Membrane topology and function of Der3/Hrd1p as a ubiquitin-protein ligase (E3) involved in endoplasmic reticulum degradation.

The endoplasmic reticulum contains a protein quality control system that discovers malfolded or unassembled secretory proteins and subjects them to degradation in the cytosol. This requires retrograde transport of the respective proteins from the endoplasmic reticulum back to the cytosol via the Sec61 translocon. In addition, a fully competent ubiquitination machinery and the 26 S proteasome are necessary for retrotranslocation and degradation. Ubiquitination of mutated and malfolded proteins of the endoplasmic reticulum is dependent mainly on the ubiquitin-conjugating enzyme Ubc7p. In addition, several new membrane components of the endoplasmic reticulum are required for degradation. Here we present the topology of the previously discovered RING-H2 finger protein Der3/Hrd1p, one of the new components of the endoplasmic reticulum membrane. The protein spans the membrane six times. The amino terminus and the carboxyl terminus containing the RING finger domain face the cytoplasm. Altogether, RING finger-dependent ubiquitination of malfolded carboxypeptidase yscY in vivo, as well as of Der3/Hrd1p itself in vitro and RING finger-dependent binding of Ubc7p, uncovers Der3/Hrd1p as the ubiquitin-protein ligase (E3) of the endoplasmic reticulum-associated protein degradation process.

The endoplasmic reticulum (ER) 1 contains a highly effective protein quality control system, which guarantees delivery of only properly folded proteins to their site of action (1). Proteins that pass through the folding process but are unable to acquire their proper conformation or cannot assemble with a respective binding partner are degraded rapidly. This process is known as ER-associated degradation or simply ER degradation (2)(3)(4). Recent studies have revealed that degradation of misfolded or unassembled secretory proteins requires retrograde transport from the ER to the cytoplasm of the cell and the ubiquitinproteasome system (2)(3)(4). The degradation of a multitude of misfolded and unassembled secretory proteins has been fol-lowed in mammalian cells (2,3). Most prominent are the degradation of unassembled T-cell receptor ␣-chains (5, 6) and a mutated version of the human cystic fibrosis transmembrane conductance regulator, which leads to the severe disease cystic fibrosis (7,8). The yeast Saccharomyces cerevisiae has turned out to be an excellent model system to study eukaryotic cell function. Here, mutant versions of the vacuolar carboxypeptidase yscY (CPY*) (9), the plasma membrane ATP-binding cassette transporter Pdr5p (Pdr5*) (10), the ER membrane-located Sec61p (11), the mating type pheromone ␣-factor (12), and down-regulated hydroxymethylglutaryl-CoA reductase (13) were found to be subject to ER degradation. As found for the mammalian proteins destined for ER degradation, removal of these proteins requires the proteasome and in most cases an intact ubiquitin conjugation system (9 -11, 14). Proteasomal substrates are usually tagged by repeated attachment of ubiquitin moieties through isopeptide bonds between the carboxylterminal glycine of ubiquitin and ⑀-amino groups of lysine residues within the substrate molecules (15). A sequence of enzymatic reactions is necessary to link ubiquitin to protein substrates: after activation through an E1 enzyme via an ATPrequiring step, which leads to a thiol ester intermediate, ubiquitin is transferred to a cysteine thiol group of a ubiquitinconjugating enzyme E2. In a third step involving a ubiquitinprotein ligase (E3), ubiquitin is linked by its carboxyl terminus to an ⑀-amino group of the target protein, forming an isopeptide linkage (15). When studying the fate of mutated, soluble vacuolar CPY* and the mutated plasma membrane protein Pdr5*, we found both proteins to be polyubiquitinated via the ubiquitin-conjugating enzymes Ubc6p and Ubc7p, whereby Ubc7p appeared to be the most prominent catalyst (9,10). Similar results have been obtained for mutated Sec61p and downregulated hydroxymethylglutaryl-CoA-reductase (11,14). The ubiquitin-protein ligase E3 participating in this process has remained a mystery. Genetic screens have uncovered a variety of new gene products embedded in the ER membrane which are necessary for ubiquitination and degradation of the respective proteins (13, 16 -18). One of these proteins, Der1p, was shown