A Ubc7p-binding domain in Cue1p activates ER-associated protein degradation

The CUE domain is dispensable for ERAD

To determine the domains within Cue1p required for ERAD we first focused on its only characterized region, the CUE domain. CUE domains were identified as a consequence of their similarity to a region of yeast Cue1p (Ponting, 2000). For several proteins this ∼40 amino acid domain is known to bind ubiquitin. In the case of gp78 it is required for its ligase function in vivo (Chen et al., 2006). Structural analyses of CUE domains from Cue2p and Vps9p (Kang et al., 2003; Prag et al., 2003) reveal a three alpha helical structure with overall similarity to the well-characterized UBA domain.

Unlike other CUE-domain proteins the ability of Cue1p to bind ubiquitin is extremely weak. This is contributed to by the lack of a Met-Phe-Pro (MFP) tripeptide at the boundary between the A and B alpha helices in ubiquitin-binding CUE domains. Mutation of the Leu-Ala-Pro (LAP) tripeptide in the corresponding position (amino acids 76-78) in Cue1p to MFP was reported to increase ubiquitin binding (Kang et al., 2003). We mutated the region encoding LAP in Cue1p to either MFP (Cue1p^MFP) or altered it to potentially further disrupt the CUE domain structure by removing the conserved proline (Cue1p^AVA; Fig. 1A). These proteins were expressed in a Δcue1 background from low copy CEN plasmids. When degradation of the commonly used Hrd1p ERAD substrate CPY^* was assessed by metabolic labeling, neither mutation had an impact on Cue1p-dependent CPY^* degradation (Fig. 1B).

Based on these findings we deleted the entire CUE domain (Cue1p^ΔCD). As with the point mutations, this was similarly without effect on CPY^* degradation (Fig. 1C). Equivalent results were obtained at 23°C where degradation proceeds more slowly (not shown). A similar lack of requirement for the Cue1p CUE domain was found when CTG^*, a transmembrane form of CPY^* (Taxis et al., 2003), was expressed in a Δcue1 strain expressing the CPY^* allele from its chromosomal locus (Fig. 1D). To evaluate whether the CUE domain was also dispensable for Doa10p-mediated ERAD we evaluated the degradation of Ste6p^*, a primarily DOA10 ligase substrate (Huyer et al., 2004). Again, deletion of the CUE domain had no effect on Ste6p^* degradation (Fig. 1E). Thus, the Cue1p CUE domain has no discernable role in the function of either ERAD E3. One possible explanation is that this domain is playing a subtle role in ERAD not evident with these substrates. However, no difference was observed between Cue1p and Cue1p^ΔCD re-expression when ER stress was induced either pharmacologically in a Δcue1 strain by tunicamycin or genetically by deletion of the ER stress response gene IRE1 (not shown).

A C-terminal Ubc7p binding region (U7BR) in Cue1p required for ERAD

We next evaluated the requirements for Cue1p binding to Ubc7p in vitro using N-terminal GST fusions (Fig. 2A). A truncation that removed the membrane-proximal region of Cue1p, as well as the CUE domain (Cue1p^110-203), bound Ubc7p as effectively as the full cytoplasmic domain (Cue1p^25-203; Fig. 2B). Moreover, N-terminal deletions up to amino acid 151 retained Ubc7p binding in vitro whereas deletions beyond that point, as well as C-terminal truncations, abrogated binding. This establishes a Ubc7p binding region (U7BR) between amino acids 151 and 203 of Cue1p. To address whether Ubc7p binding is required in vivo, Cue1p^1-173 was expressed in a Δcue1 strain. This protein is stable and membrane anchored (not shown). Cue1p^1-173 was unable to restore ERAD of either CPY^* or CTG^* as determined both by cycloheximide chase and pulse-chase metabolic labeling (Fig. 2C,D). As expression of CTG^* was lower than CPY^* under wild-type ERAD conditions, a longer exposure of CTG^* is shown to allow for the visualization of this protein (Fig. 2C, lower panel). Similar results were obtained with the DOA10 substrate Ste6p^* (not shown). Importantly however, as with the Δcue1 strain, yeast reconstituted with Cue1p^1-173 exhibited marked instability of Ubc7p (Fig. 5A, lower right panel; supplementary material Fig. S2), consistent with its inability to bind the E2 in vitro (Fig. 2B).

