coimmunoprecipitation experiments, in combination with kinetic analysis of these interactions during Yor1p biogenesis, will be required to fully map the sequential and competitive interactions between the various cytoplasmic chaperone machinery, the ER export apparatus and newly synthesized Yor1p. Furthermore, the action of these chaperones may be very complex: recent evidence suggests that HSP90 can influence CFTR biogenesis in both a positive and negative manner, likely modulated by distinct cochaperones (Wang et al., 2006 ).
In addition to HSP90s, the ER-localized HSP40s, Ydj1p and Hlj1p, are also candidates for interfering with ER export of Yor1p. Through recruitment of cytoplasmic HSP70s, these chaperones may impair maturation of misfolded proteins either by sequestering the aberrant protein in a distinct ER subdomain or by triggering the degradation pathway. In our experiments, the action of these HSP40s promoted the trypsin-sensitive conformation of Yor1p-ΔF, suggesting an active role in unfolding the cytoplasmic domains before exposure to the ubiquitination and retrotranslocation machinery. A similar role in actively unfolding cytosolic domains of misfolded proteins (both soluble and membrane-bound) before proteasomal degradation has been suggested for the Ssa1p–Ydj1p chaperone complex (Taxis et al., 2003 ; Park et al., 2007 ). This action may be required for tightly folded domains that could not be denatured solely by the 19S cap of the proteasome (Taxis et al., 2003 ). We note that despite some restoration of a more wild-type conformation of the C-terminal domain, Yor1p-ΔF still failed to enter into COPII vesicles in the HSP40 mutant, likely because the arrangement of its transmembrane segments was still aberrant.
In summary, we have developed Yor1p-ΔF as a model yeast misfolded protein, and we have defined some of the chaperones, ER export machinery, and degradation apparatus that engage both the wild-type and aberrant forms of the protein. It is clear from our data that aberrant Yor1p accesses multiple degradation pathways and that stabilization of this misfolded protein by blocking its degradation does not confer transport competence. This observation has important implications for potential treatments for diseases caused by improper ER retention and degradation. A growing body of evidence suggests that different misfolded proteins are handled differently by the multiple processes that comprise ER quality control. It seems likely that, like the multiple pathways of protein degradation from the ER, diverse mechanisms that prevent forward transport are likely to impact the efficiency of ER export in the context of protein misfolding. Such mechanisms may include active ER retention, steric hindrance of interaction with Sec24p, prevention of capture of prebudding complexes into a vesicle proper, or failure to engage additional sorting machinery. Given the wide array of ER export signals and binding sites on Sec24p, it is easy to imagine that cells use a variety of mechanisms to ensure that aberrant proteins are not deployed within the secretory pathway. A deeper understanding of the interplay between the degradative pathways and the ER export machinery will be required to fully elucidate how misfolded proteins are subjected to quality control within the secretory pathway.
We thank Scott Moye-Rowley, Jeff Brodsky (University of Pittsburgh), Randy Schekman, and Charlie Barlowe for generously providing plasmids, strains, and antibodies. We are grateful to Alenka Copic, Justine Barry, and Ray Louie for comments on the manuscript and technical assistance. Live cell imaging was performed at the Analytical Imaging Facility at the Albert Einstein College of Medicine (Bronx, NY). This work was supported by National Institutes of Health grant GM-078186 (to E.A.M.).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-01-0046) on July 5, 2007.
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org).