Use of modular substrates demonstrates mechanistic diversity and reveals differences in chaperone requirement of ERAD.

The endoplasmic reticulum (ER) harbors a protein quality control system, which monitors protein folding in the ER. Elimination of malfolded proteins is an important function of this protein quality control. Earlier studies with various soluble and transmembrane ER-associated degradation (ERAD) substrates revealed differences in the ER degradation machinery used. To unravel the nature of these differences we generated two type I membrane ERAD substrates carrying malfolded carboxypeptidase yscY (CPY*) as the ER-luminal ERAD recognition motif. Whereas the first, CT* (CPY*-TM), has no cytoplasmic domain, the second, CTG*, has the green fluorescent protein present in the cytosol. Together with CPY*, these three substrates represent topologically diverse malfolded proteins, degraded via ERAD. Our data show that degradation of all three proteins is dependent on the ubiquitin-proteasome system involving the ubiquitin-protein ligase complex Der3/Hrd1p-Hrd3p, the ubiquitin conjugating enzymes Ubc1p and Ubc7p, as well as the AAA-ATPase complex Cdc48-Ufd1-Npl4 and the 26S proteasome. In contrast to soluble CPY*, degradation of the membrane proteins CT* and CTG* does not require the ER proteins Kar2p (BiP) and Der1p. Instead, CTG* degradation requires cytosolic Hsp70, Hsp40, and Hsp104p chaperones.

The eukaryotic endoplasmic reticulum (ER) 1 harbors a highly efficient quality control system, which ensures proper folding of newly synthesized soluble and membrane-bound secretory proteins. Because failure in protein folding would lead to impaired biological function and, eventually, disease and cell death, terminally misfolded or inappropriately modified proteins are removed from the ER. This process is generally known as ER-associated degradation (ERAD) or simply ER degradation (1)(2)(3). Substrates recognized by the ER quality control system are translocated back into the cytoplasm through the Sec61 translocon (4 -8), where they become polyubiquitinated via the concerted action of the ubiquitin-conjugating enzymes (E2) Ubc1p, Ubc6p, and Ubc7p, and the ubiquitin-protein ligase (E3) complex Der3/Hrd1p-Hrd3p. Degradation is then carried out by the 26S proteasome (9 -14). Recently, the AAA-ATPase Cdc48p (p97) and its interaction partners Ufd1p and Npl4p have also been involved in the degradation process (13,(15)(16)(17)(18). Furthermore, depending on the substrate examined, additional cellular components are required for the dislocation/ degradation process. For example, BiP (Kar2p in yeast), the major lumenal Hsp70 chaperone of the ER, is crucial for the degradation of soluble CPY* (4), ␣-factor and mammalian ␣1protease inhibitor (19), but is not involved in the elimination of the membrane proteins Pdr5* (20), Vph1p (21), and Sec61-2p (22). Interestingly, cytoplasmic Hsp70 chaperones of the Ssa family are necessary for the removal of the polytopic membrane proteins CFTR (23) and Vph1p (21). However, they are dispensable for the elimination of the mutant Sec61-2 protein (23) as well as for the degradation of soluble proteins like mutant ␣-factor and mammalian ␣1-proteinase inhibitor (19). Der1p, an ER membrane protein, important for the degradation of soluble CPY*, PrA*, and KHN t (24,25) is not required for the membrane proteins Pdr5* (20) and Vph1p (21).
In this study, we introduce a set of structurally different misfolded proteins sharing the same ER-degradation signal. Data obtained by these topologically different substrates allow us to correlate the variations in the ERAD components involved and let us understand the different results concerning chaperone and Der1p requirements. CPY* was chosen as the ER-degradation recognition motif for all substrates. CPY* itself is a malfolded protein, fully glycosylated and completely translocated into the ER (26 -28). It has become the best studied ERAD model substrate in Saccharomyces cerevisae (reviewed in Refs. 29 and 30). We constructed two topologically distinct modular membrane proteins based on CPY*. The first is CT*, in which CPY* is inserted into the ER membrane via a single transmembrane domain. In the second (CTG*), the green fluorescent protein (GFP) is fused to CT* providing a cytoplasmic domain. Together with CPY*, these new ERAD substrates give us information concerning substrate-specific requirements of the ERAD machinery used. In the present study, we were able to identify a basic degradation machinery for elimination of soluble and membrane proteins with misfolded ER-lumenal domains. It consists of the ubiqitin-protein ligase complex Der3/Hrd1p-Hrd3p, the ubiquitin-conjugating enzymes Ubc1p and Ubc7p, the Cdc48-Ufd1-Npl4 complex, and the proteasome. Furthermore, we clarified the function of ER-lumenal and cytosolic chaperones with respect to their topological requirements.