The Cue1p transmembrane domain is not required for ERAD of CPY^*

To address the requirements for stabilization of Ubc7p and how this correlates with ERAD, a Δcue1Δubc7 strain was generated and reconstituted with C-terminal HA-tagged Ubc7p (Ubc7pHA) and various forms of Cue1p, all of which lack a membrane anchor. Consistent with previous findings (Ravid and Hochstrasser, 2007), Cue1p^24-203, which fractionated into the cytosolic fraction (not shown), stabilized Ubc7pHA (Fig. 3A, bottom panel). In sharp contrast, Cue1p^151-203, which includes only the U7BR, resulted in a markedly lower initial level and rapid degradation of Ubc7p whereas Cue1p^110-203 substantially stabilized Ubc7p. Unexpectedly Cue1p^24-203 not only stabilized Ubc7p but also reconstituted CPY^* degradation (Fig. 3A, upper panel). This was confirmed in a Δcue1 strain where expression of cytoplasmic Cue1p^24-203 or of full length Cue1p provided similar results (Fig. 3B). By contrast, Cue1p^151-203 did not reconstitute CPY^* degradation, as expected because of loss of Ubc7p (Fig. 3A). Variable results were found with Cue1p^110-203: CPY^* either remained relatively stable or was degraded with delayed kinetics, perhaps reflecting the incomplete stabilization of Ubc7p (Fig. 3A).

U7BR alone can activate ERAD

To dissociate issues of E2 stabilization from ERAD and to explore the significance of membrane recruitment of Ubc7p we directly anchored Ubc7p to the ER membrane (™Ubc7pHA) by fusing C-terminally HA-tagged Ubc7p to amino acids 1-44 of Cue1p (Fig. 4A). ™Ubc7pHA was stabilized when expressed in a Δcue1 strain, in contrast to Ubc7pHA, which is known to be degraded in the absence of Cue1p (Fig. 4B). ™Ubc7pHA encoded a functional E2 as degradation of CPY^* proceeded at comparable rates in Δubc7 cells expressing either Ubc7pHA or ™Ubc7pHA (Fig. 4C). Thus, direct membrane anchoring does not interfere with the function of Ubc7p. However, when assessed in a Δcue1Δubc7 strain, ™Ubc7pHA was insufficient to restore CPY^* degradation in the absence of Cue1p (Fig. 4D, left side). These findings suggest that the function of Cue1p cannot be explained simply by membrane recruitment and stabilization of Ubc7p.

Having now generated a stable, membrane anchored, active Ubc7p we re-evaluated the capacity of Cue1p^1-173 to restore ERAD. Even when Ubc7p is membrane bound and stably expressed, Cue1p lacking an intact U7BR failed to reconstitute ERAD (Fig. 5A). This demonstrates the importance of Ubc7p binding to Cue1p, independent of both membrane recruitment and stabilization of this E2. Given the apparent dispensability of the CUE domain and the necessity of the E2 binding site for the function of membrane-anchored Cue1p, we asked whether truncated forms of Cue1p, lacking the transmembrane domain, could restore ERAD in cells expressing stable membrane anchored ™Ubc7pHA. Both Cue1p^151-203 and Cue1p^110-203 as well as Cue1p^24-203 reconstituted ERAD of CPY^* in the Δcue1Δubc7 strain, as assessed by both cycloheximide chase (Fig. 5B) and pulse-chase metabolic labeling (Fig. 5C). This establishes that at least when the E2 is stabilized by tethering to the ER membrane, the C-terminal U7BR of Cue1p is sufficient to reconstitute Ubc7p-dependent ERAD.

