Quality Control in the Yeast Secretory Pathway

The yeast plasma-membrane H+-ATPase, encoded by PMA1, is delivered to the cell surface via the secretory pathway and has recently emerged as an excellent system for identifying quality control mechanisms along the pathway. In the present study, we have tracked the biogenesis of Pma1-G381A, a misfolded mutant form of the H+-ATPase. Although this mutant ATPase is arrested transiently in the peripheral endoplasmic reticulum, it does not become a substrate for endoplasmic reticulum-associated degradation nor does it appear to stimulate an unfolded protein response. Instead, Pma1-G381A accumulates in Kar2p-containing vesicular-tubular clusters that resemble those previously described in mammalian cells. Like their mammalian counterparts, the yeast vesicular-tubular clusters may correspond to specific exit ports from the endoplasmic reticulum, since Pma1-G381A eventually escapes from them (still in a misfolded, trypsin-sensitive form) to reach the plasma membrane. By comparison with wild-type ATPase, Pma1-G381A spends a short half-life at the plasma membrane before being removed and sent to the vacuole for degradation in a process that requires both End4p and Pep4p. Finally, in a separate set of experiments, Pma1-G381A was found to impose its phenotype on co-expressed wild-type ATPase, transiently retarding the wild-type protein in the ER and later stimulating its degradation in the vacuole. Both effects serve to lower the steady-state amount of wild-type ATPase in the plasma membrane and, thus, can explain the co-dominant genetic behavior of the G381A mutation. Taken together, the results of this study establish Pma1-G381A as a useful new probe for the yeast secretory system.

In eukaryotic cells, plasma membrane proteins are delivered to the cell surface by the secretory pathway, whose port of entry is the endoplasmic reticulum (ER). 1 Not surprisingly, quality control mechanisms have evolved to detect and eliminate misfolded proteins at several points along the pathway. Many proteins that fail to fold properly in the ER are eliminated by a process known as ERAD (ER-associated degradation), which transports them back into the cytoplasm to be degraded in a ubiquitin-dependent manner by the proteasome (1)(2)(3). In the event of excessive ER stress, the ER-associated degradation pathway is complemented by the unfolded protein response, a stress response that involves transcriptional induction of a set of target genes encoding ER chaperones and other products involved in ER function (4 -6). Misfolded proteins that escape the ER may encounter a second, Golgi-based quality control mechanism, recently detected in yeast, which sends them to the vacuole for degradation (7)(8)(9)(10). Finally, recognition mechanisms exist at the plasma membrane to down-regulate many transporters and receptors by endocytosis in response to specific physiological signals (reviewed by Hicke (11,12)); at least in principle, these or related mechanisms could function to remove misfolded proteins that manage to reach the cell surface (e.g. Ref. 13).
The present study has focused on the yeast plasma-membrane H ϩ -ATPase as a model to investigate quality control during biogenesis. The H ϩ -ATPase belongs to a large, widely distributed family of transporters known as the P-type cation pumps (14); it is encoded by the PMA1 gene (15) and is the most abundant protein in the yeast plasma membrane. Work in many laboratories has established that the physiological function of the H ϩ -ATPase is to pump protons out of the cell, setting up the driving force for nutrient uptake by H ϩ -dependent cotransporters (reviewed by Rao et al. (16)). Consistent with predictions from hydropathy analysis (15), there is now good evidence from cryo-electron microscopy that the 100-kDa H ϩ -ATPase is embedded in the lipid bilayer by 10 membranespanning ␣-helices, four at the N-terminal end of the molecule and six at the C-terminal end (17). Its central portion protrudes into the cytoplasm and contains ligand-binding sites needed for catalysis. These structural characteristics are supported by comparison with the 2.6-Å crystal structure of the P-type Ca 2ϩ -ATPase from rabbit sarcoplasmic reticulum (18).
The yeast H ϩ -ATPase is known to be delivered to the cell surface via the secretory pathway, since it can be trapped in the ER, Golgi, or secretory vesicles by temperature-sensitive mutations of SEC18, SEC7, or SEC6 (19 -22). More than 300 point mutations have been introduced into the PMA1 gene by sitedirected mutagenesis; others have been selected by resistance to hygromycin B (reviewed by Morsomme et al. (23)). Taken together, these mutants provide a wealth of material for exploring the relationship between H ϩ -ATPase structure and the components of various quality control mechanisms.
A particularly striking example comes from recent work on mutations of Asp-378, a key residue that is conserved in all P-type ATPases and forms a transient ␤-aspartyl phosphointermediate during ATP hydrolysis. Unexpectedly, Pma1 H ϩ -ATPases carrying point mutations of Asp-378 are retained in the ER and, when co-expressed with the wild-type ATPase, display a dominant lethal phenotype (16, 24 -27). The ER arrest of Asp-378 mutants has been traced to gross misfolding, as evidenced by the extreme sensitivity of Pma1-D378N, Pma1-D378S, and Pma1-D378A to low concentrations of trypsin (25,27). More recently, Wang and Chang (28) have shown that Pma1-D378N is ultimately ubiquitinated and degraded, presumably by the proteasome. To identify components of the ER quality control machinery, they exploited the dominant lethal phenotype of the D378N mutant to select suppressors from a mutagenized genomic library carrying random insertions of lacZ and LEU2. Indeed, one such suppressor (eps1⌬) allowed Pma1-D378N to move from the ER to the plasma membrane; the corresponding gene (EPS1) proved to encode a novel member of the protein disulfide isomerase-related family, which may function as a quality control chaperone to retain Pma1-D378N in the ER.
Chang and co-workers (7,9,10) have also used a temperature-sensitive mutant of the yeast H ϩ -ATPase, pma1-7, to probe the organization of the secretory pathway. At the restrictive temperature, this mutant protein exits the ER but is then shunted from the Golgi to the vacuole for degradation (7). Screening with a high copy genomic library has yielded two suppressors (AST1 and AST2) whose products, when overexpressed, can re-route Pma1-7 to the plasma membrane (7). In parallel, 16 different insertional suppressors of pma1-7 have been isolated (9) and are being used to define the components of two distinct pathways by which Pma1-7 is diverted to the cell surface (10). Although there is not yet any direct information on the structure of the Pma1-7 mutant protein, it must be capable of folding relatively well even at 37°C, since in the presence of an appropriate suppressor mutation, it reaches the plasma membrane, hydrolyzes ATP at a measurable rate, and supports growth (7,9,10).