EXPERIMENTAL PROCEDURES
Construction and Growth Conditions of Strains-Genetic experiments, media preparation, and methods employing molecular biology were carried out using standard methods (31,32). S. cerevisiae strains used in this study are summarized in Table I. Cells were grown at 25°C (temperature-sensitive strains) or at 30°C in synthetic complete media.
The ⌬ubc1 ⌬ubc7 strain YCT519 was derived from crossing and tetrad dissection of the respective single mutants. The ⌬ubc1 knockout strain was generated in W303-1C background using the UBC1-deletion plasmid described previously (33). Deletion of UBC1 was confirmed by Southern blotting. The ⌬ubc7 strain was described previously (9). Strain YCT549 (cdc48 Y834E ) was generated according to the gene truncation technique (34), except that CDC48 was not truncated, but mutated using the oligonucleotides: CDC48EF3, TGCATTTGGTTCTAAT-GCGGAGGAAGATGATGATTTGGAGAGTTGAGGCGCGCCACTTCT AAA and CDC48R1, TGAATTTACGATTTAAAATAAAAATATACCTG-GCATATAAGAATTCGAGCTCGTTTAAAC. Successful introduction of the point mutation was confirmed by sequencing.
CT* and CTG* were expressed from plasmids pCT67 and pMA1, respectively. Plasmid pMA1 was cloned in a stepwise process: the last transmembrane domain (TM) of Pdr5p was PCR amplified from plasmid pCKSF1 (41) using the primer set HpaI-XhoI-TM (AACGAATGG-ATCCACGGTGGTTTCTCCTTACTCGAGGATGAAAATGCCACTGAC) and TM-PacI (CCTTAATTAATTTCTTGGAGAGTTTACC). The PCR product was ligated into the HpaI site of pRS306prc1-1, resulting in plasmid pRS306-CPY*-TM-PacI. The TDH3 promoter was cloned into plasmid pFA6a-GFP(S65T)-His3MX6 (34) as a 700-bp HindIII-BamHI fragment derived from pHGpd/p82 (40), yielding pFA6-GPD-GFP. A 1.8-kb BstZ17I-PacI fragment corresponding to prc1-1-TM was obtained from plasmid pRS306-CPY*-TM-PacI and was ligated into pFA6-GPD-GFP digested with SmaI and PacI (which cut between the TDH3 promoter and GFP). The resulting plasmid pGPD-Y*-TM-GFP (P TDH3 ::prc1-1::pdr5 4332-4532 ::GFP) was digested with HindIII and EcoRI and the 4.7-kb fragment, containing P TDH3 ::prc1-1::pdr5 4332-4532 ::GFP, was cloned into the same sites of pRS316, yielding pMA1. Plasmid pCT67 was obtained by in vivo recombination in yeast, replacing the GFP coding sequence with a stop codon. The GFP sequence was removed from pMA1 by digestion with PacI and AscI. The linearized plasmid was transformed into yeast together with a double stranded oligonucleotide linker (Prc1tmtrunc1, CCTAAAAAGAACGGTAAACTCTCCAAGAAATAAGG-CGCGCCACTTCTAAATAAGCGAATTTCTTATGA, and Prc1tmtrunc2, TCATAAGAAATTCGCTTATTTAGAAGTGGCGCGCCTTATTTCTTGG-AGAGTTTACCGTTCTTTTTAGG) containing the stop codon following the transmembrane domain and ends complementary to either end of the linearized plasmid.  Cell Labeling and Immunoprecipitation-CPY* pulse-chase experiments were performed as described previously (42). Temperature-sensitive strains were grown at 25°C and shifted to restrictive temperature for the times indicated in the figure legends.
Cycloheximide Decay Experiment-Cells were grown to log phase in synthetic complete medium. Temperature-sensitive strains were then shifted to restrictive conditions for 1 h. Cycloheximide was added (0.25 mg/ml) and 2 A 600 of cells were taken at the indicated time points. Cell extracts were prepared by alkaline lysis (9) and subjected to SDS-PAGE followed by immunodetection as described in Ref. 24.