up until now to be necessary only for ER-degradation of soluble, fully translocated CPY* (16,19), whereas Cue1p, Der3/Hrd1p, and Hrd3p have more general functions in the ER degradation process. Cue1p was discovered as to be a receptor for Ubc7p in the ER membrane, activating this ubiquitin-conjugating enzyme (18). The function of Der3/Hrd1p and Hrd3p have remained unknown so far. Recently, a new family of ubiquitinprotein ligases (E3) has been discovered, which all contain a RING finger domain as a common motif (20 -27). Der3/Hrd1p is a RING-H2 domain-containing protein (13,17) that resides in the ER membrane (17). We have found previously that deletion of the RING-H2 domain or exchange of a single cysteine residue at position 399 against serine in the Der3/Hrd1 protein completely abolishes degradation of CPY* and Pdr5*, indicating the essentiality of this domain (10,17,28). We have furthermore shown that Der3/Hrd1p, Hrd3p, and Sec61p interact genetically, which led us to the proposal that these proteins formed both the retrotranslocon and an E3 complex (19). Here we examine the topology of Der3/Hrd1p in the ER membrane via N-glycosylation scanning and fusion of a topology-sensitive reporter protein domain to carboxyl-terminally truncated versions of the Der3/Hrd1 protein. We show that the carboxylterminally located RING-H2 domain of Der3/Hrd1p faces the cytoplasm. Altogether, RING finger-dependent ubiquitination of CPY* in vivo as well as self-ubiquitination of Der3/Hrd1p in vitro, and RING finger-dependent binding of Ubc7p identify Der3/Hrd1p as the ubiquitin-protein ligase (E3) of the ER degradation process.

Construction of Plasmids
Genetic experiments and molecular biological methods were carried out using standard protocols (29).
Insertion of an N-Glycosylation Site at the Carboxyl Terminus of Der3p-A carboxyl-terminal region of DER3 was PCR amplified from YEpDER3 using the primers DER3-G1 (5Ј-cccaaaaggttaccttgtggc-3Ј) and DER3-G2 (5Ј-gccgcctctagatagtactattatgctggataaatttatctgg-3Ј). The XbaI site containing the stop codon of DER3 is underlined, the bases encoding the N-glycosylation sequence Asn-Ser-Thr are in bold letters. The PCR product was digested with BglII/XbaI and inserted into YEpDER3 linearized with the same enzymes, yielding plasmid YEpDER3-G. For Western blot analysis of the protein, the plasmid was transformed into yeast strains W303-1C (16) and YJB009 (17).

Analysis of the Der3-His4C Fusion Proteins
Yeast strain STY50 (31) was separately transformed with plasmids pQ37, pE78, pR103, pQ134, pQ176, pD345, and YEp352. Transformants growing on SD plates supplemented with histidine but lacking uracil were tested for their ability to grow on minimal medium lacking uracil and containing 6 mM histidinol instead of histidine. Plates were incubated at 30°C for 3-5 days.

Split Ubiquitin Assay
Yeast strain YPD41 (see above) was transformed with plasmid pNub-Der3 and a plasmid encoding a Nub-Sed5 fusion (32). Transformants were grown overnight in SD medium lacking tryptophan, diluted with water, and spotted on minimal medium containing 1 mg/ml 5-FOA and 0.4 mM uracil. Plates were incubated at 30°C for 3-4 days.

GST Pull-down Assays
The GST-Der3 fusion proteins were expressed from plasmids pGEX-DER3 and pGEX-DER3C399S in Escherichia coli strain BL21 (Amersham Pharmacia Biotech). GST alone was expressed from plasmid pGEX-4T-1. Culturing of the cells and preparation of the bacterial protein extracts were carried out according to the manufacturer's instructions. The extracts were incubated with 100 l (bead volume) of gluthathione Sepharose 4B (Amersham Pharmacia Biotech) for 3 h at 4°C. Sepharose beads were collected (5 min at 3,000 rpm), washed three times with 1 ml of phosphate-buffered saline (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 1.8 mM KH 2 PO 4 , pH 7.3), and incubated at 4°C overnight with whole cell extracts prepared from yeast strain W303-1C, carrying the plasmid pTX146 (18) encoding a Myc-tagged version of Ubc7p. Sepharose beads were collected and washed for 5 min with 1 ml of phosphate-buffered saline containing 500 mM NaCl. Proteins from the supernatant and the wash fraction were precipitated with trichloroacetic acid and dissolved in urea buffer. Proteins bound to the Sepharose matrix were then eluted with urea buffer by incubating at 95°C for 5 min. Samples were subjected to Western blot analysis using monoclonal anti-Myc antibody (Roche).