The goal of the present study was to see whether new information about the yeast secretory pathway could be obtained with the help of a different mutant form of the H ϩ -ATPase, Pma1-G381A, which is moderately sensitive to trypsin and, thus, appears to be intermediate in structure between Pma1-D378N and Pma1-7 (25). Although immunofluorescence studies have shown that Pma1-G381A can reach the plasma membrane, it lacks detectable enzymatic activity and displays a co-dominant phenotype when co-expressed with the wild-type H ϩ -ATPase, suggesting that it may interact in an informative way with one or more components of the quality control machinery. We have carefully tracked the expression of epitopetagged Pma1-G381A as a function of time. Initially, the mutant protein induces dramatic proliferation of a morphologically well defined compartment derived from the peripheral ER, in which it transiently accumulates along with the ER-resident chaperone Kar2p (BiP). Pma1-G381A then escapes this compartment and reaches the secretory vesicles, still in a misfolded form. As a consequence, it is relatively unstable at the plasma membrane, being removed rapidly by endocytosis for degradation in the vacuole. Significantly, the mutant form is able to impose its phenotype on co-expressed wild-type ATPase, which likewise becomes retarded in the ER and then quickly degraded upon arrival at the plasma membrane.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Growth Conditions-The Saccharomyces cerevisiae strains and plasmids used in this study are listed in Table I. Standard yeast media and genetic manipulations were as described by Sherman et al. (35), and yeast transformations were performed by the lithium acetate method (36). Strains TFY25 and TFY54 were derived from a cross between NY605 and NY1174; strain TFY1 was derived from a cross between TFY25 and RH268 -1; and strains TFY15 and TFY32 were derived from a cross between TFY54 and SY2. For expression studies, derivatives of strains TFY1, TFY15, TFY32, WG4a, and W303 were constructed in which the chromosomal copy of the PMA1 gene was placed under control of the GAL1 promoter (37) by transforming yeast with the integrative plasmid YIpGAL-PMA1 (32) linearized with BstEII; URA3 served as the selectable marker in these transformations. The pma1 allele to be expressed was tagged after codon 2 with either a sequence corresponding to a 10-amino acid cMyc epitope plus linking sequences (MTASEQKLISEEDLNDTS) or a 9-amino acid HA epitope (MTYPYDVPDYADTS) (25). To generate a tagged version of LST1, the cMyc epitope was inserted after codon 13, and this construct was introduced at the LST1 genomic locus using integrative plasmid YIplac128 linearized with StuI. As shown in Table I, tagged versions of pma1 were placed under control of a heat-shock promoter (27) in integrative plasmid pRS303 (HIS3), YIplac128 (LEU2), or YIplac204 (TRP1) or centromeric plasmid Ycplac22 (LEU2). Integrative plasmids were linearized with NheI (pRS303), AflII (YIplac128), or BsgI (YI-plac204) and integrated at the genomic locus corresponding to the selectable marker. The transformants were grown at 23°C in synthetic complete medium (38) lacking uracil and the relevant auxotrophic amino acids and containing 2% galactose. As desired, expression of the GAL1-PMA1 copy was turned off by incubating the cells for 3 h in medium containing 2% glucose instead of galactose, and expression of the tagged PMA1 gene was induced by incubation of the cells at 39°C.
Immunofluorescence-Subcellular structures containing HA-or cMyc-tagged ATPase were visualized by confocal microscopy as described by DeWitt et al. (25). Cells were fixed, treated with zymolyase T-100, permeabilized by Triton X-100, and stained for immunofluorescence essentially as described by Redding et al. (39) but with the addition of a blocking step in modified WT buffer, in which the dry milk and bovine serum albumin were replaced by 5% normal goat serum (Sigma). Primary antibodies that were used included: HA polyclonal (Medical and Biological Laboratories Co., Ltd., Nagoya, Japan), diluted 1:100; HA.11 monoclonal 16B12 and cMyc monoclonal 9E10 from raw ascites fluid (Berkeley Antibody Co., Richmond, CA), diluted 1:150; V-ATPase 60-kDa subunit monoclonal 13D11-B2 from raw ascites fluid (Molecular Probes, Eugene, OR), diluted 1:150; Sec7 (large domain) and Sec12 polyclonal antibodies (provided by R. Schekman (40)), diluted 1:150 and 1:100, respectively; and Kar2 polyclonal antibody (provided by M. Rose (41)), diluted 1:2500. Goat anti-rabbit FITC and goat antimouse Texas Red IgG (Jackson Immunoresearch, West Grove, PA) served as fluorescent secondary antibodies and were diluted 1:100. For double-labeling experiments, both primary antibodies were present during the initial incubation, and both secondary antibodies were present during the subsequent incubation. To control for spurious crossreactivity of the cMyc and HA antibodies, uninduced cells were simultaneously fixed and stained with each set of antibodies. Cells were then mounted in Citifluor (Ted Pella, Reading, PA).
Cells were observed on a Zeiss L510 scanning confocal microscope using dual channel filters for simultaneous visualization of Texas Red and FITC fluorochromes. All images were taken with a 63 ϫ 1.4 NA Plan-Apochromat III DIC objective (Zeiss). In time-course experiments, the exact same settings were used throughout the experiment to obtain a semi-quantitative signal. Cross-talk between FITC and Texas Red was avoided through the use of the Zeiss L510 digital signal processor. The absence of bleed-through was confirmed by checking that the signal disappeared when viewed with single-wavelength filter blocks. Images were collected with LSM5 software (Zeiss) and modified by contrast stretching, application of pseudocolor, and merging using Adobe Photoshop 4.0 (Adobe Systems Inc., San Jose, CA).
Immunoelectron Microscopy-Early logarithmic-phase cells were collected on a 0.22-m filter and transferred without washing to 3% paraformaldehyde (Electron Microscopy Sciences, Fort Washington, PA), 0.04 M potassium phosphate, pH 6.7, 0.8 M sorbitol, 1 mM MgCl 2 , and 1 mM CaCl 2 . After fixation for 1 h at room temperature and overnight at 4°C, the cells were embedded in LR White resin (Electron Microscopy Sciences) as described by Mulholland et al. (42).
Ultrathin sections were cut using a Reichert FC4E ultramicrotome and collected on Formvar/carbon-coated grids. Nonspecific binding sites were quenched using 1% fish skin gelatin (Sigma) in phosphate-buffered saline (PBS). The grids were incubated with anti-HA antibody (HA.11 monoclonal 16B12; Berkeley Antibody Co.) at a dilution of 1:40 in PBS, 1% fish skin gelatin, then washed in PBS and incubated with rabbit anti-mouse antibody (Cappel, ICN Pharmaceutical, Aurora, OH) at a dilution of 1:50. Finally, they were incubated with protein A-5-nm gold complex (purchased from J. Slot, Utrecht, Netherlands) at a dilution of 1:70. After final washes in PBS, the sections were fixed for 5 min in 1% glutaraldehyde (Electron Microscopy Sciences) in PBS. The sec-  (34) tions were contrasted with 5% aqueous uranyl acetate and lead citrate and examined in a Philips 410 electron microscope.