RESULTS
Recent studies have revealed that degradation of soluble and membrane-bound ERAD substrates involve different components of the ERAD machinery (29). To study these different degradation mechanisms used by ERAD substrates, we constructed a set of misfolded proteins sharing CPY* as the degradation motif. Therefore, we fused a transmembrane domain or a GFP-tagged transmembrane domain to the C terminus of CPY*. We named the fusion proteins CT* (CPY*-transmembrane domain) and CTG* (CPY*-transmembrane domain-GFP). Schematic drawings of these proteins are shown in Fig. 1.
First, we characterized the localization and topology of the fusion proteins. Membrane insertion of CT* and CTG* was determined with solubilization experiments. Crude cell extracts were treated with urea, potassium acetate, or sodium carbonate, all known to remove peripheral membrane proteins. FIG. 2. CTG* and CT* are glycosylated integral membrane proteins of the ER. A, crude extracts from CTG* expressing WT (W303-1C) cells were treated with buffer, or with buffer containing either 2.5 M urea, 0.8 M potassium acetate, 0.1 M sodium carbonate, pH 11.6, 1% SDS or 1% Triton X-100 followed by centrifugation at 20,000 ϫ g. Soluble (S) and pellet (P) fractions were analyzed by immunoblotting using CPY and Sec61 antibodies. Treatment with EndoH was done for 1 h at 37°C after immunoprecipitation of CTG* with CPY antibody. B, protease treatment of CTG* expressing WT (W303-1C) cells. Trypsin (0.5 mg/ml) and Triton X-100 were added as indicated. Immunoblots were analyzed either with CPY or GFP antibodies. The positions of CTG* and a proteolytic fragment (arrow) are indicated. C, colocalization of CTG* and the ER-membrane protein Sec61p was performed with W303 ⌬C cells expressing CTG*. GFP fluorescence shows CTG* localization. Sec61p is visualized by indirect immunofluorescence using anti-Sec61 primary and Cy3-conjugated secondary antibodies (Nom: Nomarski optics, 4Ј,6Ј-diamidino-2-phenylindole (DAPI): nuclear staining). D, membrane insertion of CT* was determined as described in A. Soluble and pellet fractions of WT (W303-1C) and WT expressing plasmid encoded CT* were analyzed by immunoblotting using CPY and Sec61 antibodies. E, indirect immunofluorescence of W303 ⌬C cells expressing CT* shows colocalization of CT* and Sec61p. Proteins were visualized using an anti-CPY/Alexa-conjugated anti-mouse and an anti-Sec61/fluorescein isothiocyanate-conjugated anti-rabbit antibody sandwich, respectively. But, CT* and CTG* could only be solubilized after treatment with detergents like Triton X-100 or SDS (Fig. 2, A and D). Next, we checked whether CTG* and CT* are glycosylated, because glycosylation of CPY* is important for its degradation (44 -47). Therefore, we treated CTG* and CT* with endoglucosidase H, which removes N-linked sugar chains. Both molecules showed a decrease in molecular mass, indicating that the N-glycosylation sites of the CPY* moiety of the fusion proteins were recognized by the ER-lumenal glycosylation machinery (Fig. 2, A and D). In addition, protease protection assays with intact microsomes showed that the CPY* moiety of CTG* and CT* were protected from trypsin digestion ( Fig. 2B and not shown). In the case of CTG*, the molecular mass decreased after protease treatment and the protein was not recognized in Western blots by the anti-GFP antibody any more (Fig. 2B). This proves that the GFP domain is located in the cytosol and, therefore, is protease accessible. Furthermore, fluorescence microscopy showed colocalization of CTG*, CT*, and the ERmembrane protein Sec61p (Fig. 2, C and E). In summary, these experiments show that CTG* and CT* are integral type I membrane proteins with their glycosylated N-terminal CPY* moiety located in the ER lumen and their C terminus in the cytosol as depicted in Fig. 1.
Next, we measured the turnover of CT* and CTG* in wild type cells by cycloheximide decay and pulse-chase experiments. We found that both proteins are unstable with half-lives of about 20 (CT*) and 40 min (CTG*) at 30°C (see Figs. 3 and 4). As CTG* has two topologically diverse domains, one residing in the ER-lumen and the other in the cytoplasm, we tested whether both domains were degraded concomitantly. Pulsechase analysis using antibodies recognizing either CPY* or GFP showed that both protein domains were degraded with similar kinetics in wild type cells, indicating that the fusion protein is degraded as a single entity (not shown).
