Ube2g2–gp78-mediated HERP polyubiquitylation is involved in ER stress recovery

A cell or a multicellular organism often needs to adjust its physiological status to respond to and defend against various stress conditions such as high or low temperature, oxidative and reductive environments. Adaptation to a stress condition can be achieved by enhancement of certain biological traits, and this is often associated with reduction in other traits that are dispensable to survive under this particular stress condition (Bennett and Lenski, 2007). This kind of ‘trade-off’ is often viewed as a cost or constraint associated with adaptation (Bennett and Lenski, 2007; Novak et al., 2006). Thus once stress is attenuated, the cell needs to adjust its status back to the normal physiological condition in order to resume division, growth or prepare to face other challenges (Qian et al., 2006). In general, during stress response, transcription and translation are two major ways to increase cellular stress tolerance capacity (Spriggs et al., 2010), but in the stress recovery stage, post-translational modification and degradation of protein appear to play key roles in eliminating stress-induced proteins (Kamura et al., 2000; Majmundar et al., 2010; Qian et al., 2006; Roobol et al., 2009). There are two major degradation systems in eukaryotic cells: autophagy, which is involved in the degradation of long-lived proteins and organelles, and the ubiquitin–proteasome system, which usually targets short-lived proteins for degradation (Mizushima and Komatsu, 2011). Because stress recovery usually occurs in a short time frame (Kamura et al., 2000; Qian et al., 2006), the ubiquitin–proteasome system is more likely to play essential role in this process.

It has been well established that upregulation of some heat shock proteins (HSPs, molecular chaperones) is needed for cells to survive many stress conditions (Monaghan et al., 2009; Qian et al., 2006). But cells seem to maintain some of these chaperones purposely at low levels under physiological conditions to permit constitutive cellular activities to proceed. The degradation of Hsp70 is mediated by carboxy terminus of Hsp70-binding protein (also known as E3 ubiquitin-protein ligase CHIP, and STIP1 homology and U-box containing protein 1; STUB1)-dependent polyubiquitylation during the stress recovery process (Qian et al., 2006). In addition, another short-lived protein, hypoxia-inducible transcription factor HIF1a, is also rapidly ubiquitylated by the ubiquitin ligase VHL (von Hippel-Lindau) and degraded by proteasomes under normoxic conditions (Kamura et al., 2000; Majmundar et al., 2010). However, whether the ubiquitin–proteasome system participates in other stress recovery processes such as ER stress recovery remains unknown.

Various stimuli, such as reductive reagents, oxidative reagents and Ca²⁺ overload can damage ER functions, leading to the accumulation of misfolded proteins and ER stress (Ellgaard and Helenius., 2001; Rutkowski and Kaufman, 2004; Shen et al., 2004). To cope with ER stress, the cell employs a quality control system named unfolded protein response (UPR) (Kaufman, 1999; Schulze et al., 2005) to attenuate translation and accumulation of misfolded proteins and to increase chaperone expression. HERP (homocysteine-induced ER stress protein) is a chaperone-like protein that is strongly upregulated by ER stress (Kokame et al., 2001; Schröder and Kaufman, 2005) and then rapidly degraded (Hori et al., 2004; Kim et al., 2008; Sai et al., 2003). Unlike other stress-induced cytoplasmic chaperones, HERP is an integral membrane protein with both its N- and C-termini facing the cytoplasm (Nogalska et al., 2006; Sai et al., 2002). The function of HERP is not fully understood, but accumulating evidence suggests that it has an essential role in ER-membrane-associated protein degradation (ERAD), which functions to retrotranslocate ubiquitylated proteins from the ER to proteasomes for degradation (Schulze et al., 2005). Knockdown or knockout of Herp leads to stabilization of several ERAD substrates (Hori et al., 2004; Okuda-Shimizu and Hendershot, 2007). It has been reported that upregulation of HERP can protect cells from ER-stress-induced apoptosis, mainly by forming an ERAD complex together with p97 homohexamer, Derlin1, VIMP and the ubiquitin E3 ligase HRD1 (Jarosch et al., 2002; Schröder and Kaufman, 2005; Schulze et al., 2005; Ye et al., 2001), which stimulates HRD1-mediated ubiquitylation and degradation of aberrant ER proteins (Kny et al., 2011).