In Vitro Ubiquitination Experiments
Ubiquitination reactions were performed in 25 mM Tris/HCl, pH 7.5, 50 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 10 mM ATP, and 0.5 g/l ubiquitin (yeast, Sigma) and contained one or more of the following components as indicated in Fig. 6: 0.5 g of E1 (yeast, Affiniti), 1 g of E2 (human UBCH1, Affiniti), and approximately 1 g of GST-Der3 fusion protein (see above). Incubation of the reactions for 60 min at 30°C was followed by dilution with phosphate-buffered saline and precipitation of GST-Der3 species with Sepharose beads (3 h at 4°C). The beads were collected, washed three times with phosphate-buffered saline, and eluted with urea buffer. Samples were subjected to Western blot analysis using monoclonal anti-ubiquitin antibody (Calbiochem) or polyclonal anti-Der3 antiserum (17), respectively.

Immunoprecipitation of Ubiquitinated CPY*
For in vivo ubiquitination of CPY*, HA-tagged ubiquitin was expressed from plasmid YEp112 (33) by induction with copper as indicated in Fig. 5. The experiment was performed in a DER3 wild type strain (W303-1C) and a der3 deletion strain (YJB009) expressing the C399S mutant of Der3p from plasmid YCpDER3C399S (17). Immunoprecipitation and detection of ubiquitinated CPY* species were done as described previously (9).

Western Blot Analysis
For Western blot analysis of complete cellular protein, 3 A 600 units of cells were subjected to alkaline lysis, and proteins were precipitated with trichloroacetic acid and resuspended in urea buffer. Protein samples were separated on 6 -12% SDS-polyacrylamide gels and subjected to immunoblotting as described before (19).

Der3/Hrd1p Spans the ER Membrane Six Times; Its Amino and Carboxyl
Termini Face the Cytoplasm-Sequence analysis of Der3/Hrd1p had indicated that the protein contains a hydrophobic amino-terminal region with five putative membrane domains and a long hydrophilic carboxyl-terminal tail containing the RING-H2 motif (13,17). Protease protection and immunolocalization experiments had localized the carboxyl-terminal RING-H2 finger domain to the lumen of the ER. However, because the ubiquitin-conjugating enzyme in charge, Ubc7p, as well as the proteasome reside in the cytoplasm, it was hard to reconcile the fact that the RING-H2 domain of the protein resided in the ER lumen with a function for Der3/ Hrd1p as a ubiquitin-protein ligase, unless one assumed a process by which the carboxyl terminus with the RING-H2 domain and bound substrate reached back through the translocon into the cytoplasm of the cell (19).
We therefore undertook a new set of experiments to localize the carboxyl-terminal RING finger domain of Der3/Hrd1p and to uncover the membrane topology of the protein. First we undertook an N-glycosylation scanning experiment. Der3/ Hrd1p contains two N-glycosylation sites at positions 58 and 137 of its sequence (17). However, these sites are not accessible to the glycosylation machinery because they are predicted to be located in a transmembrane span and on the cytosolic side of the ER membrane, respectively. Therefore, we introduced a new potential N-glycosylation site between the penultimate and last amino acid of Der3/Hrd1p as done with CPY* (34). When expressing this modified Der3/Hrd1 protein either in a DER3 wild type or a der3 deletion background, no shift of the molecule to higher molecular mass was visible although a similarly modified CPY* molecule was shifted to higher molecular mass due to additional glycosylation (Fig. 1). This may indicate that the carboxyl terminus of the ER-localized Der3/Hrd1 protein does not reach the lumen of the organelle where core glycosylation occurs. However, this experiment does not give final proof for cytoplasmic localization of the carboxyl terminus of Der3/Hrd1p.