Because both CT* and CTG* contain misfolded CPY* as the degradation motif, we investigated whether they are degraded via ERAD. Cycloheximide decay experiments showed that degradation of CTG* and CT* was retarded in proteasomal mutants defective in subunits of the 20S core particle (Fig. 3, A  and B). Degradation of both substrate proteins was nearly abolished in a Rpt6 (cim3-1) mutant of the 19S regulatory particle of the proteasome (Fig. 3, C and D). Moreover, deletion of the ubiquitin-conjugating enzymes Ubc1p and Ubc7p, known to be crucial for ERAD of CPY*, led to impaired degradation of CTG* and CT* (Fig. 3, E and F). Next, we tested the requirement of the ubiquitin-protein ligase Der3/Hrd1p (12) by pulsechase analysis. As shown in Fig. 4, A and B, Der3/Hrd1p is indeed involved in the degradation of CTG* and CT*. Finally, we could also show that Hrd3p, a protein that interacts with and stabilizes Der3/Hrd1p (6,48), is also necessary for the degradation of CTG* and CT* (not shown). In summary, these experiments demonstrate that CTG* and CT* share the same basic ERAD machinery as CPY*: degradation is carried out by the 26S proteasome following polyubiquitination by the ubiquitin-conjugating enzymes Ubc1p and Ubc7p and the ubiquitin-protein ligase Der3/Hrd1p (Figs. 3 and 4).
Another protein involved in ERAD is Der1p. Up till now, it was found only to be required for the degradation of a subset of proteins like soluble CPY*, PrA* (24), and KHN t (25), but not for membrane-bound proteins like Pdr5* (20). From this, one could hypothesize that Der1p is either involved in the recognition of a specific subset of misfolded proteins or is solely involved in the degradation of soluble substrates. In fact, as shown in Fig. 4, A and B, CTG* and CT* degradation kinetics were not changed in ⌬der1 cells. This result shows that Der1p is not specifically involved in recognition of the misfolded CPY* moiety. Instead, one may conclude that it is required only for the degradation of soluble ERAD substrates like CPY*, PrA*, and KHN t (24,25).
Because CTG* and CT* are still degraded to some extent in Der3/Hrd1p-deleted cells (Fig. 4), we reasoned that this might be because of the action of an additional E3 enzyme. Therefore, we also tested the involvement of the ubiquitin-protein ligase Ssm4/Doa10p in the degradation of both proteins. Ssm4/ Doa10p has been shown previously to be an E3 enzyme necessary for the ERAD of Ubc6p (49). However, half-lives of CT* and CTG* were not altered in ⌬ssm4/doa10 or in ⌬ssm4/doa10 ⌬der3/hrd1 cells (not shown), indicating that other components may be involved in the degradation process of these proteins.
Next, we investigated whether the Cdc48-Ufd1-Npl4 complex, which was described to be necessary for ERAD of soluble and membrane proteins (13,(15)(16)(17)(18), also acts in the degradation of CTG* and CT*. Therefore, we determined the half-life of both proteins in cdc48 Y834E cells, where Cdc48 Y834E p is perma- Cycloheximide decay experiments were performed in WT, proteasomal mutant, and ⌬ubc1 ⌬ubc7 cells at 30°C. Cycloheximide was added (t ϭ 0 min), samples were collected at the indicated time points and subjected to SDS-PAGE, followed by immunoblotting. Immunoblots were analyzed with CPY antibody. Either Sec61p or Kar2p were detected using Sec61 or Kar2 antibodies, respectively, to show equal protein levels in each lane. Degradation of CTG* and CT*: A and B, in WT (WCGY4a) and proteasomal mutant (pre1-1 pre4-1) cells; C and D, in WT (YPH499Y) and proteasomal mutant (cim3-1) cells; and E and F, in WT (W303-1C) and ⌬ubc1 ⌬ubc7 cells.
nently localized to the nucleus (50). Pulse-chase experiments showed that degradation of CTG* and CT* was delayed to a considerable degree in this mutant (Fig. 5, A and B). Likewise, cycloheximide decay experiments with ufd1-1 and temperature-sensitive npl4-1 cells showed that degradation of CTG* and CT* was also affected in these mutants (Fig. 5, C-F). These findings indicate that action of the Cdc48-Ufd1-Npl4 AAA-ATPase complex in ERAD is independent of substrate topology.