In contrast to its function in ER stress, the fate of HERP after ER stress is still largely unknown. Here we show that after ER stress, HERP is quickly degraded by the Ube2g2–gp78-mediated ubiquitin–proteasome system, and the interaction between the ubiquitin-like (UBL) domain of HERP and the coupling of the ubiquitin conjugation to the ER degradation (CUE) domain of gp78 is essential for HERP polyubiquitylation and subsequent degradation. Overexpression of HERP in cells improved tolerance to ER stress. Surprisingly, HERP overexpression reduces cell viability under some oxidative stress conditions. We provided further evidence to show that this mechanism is evolutionarily conserved from yeast to mammalian cells. Our results suggest that the ubiquitin–proteasome system plays an important role in the ER stress recovery process, which is essential for adjusting the cellular physiological status to survive multiple challenges in changing environments.

It was reported that HERP is strongly upregulated both at the mRNA and protein level by homocysteine and other ER stress inducers (Hori et al., 2004; Kokame et al., 2000; Rubel et al., 2013). We firstly confirmed that ER stress inducers such as homocysteine, β-mercaptoethanol, tunicamycin, thapsigargin and dithiothreitol (DTT) but not oxidative ER stress inducers such as H₂O₂ and Paraquat caused the accumulation of HERP protein in HEK293T, HeLa and HCT116 cells by immunoblotting (Fig. 1A,B; supplementary material Fig. S1A). HERP was rapidly induced and the protein level peaked at 4 hours after DTT treatment in both HEK293T and HeLa cells, but subsequently, the protein level reduced quickly (Fig. 1C; supplementary material Fig. S1B). Presumably this was due to the inactivation of DTT by oxidation, because the HERP level did not decrease at all during the experiment when the cells were induced by relatively stable reductive inducers such as thapsigargin and homocysteine (compare supplementary material Fig. S1B–D,F). Thus, this stage could represent the ER stress recovery phase. To test this possibility, we first treated HEK293T cells with different reductive ER stress inducers for 4 hours, then transferred cells to fresh medium without any reductive ER stress inducers and tested the HERP protein level (Fig. 1D; supplementary material Fig. S1E,G). We found a similar result in all of the tested cell lines: HERP was quickly degraded and returned to the normal level in 8 hours after ER stress (Fig. 1D; supplementary material Fig. S1C), suggesting that after 4 hours incubation, cells consumed DTT and started the recovery process. Because ER stress can be easily induced by DTT without changing the medium, it was selected for most of the following experiments.

To test whether autophagy or the ubiquitin–proteasome pathway mediates the HERP turnover during ER stress recovery, we treated HEK293T cells with inhibitors that specifically block these two pathways. Only the proteasome inhibitor MG132 was found to strongly inhibit HERP degradation under normal condition as well as during ER stress recovery (Fig. 1E,F). By contrast, none of the four autophagy or lysosome inhibitors such as ammonium chloride (NH₄Cl), 3-methyladenine, bafilomycin A1 or chloroquine affected HERP degradation. These results suggest that HERP is degraded through the ubiquitin–proteasome system rather than by autophagy.

Because HERP is an ER-membrane-associated protein, the E2 and E3 enzymes that mediate HERP polyubiquitylation should also be ER-membrane-associated proteins. It has been reported that HRD1 and gp78 are two ERAD-related E3 ubiquitin ligases that are associated with the ER membrane (Chen et al., 2006; Kostova et al., 2007). Hrd1 forms a complex with SEL1L and mainly mediates the degradation of soluble, ER-luminal substrates and integral membrane proteins (Rubenstein et al., 2012; Shmueli et al., 2009). gp78, which was the first described and is the best documented human ERAD E3, is a multi-spanning membrane protein with its catalytic domain facing the cytosol (Chen et al., 2006; Fang et al., 2001; Li et al., 2009). Both proteins were shown to interact with HERP, but only gp78 could ubiquitylate HERP efficiently in vitro (Kny et al., 2011; Li et al., 2007; Schulze et al., 2005), raising the possibility that gp78 mediates HERP ubiquitylation and degradation. To test this idea, we created a HCT116 gp78 knockout (KO) cell line by homologous recombination and subsequent Cre–loxP-based selection marker excision (Zhang et al., 2011). The lack of gp78 expression in these cells was verified by immunoblotting (supplementary material Fig. S2) and HERP degradation during ER stress recovery were examined by immunoblotting. gp78 knockout significantly reduced HERP turnover compared with that in the wild-type cell line, regardless of whether or not DTT was present (Fig. 2A,B). Thus, we conclude that gp78 is an E3 required for HERP turnover during ER stress recovery.