Isolation of Secretory Vesicles and Limited
Trypsinolysis-Secretory vesicles were isolated from cells carrying the temperature-sensitive sec6-4 mutation as described previously (43) with the substitution of a multistep discontinuous gradient for the two-step gradient used in the original method. Cells grown to early logarithmic phase in synthetic complete medium lacking uracil and tryptophan and containing 2% galactose were shifted for 3 h to medium lacking galactose but containing 2% glucose. The cells were then incubated for 2 h at 39°C to induce the expression of HA-tagged ATPase and the accumulation of secretory vesicles. After conversion to spheroplasts (32) and incubation with concanavalin A (0.8 mg/ml), the cells were broken by Dounce homogenization and centrifuged for 10 min at 14,500 ϫ g to remove concanavalin A-coated plasma-membranes, mitochondria, and unbroken cells. The supernatant fraction (cell lysate) was centrifuged for 35 min at 160,000 ϫ g, and the resulting pellet was suspended in 3 ml of 12.5% sucrose (w/v), 20 mM triethanolamine/acetic acid, pH 7.2, and 1 mM EDTA containing 1 mM diisopropyl fluorophosphate, leupeptin (1 g/ ml), pepstatin (1 g/ml), aprotinin (1 g/ml), and chymostatin (2 g/ml). This fraction has been shown in control experiments to contain little, if any, plasma-membrane-associated ATPase. It was further fractionated by loading 2 ml onto a discontinuous sucrose gradient (15-60%, consisting of 2.7 ml of 15,18,22,26,30,34,38,42,46,50,54, 60% (w/v) sucrose, each in 10 mM HEPES, pH 7.5, and 1 mM EDTA) and centrifuging at 100,000 g for 90 min. Fractions were collected and analyzed by immunoblotting or enzyme assays. Trypsinolysis experiments were carried out as described by Nakamoto et al. (27), using a trypsin to protein ratio of 1:4.
Total Protein Extracts and Immunoblotting-Total protein extracts were prepared as described by Volland et al. (44). Logarithmic-phase cells (A 600 nm ϭ 0.5) were suspended in 500 l of water and broken by the addition of 50 l of 1.85 M NaOH, 5% (v/v) ␤-mercaptoethanol, and proteins were precipitated by the addition of 50 l of 50% (m/v) trichloroacetic acid. The resulting pellets were resuspended in 10 l of 1 M Trizma (Tris base) and 20 l of Laemmli SDS-PAGE loading buffer (45), resolved on 10% polyacrylamide gels (SDS-PAGE), and transferred to a polyvinylidene difluoride membrane. HA-tagged and cMyc-tagged AT-Pases were quantitated by immunoblotting with rabbit anti-cMyc or anti-HA polyclonal antibody (Medical and Biological Laboratories, Nagoya, Japan), diluted 1/2000 and Kar2p by immunoblotting with rabbit Kar2 polyclonal antibody (provided by M. Rose, 41), diluted 1/5000, followed by incubation with 125 I-protein A and fluorography. For quantitative immunoblotting of the HA-tagged ATPase in the sucrose gradient fractions (see above), trichloroacetic acid-precipitated proteins from 100-l samples were assayed in the same manner.
Enzyme Assays-GDPase activity used as a Golgi marker was measured according to Abeijon et al. (46). NADPH cytochrome c oxidase served as an ER marker and was assayed by the method of Feldman et al. (47). Assays of ATPase hydrolysis and protein were carried out as described previously (48).
Immunoprecipitation and Detection of Ubiquitin-conjugated ATPase-Cells growing in selective synthetic complete medium containing 2% galactose were induced to express wild-type or mutant ATPase by heat shock at 39°C as described above (see "Strains, Plasmids, and Growth Conditions"). HA-tagged ATPase was immunoprecipitated from solubilized yeast membranes using a modification of the method described by Galan et al. (49). Lysis buffer (0.1 M Tris-HCl, pH 7.5, 0.15 M NaCl, 5 mM EDTA) included N-ethylmaleimide to 25 mM, phenylmethylsulfonyl fluoride to 0.5 mM, and a mixture of leupeptin, aprotinin, pepstatin, and chymostatin (each at 2 g/ml final concentration). Cell homogenates, obtained by vortexing cells with chilled glass beads, were centrifuged at 3,000 rpm in an Eppendorf microcentrifuge for 5 min at 4°C. Supernatants were collected and centrifuged at 14,000 rpm for 45 min. The membrane-enriched pellets were resuspended in 0.5 ml TNET buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) supplemented with the same protease inhibitors used above for the lysis buffer. Insoluble aggregates were removed by centrifugation at 14,000 rpm for 45 min. Solubilized membranes were pre-cleared by adding 50 l of protein A-Sepharose CL-4B (Amersham Biosciences) equilibrated at 0.125 g/ml in 25 mM Tris-HCl, pH 7.4, 20 mM NaCl, 0.01% sodium azide, and incubating on a rotator at 4°C for 90 min. The affinity matrix was removed by centrifugation at 3,000 rpm for 5 min, then each cleared supernatant received a fresh 50 l of protein A-Sepharose CL-4B slurry along with 5 l of either anti-HA (HA.11) monoclonal 16B12 or anti-cMyc monoclonal 9E10 raw ascites fluid (Berkeley Antibody Co., Richmond, CA). After overnight incubation at 4°C, the affin-ity matrix was recovered by centrifugation at 5,000 rpm and washed a total of four times, twice with TNET, once with TNET containing 0.5 M NaCl, and finally with TNET. Immunoprecipitates were recovered from the affinity matrix by adding 50 l of Laemmli SDS-PAGE loading buffer and incubating at 30°C for 15 min. Samples were separated on 8% polyacrylamide gels and immunoblotted as described above. HAtagged or ubiquitinated proteins were detected using rabbit polyclonal antibodies against the appropriate epitope followed by incubation with horseradish peroxidase-conjugated anti-rabbit antibodies. Immunoblots were developed using the ECLϩ detection system (Amersham Biosciences).
Solubility of ATPase in 1% Triton X-100 -The solubility of the ATPase in 1% Triton X-100 was assessed essentially as described by Gong and Chang (50). Cells were lysed by vortexing with glass beads in 50 mM Tris-HCl, pH 7.5, containing 0.3 M sucrose, 5 mM EDTA, 1 mM EGTA and supplemented with the same protease inhibitor mixture used for immunoprecipitation (see above). After centrifugation at 3,000 rpm for 5 min, a volume of lysate supernatant equivalent to 5 A 600 nm units of cells was extracted with an equal volume of ice-cold 2% Triton X-100 (v/v) and incubated on ice for 30 min. Samples were centrifuged at 100,000 ϫ g for 1 h, then pellets (resuspended in 1% SDS) and supernatants were analyzed by immunoblotting. HA-tagged wild-type and Pma1-G381A ATPases were detected as described above (see "Total Protein Extracts and Immunoblotting").