We set up further experiments to prove this result and to elucidate the exact membrane topology of Der3/Hrd1p. The hydropathy profile of Der3/Hrd1p predicts the existence of five or six transmembrane domains ( Fig. 2A). We created fusion constructs of Der3/Hrd1p with a truncated version of the His4 protein (His4C) containing a fragment of invertase (Suc2p) which we used as topology-sensitive reporters. His4C harbors histidinol dehydrogenase activity and is translocated through the ER membrane when fused to a signal sequence (35). Yeast his4 mutant strains expressing a His4C fusion are able to grow on minimal medium containing histidinol when the catalytic domain is present on the cytoplasmic side of the ER membrane. In this case, histidinol is metabolized to histidine resulting in a His ϩ phenotype. When the catalytic domain is targeted to the ER lumen, histidinol cannot be converted to histidine, leading to a His Ϫ phenotype. However, when located in the ER lumen, the protein becomes extensively glycosylated because of the presence of the SUC2 sequence and thus several glycosylation sites (31). We constructed a series of fusion proteins consisting of carboxyl-terminally truncated versions of Der3/Hrd1p and the His4C protein domain ( Fig. 2A and Table I), which allowed us to identify the orientation of the transmembrane domains of Der3/Hrd3p and, in addition, the location of the carboxyl-terminal RING-H2 finger domain. The invertase sequence contained within these constructs also introduced an epitope for immunodetection, thus making it possible to visualize the fusion proteins using anti-invertase antibodies.
We first determined the orientation of the amino terminus of Der3/Hrd1p using the fusion construct Der3/Hrd1 Q37 containing the first transmembrane domain ( Fig. 2A and Table I). As can be seen in Fig. 2B, Der3/Hrd1 Q37 does not support growth on histidinol. Instead, the construct is heavily glycosylated; endoglycosidase H treatment reduces the molecular mass of the fusion protein considerably, leading to its calculated value (Fig. 2C, lanes 1 and 2). These results indicate that the amino terminus of Der3/Hrd1 Q37 resides on the cytoplasmic side of the ER membrane. However, there is still the possibility that the FIG. 1. An N-glycosylation site introduced carboxyl-terminally into Der3/Hrd1p fails to be glycosylated. The amino acids Asn-Ser-Thr were introduced between the last two amino acids of Der3/Hrd1p, and the resulting protein (Der3-G) was expressed in a DER3 wild type and a ⌬der3 deletion strain and subjected to Western blot analysis. In contrast to a CPY* species modified in the same manner (Prc1-G5) (34), no increase of the molecular mass resulting from N-glycosylation is observed. Transformants were streaked on selective media supplemented with histidine (left panel) or histidinol (right panel) and incubated for 3-5 days at 30°C. C, analysis of the N-glycosylation state of the Der3/Hrd1-His4C fusion proteins. The proteins were immunoprecipitated from whole cell extracts of STY50 transformed with plasmids pQ37 to pD345 (lanes 1-12) or plasmid YEp352 (lane 13) using polyclonal anti-invertase antibody. Immunoprecipitates were treated with endoglycosidase H (Endo H) as indicated and separated on 6% SDS-polyacrylamide gels. Western blot analysis was performed using anti-invertase antibody. first transmembrane region functions as a cleavable signal sequence or that the entire construct fully translocates into the ER lumen. The latter possibility is ruled out on the basis that Der3/Hrd1 Q37 is detected in the fraction corresponding to the membrane pellet when anti-invertase antibodies are used. Der3/Hrd1 Q37 cannot be solubilized with urea, high salt, or carbonate. Only after treatment with Triton X-100 was Der3/ Hrd1 Q37 transferred into the soluble fraction (data not shown). Furthermore, we had shown previously that