To identify the cognate E2 of gp78 for HERP degradation during ER stress recovery, we tested a collection of mammalian E2s using an in vitro ubiquitylation assay (Jin et al., 2007), and found that at least six of them were able to cooperate with gp78 in polyubiquitylation (supplementary material Fig. S3). Because only E2G1, E2G2, E2J1 and E2J2 are ER-membrane-associated E2s (Ye and Rape, 2009), we focused on these four E2s. In agreement with previous results (Cao et al., 2007; Li et al., 2007), only Ube2g2 could function with gp78 but not HRD1 to efficiently assemble either free polyubiquitin chains, or on HERP (Fig. 2C,D and supplementary material Fig. S4). These results suggest that the gp78–Ube2g2 pair are involved in HERP polyubiquitylation during the ER stress recovery process.

Substrate specificity is usually determined by E3 (Pickart, 2001; Rubel et al., 2013). We therefore tested whether there was a physical interaction between gp78 and HERP. GST–gp78c (the cytosolic domain of gp78) purified from Escherichia coli was used to pull down His-tagged recombinant HERPc, the cytosolic domain of HERP. After Coomassie staining, we found HERPc was efficiently co-precipitated, indicating a direct interaction between these proteins (Fig. 3B, lane 3). The cytosolic segment of gp78 contains four domains: RING, CUE, G2BR and VIM (Chen et al., 2006; Donaldson et al., 2003; Shih et al., 2003). To identify which domain was responsible for its interaction with HERP, we generated gp78 mutants as indicated in Fig. 3A. We purified these proteins as GST-tagged proteins from E. coli. GST pull-down experiments showed that when the CUE domain was deleted gp78c could not bind HERPc. By contrast, deletion of either G2BR or VIM did not affect gp78 binding to HERPc (Fig. 3B). These results suggested that the CUE domain in gp78 is required for the interaction between gp78 and HERP. Importantly, when the CUE domain was deleted, gp78c could not efficiently assemble polyubiquitin chains on HERPc (Fig. 3C,D), suggesting this interaction is functionally important. Furthermore, gp78 appeared to assemble lysine-48-linked ubiquitin chains on HERPc because ubiquitylation with a K48R ubiquitin mutant only supported monoubiquitylation of HERPc (Fig. 3E). It has been reported that Lys48-linked ubiquitin chains target proteins for degradation by the 26S proteasome (Chau et al., 1989; Ye and Rape, 2009), which agrees with our in vivo studies showing that HERP is an unstable protein degraded by the proteasome.

To identify the domain in HERP that is responsible for interaction with gp78, we generated several HERPc mutants (Fig. 3F) and purified them from E. coli. The binding to gp78c was tested by a GST pull-down experiment. We found that deletion of either the UBL domain or a small C-terminal fragment abolished the binding between HERPc and gp78c (Fig. 3G). This is consistent with the established ubiquitin-binding function of the CUE domain (Chen et al., 2006; Donaldson et al., 2003; Shih et al., 2003). An in vitro ubiquitylation experiment showed that these mutants could not be ubiquitylated by gp78c (Fig. 3H). This result, together with our previous observations that lysine 61 in the UBL domain is the ubiquitylation site for gp78-mediated polyubiquitylation (Li et al., 2007), indicates that the UBL domain of HERP is one of the key sites to interact with gp78, further resulting in the ubiquitylation of HERP.

To test whether the interaction between the CUE and UBL domains is physiologically important for HERP turnover in cells, we reintroduced wild-type gp78 or the gp78 ΔCUE mutant together with a control E3 HRD1 into gp78-deficient HCT116 cells and tested whether they could rescue the HERP degradation defect (Fig. 4A). We treated these transfected cells with DTT for 4 hours, then transferred them to fresh medium without DTT. The turnover of HERP was monitored by immunoblotting. Wild-type gp78 transfection successfully rescued the HERP degradation defect, whereas the gp78 ΔCUE mutant and the control E3 HRD1 failed to do so (Fig. 4B). These results indicate that the CUE domain is essential for gp78 function in the degradation of HERP during ER stress recovery.