RESULTS
Expression and Subcellular Localization of Pma1-G381A-To track the biogenesis of Pma1-G381A, we modified an expression system developed previously by Nakamoto et al. (32) in which the wild-type PMA1 gene (which is required for growth (15)) is placed under control of the inducible GAL1 promoter, and the mutant pma1 allele is introduced under control of a heat-shock promoter (2HSE). Thus, expression can be switched from the wild-type to the mutant ATPase by transferring the cells from galactose medium at 23°C to glucose medium at 39°C. For the present study, Pma1-G381A was tagged at the N-terminal end with an HA epitope to follow the expression and subcellular localization of the mutant ATPase (25). The tagged gene was cloned in-phase with the 2HSE promoter and integrated at the HIS3 locus to give a homogeneous population of cells carrying a single copy of the mutant allele. After growth on galactose medium at 23°C, the culture was transferred to glucose medium at 23°C for 3 h to turn off expression of the wild-type gene and empty the secretory pathway of the corresponding protein. At that point (defined as time zero), the culture was incubated for 15 min at 39°C to induce a short pulse of Pma1-G381A-HA synthesis and then returned to 23°C. As a control, a parallel experiment was carried out with cells in which HA-tagged wild-type PMA1 had been integrated at the HIS3 locus, again under control of the 2HSE promoter.
To track the time course of expression under these conditions, total protein extracts were prepared at intervals and immunoblotted with anti-HA polyclonal antibody. As shown in Fig. 1A, the wild-type form was expressed maximally by 15 min of incubation at 39°C and remained fully stable for at least 240 min. Pma1-G381A was also readily detectable at 15 min and rose slightly by 30 min, but it then decayed with a half-life of ϳ60 min, as estimated by quantitative densitometry. The degradation pathway for the mutant form is discussed in a later section.
Another striking difference between the wild-type and mutant forms became apparent when confocal microscopy was used to localize the HA-tagged proteins within the cell. As shown in Fig. 1B (upper panels), the wild-type form could be detected as a diffuse signal, spread throughout the cytoplasm at 30 min after induction; most of the signal had reached the plasma membrane by 60 min and was still visible there at 180 min. Thus, the tagged wild-type ATPase appears to move smoothly through the secretory pathway, with no obvious re-tention in any intracellular compartment. By contrast, Pma1-G381A accumulated in punctate intracellular structures, which could be seen as early as 30 min and were very prominent at 60 min (Fig. 1B, lower panels). By 120 min, most of the punctate structures had disappeared, and Pma1-G381A was clearly visible at the plasma membrane. By 180 min, consistent with the results in Fig. 1A, the mutant form could barely be detected.
The results presented in Fig. 1 illustrate two clear abnormalities in expression of the G381A ATPase; they are a transient intracellular accumulation and an accelerated rate of degradation once it reached the plasma membrane. To ensure that neither abnormality was simply a consequence of the short incubation at 39°C needed to activate the 2HSE promoter, control experiments were carried out using a different promoter (GAL1), which does not permit the same time resolution but can be activated at 23°C by transferring cells to galactose as sole carbon source. Cells carrying GAL1 pr -HA-G381A on a centromeric plasmid (25) were grown on 4% raffinose and then transferred to medium containing 4% galactose to induce ex-pression of the ATPase. Once again, confocal microscopy revealed prominent intracellular punctate bodies at 60 -120 min in cells expressing G381A but not in the wild-type control. Furthermore, when the GAL1 promoter was turned off by transferring the cells to 3% glucose, immunoblot analysis of total protein extracts showed once again that the G381A-HA ATPase decayed much more rapidly than its wild-type counterpart, with little or no mutant ATPase visible by 180 min (not shown). Thus, both the transient intracellular accumulation and the rapid degradation appear to be properties inherent to the mutant polypeptide, independent of the brief heat shock used in the experiment of Fig. 1.
Pma1-G381A Is Arrested Transiently in the ER but Does Not Stimulate the Unfolded Protein Response-To pinpoint the compartment in which the mutant ATPase is transiently retained, confocal microscopy was used to compare the localization of the ATPase with that of ER and Golgi markers. At 60 min post-induction, as shown in Fig. 2, A and B, the intracellular Pma1-G381A signal was coincident with that of Kar2p (a specific ER marker (41)) but not with that of Sec7p, a late Golgi FIG. 1. Expression and subcellular localization of the wild-type and Pma1-G381A ATPases as a function of time. WG4a WT-HA and WG4a G381A-HA cells were grown to early logarithmic phase in medium containing 2% galactose at 23°C and transferred for 3 h to medium containing 2% glucose to repress expression of the untagged Pma1 ATPase. The cells were then shifted to 39°C at t ϭ 0 min to induce expression of the HA-tagged form and returned to 23°C at t ϭ 15 min. A, total protein extracts were prepared as described under "Experimental Procedures," subjected to SDS-PAGE, immunoblotted with polyclonal HA antibody, and detected by 125 I-protein A and autoradiography. B, cells were fixed with formaldehyde and processed for immunofluorescence using anti-HA monoclonal antibodies and Texas Red-conjugated goat anti-mouse IgG, as described under "Experimental Procedures." At each time point, the same field of cells was analyzed by confocal microscopy under exactly the same conditions to obtain a semi-quantitative signal. Bar, 5 m. marker (40). It was not possible to visualize two other ER membrane-associated proteins, Sec12p (a guanine nucleotide exchange factor (51)) and Lst1p (a plasma membrane H ϩ -ATPase-specific homolog of Sec24p (52)) at the same time point, presumably because protein levels were too low or the proteins were too dispersed to detect by confocal microscopy (not shown). At 15 min post-induction, however, Lst1-cMyc co-localized in small regions of punctate labeling with both wild-type and mutant forms of the ATPase (Fig. 2, C and D); given the early time at which these structures were observed, they may represent specific sites of ATPase incorporation into specialized COPII vesicles.
At later times, the morphology of the ER became severely altered during Pma1-G381A expression. In the experiment of Fig. 3A, cells producing either wild-type or mutant ATPase were compared at 60 and 120 min after induction. At both time points, cells expressing wild-type ATPase showed Kar2p labeling surrounding the nucleus, with distinct extensions to the cell periphery, consistent with the typical pattern for yeast ER (53). By contrast, in cells expressing the mutant ATPase, Kar2p first co-localized in peripheral punctate structures with Pma1-G381A (Fig. 3A, 60 min). By 120 min, when Pma1-G381A had finally reached the plasma membrane, the punctate bodies had largely disappeared, and Kar2p was again distributed in the normal ER pattern.