U7BR activates ubiquitylation of Ubc7p-Hrd1pC

Bazirgan and Hampton have recently shown (Bazirgan and Hampton, 2008) that the complete cytoplasmic domain of Cue1p can activate Ubc7p in vitro, in the absence of a known ligase domain, enhancing the known potential of this E2 to form polyubiquitin chains that can be thiolester linked to the active site cysteine of Ubc7p (Ravid and Hochstrasser, 2007). To corroborate and extend these findings, in vitro ubiquitylation assays were employed using components that were independently expressed in bacteria and purified. In Fig. 6A various fragments of Cue1p were added to glutathione-Sepharose-immobilized GST-Ubc7p and ubiquitylation reactions carried out without added E3. Ubiquitin that remained bound to immobilized GST-Ubc7p after washing of beads was analyzed following elution of bead-bound material by heating in SDS-PAGE sample buffer containing reducing agent to disrupt thiolester bonds. As reported (Bazirgan and Hampton, 2008), Cue1p^25-203 resulted in an increase in chains of three or four ubiquitins (Fig. 6A, compare lower molecular mass regions of lanes 1 and 2). Most striking, however, was a dramatic increase in ubiquitin conjugated to GST-Ubc7p in a stable, non-reducible form when Cue1p^25-203 was added to the reaction. Importantly, no substantial Cue1p^25-203-induced activation was seen with GST-UbcH5b, a close relative of the yeast E2s Ubc4p and Ubc5p (Fig. 6A, lanes 7 and 8) (Hatakeyama et al., 1997).

Consistent with our in vivo findings, the ability to activate Ubc7p did not require the CUE domain (Cue1p^25-ΔCD-203) but did require an intact U7BR (compare lanes 2, 4 and 5). Notably, despite the ability of Cue1p^151-203 to restore ERAD in cells, the U7BR was insufficient to cause this dramatic in vitro activation of GST-Ubc7p even at eight-fold higher concentration than the other Cue1p fragments (Fig. 6A, lane 3). Qualitatively similar results were obtained when ubiquitylation was assessed using soluble Ubc7p and immobilized GST fusions of Cue1p (not shown).

E3-dependent ubiquitylation was then evaluated as described (Fang et al., 2001) using a GST fusion of the cytoplasmic RING-finger-containing domain of Hrd1p (Hrd1pC), soluble Ubc7p and forms of Cue1p that were expressed and purified separately. Consistent with previous findings (Bazirgan and Hampton, 2008), Cue1p^25-203 activated Hrd1pC-mediated ubiquitylation (Fig. 6B). As with the E3-independent ubiquitylation, this was not observed with UbcH5b (lanes 2 and 3), was independent of the CUE domain and absolutely dependent on the U7BR (lanes 5-7). However, unlike the E3 independent reaction, the U7BR of Cue1p (Cue1p^151-203) activated ubiquitylation in a dose-dependent manner, albeit to a lesser extent than the full cytoplasmic domain (Fig. 6C). This finding provides an in vitro correlate for the capacity of the U7BR (Cue1p^151-203) to restore Ubc7p-dependent ERAD in vivo.

Yeast strains

Yeast cells were cultured at 30°C in YPD or SD medium supplemented as necessary. The Saccharomyces cerevisiae strains Δcue1 (Plemper et al., 1999), Δubc7 (Hiller et al., 1996) and Δire1 (Taxis et al., 2002) are derived from W303-1C (MATa, ura3-1, leu2-3,112, his3-11,15, ade2-1, trp1-1, can1-100, prc1-1) and were a generous gift from Dieter H. Wolf. Mating of the respective haploid strains generated Δcue1Δubc7 and Δcue1Δire1. Yeast transformations were carried out using standard methods.