To test the function of HERP UBL domain on its degradation, we overexpressed wild-type HERP, HERP ΔUBL or HERP K61R mutants in HEK293T cells (supplementary material Fig. S5). Cycloheximide-chase experiment showed that wild-type HERP was rapidly degraded in HEK293T cells, whereas the HERP ΔUBL and HERP K61R mutants were significantly stabilized (Fig. 4C). These results suggest that the UBL domain regulates HERP stability in cells and that gp78-dependent polyubiquitylation on lysine 61 of HERP mediates its proteasome degradation.

We next wished to study why HERP is rapidly eliminated during ER stress recovery. In the natural condition, only when a cell can perform multiple tasks, can it survive numerous stresses (Shoval et al., 2012). In general, adaptation to a specific stress environment is often accompanied by increased sensitivity to another stressor (Ackerman and Gems, 2012; Bennett and Lenski, 2007; Casanueva et al., 2012; Shoval et al., 2012). We hypothesized that ER stress survival is accompanied by deterioration in other traits that render cells sensitive to a different stress inducer. To test this idea, we ectopically expressed HERP and an irrelevant control membrane protein, gp78, in which the trans-membrane domain was fused with GFP (gp78-TM–GFP) in HEK293T cell. The overexpression level of HERP is roughly equal to that induced by ER-stress inducers (supplementary material Fig. S6A), and the transfection of gp78-TM–GFP did not affect HERP degradation during the ER stress recovery process (compare supplementary material Fig. S6B with C). We then treated the cells with DTT at various concentrations and found that the viability of HERP-overexpressing cells was significantly higher than the gp78-TM–GFP-overexpressing cells, consistent with the notion that HERP overexpression improves adaptation to ER stress (Fig. 5A). Interestingly, when HERP-expressing cells were treated with hydrogen peroxide (H₂O₂) to induce oxidative stress, they were more sensitive to H₂O₂-induced cell death compared with the gp78-TM–GFP-overexpressing cells (Fig. 5B). We then tested some other stress conditions such as reductive stress (homocysteine treatment), oxidative stress (Paraquat treatment) and DNA damage (UV irradiation). Consistent with our model, the ability to grow under reductive stress condition was significantly higher in HERP-overexpressing cells than in gp78-TM–GFP-overexpressing cells (Fig. 5C and supplementary material Fig. S6D), but the oxidative stress tolerance ability of HERP-overexpressing cells was impaired (Fig. 5C and supplementary material Fig. S6E). However, no difference was found in viability between HERP-expressing and the gp78-TM–GFP-overexpressing cells after UV irradiation (Fig. 5C). Furthermore, we expressed HERP ΔUBL in HEK293T cells and treated them with H₂O₂. We found that HERP ΔUBL-overexpressing cells were even more sensitive to oxidative stress than those overexpressing full-length HERP (Fig. 5D). This result suggests that the proteotoxicity of HERP under oxidative stress condition is independent of the UBL domain. Because the UBL domain promotes gp78-mediated degradation of HERP, the increased cell death of HERP ΔUBL-expressing cells under oxidative stress conditions is probably due to higher expression of this mutant. Together, our results suggest the existence of a trade-off during cellular adaptation to reductive stress and oxidative stress: the induction of HERP by reductive stress improves cell viability under this particular stress condition, but compromises the ability of the cell to deal with oxidative stress. This model provides a plausible explanation for why HERP needs to be rapidly returned to its basal level after ER stress.

We next tested whether the aforementioned mechanism is evolutionarily conserved (Fig. 6A,B). In yeast, a UBL-containing protein named Usa1p was proposed to be the HERP homolog (Carvalho et al., 2010). To test whether Usa1p and HERP are really functional homologs, we created a USA1 (yeast HERP homolog)-deleted strain. As expected, the USA1 knockout strain was sensitive to ER stress compared with an isogenic wild-type strain (supplementary material Fig. S7A). We then expressed either USA1 or HERP in the USA1-deficient strain under the control of USA1 promoter. The expression of USA1 rescued the reductive stress sensitivity of Δusa1 cells back to the wild-type level, whereas HERP could only partially rescue the ER stress sensitivity (supplementary material Fig. S7B), suggesting that Usa1p is a functional ortholog of HERP.