There is abundant evidence that Kar2p/BiP functions as a molecular chaperone during ER quality control (reviewed by Gething (54)); indeed, Kar2p/BiP is frequently overexpressed as part of the "unfolded protein response" that is triggered by   FIG. 2. Delayed export of Pma1-G381A mutant from the ER. WG4a G381A-HA cells were treated as described in the legend to Fig. 1 and fixed with formaldehyde at t ϭ 60 min following heat-shock induction. The cells were processed for double-label immunofluorescence as described under "Experimental Procedures" using antibodies to HA (left panels) combined with antiserum to either Kar2p (A) or Sec 7p (B) (middle panels) and detected by FITC-and Texas Red-conjugated secondary antibodies. To assess their COPII vesicle-packaging behavior, mutant and wild-type ATPases were tested for co-localization with Lst1p. Cells of strains BMY451 and BMY452 were treated as described in the legend of Fig. 1. Immediately before and at various time points after induction at 39°C, 5-ml aliquots were processed for confocal microscopy as described above. Panels C and D show Pma1-G381A-HA and wild-type-HA ATPases, respectively, detected 15 min post-induction using rabbit polyclonal anti-HA antibodies and Texas Red-conjugated secondary antibodies. Lst1p-cMyc was detected with rabbit polyclonal anti-cMyc antibodies and FITC-conjugated secondary antibodies. Staining of both fluorochromes was visualized simultaneously by confocal microscopy using dual channel filters, and the images were merged using Adobe Photoshop (right panels). Bars, 5 m.

FIG. 3. Subcellular localization of HA-tagged ATPase and
Kar2p/BiP as a function of time. WG4a WT-HA and WG4a G381A-HA cells were treated as described in the legend to Fig. 1. A, the cells were fixed with formaldehyde at t ϭ 60 min or 120 min after heat-shock induction, processed for double-label immunofluorescence as described in the legend to the accumulation of poorly folded proteins in the ER (55,56). To test whether Pma1-G381A stimulates the expression of Kar2p, quantitative immunoblotting was carried out on total protein extracts. As shown in Fig. 3B, there was no significant rise in the level of Kar2p in Pma1-G381A-expressing cells at 60 min (when the punctate bodies were most prominent) nor did Kar2p fall at 120 min (when the punctate bodies had disappeared). Thus, the morphological changes seen in Fig. 3A represent a redistribution of Kar2p, with no detectable change in its amount.
Proliferation of Specific Membrane Compartments in Cells Expressing Pma1-G381A-As a follow-up to confocal microscopy results (Figs. 1-3), electron microscopy was carried out to explore the nature of the abnormal structures observed in cells expressing Pma1-G381A. Fig. 4, A-E, reveals a pronounced swelling of the ER lumen at 60 min, similar to that seen when a temperature-sensitive mutation such as bet1-3 blocks the secretory pathway between the ER and Golgi (57). Even more striking were the conspicuous clusters of vesicular and tubular elements, which appeared to form a network continuous with the peripheral ER (Fig. 4, A-E). These clusters had a similar morphology in all cells expressing Pma1-G381A, but they varied in size from discrete structures located mainly at the cell periphery (seen in most cells; e.g. Fig. 4, A and C) to larger structures that invaded much of the cytoplasm (seen in 10 -30% of the cells; Fig. 4E). Immunogold staining confirmed that they contained both Pma1-G381A (Fig. 4, A-E) and Kar2p/BiP (not shown). Significantly, the clusters were much smaller and less numerous in cells expressing wild-type ATPase (Fig. 4F), and they disappeared almost completely after 2 h in cells expressing Pma1-G381A. Taken together, the results suggest that this transiently formed compartment may be the place at which Pma1-G381A exits the ER; thus, it may be equivalent to the vesicular-tubular clusters (VTCs) that constitute an ER-to-Golgi intermediate compartment in mammalian cells (58,59).
Pma1-G381A Avoids ER-associated Protein Degradation and Escapes from the ER-ER-associated protein degradation is a well characterized pathway that involves retro-translocation of misfolded proteins from the ER into the cytoplasm, multiubiquitination (1, 3), and digestion by the proteasome (2). In the case of Pma1-G381A, which can reach the plasma membrane (Fig. 1), this pathway might be considered unlikely. It seemed important, however, to examine the possibility that part of the Pma1-G381A population might be targeted directly for degradation by the proteasome after transient retention in the vesicular-tubular compartment. For this purpose, Pma1-G381A was expressed in a pre1-1, pre2-2 mutant strain that lacks the chymotrypsin-like activity of the proteasome (60). Such a strain displays enhanced sensitivity to stress conditions such as prolonged heat shock or treatment with canavanine, a toxic amino acid analog that causes the accumulation of unfolded proteins (60). In the experiment of Fig. 5A, pre1-1, pre2-2 or normal cells were grown at 23°C, pretreated with canavanine (20 g/ml) for 90 min, transferred to 39°C for 15 min to induce the expression of HA-tagged Pma1-G381A ATPase, and then returned to 23°C. At intervals, total protein extracts were prepared and assayed for HA-tagged ATPase by quantitative immunoblotting. As shown in Fig. 5A, the pre1-1, pre2-2   mutation had no detectable effect on the time course of degradation of Pma1-G381A. Moreover (not shown), there was no change in time course when Pma1-G381A was expressed in cells carrying a deletion of UBC1, which encodes one of the two major ubiquitin-conjugating enzymes of the ER-associated protein degradation pathway (4). These findings argue against a significant role for the proteasome in the degradation of Pma1-G381A.
Independent evidence came from experiments designed to detect ubiquitinated forms of wild-type or Pma1-G381A ATPase. For this purpose, HA-tagged ATPase was immunoprecipitated with anti-HA monoclonal antibodies at various time points after heat-induced expression and then analyzed for modification by immunoblotting with anti-ubiquitin polyclonal antibodies. Ubiquitinated forms of ATPase were detected only when the amounts of immunoprecipitate loaded were increased 30-fold (see Fig. 5B legend); although this result probably indicates a low level of protein ubiquitination, it may also reflect differences in the reactivities of anti-HA and anti-ubiquitin polyclonal antibodies. Such differences can be corrected for by considering ubiquitin-to-ATPase ratios. As shown in Fig. 5B, the ubiquitin-to-ATPase ratio was similar in both wild-type and Pma1-G381A ATPases, and the ratio decreased as a function of time, suggesting early ubiquitination may simply reflect a low basal level of protein misfolding. These results provide added evidence against a quantitatively significant role for ER-associated protein degradation in Pma1-G381A degradation.