Der3/Hrd1p genetically interacts with another ER membrane protein necessary for ER degradation, Hrd3p (19). Hrd3p spans the membrane once, its carboxyl terminus being exposed to the cytosol (19). Thus it was feasible that the amino terminus of Der3/ Hrd1p was in close proximity to the carboxyl terminus of Hrd3p. To elucidate this possibility and to show that the amino terminus of Der3/Hrd1p indeed faces the cytoplasm, we made use of the split ubiquitin technique (32,36). This technique monitors interactions between proteins in vivo and is based on the self-reassembly of the amino-and carboxyl-terminal halves (Nub and Cub) of ubiquitin. The reassembled ubiquitin is recognized by ubiquitin-specific proteases that cleave any carboxyl-terminally attached polypeptide from Cub (32,36). Cub, extended with an amino-terminally modified version of the enzyme Ura3p (RUra3p), was linked to the cytosolic carboxyl terminus of Hrd3p. Nub was linked to the amino terminus of Der3/Hrd1p. If Cub and Nub interacted, RUra3p would be cleaved off by the ubiquitin-specific proteases. Because the amino-terminal residue of RUra3p is an arginine, rapid degradation of RUra3p by the enzymes of the N-end rule pathway leads to a Ura Ϫ phenotype. 5-FOA is converted by Ura3p into 5-fluorouracil, which is toxic for the cells. However, degradation of RUra3p rescues cells and allows growth on medium containing 5-FOA (32). As can be seen in Fig. 3, cells expressing the carboxyl-terminally modified Hrd3-Cub-RUra3 and the amino-terminally modified Nub-Der3/Hrd1 can grow on 5-FOA, indicating close proximity or interaction between both proteins and at the same time a cytoplasmically exposed amino terminus of Der3/Hrd1p. N-Glycosylation scanning with a newly introduced N-glycosylation site at the carboxyl terminus of Der3/Hrd1p had indicated that this part of the protein does not reach the ER lumen and is localized in the cytoplasm. When the construct Der3/ Hrd1 D345 was tested, which contains His4C in the carboxylterminal region, growth of cells on histidinol (Fig. 2B) and the absence of any glycosylation (Fig. 2C, lanes 11 and 12) were observed. A protein band of calculated molecular mass was detected upon Western blot analysis. This strongly suggests that the carboxyl terminus of Der3/Hrd1p is located on the cytoplasmic side of the ER. This again indicates that the RING-H2 domain of Der3/Hrd1p is cytoplasmic. Orientation of the amino and carboxyl terminus of Der3/Hrd1p to the same side, the cytoplasm, suggests the presence of an even number of transmembrane domains in Der3/Hrd1p. We therefore analyzed the fusion proteins Der3/Hrd1 E78 , Der3/Hrd1 R103 , Der3/ Hrd1 Q134 , and Der3/Hrd1 Q176 . The constructs Der3/Hrd1 E78 and Der3/Hrd1 Q134 promoted growth of cells on histidinol (Fig.  2B) and exhibited a weight comparable to their calculated molecular masses with no glycosylation occurring (Fig. 2C,  lanes 3 and 4 and lanes 7 and 8). This points to a cytoplasmic localization of both His4C domains arguing for an identical orientation of both transmembrane helices II and IV. This indicates that the hydrophobic sequence III ( Fig. 2A) also constitutes a short helix, which dips into the ER membrane in the opposite direction. Indeed, the construct Der3/Hrd1 R103 did not promote growth on histidinol (Fig. 2B) and was heavily glycosylated (Fig. 2C, lanes 5 and 6). As expected, the fusion construct Der3/Hrd1 Q176 did not promote growth on histidinol either (Fig. 2B), and it was heavily glycosylated (Fig. 2C, lanes  9 and 10), supporting the view of ER luminal orientation of the His4C portion and of helix V spanning the ER membrane in the opposite direction to transmembrane helices II, IV, and VI. From these studies, we predict Der3/Hrd1p to be a protein with six membrane-spanning helices, whereby the amino and car-   boxyl termini of the protein face the cytoplasm (Fig. 4).