Next we tested whether Usa1p also plays an important role in the trade-off during cellular adaptation to reductive and oxidative stresses. We then overexpressed USA1 under the control of a strong TDH3 promoter in the wild-type strain. The overexpression of USA1 slightly retarded the growth of yeast cells (supplementary material Fig. S7C), suggesting there is some proteotoxicity of Usa1p. We then treated these cells with either DTT or H₂O₂, and consistent with the function of HERP in mammalian cells, USA1 overexpression also helped yeast cells resist the reductive stress, but impaired their oxidative stress tolerance ability (Fig. 6C,D). These results suggest that the role of Usa1p and HERP in the trade-off during cellular adaptation to reductive and oxidative stresses is evolutionarily conserved.

Because the UBL domain is conserved in all the homologs of HERP, it raises the possibility that Usa1p may also be degraded by the ubiquitin–proteasome system during the ER stress recovery process. To test this idea, we added a GFP tag to endogenous Usa1p at the C-terminus. After treating with an ER stress inducer such as DTT, Usa1p was induced and reached the peak at around 4 hours and then degraded rapidly, which was the same as the behavior of HERP in mammalian cells (Fig. 6E). The degradation of Usa1p could also be significantly inhibited by MG132 (Fig. 6F). Thus, we conclude that HERP degradation during the ER stress recovery process is also evolutionarily conserved.

In the present study, we investigated the fate of HERP after ER stress. We found that HERP was quickly eliminated by an ubiquitin–proteasome-dependent pathway during the ER stress recovery process. It has been reported that HERP can be polyubiquitylated by an E3 ubiquitin ligase POSH, but the ubiquitin chains formed by POSH were lysine-63-linked and they seem to regulate HERP localization in response to ER stress (Tuvia et al., 2007). Our previous work showed that gp78 can ubiquitylate HERP in vitro (Li et al., 2007; Li et al., 2009), but whether it can catalyze HERP turnover during ER stress recovery is unclear. Here we show that gp78 is a major E3 that mediates lysine-48-linked ubiquitylation of HERP and its degradation in cells. gp78-mediated HERP turnover represents another means of controlling the HERP protein level in addition to transcriptional upregulation by ER stress inducers. Unlike other molecular chaperones such as Hsp70, which could be directly ubiquitylated by an associated E3 ligase CHIP after substrate is depleted (Qian et al., 2006), HERP associates with the HRD1–SEL1L complex, stimulating HRD1-mediated ubiquitylation and degradation of aberrant ER proteins (Kny et al., 2011), but itself is ubiquitylated by another E3, gp78. Thus, when compared with the CHIP system, the mechanism that regulates HERP expression and activity involves another layer of complexity.

Previously, it was reported that the yeast homolog of HERP, Usa1p, is involved in eliminating un-partnered Hrd1p by mediating its self-ubiquitylation and ultimate degradation (Carroll and Hampton, 2010). gp78-mediated HERP polyubiquitylation and subsequential degradation is conceptually similar to the above mentioned ERAD components controlling the degradation of each other. The self-degradation of Hrd1p requires the UBL domain of Usa1p in yeast, but at present we do not know whether the degradation of excess HRD1 depends on HERP or not in mammalian cells. However, the ubiquitylation of HERP also depends on a physical interaction between its UBL domain and the CUE domain of gp78. The primary function of the CUE domain is to bind ubiquitin and preferentially polyubiquitin chains (Chen et al., 2006; Donaldson et al., 2003; Shih et al., 2003). Thus, it is possible that during ER stress HERP assists HRD1 in mediating ubiquitylation of aberrant ER proteins. gp78 may cooperate with HRD1 in ubiquitin chain assembly. After substrates are eliminated, the UBL domain in HERP then works as a ubiquitin moiety and is recognized by the CUE domain of gp78. As a result, HERP is ubiquitylated and degraded, thus initiating the ER stress recovery process.