Plasmid constructs

A ∼2 kb fragment derived from pGEM5Z-27k (from Thomas Sommer) spanning the wild-type CUE1 gene and promoter was subcloned into the BamHI site of pRS314 and pRS316 to yield pRS314-CUE1 and pRS316-CUE1, respectively. The cue1^AVA, cue1^MFP, cue1^ΔCD, cue1^1-173 and cue1^24-203 (superscripts indicate amino acids name or numbers) mutants were created using Stratagene's QuikChange® XL Site-Directed Mutagenesis Kit. A 1.2 kb SacII-ClaI fragment of UBC7 was PCR amplified from yeast genomic DNA and cloned into pRS314 and a C-terminal HA tag was subsequently introduced by PCR. pRS316-™UBC7HA was created by replacing the CUE1 sequence in pRS316-CUE1 starting from the Bsu36I site (corresponding to amino acid 44 of Cue1p) with PCR amplified UBC7HA. PCR-amplified DNA fragments corresponding to cue1^110-203 and cue1^151-203 were inserted behind the CUE1 promoter into NcoI-ClaI-digested pRS314CUE1, thus replacing the wild-type CUE1 sequence. To generate pRS424-GFPcue1^110-203 and pRS424-GFPcue1^151-203, the GFP gene was PCR amplified from pCTG^*, inserted between the CUE1 promoter and the cue1^110-203 or cue1^151-203 coding regions of pRS314-GFPcue1^110-203 or pRS314-GFPcue1^151-203, respectively, then transferred into pRS424. pSM1911 (2 μ, URA3, P_PGK::ste6-166::HA), encoding Ste6p^*HA, was a gift from Susan Michaelis (Huyer et al., 2004). pCTG^* (pRS316-P_CUE1::prc1-1::pdr5_4332-4532::GFP), obtained from Dieter H. Wolf, encodes the membrane-bound and GFP-tagged CPY^* variant CTG^* (Taxis et al., 2003) expressed from the CUE1 promoter (Kostova and Wolf, 2005). Constructs for bacterial protein expression of the various Cue1p mutants, Hrd1pC, Ubc7p and UbcH5b were generated by PCR amplification of the respective genes and ligation into pGEX-4T-1 or pGEX-KG (Amersham Biosciences). The identity of all constructs was determined by sequencing of the individual clones. Further cloning details are available upon request. Plasmid constructs used in this study are summarized in Table 1.

Cycloheximide chase

Yeast cells were grown to late-log phase in selective media. Cycloheximide was added to a concentration of 50 μg/3 OD₆₀₀ of cells and aliquots of 3 OD₆₀₀ equivalents were removed into an equal volume of 30 mM NaN₃ at the indicated time points. Cell extracts were prepared by alkaline lysis (1.85 M NaOH, 7.5% β-mercaptoethanol) on ice for 10 minutes with intermittent vortexing. Following TCA precipitation, the protein pellet was resuspended in 100 μl urea sample buffer (8 M urea, 5% SDS, 200 mM Tris-HCl pH 6.8, 0.1 mM EDTA, 0.03% Bromophenol blue, 1.5% β-ME) by shaking at 37°C. Proteins separated by SDS-PAGE were analyzed by immunoblotting with the appropriate antibodies. Pgk1p (phosphoglycerate kinase) was used to assess loading.