One may argue that the membrane anchor in CT* and CTG* might represent a degradation motif by itself, which is recognized by the ER quality control system, independently of the point mutation in the CPY* moiety of the proteins. Therefore, we generated fusion molecules where the mutant CPY* moiety was exchanged against wild type CPY denoted as CT and CTG. First, we tested CPY activity of the fusion proteins in an in vitro assay (51). Indeed, CT and CTG were able to cleave the CPY-specific substrate N-benzoyl-L-tyrosine p-nitroanilide, whereas CPY*, CT*, and CTG* were unable to do so (not shown). Furthermore, Western blotting revealed processing of CT and CTG yielding proteins with a molecular mass similar to that of matured CPY, whereas no processing of CPY*, CT*, and CTG* could be observed (not shown). This suggests that CT and CTG reach the vacuole where they are processed to mature CPY. Therefore, we conclude that the transmembrane domain does not interfere with the folding of CPY and does not act as an ER-degradation motif by itself. Thus, only the CPY* moiety of CT* and CTG* renders the fusion proteins unstable.
We were further interested in the role played by the ERlumenal chaperone Kar2p. Kar2p is a member of the highly conserved Hsp70 family, involved in protein import into the ER (52). Additionally Kar2p activity is necessary for the degradation of misfolded, soluble proteins like CPY* or mutant ␣-factor (4,19). In contrast, the chaperone is not involved in degradation of polytopic membrane proteins like Pdr5* or CFTR (20,23). Kar2p activity is believed to keep CPY* in a soluble form, to make its dislocation into the cytosol possible (22). Additionally, Kar2p may be involved in the recognition of the misfolding in CPY*. In this case, degradation of CT* and CTG* should also be dependent on Kar2p. But, in contrast to CPY*, the degradation kinetics of CT* and CTG* were not changed in kar2-159 temperature-sensitive cells at restrictive conditions (Fig. 6). No degradation intermediates because of partially clipped CT* or CTG* proteins were detected (Fig. 6, B and C). These experiments indicate that Kar2p activity is important for the degradation of soluble proteins, but that it is not involved in the recognition of the unfolded state of CPY*.
Recently, several groups reported the dependence of CPY* degradation on ER to Golgi protein transport (25, 42, 53), whereas the degradation of the membrane protein Sec61-2p was unaffected (10,25). To clarify the question whether the degradation defect of CPY* in transport mutants is because of a specific misfolding in CPY* or caused by CPY* being a soluble protein, we tested CTG* degradation in a strain carrying the ufe1-1 allele. This mutant is defective in ER to Golgi transport (54) and degradation of CPY* is impaired (42). In contrast to CPY*, we found that CTG* degradation is not affected in ufe1-1 cells at restrictive conditions (not shown). This shows that the degradation defect in ER to Golgi transport mutants is simply because of CPY* being a soluble protein. One plausible explanation of these findings is that Kar2p is mislocalized in these ER-Golgi transport mutants (55) and, therefore, indirectly affects solely the degradation of soluble CPY*.