Appearance of Misfolded Pma1-G381A in the Secretory Vesicles-The next step was to look directly at the folding of Pma1-G381A ATPase that has left the ER and is moving toward the plasma membrane. For this purpose, secretory vesicles were isolated from sec6-4 cells expressing either wild-type or G381A ATPase. Cells were heat-shocked for 2 h to induce ATPase expression and the accumulation of secretory vesicles, and the vesicles were isolated by differential centrifugation followed by centrifugation on a multi-step discontinuous gradient. Immunoblotting revealed a prominent peak of HAtagged ATPase in fraction 8 of the wild-type gradient and a slightly broader peak in fractions 8 -10 of the Pma1-G381A gradient (Fig. 6, A and B). Control experiments established that this peak corresponded to the secretory vesicles, since it was not seen in cells carrying a normal copy of the SEC6 gene (not shown). Wild-type ATPase was fully active in the secretory vesicles, confirming an earlier observation by Nakamoto et al. (32), but Pma1-G381A lacked measurable ATPase activity (see the legend to Fig. 6). In both cases, there was little if any contamination of the secretory vesicle fraction by ER or Golgi, for which NADPH-cytochrome c reductase and GDPase served as markers (Fig. 6, A and B).
In the experiment of Fig. 6C, fractions 7 and 8 from the wild-type gradient and fractions 8 and 9 from the Pma1-G381A gradient were incubated with a low concentration of trypsin to compare the folding of the wild-type and mutant ATPases in the secretory vesicles. As in previous studies, the wild-type ATPase was degraded slowly, with the transient appearance of 96-, 72-, and 62-kDa bands; even after 10 min, a considerable amount of the 62-kDa fragment could still be seen. By contrast, Pma1-G381A was extremely sensitive to trypsin, as reported previously for assays of total membrane fractions (25); the 96-kDa band disappeared rapidly after the 0.5-min time point, and smaller fragments were not detectable. Thus, it seems clear that Pma1-G381A is still poorly folded when it reaches the secretory vesicles.
Degradation of PMA1-G381A after Arrival at the Plasma Membrane-Is Pma1-G381A targeted to the vacuole for degradation, and does this occur before or after fusion of the secretory vesicles with the plasma membrane? We addressed these questions by asking whether the mutant ATPase is protected by temperature-sensitive blocks introduced at the latter stages of the export pathway. As shown in Fig. 7A, Pma1-G381A degradation was slowed markedly in cells bearing either the temperature-sensitive sec6-4 mutation (to interrupt fusion of the secretory vesicles with the plasma membrane (61)) or a temperature-sensitive end4 mutation (to interrupt endocytosis from the plasma membrane (29)). Furthermore, as shown in Fig. 7B, Pma1-G381A degradation was also inhibited fully in cells carrying a pep4 deletion, which eliminates the aspartyl protease (proteinase A) that is required for post-translational activation of vacuolar hydrolases (62). In parallel, confocal microscopy was used to examine the localization of Pma1-FIG. 5. Pma1-G381A is not degraded by the proteasome. A, WG4a G381A-HA (filled circles) and WG4a-11/22a G381A-HA (open circles) cells were grown to early logarithmic phase in medium containing 2% galactose at 23°C and transferred for 1.5 h to medium containing 2% glucose. Canavanine was added to the culture (20 g/ml), and incubation was continued for 1.5 h. Expression of the HA-tagged ATPase was induced by shifting the cells to 39°C for 15 min. Total protein extracts were prepared as described under "Experimental Procedures," subjected to SDS-PAGE, immunoblotted with antiserum to HA, and detected by 125 I-protein A and autoradiography. B, BMY401 (WT) and BMY402 (G381A) cells were grown as for strains described in the legend of Fig. 1. Immediately before induction of expression (t ϭ 0) and at 30-min intervals thereafter, 25-ml samples of cells were harvested and lysed, and the plasma membrane H ϩ -ATPase was immunoprecipitated from detergent-solubilized membranes as described under "Experimental Procedures." The 100-kDa HA-tagged ATPase was detected in samples equivalent to 0.5 A 600 units of cells (1ϫ) by Western blotting using rabbit anti-HA antibody. By contrast, ubiquitinated (Ub) forms of ATPase, identified using rabbit anti-ubiquitin antibody, were detected only when protein samples equivalent to 15 A 600 units of cells (30ϫ) were loaded. Blots were developed using horseradish peroxidaseconjugated anti-rabbit antibodies and the ECLϩPlus chemiluminescent detection system (Amersham Biosciences).
G381A after a 180-min incubation in pep4⌬ cells (Fig. 7C). Under these conditions, HA-tagged Pma1-G381A could be seen clearly in the plasma membrane and the vacuole, whereas in the PEP4 control strain, it had disappeared completely from the cell. As expected, HA-tagged wild-type ATPase maintained a stable association with the plasma membrane, even in the pep4⌬ strain.  Fig. 1 except that the temperature was maintained at 39°C from t ϭ 0 min. Total protein extracts were separated by SDS-PAGE, immunoblotted with antiserum to HA and detected by 125 I-protein A and autoradiography. W303-1B/D G381A-HA (filled circles) and W303-1B-pep4⌬ G381A-HA (open circles) cells were treated as described in the legend to Fig. 5. B, total protein extracts were subjected to SDS-PAGE, immunoblotted with antiserum to HA, and detected by 125 I-protein A and autoradiography. C, W303-1B-pep4⌬ G381A-HA (upper panels) and W303-1B/D G381A-HA cells (middle panels) were fixed with formaldehyde at t ϭ 180 min and processed for double-label immunofluorescence as described under "Experimental Procedures" using antiserum to HA (left panels), combined with antibodies to the 60 kDa V-ATPase subunit (middle panels), and detected by FITC-and Texas Red-conjugated secondary antibodies. As a control, W303-1B-pep4⌬ WT-HA cells (lower panels) were fixed and processed for immunofluorescence under the same conditions. Merged images are displayed in the right panels. Staining and visualization were as in the legend to Fig. 2. Bar, 5 m. Taken together, the results described up to this point indicate that the bulk of misfolded Pma1-G381A is not sent from the ER for degradation in the proteasome; rather, it progresses via the secretory vesicles to the plasma membrane, where a different quality control system dictates its removal by endocytosis and delivery to the vacuole for degradation. One way for such a system to recognize the misfolded mutant ATPase is suggested by the recent finding that wild-type Pma1 becomes associated with glycosphingolipid-and ergosterol-containing lipid rafts en route to the cell surface (63). If Pma1-G381A failed to enter lipid rafts, it could conceivably be more vulnerable to the endocytic process, as has recently been reported for another Pma1 mutant (Pma1-10 (50)).