In sharp contrast to the many studies about how cells respond to various stresses, the processes of stress recovery are still far from well characterized. CHIP-mediated Hsp70 degradation is one of the best investigated. CHIP is involved in proteasome-dependent clearance of unfolded cytosolic proteins during stress responses. At the same time, it is also required for recovery from the stress response by promoting Hsp70 ubiquitylation and subsequent degradation (Qian et al., 2006). Another analogous example can be found in hypoxia. After stress, the SUMOylated hypoxia-inducible factor 1a (HIF1a) binds to a ubiquitin ligase, von Hippel-Lindau (VHL) protein, leading to its rapid ubiquitylation and degradation by the proteasome (Kamura et al., 2000). Recently, it was reported that in response to DNA damage, cells inhibit protein synthesis by activating the eukaryotic elongation factor 2 kinase (eEF2K), whereas during the recovery process, eEF2K is ubiquitylated by the SCF (Skp, Cullin, F-box-containing) βTrCP (β-transducin repeat-containing protein) E3 ubiquitin ligase and then degraded by the proteasome, thus allowing the restoration of peptide chain elongation (Meloche and Roux, 2012). In this study, we found that during the ER stress recovery process, HERP was quickly degraded through a Ube2g2–gp78-dependent ubiquitin–proteasome system, and the yeast homolog of HERP, Usa1p, was similarly eliminated by the ubiquitin–proteasome system. Based on these results, we proposed that, ubiquitin–proteasome-mediated degradation of certain stress response molecules may generally occur during stress recovery processes, which may involve different E2 and/or E3 enzymes under different stress conditions.

The assumption of cost associated with gain or trade-offs between traits has been postulated for centuries as either a philosophical or a biological premise (Bennett and Lenski, 2007), but the evidence in support of this notion is still lacking because most comparative studies on different taxa are essentially correlational without functional validation. Recently, some bacteria such as E. coli were used to prove the trade-off concept by imposing selection on one trait and then measuring correlated changes in other traits such as adaption to either low or high temperature (Bennett and Lenski, 2007; Ibarra et al., 2002; Novak et al., 2006). In the last year, it was reported that a trade-off occurs in C. elegans between stress resistance and reproduction/growth fitness (Ackerman and Gems, 2012; Casanueva et al., 2012). However, most of these studies are still far from determining the molecular mechanism underlying these trade-offs. In the current study, we not only show a clear trade-off between reductive and oxidative stress tolerance from yeast to mammalian cells, but also demonstrate HERP and its homologs play a key role in this trade-off. In addition, we found that the Ube2g2–gp78-mediated ubiquitin–proteasome system regulates this trade-off by promoting HERP polyubiquitylation and subsequent degradation, thus providing an elegant molecular mechanism that modulates two stress adaptation programs.

Anti-FLAG, anti-Myc and anti-β-actin antibodies were purchased from Abmart, and anti-RGS-His antibody was purchased from Qiagen. Ube2g2, gp78, HERP and GFP polyclonal antibodies were generated in rabbits using the corresponding recombinant proteins as antigen. FLAG–ubiquitin and GST–E1 were purchased from Boston Biochem.

pET28-Ube2g2, pQE9-HERPc constructs have been described previously (Li et al., 2007). All the other 24 mammalian E2s were constructed by amplifying the ORFs from mouse cDNA and cloning into the pET28a vector. The gp78 deletion mutants and HERPc mutants were generated by site-directed mutagenesis from pGEX-gp78 or pQE9-HERPc vectors. pRK-HERP and pRK-HERP ΔUBL were constructed by cloning the coding sequence into the SalI and NotI sites of the pRK vector. pRS425-USA1 and -HERP were constructed by cloning the coding sequence of USA1 or Herp + 850 bp 5′ and 600 bp 3′ of USA1 into the XhoI and SacI sites of the pRS425 vector. pRS315-TDH3 USA1 and pRS315-TDH3 HERP were constructed by cloning the TDH3 promoter and USA1 or HERP coding sequence, digestion (XhoI–EcoRI for TDH3, EcoRI–SacI for USA1 and HERP) and ligating into the pRS315 vector.

Purification of Ube2g2 and gp78c has been described previously (Ye et al., 2003). Purified E2 and E3 variants were further fractionated by size exclusion chromatography on Superdex200 and Superrose 6 columns, respectively in 50 mM Tris-HCl (pH 8.0), 150 mM potassium chloride, 5% glycerol and 2 mM magnesium chloride. RGS-His-tagged HERPc mutants were purified under native conditions using Ni-NTA beads.