Der3/Hrd1p Is a Ubiquitin-Protein Ligase (E3)-We had shown that Der3/Hrd1p is required for degradation of CPY* and the polytopic membrane protein Pdr5*. We furthermore provided evidence for the RING-H2 domain as being crucially involved: deletion of the RING-H2 domain or mutation of one of the cysteine residues (Cys-399) to serine within this domain abolished the function of the protein (10,17,28). Recent studies have uncovered a new family of ubiquitin-protein ligases (E3 enzymes) which contain a RING-H2 finger domain that is necessary for function (20 -27). It was thus very likely that Der3/ Hrd1p constituted the ubiquitin-protein ligase involved in the degradation of the ER proteins tested so far (13,17). As shown in Fig. 5, the C399S mutation in Der3/Hrd1p indeed abolishes ubiquitination of CPY*. It is an outstanding feature of ubiquitin-protein ligases to self-ubiquitinate in vitro in a RING finger-dependent manner in the absence of other substrate proteins (20,23,27). We tested Der3/Hrd1p for this property. The soluble RING-H2 finger-containing carboxyl-terminal domain of Der3/Hrd1p (amino acids 208 -551) was expressed as a GST fusion protein and incubated with ubiquitin, E1, and E2 enzymes and Mg 2ϩ -ATP in vitro. Indeed, the GST-Der3/Hrd1 protein ubiquitinated itself to some extent in an E1-and E2dependent fashion. As expected, formation of the ubiquitin conjugates was completely abolished in the Der3/Hrd1p C399S mutant (Fig. 6).
It has been demonstrated that the RING domain mediates the recruitment of the respective ubiquitin-conjugating enzyme E2 involved in the degradation reaction (25,37). In the case of the ubiquitin-protein ligase c-Cbl, the RING finger motif has been shown to be essential for binding of the E2 enzyme (37,38). We tested the ability of Der3/Hrd1p to bind the ubiquitinconjugating enzyme Ubc7p, which is known to be necessary for all ubiquitin conjugation reactions dependent also on Der3/ Hrd1p (10,13,14,17). The soluble GST-fused RING-H2 fingercontaining domain of Der3/Hrd1p was bound to glutathione-Sepharose. Application of extracts of cells expressing a functional Myc-tagged Ubc7p (18) to the GST-Der3/Hrd1 Sepharose beads resulted in specific binding of Ubc7p to the protein (Fig. 7A). Interestingly, the inactive C399S mutant of Der3/Hrd1p is also defective in binding of Ubc7p (Fig. 7B). Taken together, the lack of CPY* ubiquitination in cells carrying a mutated Der3/Hrd1 C399S protein, the ability to ubiquitinate itself, and its interaction with Ubc7p clearly identify Der3/Hrd1p as the ubiquitin-protein ligase (E3) of the ER degradation process. The results also show that the RING-H2 domain of the ligase is crucial for recruitment of the ubiquitinconjugating enzyme Ubc7p. DISCUSSION This study presents a first analysis of the complete membrane topology of Der3/Hrd1p, an essential component of the ER degradation machinery. We furthermore provide evidence that Der3/Hrd1p is the ubiquitin-protein ligase (E3) of the ER degradation process. Using a topology-sensitive reporter protein domain, we show that Der3/Hrd1p contains six transmembrane domains. The amino terminus and the carboxyl terminus of the protein face the cytoplasmic side of the ER membrane. The newly established cytoplasmic localization of the carboxyl- Gluthatione-Sepharose was added to the reaction mixtures, and bound proteins were subjected to gel electrophoresis. Subsequent Western blot analysis using an anti-ubiquitin antibody shows a series of multiubiquitinated species of higher molecular weight than the GST-Der3 fusion protein itself. Formation of these conjugates was completely abolished when the C399S mutant of Der3/Hrd1p was used or when either Der 3/Hrd1p or E1 and E2 were omitted. To visualize the amounts of GST-Der3 fusion protein in the different reactions, the samples were subjected to Western blot analysis using anti-Der3/Hrd1p antibodies.
terminal domain of Der3/Hrd1p corrects our previous preliminary finding, in which upon protease protection experiments and subsequent immunolocalization studies with an antibody directed against the carboxyl terminus of Der3/Hrd1p, an ER luminal localization of the carboxyl-terminal domain of the protein had been suggested (17). This may have been the result of conditions leading to an artifactual burying of the carboxyl terminus of Der3/Hrd1p because of membrane aggregation and occlusion of membrane surfaces (39) or a carboxyl terminus resistant to protease attack. While our work was nearly completed, a paper appeared which also reported a cytoplasmic localization for the carboxyl-terminal domain of Der3/Hrd1p (40).