To explore this possibility, we assayed total cell lysates for the solubility of Pma1 ATPase in 1% Triton X-100 and found that, like the wild-type ATPase, Pma1-G381A was not extracted into the supernatant under these conditions at any of the time points tested (Fig. 8). Thus, by contrast with Pma1-10, Pma1-G381A appears to be able to associate with lipid rafts, at least as measured by the standard Triton insolubility assay. Further work will be required to learn whether this association is completely normal or whether it may be altered in some way that makes the mutant ATPase a target for the endocytic machinery.
Effects of Pma1-G381A on Co-expressed Wild-type ATPase-In the study by DeWitt et al. (25), the G381A mutation was shown to behave in a co-dominant fashion, slowing but not preventing the growth of cells that were also expressing wild-type ATPase. To explore the mechanism of co-dominance in the light of the new information (described above) on Pma1-G381A biogenesis, the wild-type and mutant forms were epitope-tagged with cMyc and HA, respectively, placed under control of the 2HSE promoter, and chromosomally integrated at the HIS3 and LEU2 loci. Expression of both was induced simultaneously by means of a 15-min incubation at 39°C.
Under these conditions, the wild-type and mutant ATPases co-localized in punctate bodies at the 60-min time point, indicating that the biogenetic phenotype of the Pma1-G381A mutation was imposed on the wild-type ATPase (Fig. 9). This was not simply a consequence of ATPase overexpression, since no intracellular retention was observed in control cells expressing both cMyc-and HA-tagged wild-type ATPase under the same conditions (not shown). As described for cells expressing the mutant form alone, the locations of Pma1-G381A and wild-type ATPase changed as a function of time, progressing from the ER at 60 min to the plasma membrane at 180 min.
The clear effect of Pma1-G381A on the biogenesis of the wild-type ATPase made it important to ask whether there was also an effect on degradation. Figs. 10, A and B, present the results of an immunoblotting experiment designed to answer this question. In fact, although there was no significant degradation of the wild-type ATPase when HA-tagged and cMyctagged versions were expressed together over a 180-min time FIG. 8. Pma1-G381A is insoluble in 1% Triton X-100. Strains BMY401 (WT) and BMY402 (G381A) were grown as described in legend of Fig. 1. Immediately before induction (t ϭ 0) and at 60 and 120 min thereafter, protein lysates were made from 25-ml cell samples (see "Experimental Procedures"). After extraction with 1% Triton X-100 and centrifugation at 100,000 ϫ g, supernatant (S) and pellet (P) fractions representing 25% of total sample (i.e. 5 A 600 units cells) were immunoblotted to detect HA-tagged WT or Pma1-G381A ATPase using polyclonal anti-HA antiserum.
FIG. 9. Co-localization of the Pma1-G381A mutant with the wild-type ATPase. WG4a G381A-HA/WT-cMyc cells were treated as described in the legend to Fig. 1. Co-expression of the cMyc-tagged wild-type form (WT-cMyc) and the HA-tagged Pma1-G381A mutant form (G381A-HA) was induced by shifting the cells to 39°C for 15 min. The cells were fixed with formaldehyde and processed for double-label immunofluorescence as described under "Experimental Procedures" using anti-HA (top panels) and anti-cMyc (middle panels) and detected by Texas Red-and FITCconjugated secondary antibodies, respectively. Staining and visualization were as in the legend to Fig. 2. Bar, 5 m. course (panel A), cMyc-tagged wild-type ATPase was significantly degraded when co-expressed with HA-tagged Pma1-G381A (panel B). Quantitative densitometry gave rate constants of 60 min Ϫ1 for both forms, essentially identical to the rate constant for Pma1-G381A when expressed alone (see Fig. 1).
Once again, the co-degradation of the Pma1-G381A and wildtype forms required delivery to the plasma membrane followed by endocytosis. Degradation was fully inhibited when both forms were expressed in a sec6-4 background to block fusion of the secretory vesicles with the plasma membrane (Fig. 10C) or when endocytosis was blocked by the addition of 4% ethanol to the culture before induction (Fig. 10D; Ref. 64). Thus, just as it affects the biogenesis of the wild-type ATPase, Pma1-G381A can also impose its degradation phenotype on the wild-type form.

DISCUSSION
In this study, a misfolded mutant of the Pma1 H ϩ -ATPase has been used to explore quality control mechanisms in the yeast secretory pathway. The mutant protein, Pma1-G381A, behaves in a highly distinctive way at two different points along the pathway; first, at the ER-to-Golgi transition, where it is retained transiently in Kar2p-containing vesicular and tubular bodies but avoids any significant level of degradation, and later, at the plasma membrane, where it is removed rapidly by endocytosis and delivered to the vacuole for degradation (see Fig. 11). Both steps provide insights into the functioning of the secretory pathway and will be discussed in turn.
Transient Retention of Pma1-G381A in the ER-At the outset of this work, there was already abundant evidence that the yeast ER can proliferate abnormally when membrane trafficking is "jammed," either by excessive amounts of a normal cargo protein (e.g. Refs. 66 -68) or by synthesis of a defective (24 -27) or heterologous (69 -78) cargo protein that fails to escape from the ER. ER-derived structures also accumulate when ER-to-Golgi transport is blocked by overexpressing Sec12p (79). Depending upon the specific experimental conditions, the structures described by other authors have ranged from stacks of membranes closely associated with the nuclear envelope (termed "karmellae" by Wright et al. (66)) to "sausage-like" bodies (67,77,78) and "BiP bodies" (79) that are more widely dispersed in the cytoplasm. Karmellae probably arise from a block in trafficking at a relatively early point in the secretory pathway, whereas sausage-like structures and BiP bodies may reflect a later block at or near the point of exit from the ER (79).
The results of the present study reinforce the position that BiP bodies may be associated with the exit process. Most of the cells (70 -90%) expressing the Pma1-G381A mutant ATPaseformed punctate, Kar2p-containing structures that were shown by electron microscopy to consist of vesicular and tubular elements, derived from (and sometimes still physically connected to) the peripheral ER. Significantly, these structures formed only transiently; when Pma1-G381A escaped from the ER and moved toward the plasma membrane, the ER recovered a characteristic ring-like shape around the nucleus. It seems likely that these ER-derived BiP bodies are analogous to the mammalian cell VTCs described by Balch and co-workers (58,59), which comprise a dynamic ER-Golgi intermediate compartment (ERGIC) that is visible even under normal conditions (80). By contrast, the ability to see VTCs in yeast required perturbation of the secretory pathway by expression of Pma1-G381A. Under normal circumstances, there is no obvious spatial differentiation of the secretory pathway in S. cerevisiae; instead, transitional vesicles bud throughout the ER, and Golgi cisternae are scattered throughout the cytoplasm (81).