Previously, it was believed that the ATPases of the proteasomal 19S cap might support unfolding of misfolded ER proteins before degradation. However, it was recently discovered that turnover of membrane proteins like CFTR or the variant CFTR⌬F508 requires cytosolic Hsp70 chaperones (23,56,57). Hsp70s act in a system composed of molecular chaperones including Hsp40, Hsp90, and other associated proteins. It was found that biogenesis of CFTR depends also on the Hsp90 chaperone family (56,58). Together with the Hsp70 chaperones and carboxyl terminus of Hsc70-interacting protein they target misfolded CFTR⌬F508 for proteasomal degradation (57,59). Additionally, it was shown that Hdj-1, a mammalian member of the Hsp40 family, is involved in the turnover of CFTR in vivo (60). In yeast it has been shown that degradation of the membrane proteins Vph1p, Ste6p*, and of heterologously expressed CFTR also depends on Hsp70 chaperones of the cytosolic Ssa family (21,23). In contrast, only a minor effect on the degradation of Sec61-2p was observed using ssa mutants (23). From these observations the authors proposed that Hsp70 action is needed for the degradation of proteins exhibiting prominent cytoplasmic domains. These studies depend on very different types of malfolded proteins. Therefore, the experiments cannot unequivocally distinguish which of the proteins' domains require additional chaperone activity for degradation. We, therefore, tested the degradation of our topologically well defined substrates CPY*, CT*, and CTG* with respect to cytosolic chaperone requirements. The cytosolic Hsp70 Ssa family in yeast consists of four highly homologous proteins, Ssa1p, Ssa2p, Ssa3p, and Ssa4p. They are known to be involved in protein import into the ER and mitochondria (38). We measured degradation of CPY*, CT*, and CTG* in mutants lacking three Ssa family members and the fourth being either wild type or a temperature-sensitive allele of SSA1 (ssa1-45). Pulsechase analysis showed that degradation of CPY* and CT* was not affected at restrictive conditions, whereas the half-life of CTG* was dramatically prolonged (Fig. 7, A and B). No intermediate products resulting from partial proteolysis of the GFP domain were observed, indicating that CTG* is degraded as a whole entity (Fig. 7B). A strain carrying all four wild type members of the Ssa family (SSA1, SSA2, SSA3, and SSA4) is indistinguishable from an SSA1 (SSA1, ⌬ssa2, ⌬ssa3, ⌬ssa4) strain with respect to CPY*, CT*, or CTG* degradation (not shown). The Ssa family of chaperones is also involved in posttranslational import of some proteins into the ER (38). Therefore, we targeted CTG* for cotranslational import by exchanging the import sequence of CPY with the invertase import sequence (61). As expected, degradation of this protein was still dependent on Ssa1p activity (not shown). This result indicates that the degradation defect observed in ssa1-45 mutant cells is not because of impaired import of CTG* into the ER. Additionally, we found that the requirement of chaperones for proteasomal degradation of the tightly folded GFP domain is not restricted to ERAD substrates only. Also, degradation of Deg1-GFP, a cytosolic substrate of the proteasome, is decreased in ssa1-45 mutant cells (not shown).
Hsp70 chaperones function in a complex with co-chaperones of the Hsp40 family, which modulate the substrate specificity of the Hsp70s (62-64). Hsp40s bind and activate Hsp70 proteins with their N-terminal DnaJ domain (65,66). Yeast contain around 20 proteins with a DnaJ domain motif (YPD TM data base (67)). Fig. 8 shows the alignment of the DnaJ domains of Escherichia coli DnaJ, human Hdj-1, and the S. cerevisiae proteins Hlj1, Ydj1, Ypr061c, and Cwc23. Amino acid residues FIG. 5. The Cdc48-Ufd1-Npl4 complex is necessary for the degradation of CTG* and CT*. A and B, CTG* and CT* turnover in WT (W303-1C) and cdc48 Y834E cells. Pulse-chase analysis was performed as described in the legend to Fig. 4. C and D, CTG* and CT* degradation in WT (YCT397) and ufd1-1 cells.
Cycloheximide decay experiments were done as described in the legend to Fig. 3. E and F, turnover of CTG* and CT* in WT (FY23) and npl4-1 cells. Cycloheximide decay analysis was performed as described in the legend to Fig. 3, except that the cells were shifted to 37°C 1 h prior the addition of cycloheximide.
important for the J domain (65,66) are conserved in all given sequences (Fig. 8). One family member, Ydj1p, had been analyzed for its involvement in CFTR degradation, but was found to be dispensable (23). We measured CPY*, CT*, and CTG* degradation in cells carrying the temperature-sensitive allele ydj1-151. At restrictive conditions, no alteration of the degradation kinetics of the three ERAD substrates was observed (not shown). Furthermore, we performed cycloheximide decay experiments with yeast strains lacking other DnaJ proteins to identify Hsp40s, which are involved in CTG* degradation (see "Experimental Procedures"). We observed decreased turnover of CTG* in mutants lacking Hlj1p, Cwc23p, or Ypr061c proteins (not shown). As Ypr061c was a previously uncharacterized protein, we named it Jid1p (DnaJ protein involved in ER-associated degradation). Subsequently, these strains were studied in pulse-chase analysis. Degradation of CTG* was indeed affected in ⌬hlj1, ⌬cwc23, and ⌬jid1 mutant cells (Fig. 9). The double mutants ⌬cwc23 ⌬hlj1 and ⌬cwc23 ⌬jid1 showed no additional defect (not shown). We were unable to obtain a viable triple mutant. Our data suggest that J domain proteins, of previously unknown cellular function, are involved in degradation of CTG*. We also tested whether components of the Hsp90 chaperone system are involved in degradation of ERAD substrates. The yeast Hsp90 family consists of two proteins, Hsc82p and Hsp82p (40). Hsp90s are associated with the cochaperones Sba1p and Sti1p (68). We tested turnover of CPY* in ⌬hsc82 hsp82 D170G , ⌬sba1Ϫ1, and ⌬sti1Ϫ1 cells. Degradation of CPY* was not affected in any of these mutants (not shown). Additionally, we assessed the half-life of CTG* in ⌬hsc82 hsp82 D170G mutant cells. However, degradation of CTG* was not altered (not shown).