In vivo labeling and immunoprecipitation

Yeast strains grown in selective media to logarithmic phase were labeled for 20 minutes (CPY^* and CTG^*) or 10 minutes (Ste6p^*) with 20 μCi/1 OD₆₀₀ Easytag [³⁵S]methionine (Perkin Elmer). The chase was initiated by addition of an excess of non-radioactive methionine and 2.5 OD₆₀₀ equivalents were removed into ice-cold TCA at the indicated time points. Cells were broken by glass bead lysis in breaking buffer (6 M urea, 50 mM Tris-HCl pH 7.5, 1% SDS, 1 mM EDTA) then diluted tenfold with IP buffer (50 mM Tris-HCl pH 7.5, 190 mM NaCl, 6 mM EDTA, 1.25% Triton X-100). Following removal of cell debris, the supernatant was incubated with anti-CPY antibodies for 1 hour at 4°C and immunoprecipitated using Protein A-Sepharose 4B (Zymed) beads for 1 hour at 4°C. Immunoprecipitated material was washed with IP buffer, eluted in urea sample buffer and analyzed by autoradiography following SDS-PAGE. Labeling and immunoprecipitation of Ste6p^*HA was done according to the procedure described by Huyer et al. (Huyer et al., 2004). Radioactive material was quantified using Storm Phosphorimager and ImageQuant software (GE Health Care Life Sciences).

In vitro binding assays

GST-Cue1p truncation mutants (schematized in Fig. 2A) were expressed in E. coli BL21 cells and quantified by Coomassie Blue staining (supplementary material Fig. S1). Ubc7p was in vitro translated from pDEST47-UBC7 in a rabbit reticulocyte lysate system (Promega) using [³⁵S]methionine (GE Healthcare). Binding assays were carried out overnight at 4°C by incubating the GST-Cue1p proteins prebound to glutathione-Sepharose™ 4B beads (Amersham Pharmacia Biosciences) with ³⁵S-labeled Ubc7p in binding buffer (25 mM Tris-HCl pH 7.4, 50 mM NaCl, 5 mM DTT, 0.5% NP-40). Beads were then washed with binding buffer, eluted from beads, resolved by SDS-PAGE and visualized using a Storm Phosphorimager.

In vitro ubiquitylation assays

For the E3-independent ubiquitylation assays GST-Ubc7p and GST-UbcH5b expressed in E. coli BL21 were bound to glutathione-Sepharose beads. Free Cue1 proteins were generated by thrombin cleavage of the respective GST-fusions. Protein concentrations were calculated by Coomassie staining and densitometry, by comparison to BSA standards. Approximately 80 pmol GST-Ubc7p or GST-UbcH5b and 150 pmol GST were incubated with the Cue1 proteins (30 pmol of Cue1p^25-203, Cue1p^25-173 or Cue1p^25-ΔCD-203 and 240 pmol Cue1p^151-203) for 90 minutes at 30°C in 50 μl reactions containing 50 nM human E1 (Boston Biochem) and 1 μg ubiquitin (Sigma) in 1× ubiquitylation buffer (50 mM Tris-HCl pH 7.4, 0.2 mM ATP, 0.5 mM MgCl₂, 0.1 mM DTT, 1 mM phosphocreatine). For ubiquitylation reactions in the presence of E3, GST-Hrd1pC was expressed in bacteria and bound to glutathione-Sepharose beads. 100 μl reactions were assembled in 1× ubiquitylation buffer containing 5 pmol GST-Hrd1pC, thrombin-cleaved Ubc7p or UbcH5b (∼30 pmol) and Cue1 proteins (∼30 pmol except were indicated), human E1 (30 nM), and ubiquitin (1 μg). Following incubation, beads were washed in 1× TBS and bound material was eluted in SDS-reducing sample buffer. Reaction products were analyzed by SDS-PAGE and anti-ubiquitin immunoblotting.

Antibodies

Mouse monoclonal anti-CPY and anti-PGK (Molecular Probes), and rat monoclonal anti-HA-peroxidase (Roche) were used for immunoblotting according to the manufacturers' recommendations. Rabbit polyclonal anti-CPY (Rockland) and rat monoclonal anti-HA Affinity Matrix (Roche) were used for immunoprecipitation. Rabbit polyclonal anti-Cue1p and anti-ubiquitin antibodies were generated by immunizing rabbits with bacterially expressed cytosolic Cue1p^25-203 or with ubiquitin, respectively.