Recent evidence suggests that selective processes are oper-ating during export from the ER and that at least some cargo proteins, including Pma1p (52) and the glycophosphatidylinositol-anchored protein Gas1 (82), are incorporated into distinct subsets of COPII transport vesicles. Thus, the proliferative regions of peripheral ER defined by the delayed exit of Pma1-G381A may reflect the presence of an export subdomain. The brief but highly localized coincidence of both WT and mutant ATPases with Lst1p at only 15 min after induction and before accumulation of Pma1-G381A in BIP bodies may represent exit sites that are utilized preferentially by the plasma membrane H ϩ -ATPase for loading into COPII vesicles (Fig. 2, C and D).
Because the mutant ATPase showed an Lst1p labeling pattern and time course very similar to that of the wild-type protein, it seems unlikely that Pma1-G381A is grossly aberrant in its vesicle-packaging behavior. Furthermore, the apparently normal association of Pma1-G381A with lipid rafts (Fig. 8) suggests that its subsequent passage to the plasma membrane is comparable with that of the wild-type ATPase.
At the molecular level, it is not surprising that Pma1-G381A is recognized in the ER as an incorrectly folded protein, since in vitro experiments show it to be cleaved by trypsin under conditions that leave wild-type H ϩ -ATPase intact (25). However, the structural defect in Pma1-G381A is less severe than in many other misfolded H ϩ -ATPase mutants, which are degraded almost instantaneously by trypsin under the same conditions (25,27). The intermediate nature of the defect may explain why Pma1-G381A causes VTCs to proliferate and Kar2p/BiP to be redistributed without triggering a typical unfolded protein response. In particular, if Pma1-G381A is not sufficiently unfolded to bind Kar2p/BiP, it may not bring about the decrease in free lumenal concentration of Kar2p/BiP that is believed to be necessary for a fully fledged unfolded protein response (reviewed in Refs. 6 and 55). This, in turn, may underlie the eventual escape of Pma1-G381A from the quality control machinery of the ER.
Retrieval of Pma1-G381A from the Plasma Membrane and Degradation in the Vacuole-At the plasma membrane, Pma1-G381A falls prey to another as yet uncharacterized quality control system that recognizes the mutant polypeptide as abnormal and sends it to the vacuole for degradation. Evidence for the re-direction of Pma1-G381A to the vacuole via the plasma membrane comes from the clear-cut protection conferred upon Pma1-G381A by mutations blocking secretory vesicle fusion with the plasma membrane (sec6), endocytosis (end4), or vacuolar aspartyl protease activity (pep4⌬). The end result is a much reduced half-life of the Pma1-G381A mutant protein (ca.1 h) compared with the wild-type Pma1 ATPase (Ͼ11 h; Ref. 83) at the plasma membrane. Thus, once delivered to the cell surface, Pma1-G381A follows the pathway of shortlived plasma-membrane proteins that are down-regulated by endocytosis (reviewed by Hicke (11,12)).
Work in other laboratories has shown that the wild-type Pma1 ATPase can be artificially targeted to the vacuole in at least two ways; they are by adding either the PEST-like sequence of Ste3p, the yeast a-factor receptor (84), or the linker region of the ABC-transporter Ste6p (85). In both cases, ubiquitination has been suggested to play a role in the degradative pathway, since prominent higher molecular weight ATPase bands could be seen upon immunoblotting and increased in amount when endocytosis was blocked in an end4 strain.
By contrast, ubiquitination appears not to be responsible for the endocytosis and vacuolar degradation of the Pma1-G381A ATPase. Ubiquitinated forms were difficult to detect in the present study; they appeared at a very low level almost immediately after induction and then decreased over time. Furthermore, there was no appreciable difference in the amount or time course of ubiquitination between Pma1-G381A and the wild-type control. Thus, it seems likely that the modification by ubiquitin occurs before or during transit of the ATPase from ER-to-Golgi rather than at the plasma membrane and that it reflects a low basal level of unfolding seen in a minor fraction of newly synthesized ATPase. Small amounts of ubiquitinated wild-type Pma1p were observed in the earlier study by Kolling and Losko (85) and may have the same explanation.
Gong and Chang (50) recently described a temperature-sensitive H ϩ -ATPase mutant, Pma1-10, which was degraded after less than 30 min at the plasma membrane. Like Pma1-G381A, Pma1-10 can be protected by expression in sec6-4 or end4-1 cells, indicating that it is removed from the plasma membrane by endocytosis before degradation. Pma1-10 is soluble in Triton X-100 and, thus, appears unable to associate successfully with lipid rafts; the authors have proposed that this defect may underlie its instability at the plasma membrane. By contrast, Pma1-G381A is insoluble in 1% Triton X-100 (Fig. 8), suggesting that another mechanism must exist to discriminate between wild-type and G381A ATPases at the plasma membrane. This ultimate quality control step may involve interactions with other proteins that serve to stabilize Pma1p at the cell surface in a manner analogous to that by which spectrin stabilizes the Na,K-ATPase of epithelial cells (86,87). Mop2p (End4p/Sla2p) has been shown previously to regulate the amount of wild-type ATPase at the plasma membrane; indeed, Mop2p overexpression can even allow for overexpression of an otherwise toxic level of Pma1p (88). Because the N-terminal and central coiled-coil domains of Mop2p/End4p are required for endocytosis and interaction with actin-associated proteins (89), it seems plausible that Mop2p could play a dual role in facilitating both the insertion and the removal of Pma1p from the yeast plasma membrane.
Effect of Pma1-G381A on the Biogenesis and Degradation of Wild-type Pma1 ATPase-Finally, this work has shown that Pma1-G381A severely affects the biogenesis of co-expressed wild-type ATPase, causing the wild-type protein to be retained transiently in the ER and then to be degraded rapidly after delivery to the plasma membrane. Given the exact coincidence between the behavior of the mutant and wild-type polypeptides at two different checkpoints, the most likely explanation is that they travel together as mixed oligomers. Indeed, the closely related H ϩ -ATPase of Neurospora crassa has been suggested to be dimeric based on both kinetic analysis (90) and radiation inactivation measurements (91), and hexamers have been seen by cryoelectron microscopy after detergent treatment (17). Future experiments will look directly for oligomerization of mutant and wild-type yeast H ϩ -ATPases using the epitope-tagged constructs. In the meantime, the results of the present study provide a clear explanation for the fact that G381A behaves genetically as a co-dominant mutation, slowing but not completely inhibiting growth when co-expressed with the wild-type PMA1 allele (25); presumably it does so by reducing the rate at which functional ATPase is delivered to the cell surface and by accelerating degradation at the plasma membrane.