Another chaperone, known to work together with the cytosolic Hsp70s of the Ssa family, is Hsp104p (69). This protein belongs to the family of Hsp100 chaperones, which are part of the AAA-ATPase superfamily (70,71). Hsp100s are known to unfold proteins, either after heat shock (69) or prior to hydrolysis (72). Additionally, they bind in an ATP-dependent manner to the Ssa1p-Ydj1p complex (69). Therefore, we measured the degradation kinetics of CTG* in pulse-chase experiments in ⌬hsp104 cells. Interestingly, degradation of CTG* is clearly delayed in ⌬hsp104 cells (Fig. 10). In contrast, degradation of CPY* or CT* is not affected in this mutant (not shown).

DISCUSSION
The current model for ERAD has evolved over the past 20 years and is based on data accumulated using a wide range of substrates with different structural and functional characteristics studied in different eukaryotic systems. An undisputed advantage is that we now have a broad knowledge what ERAD is and how it functions in general. On the other hand, it was hard to achieve a clear assignment of components of the degradation machinery to topologically diverse domains of malfolded substrate proteins. To address this question, we created a set of modular substrates with topologically defined domains carrying the same degradation motif. The set consists of a soluble misfolded protein of the ER lumen (CPY*), CPY* linked to a transmembrane domain (CT*), and CPY* fused to a transmembrane domain followed by the green fluorescent protein (CTG*). Our studies revealed that, as expected, CT* and CTG* span the ER membrane. The N-terminal CPY* moiety is located in the ER lumen, whereas the C termini of both proteins are in the cytoplasm. Fluorescence of the cytosolic GFP domain of CTG* shows that it is biologically active and, therefore, correctly folded (Fig. 2). As expected, the misfolded CPY* domain prevents transport of the proteins out of the ER. As shown previously for soluble CPY* (6,9,12,16,73), degradation of CTG* and CT* is also dependent on ubiquitination by the ubiquitin-conjugating enzymes Ubc1p and Ubc7p, together with the ubiquitin-protein ligase complex Der3/Hrd1-Hrd3. Further components of the degradation apparatus are the AAA-ATPase complex Cdc48-Ufd1-Npl4 and the 26S proteasome (Figs. [3][4][5]. We have come to the conclusion that the components mentioned above constitute the basic machinery for ERAD of soluble and membrane proteins alike, which display malfolded domains in the ER lumen. Further studies are necessary to unravel whether other types of misfolded config-urations in ER-lumenal domains exist, which would recruit a different degradative machinery. The machinery used for degradation differs, in most cases, in the E3 used for polyubiquitination. Degradation of substrates with misfolding in the ER lumen, like CPY*, CT*, CTG*, and Pdr5*, depend on the E3 complex Der3/Hrd1-Hrd3 (6,12,20). However, these substrates do not show complete stabilization in ⌬der3/hrd1 cells, which implies the involvement of another E3 or of a different pathway in degradation. In contrast, degradation of Ubc6p, a protein without an ER-lumenal domain, is completely independent of the ubiquitin-protein ligase complex Der3/Hrd1-Hrd3 (49,74), but has been shown to depend on the ubiquitin-protein ligase Ssm4/Doa10p (49,74). In the case of other ERAD substrates, like unassembled Vph1p, there is controversy on whether the degradation is dependent on the E3 complex Der3/Hrd1-Hrd3 (21,75). Degradation of UP*, a mutated version of the yeast uracil permease, is independent of the Der3/Hrd1-Hrd3 complex. Interestingly, the mutation in UP* is in a predicted cytosolic loop (76). A likely explanation, that we favor, is that misfolded cytosolic regions may attract a ubiquitin-protein ligase different from that required for misfolded lumenal domains.