The carboxyl-terminal domain of Der3/Hrd1p contains a RING-H2 finger motif that is crucial for its function: deletion of the motif or exchange of one of the cysteine residues (Cys-399) of the RING-H2 domain to serine completely abolishes degradation of mutated and malfolded ER proteins (17,28). RING finger-containing proteins have been identified as a new class of ubiquitin-protein ligases (E3 enzymes) necessary for ubiquitination of substrates destined for degradation via the proteasome (20 -27). We show here that cells carrying a mutated Der3/Hrd1p C399S version are unable to ubiquitinate CPY* in vivo (Fig. 5). A common feature of RING finger-containing ubiquitin-protein ligases is their ability to ubiquitinate themselves in vitro in the absence of substrates (20,23,27). We found that this also holds true for Der3/Hrd1p (Fig. 6). Function of a protein as an E3 requires recruitment of the respective ubiquitin-conjugating enzyme (E2). Protein-protein interaction studies using a GST-bound carboxyl-terminal fragment of Der3/Hrd1p containing the RING-H2 domain demonstrated that Ubc7p, the major ubiquitin-conjugating enzyme of the ER degradation process, selectively bound to the protein. Altogether, these properties identify Der3/Hrd1p as the ubiquitin-protein ligase (E3) of ER degradation as suggested previously (19). Mutation of a crucial cysteine (C399S) in the RING-H2 finger domain completely abolished binding of Ubc7p, suggesting that the RING-H2 domain is directly involved in the binding of E2. Cue1p, another ER membrane protein, had been shown to bind Ubc7p and to be necessary for Ubc7p activity (18). Our pull-down experiment clearly demonstrates that Der3/Hrd1p carries an intrinsic binding activity toward Ubc7p. The additional binding of Ubc7p by Cue1p might result in a tightening of Ubc7p binding to the membrane and by this inducing the formation of the "active" structure of Ubc7p.
We had shown previously that Der3/Hrd1p and Hrd3p, another essential ER membrane protein involved in ER degradation, interact genetically, indicating complex formation between these two proteins (19). Using the split ubiquitin system, we could show that the cytoplasmically localized amino terminus of Der3/Hrd1p and the carboxyl terminus of Hrd3p are in close proximity, which might indicate that they physically interact (Fig. 3). An interaction between both proteins has also been presented recently by others (40). The fact that the carboxyl terminus of Der3/Hrd1p with its RING-H2 domain is localized to the cytoplasm and never sees the ER lumen makes the previous proposal of Der3/Hrd1p as being a binding partner of the substrate in the ER lumen, which delivers it to the cytoplasmic degradation machinery, unlikely. It also makes a function of Hrd3p as a recycling molecule for the carboxyl terminus of Der3/Hrd1p from the cytoplasm back into the ER lumen improbable (19). A complex formation of Der3/Hrd1p and Hrd3p must have different functions. As induction of Ubc7pdependent Der3/Hrd1p degradation upon Hrd3p deletion indicates the presence of a fully competent ubiquitination and degradation machinery in ⌬hrd3 cells, the defective degradation of ER substrates in the absence of Hrd3p must have other reasons. It would be plausible if Hrd3p had substrate recognition and signaling functions, which lead to integration of processes such as substrate delivery via the Sec61 translocon, ubiquitination via Der3/Hrd1p and Ubc7p, and degradation via the proteasome in a fashion that makes the overall process highly regulated and processive. Similar functions for the Der3/ Hrd1p-Hrd3p interaction are also discussed by Gardner et al. (40). Future experiments will uncover the functional relationship of both proteins. FIG. 7. The Der3/Hrd1p RING-H2 domain binds Ubc7p. A, GST pull-down assay confirms protein-protein interaction between Der3/ Hrd1p and Ubc7p. A fusion protein of the soluble carboxyl-terminal part of Der3/Hrd1p containing the RING-H2 domain with GST was bound to glutathione-Sepharose and incubated with a whole cell extract prepared from a strain expressing a functional Myc-tagged version of Ubc7p. The nonbound supernatant (S) was removed and the gel beads washed with incubation buffer containing 500 mM NaCl (W). Bound proteins were eluted with urea buffer (E), and samples were analyzed by Western blot analysis using monoclonal anti-Myc antibodies. As a control, GST alone was attached to the Sepharose beads. B, when the same experiment was performed using the mutated C399S species of Der3/Hrd1p, specific binding of Myc-Ubc7p to the GST fusion protein was completely abolished.