Somatic cell gene targeting was conducted as described previously (Zhang et al., 2011). Briefly, the targeting AAVs were packaged in HEK293T cells by transfecting equal amounts of gp78 KO targeting vector (supplementary material Fig. S2), pHelper and pRC plasmids (1 µg each). After 72 hours, the transfected cells were scraped from the plates and suspended in sterile phosphate-buffered saline. The suspension was then centrifuged at 500 g, and the pellet was frozen and thawed three times. Finally, the lysate was centrifuged to remove cell debris and the supernatant containing rAAV was divided into several aliquots and frozen at −80°C. HCT116 cells were infected with the gp78-targeting viruses and selected with neomycin for 2 weeks. The Geneticin-resistant clones were then screened for homologous recombination by genomic PCR with primers derived from the neomycin resistance gene (5′-GTTGTGCCCAGTCATAGCCG-3′) and the upstream region of the left homologous arm (5′-GGGCCGTATAAGGAATTTGC-3′). Positive clones were confirmed by genomic PCR, with primers derived from the neomycin resistance gene (5′-TCTGGATTCATCGACTGTGG-3′) and the downstream region of the right homologous arm (5′-AACACCTAACTTCGGCATGG-3′). Correctly targeted clones were infected with adenoviruses expressing Cre recombinase to delete the selectable drug marker. To select clones with successful deletion of the selectable drug marker, genomic PCR was employed to amplify an approximate 250 bp fragment in which the loxP site was inserted, using specific primers (5′-CATGATGCCACATTCACTGC-3′ and 5′-GCCCAGTTTTACCTGTGTAGGA-3′).

The heterozygous KO clones were infected with the same targeting virus to target the second allele, and the neomycin resistance gene was excised as described earlier. Final confirmation in the generation of the KO cell lines was performed using western blotting.

To identify intracellular protein degradation systems, cells were treated with 25 mM NH₄Cl, 10 mM 3-methyladenine, 10 µM MG132, 100 nM bafilomycin A1 or 25 µM chloroquine for 8 hours or 24 hours to inhibit autophagy or ubiquitin–proteasome pathways. Then the HERP level was measured by western blotting (Mizushima et al., 2010).

In other cases, HEK293T cells transfected with HERP, HERP ΔUBL or HERP K61R were treated with cycloheximide (10 µg/ml) for the indicated times to inhibit de novo protein synthesis, and then degradation of the HERP mutants was examined by western blotting.

Ubiquitylation experiment was performed as described previously (Ye et al., 2003). Briefly, E1 (60 nM), Ube2g2 (200 nM) and gp78c (300 nM) were incubated with FLAG-tagged ubiquitin (10 µM) at 37°C in buffer containing 25 mM Tris-HCl (pH 7.4), 2 mM magnesium ATP and 0.1 mM DTT. Samples taken at different time points were quenched with Laemmli buffer in the absence of reducing reagent. For reducing condition, samples were treated with 100 mM DTT before SDS-PAGE. Ub chains were detected by immunoblotting with anti-FLAG antibody. Ubiquitylation of HERPc was conducted using the conditions described above with the addition of HERPc (500 nM). Ubiquitylated HERPc was detected by immunoblotting with anti-RGS-His antibody (Qiagen).

HEK293T and HeLa cells were maintained in DMEM containing 10% FBS. HCT116 WT and HCT116 gp78 KO cells were cultured in McCoy's 5A medium containing 10% FBS. ER stress was induced by treating the cells with 10 mM homocysteine, 10 mM β-mercaptoethanol, 12 µM tunicamycin, 1 µM thapsigargin or 500 µM dithiothreitol for 4 hours.

Cell viability and cell death under reductive stress or oxidative stress conditions were measured using the 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyltetrazoliumbromide (MTT) assay. HEK293T cells were placed into 96-well plates, and transfected with pRK-HERP, pRK-HERP ΔUBL or pRK vector. Different concentrations of reductive or oxidative stress inducers were added for 24 hours, then MTT was added and the mixture incubated for 2 hours. DMSO was then added to dissolve the formazan. After 10 minutes the OD was measured at 490 nm.

Cells were lysed in RIPA buffer containing 50 mM Tris pH 7.5, 150 mM NaCl, 0.1% SDS, 1% NP40, 2 mM EDTA, 0.5% sodium deoxycholate and protease inhibitors (Roche). Samples were then subjected to western blotting with anti-HERP antibody, anti-β-actin antibody, anti-FLAG antibody or anti-RGS-His antibody.

Yeast strains were grown to high density (A₆₀₀>1.2). Beginning with an A₆₀₀ of 0.126, 10-fold serial dilutions were made and spotted onto the appropriate selected SC plates followed by incubation at 30°C for 2 days.

We thank Chunsheng Han and Qing Li for critical reading of the manuscript.