Soluble and membrane-bound CPY* differ in their requirement for the ER membrane protein Der1p and the ER-lumenal Hsp70 chaperone Kar2p (Figs. 4 and 6): these proteins are only necessary for degradation of soluble CPY* (4,24). Kar2p, together with Jem1p and Scj1p, was shown to solubilize CPY* in the ER lumen (22). Our data indicate that Der1p and Kar2p are not involved in the recognition of misfolded protein domains in the ER lumen or in further unfolding of misfolded proteins prior to dislocation into the cytosol. They may function in localizing soluble substrates to the lumenal face of the ER membrane and/or to the vicinity of the translocon and keep them in a dislocation competent state (4,5,22).
We, furthermore, studied the requirement of cytosolic chaperones for the degradation of CPY*, CT*, and CTG*. Hsp70 chaperones have been tested for their involvement in the degradation of several ERAD substrates yielding different results. Degradation of mutant ␣-factor and mutated membrane protein Sec61-2p was shown to be independent of Hsp70 chaperones (19,23). In contrast, hydrolysis of proteins containing larger cytosolic domains like CFTR or Vph1p was dependent on Hsp70 chaperone activity (21,23). In this study, we show that the breakdown of soluble CPY* or of the membrane-bound variant CT* is also independent of Hsp70 chaperones of the Ssa family (Fig. 7). In contrast, we find that the degradation of CTG*, containing the tightly folded cytoplasmic GFP domain, is strongly dependent on the Ssa family activity: hydrolysis of CTG* is almost completely blocked in a mutant defective in Ssa chaperone activity (Fig. 7).
We further addressed the question if degradation of the GFP domain requires action of the Ssa chaperone family only in the context of ER degradation or not. This was answered by testing the soluble, unstable Deg1-GFP fusion protein (77) in the Ssa mutant strain. Interestingly, just like CTG*, proteasomal degradation of Deg1-GFP also depends on the action of the Ssa family (not shown).
Substrate specificity and activity of Hsp70 chaperones is modulated by Hsp40 co-chaperones. These co-chaperones bind to Hsp70 proteins via their DnaJ domain (62,64,66). Because Hsp70-Hsp40 chaperone complexes are thought to be involved in preventing protein aggregation rather than in protein unfolding (78), we were further interested in chaperones with unfolding activity. Hsp100 chaperones are known to have such an activity (69,72). Glover and Lindquist (69) showed that Hsp70s form complexes with Hsp40 and Hsp100 chaperones to protect heat-shocked cells from damage by protein aggregation. We first tested the involvement of the well characterized DnaJ homologue Ydj1p in the degradation of CTG*, but did not find any effect. After omitting essential, mitochondrial and ERlocalized potential DnaJ proteins, we analyzed involvement of the 12 remaining J domain proteins in degradation of CTG*. Absence of three of these proteins (Cwc23p, Jid1p, and Hlj1p) led to a small but reproducible effect (Fig. 9). Double mutants, lacking two of these three DnaJ domain proteins did not exhibit a slower CTG* degradation rate (not shown). The moderate defect observed in single or double mutants of the three analyzed J domain proteins might be because of their overlapping specificity. Next, we tested Hsp104p, a member of the yeast Hsp100 chaperones. Interestingly, we found that it is also involved in CTG* degradation (Fig. 10).
Our data show that only CTG*, a malfolded membrane protein with a tightly folded cytoplasmic GFP domain, requires activity of Hsp70, Hsp40, and Hsp104 chaperones for proper degradation (Figs. 7, 9, and 10). Neither soluble CPY* nor CT*, lacking a cytosolic domain, require these chaperones for their removal. These findings imply that the ATPase subunits of the 19S cap of the proteasome alone might not be sufficient to unfold tightly folded domains. Indeed it has been recently shown that the 26S proteasome is unable to degrade the tightly folded GFP domain in vitro and it was suspected that additional factors are required for efficient unfolding (79). Obviously, in vivo the cytoplasmic Hsp70 machinery and other chaperones are such additional factors. The difference in half-life of CTG* compared with CPY* or CT* may be because of the need to unfold the GFP moiety of CTG* prior to proteasomal degradation. CPY* and CT* can be degraded directly after leaving the ER.